1991: Proceedings of the Appalachian karst symposium: Radford, Virginia, March 23-26, 1991

Material Information

1991: Proceedings of the Appalachian karst symposium: Radford, Virginia, March 23-26, 1991
Kastning, Ernst H., 1944---
National Speleological Society
Publication Date:


Subjects / Keywords:
Geology ( local )
Conference Proceeding
serial ( sobekcm )
United States


General Note:
Sponsored by the National Speleological Society in celebration of its 50th anniversary; hosted by the Department of Geology, Radford University. Contents: Karst Erosion Surfaces in the Appalachian Highlands / William B. White and Elizabeth L. White -- Structural Controls on Drainage Beneath Droop Mountain, Pocahontas County, West Virginia / Douglas M. Medville and Hazel E. Medville -- Karst and Caves of Mercer and Summers Counties, West Virginia / Joseph W. Saunders and William M. Balfour - Hydrochemical and Structural Controls on Speleogenesis in the Appalachian Foldbelt / Kass Kastning -- Concepts and Classification of Cave Breakdown: An Analysis of Patterns of Collapse in Friars 35 Hole Cave System, West Virginia / Roy A. Jameson -- Mud Flow in a Karst Setting / Charles A. Lundquist and William W. Varnedoe. Jr. -- Mud Pot: A New Thermal Water Cave in Alleghany County, Virginia / Keith E. Goggin -- Travertine-Marl: The "Doughnut-Hole" of Karst / David A. Hubbard. Jr. and Janet S. Herman -- Meteorology of the Butler Cave-Sinking Creek System / Fred L. Weier -- Emerged Sea Caves and Coastal Features as Evidence of Glacio-Isostatic Rebound, Mount Desert 75 Island, Maine / Paul A. Rubin -- Cave Development in the Glaciated Appalachian Karst of New York: Surface-Coupled or 85 Saline-Freshwater Mixing Hydrology / John E. Mylroie -- Modification of Preglacial Caves by Glacial Meltwater Invasion in East-Central New York /Paul A. Rubin -- Flow Characteristics and Scallop-Forming Hydraulics within the Mill Pond Karst Basin, East-Central New York / Paul A. Rubin -- Replacement Mechanisms among Carbonates, Sulfates, and Silica in Karst Environments: 109 Some Appalachian Examples / Arthur N. Palmer and Margaret V. Palmer -- Fracture Controls on Groundwater Flow and Cave Development in Nonhern Greenbrier and 116 Southern Pocahontas Counties, West Virginia (Abstract) / Roy A. Jameson -- Surface and Subsurface Drainage Basin Asymmetry: Ramifications for Karst Development in the Appalachian Plateaus/ Ira D. Sasowsky -- Environmental Education Regarding Karst Processes in the Appalachian Region / Ernst H. Kastning and Karen M. Kastning -- Regional Karst Studies: Who Needs Them? / David A. Hubbard. Jr. -- Spatial-Temporal Characteristics of Karst Subsidence in the Lehigh Valley of Pennsylvania / Percy H. Dougherty -- Illegal Disposal in Sinkholes: The Threat and the Solution. Ronald A. Erchul -- Evaluating a Landfill Expansion in Karst Terrain / Raymond A. DeStephen and Brian Milner -- Predicting Sinkhole Flooding in Cookeville, Tennessee, Using SWMM and GIS / Hugh H. Mills. DB. George, HN. Taylor, Albert E. Ogden. Y. Robinet-Clark. and R. Forde -- Application of Dye Tracing to Evaluation of a Landfill Site in a Karst Terrane in the Tennessee Appalachians (Abstract) / James F. Quinlan. Joseph A. Ray, and Geary M. Schindel -- Computer Enhancement of Downhole-Video Borehole Logs / Malcolm S. Field and Michael Critchley -- An Attempt to Model an Appalachian Karst Aquifer Using MODFLOW. Sara A. Heller Hydrochemical Characteristics of the Greenbrier Limestone Karst of East-Central West Virginia / Eberhard Werner -- Nitrate Levels in the Karst Groundwaters of Tennessee / Albert E. Ogden, Kristie Hamilton, Edward P. Eastburn, Teresa L. Brown, and Thomas E. Pride. Jr. -- Impacts of Barnyard Wastes on Groundwater Nitrate-N Concentrations in a Maturely Karsted Carbonate Aquifer of South-Central Kentucky. Craig J. Brown and Ralph O. Ewers Preliminary Assessment of the Impact of Class V Injection Wells on Karst Groundwaters / Albert E. Ogden. Ronald K. Redman, and Teresa L. Brown -- The Carbonate Aquifer of the Northern Shenandoah Valley of Virginia and West Virginia. / K. Jones -- Influence of Hydrogeologic Setting and Lineaments on Water-Well Yield in the Great Valley Karst Terrane of Eastern West Virginia. Brad T. Zewe and Henry W. Rauch -- On Calculating the Risk of Sinkhole Collaps / Barry F. Beck -- A Suggested Strategy for Characterizing the Hydrogeologic Regime of Karst Terranes in the Valley and Ridge Province / Larry Mata and John T. Haynes.
Open Access - Permission by Publisher
Original Location:
Northup Database Collection
General Note:
See Extended description for more information.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
K26-00098 ( USFLDC DOI )
k26.98 ( USFLDC Handle )
4557 ( karstportal - original NodeID )
26132817 ( OCLC )

USFLDC Membership

Karst Information Portal

Postcard Information



This item has the following downloads:

Full Text
Sponsored by the National Speleological Society in
celebration of its 50th anniversary; hosted by the Department
of Geology, Radford University.
Contents: Karst Erosion Surfaces in the Appalachian
Highlands / William B. White and Elizabeth L. White --
Structural Controls on Drainage Beneath Droop Mountain,
Pocahontas County, West Virginia / Douglas M. Medville and
Hazel E. Medville --
Karst and Caves of Mercer and Summers Counties, West
Virginia / Joseph W. Saunders and William M. Balfour -
Hydrochemical and Structural Controls on Speleogenesis in the
Appalachian Foldbelt / Kass Kastning --
Concepts and Classification of Cave Breakdown: An
Analysis of Patterns of Collapse in Friars 35 Hole Cave
System, West Virginia / Roy A. Jameson --
Mud Flow in a Karst Setting / Charles A. Lundquist and
William W. Varnedoe. Jr. --
Mud Pot: A New Thermal Water Cave in Alleghany County,
Virginia / Keith E. Goggin --
Travertine-Marl: The "Doughnut-Hole" of Karst / David
A. Hubbard. Jr. and Janet S. Herman --
Meteorology of the Butler Cave-Sinking Creek System /
Fred L. Weier --
Emerged Sea Caves and Coastal Features as Evidence of
Glacio-Isostatic Rebound, Mount Desert 75 Island, Maine /
Paul A. Rubin --
Cave Development in the Glaciated Appalachian Karst of
New York: Surface-Coupled or 85 Saline-Freshwater Mixing
Hydrology / John E. Mylroie --
Modification of Preglacial Caves by Glacial Meltwater
Invasion in East-Central New York /Paul A. Rubin --
Flow Characteristics and Scallop-Forming Hydraulics
within the Mill Pond Karst Basin, East-Central New York /
Paul A. Rubin --
Replacement Mechanisms among Carbonates, Sulfates, and
Silica in Karst Environments: 109 Some Appalachian Examples /
Arthur N. Palmer and Margaret V. Palmer --
Fracture Controls on Groundwater Flow and Cave
Development in Nonhern Greenbrier and 116 Southern Pocahontas
Counties, West Virginia (Abstract) / Roy A. Jameson --
Surface and Subsurface Drainage Basin Asymmetry:
Ramifications for Karst Development in the Appalachian
Plateaus/ Ira D. Sasowsky --
Environmental Education Regarding Karst Processes in
the Appalachian Region / Ernst H. Kastning and Karen M.
Kastning --
Regional Karst Studies: Who Needs Them? / David A.
Hubbard. Jr. --
Spatial-Temporal Characteristics of Karst Subsidence in
the Lehigh Valley of Pennsylvania / Percy H. Dougherty --
Illegal Disposal in Sinkholes: The Threat and the
Solution. Ronald A. Erchul --
Evaluating a Landfill Expansion in Karst Terrain /
Raymond A. DeStephen and Brian Milner --
Predicting Sinkhole Flooding in Cookeville, Tennessee,
Using SWMM and GIS / Hugh H. Mills. DB. George, HN. Taylor,
Albert E. Ogden. Y. Robinet-Clark. and R. Forde --
Application of Dye Tracing to Evaluation of a Landfill
Site in a Karst Terrane in the Tennessee Appalachians
(Abstract) / James F. Quinlan. Joseph A. Ray, and Geary M.
Schindel --
Computer Enhancement of Downhole-Video Borehole Logs /
Malcolm S. Field and Michael Critchley --
An Attempt to Model an Appalachian Karst Aquifer Using
MODFLOW. Sara A. Heller Hydrochemical Characteristics of the
Greenbrier Limestone Karst of East-Central West Virginia /
Eberhard Werner --
Nitrate Levels in the Karst Groundwaters of Tennessee /
Albert E. Ogden, Kristie Hamilton, Edward P. Eastburn, Teresa
L. Brown, and Thomas E. Pride. Jr. --
Impacts of Barnyard Wastes on Groundwater Nitrate-N
Concentrations in a Maturely Karsted Carbonate Aquifer of
South-Central Kentucky. Craig J. Brown and Ralph O. Ewers
Preliminary Assessment of the Impact of Class V Injection
Wells on Karst Groundwaters / Albert E. Ogden. Ronald K.
Redman, and Teresa L. Brown --
The Carbonate Aquifer of the Northern Shenandoah Valley
of Virginia and West Virginia. / K. Jones --
Influence of Hydrogeologic Setting and Lineaments on
Water-Well Yield in the Great Valley Karst Terrane of Eastern
West Virginia. Brad T. Zewe and Henry W. Rauch --
On Calculating the Risk of Sinkhole Collaps / Barry F.
Beck --
A Suggested Strategy for Characterizing the
Hydrogeologic Regime of Karst Terranes in the Valley and
Ridge Province / Larry Mata and John T. Haynes.


AppalachianProceedingsoftheAppalachianKarstSymposiumRadford Virginia,March23-26, 1991 Ernst H. KastningandKaren M. Kastning, Editors SponsoredbytheNationalSpeleological SocietyInCelebrationofIts50thAnniversary HostedbytheDepartmentofGeology RadfordUniersityNational Dign andLaut b rnstH,KatningandKaren Produ tion bNSSSp alPubliattonnUllitteeDa 'dastninglurglChairman


AppalachianKarstProceedings oftheAppalachian Karst Symposium Radford Virginia, March 23-26, 19911991 National Speleological Society, Inc. Published by National Speleological Society, Inc. 2813 Cave Avenue Huntsville, Alabama 35810 Phone 205 852-1300Allrights reserved includingtherighttoreproduce this book or portions thereof inanyform or byanymeans, electronic or mechanical, including photocopying, recording, orbyanyinformation storageandretrieval system,withoutpermissioninwriting from the publisher.Allinquiries should be addressed totheNational Speleological Society. Library of Congress Catalog Card Number 91-061175ISBN0-9615093-5-0 Cover photo: Lookingoutoftheentrance, Sinks of Gandy, West Virginia


Cover: The photograph on the coverofthis volume is a view looking out from the upstream entranceofthe SinksofGandy, a 1.5-mile-Iong cave in southeastern Randolph County, West Virginia. Gandy Creek sinks at this point and resurges at the northern endofthe cave.Asthis picture shows, the realmofkarst encompasses both the surface and the subsurface. These zones are linked by water flow, either that now moving through solutional cavities and conduits or that which excavated these openingsinthe geologic past. The SinksofGandy was an important caveinthe early historyofthe National Speleological Society. Membersofthe DistrictofColubia Speleological Society made frequent tripstothe caveinthe 1930's and Jack Prebble, an early NSS member from Ohio, popularized the cave by founding the Ibinthruthesinks Clubin1937(seemembership card,p.34ofthe January 1991NSS News).The caverinthis view is Kass Kastning, oneofthe authorsinthis volume.PhotographbyErnst and Karen Kastning.iv


Appalachian Karst Symposium, 1991PrefacePrefaceThis volume,Appalachian Karst,is the official Proceedingsofthe Appalachian Karst Symposium, held at Radford University, Radford, Virginia on 23-26 March 1991. This was the first professional meeting to address the karst region extending along the Appalachian Mountain Range from AlabamatoMaine and including the Valley and Ridge, Appalachian Plateau, New England Upland and Coastal Plain Provinces. The scopeofthe conference included geologic and geographic studies on all componentsofkarst, from surficial landforms and processestothose in the subsurface. Furthermore, the papers presented herein reflect a balance between basic researchofkarst processesand applicationofkarst science toward understanding and solving environmental problems of karst terranes. Thirty-three papers on Appalachian karst were presented at the meeting by fifty-one authors and co authors. Thirty-oneofthese papers are published here in their entirety. Manuscripts were not received by press time for the remaining two, however, the abstracts for these are included in this volume. Additionally, the morning sessionofthe first dayofpapers included several presentations that were either not relevanttothe Appalachian region or were proposalsofvarious kinds. These are not included in the Proceedings. Attendees at the Appalachian Karst Symposium, Radford University, 23-26 March 1991.Photograph by Ira D. Sasowsky.The impetus for the Appalachian Karst Symposium was three-fold. First, the conference was heldincommemorationofthe 50th Anniversaryofthe foundingofthe National Speleological Society (NSS)in1941. The NSS BoardofGovernors sanctioned the Symposium asanofficial event in the 1991 year long celebrationofthe Golden Anniversaryofthe Society.v


Preface Appalachian Karst Symposium. 1991Second, it was appropriate to devote a meeting specifically to the Appalachians, as the beginningsofthe Society were firmly rooted in this region. The DistrictofColumbia Speleological Society (DCSS) and the Spelunkers ClubofNew England (SCNE) were both independently organized in the 1930s. Membersofthe DCSS actively explored and studied caves in the mountains and valleysofVirginia and West Virginia. Membersofthe SCNE went "spelunking" in the cavesofNew England andNew York. The two groups merged by mutual agreementin1941 by chartering the National Speleological Society in January 1941. Additional information on the formative yearsofthe NSScanbe found in commemorative articles in the December 1990 and January 1991 issuesoftheNSS News.A complete historyofthe National Speleological Society,Caving in America: The Storyofthe National Speleological Society,1941-1991, has recently been published (June 1991). The third motivation for the meeting was to convene the FriendsofKarst (FOK). TheFOKis an informal organizationofindividuals interested in researchinphysical speleology. Most "members" are active within the Geology and Geography Sectionofthe NSS. TheFOKmeets irregularly, but never more than a few years apart. Many membersofthe FOK living in the eastern halfofthe United States attended the Symposium and presented papers. As editors, we express our deep appreciation to allofthe contributors to the meeting. It is interesting to note that every paper presented was volunteered and the meeting organizers did not have to solicit a single contribution!Wealso thank Drs. Chester F. "Skip" Watts and Stephen W. Lenhartofthe DepartmentofGeology at Radford University and David Hubbardofthe VirginiaDivisionofMineral Resources for driving vans on the opening-day fieldtrip. Michael and Lori Kilgore contributed considerable time and energy at the registration desk and in making refreshments available during the meeting. The Administration at Radford University fully supported the conference by providing facilities for meeting. The DepartmentofGeology at Radford University(Dr.RobertC.Whisonant, Chairman) hosted the conference. Mr. JerroldL.PerryofRadford University Food Services made arrangements for the banquet. We thank Dr. William B. White for his Banquet presentation,The Appalachian Karst: Historic and Futuristic Impressions.The NSS BoardofGovernors approved Society funding for publicationofthis volume. David McClurg, NSS Special Publications Committee Chairman, served as editorial consultant and as liaison with the printer.Weareespeciallygrateful for his patience as we struggled with our editorial tasks.ErnstH.Kastning and KarenM.Kastning Radford University, Radford, Virginiavi


AppalachianKarstSymposium.1991ContentsAPPALACHIAN KARSTProceedingsofthe AppalachianKarstSymposiumRADFORD,VIRGINIA,23-26MARCH1991TableofContentsPrefaceKarst Erosion Surfaces in the Appalachian Highlands.William B. WhiteandElizabethL.Whitev Structural Controls on Drainage Beneath Droop Mountain, Pocahontas County, West Virginia.11DouglasM. MedvilleandHazel E. MedvilleKarst and CavesofMercer and Summers Counties, West Virginia.JosephW.Saundersand19William M.BalfourHydrochemical and Structural Controls on Speleogenesisinthe Appalachian Foldbelt.Kass25 KastningConcepts and ClassificationofCave Breakdown: An AnalysisofPatternsofCollapseinFriars 35 Hole Cave System, West Virginia.RoyA. JamesonMud Flow in a Karst Setting.CharlesA.LundquistandWilliamW.Varnedoe. Jr.45Mud Pot: A New Thermal Water CaveinAlleghany County, Virginia.Keith E. Goggin51Travertine-Marl: The "Doughnut-Hole"ofKarst.David A. Hubbard. Jr.andJanetS.Herman 59Meteorologyofthe ButlerCave-Sinking Creek System.FredL.Weier 65Emerged Sea Caves and Coastal Features as EvidenceofGlacio-Isostatic Rebound, Mount Desert 75 Island, Maine.PaulA. RubinCave Developmentinthe Glaciated Appalachian KarstofNew York: Surface-Coupled or 85 Saline-Freshwater Mixing Hydrology?JohnE.MylroieModificationofPreglacial Caves by Glacial Meltwater InvasioninEast-Central New York.91PaulA.RubinFlow Characteristics and Scallop-Forming Hydraulics within the Mill Pond Karst Basin, East-101Central New York.Paul A. RubinReplacement Mechanisms among Carbonates, Sulfates, and Silica in Karst Environments: 109 Some Appalachian Examples.Arthur N. PalmerandMargaretV.PalmerFracture Controls on Groundwater Flow and Cave Development inNonhernGreenbrier and 116 Southern Pocahontas Counties, West Virginia (Abstract).RoyA.Jamesonvii


Contents Appalachian Karst Symposium, 1991Surface and Subsurface Drainage Basin Asymmetry: Ramifications for Karst Development in117 the Appalachian Plateaus.Ira D. SasowskyEnvironmental Education Regarding Karst Processes in the Appalachian Region.ErnstH.123KastningandKarenM.KastningRegional Karst Studies: Who Needs Them?David A. Hubbard. Jr.135SpatialTemporal CharacteristicsofKarst Subsidencein the Lehigh ValleyofPennsylvania.139PercyH.DoughertyIllegal Disposal in Sinkholes: The Threat and the Solution.Ronald A. Erchul147Evaluating a Landfill ExpansioninKarst Terrain.Raymond A. DeStephenandBrian Milner153Predicting Sinkhole Flooding in Cookeville, Tennessee, Using SWMM and GIS.Hugh H.159Mills.DB.George,HN.Taylor, AlbertE.Ogden.Y.Robinet-Clark.andR. FordeApplicationofDye Tracing to Evaluationofa Landfill Siteina"Karst Terranein the Tennessee168Appalachians (Abstract).James F. Quinlan. Joseph A. Ray,andGearyM.SchindelComputer EnhancementofDownhole-Video Borehole Logs.MalcolmS.FieldandMichael169Critchley AnAttempttoModel an Appalachian Karst Aquifer Using MODFLOW.Sara A. Heller177Hydrochemical Characteristicsof the Greenbrier Limestone KarstofEast-Central West Virginia.187Eberhard WernerNitrate Levelsin the Karst GroundwatersofTennessee.AlbertE.Ogden, Kristie Hamilton,197EdwardP.Eastburn, Teresa L. Brown,andThomasE.Pride. Jr.ImpactsofBarnyard Wastes on Groundwater Nitrate-N Concentrationsina Maturely Karsted 205 Carbonate AquiferofSouth-Central Kentucky.CraigJ.BrownandRalphO.EwersPreliminary Assessmentofthe ImpactofClass V Injection Wells on Karst Groundwaters.211AlbertE.Ogden. RonaldK.Redman,andTeresa L. BrownThe Carbonate Aquiferof the Northern Shenandoah Valley of Virginia and West Virginia. 217WilliamK.JonesInfluenceofHydrogeologic Setting and Lineaments on Water-Well Yieldin the Great Valley 223 Karst Terrane of Eastern West Virginia.BradT.ZeweandHenryW.RauchOn Calculating the RiskofSinkhole Collapse.Barry F. Beck231A Suggested Strategy for Characterizing the Hydrogeologic RegimeofKarst Terranesin the 237 Valley and Ridge Province.Larry MataandJohnT.HaynesPlates 58, 84, 100 146, 152 186, 204 viii


Appalachian Karst Symposium, 1991 White and WhiteKarst Erosion Surfaces in the Appalachian HighlandsWilliamB.Whiteland ElizabethL.White21 DepartmentofGeosciences 2DepartmentofCivil Engineering The Pennsylvania State University University Park, PA 16802ABSTRACTAmong the various "peneplain" surfaces identifiedinthe Appalachian Highlands by the early geomor phologists, the Harrisburg Surface is oneofthe most important. The surface variesinelevation from 200 meters in the Cumberland Valley at Harrisburg, Pennsylvania to 760 metersinVirginia and West Virginia.Thebest preserved remnantsofthe Harrisburg Surface are doline karst plains. Internal drainage through the dolines prevents dissection by surface streams and maintains an easily recognizable surface. Those portionsofthe Harrisburg Surface preserved as doline karst are continuously lowered by dissolutionofthe limestone and its removal by internal runoff.Therateoflowering based on present day denudation rates is 20to40 meters/megayear with 30 m/Ma selected as a representative value. Various representative segmentsofthe Harrisburg Surface have been reconstructed backward in time by "replacing" the missing limestone. Such reconstructions are often blocked by stratigraphicorstructural barriers thus giving a limiting age for the Harrisburg Surface. In general, the Harrisburg Surface appears to be younger than the classical estimates with a late-Tertiary (perhaps Pliocene) age rather than the Early Eoceneorlate Cretaceous ages proposedinsomc recent work.IntrOductionResearch on karst is turning from descriptive studiesofindividual karst areas to either small scale analysisofkarst processesorto large scale interpretationsofentire drainage basinsorphysiographic provinces. Among the latter investigations the questionofageand time scale must arise.Thatis, we would like to know the agesofpresent day featuresinthe karst landscape and we would like to know the lengthoftime required for their developmenloIn a very rough way, time scales that must be dealt with by investigatorsofcontemporary karst landscapes (as distinguished from paleokarst) canbedivided into three groups. (i) Karst phenomena that span a few hundred thousand years. These include developmentofmost active karst drainage systems, depositionofspeleothems, and transportofclastic sediments.Thereis a substantial choiceofdating methods includingU/Th, ESR and thermolumines cence, amino acid racemization, and for the youngest events, radiocarbon(seeWhite, 1988, Chapter 10 for review). (ii) Karst phenomena that span a few million years. This includes developmentoflarge cave systems and developmentofmost karst landscapes. The only absolute dating tool is the patternofmagnetic normals and reversals recordedincave clastic sediments (Schmidt, 1982). Most dating depends on geomorphic context. (iii) Karst phenomena that relate to the oldest survi ving featuresofthe landscape.Thetime scale could be tensofmillionsofyears and might span the entire Ceno zoic Era. Mostly, these features are the doline karsts or sinkhole plains which have descended from equivalent features whose ultimate origins lie far backinTertiary or even late Cretaceous time. The present paper is concerned with karst features on the intermediate time scalesofa few million years and also on what may be the oldest karst features preserved. These are the doline plainsinthe Appalachian Highlands. The paperisconcerned with their relation to the Harrisburg erosion surface and the time scale for their evolution.TheHarrisburgSurfaceIntrOduction of theHarrisburgSurfaceThe peneplain concept has its originsinthe Appala chian Highlands where it was introduced by William Morris Davis (1889)inhis classicRivers and Valleysof


White and White Pennsylvania.The idea that the landscape is eroded downtoa flat plain which is then elevated and dissected again has taken some abuse over the years although recently it may be said to have made a modest comeback(see e.g.Morisawa, 1989).Ofthe multitudinous erosion surfaces introduced in the haydayofthe peneplain, two have the greatest acceptance and to some extent have survived the turmoil. One is the Schooley Peneplain, which is repre sented by accordant mountain summits and ridge lines in the Appalachians. The Schooley surface is so highly dis sected and fragmented that little can be said about it. The second is the Harrisburg Surface, represented throughout the Appalachians by valley uplands, which is the most pervasiveofthe erosion surfaces. Muchofthe Harrisburg Surface is karst. Descriptionsofthe Harrisburg Surfaceinthe traditional style are given by Fenneman (1938) and by Thornbury(1%5).Tracing the Harrisburg Karst Surface Through the AppalachiansThe tracingofthe Harrisburg Surface given here is to some extent schematic and depends on the correlationsofothers. It is quite clear, however, that the valley upland surface that canbetraced along the major rivers in the Appalachians rises toward the headwaters and, whatever its regional extent, it is not a horizontal surface. In the "type locality", the broad Cumberland Valley near Harrisburg, Pennsylvania, the rolling upland surface is at an elevationofabout 520 feet (160 m). The Cumber land Valley rises somewhat to the south reaching 700 feet (210 m) northofHagerstown, Marylandandthen falls again to elevations near 500 feet along the Potomac River. WestofHarrisburg, there is a pronounced valley upland forming accordant hill tops above the incised valleyofthe Juniata River. The upland surface reaches an elevationof1000 feet (305 m) in Huntingdon Countyand1200 feet (365 m) at the basin divide in the broad Nittany Valley near State College, Pennsylvania. The upland surface is highly dissected where underlain by shales and well preserved where underlain by limestone. In the Great Valley Sub-Province southofthe Poto mac, the upland surface can be traced along the Shenandoah River as a valley upland that reaches 2000 feet (600 m) at the drainage divide. Cave development in the Shenandoah Valley can be correlatedatleast roughly with the upland surface (White and White, 1974). The Great Valley Sub Province is less clearly defined along the transverse valleysofthe James River and the New River but an upland sur face canbetracedatabout 2000 feet (600 m). Remnantsofvalley uplands canbefollowed into the headwatersofthe various tributaries. Burnsville Cove, in the upper reachesofthe James River drainage, has a remnantofa karst surface at the upstream divideatan elevationof2500 feet (760 m).Theextensive karst in the Greenbrier River drainage, partofthe New River BasininWestVirginia, 2Appalachian Karst Symposium. 1991has been claimedtorepresent the Harrisburg Surface (price and Reger, 1929, plate 16B). In southwestern Virginia, eastern Tennessee, and nor thern Alabama in the drainageofthe Tennessee River are also upland surfaces. These are not considered in the pre sent paper. However, the Highland Rim surfaceofcentral Tennessee, below the escarpmentofthe western marginofthe Cumberland Mountains, has been correlated with the Harrisburg Surface by Fenneman (1938).Recent Estimatesofthe Ageofthe Harrisburg SurfaceUnlike rocks, whichinthe first place were formedina specific depositional event and in the second place carry signaturesoftheir ageinthe formoffossils, radioactive minerals, and other indicators, "surfaces"or"topography" (beingina certain sense the absenceofrocks), are difficult to assign an age. In the first place, topography is contin uously evolving; it is not formed in a single discreet event. In the second place, age indicators that exist are fragmentary and often inconclusive. Some indicationofthe rangeofopinion is given by the following quotes, taken from Sevon (1985): "Littleofthe earth's topography is older than the Ter tiary and mostofitisno older than the Pleistocene." Thornbury, 1969 .... one can conclude that the great bulkoferosionofthe Appalachians took placeinthe Cretaceousorear lier, and that the present mountains have had a relief not much less than the present since well backinthe Tertiary, perhaps since the Eocene." Rogers, 1967 The Harrisburg Surface itself has the best recordofthe many Appalachian erosion surfaces in part becauseofsaprolites, mineral deposits, and various residual sedi ments. Sevon's (1985) thoughtful analysisofthe Harris burg SurfaceinPennsylvania concludes that the surface is very old. Sevon (p. 22) says: "The Late Cretaceous, Paleocene, and Early Eo cene subjected Pennsylvania to an extended periodofhumid subtropical climate during which chemical wea thering wasatan extreme while physical appearanceofthe landscape changed slowly.Thevery long periodofextreme climatic conditions aliowed the land scape to achieve a stateofadjustment (dynamic equili brium?) which resultedina climax landscape -the Harrisburg erosion surface." "The timingofevents related to the Harrisburg Surface is not by any means exact. The developmentofthe surface occurred over a periodoftime which may have been as long as 45 m.y."


Appalachian Karst Symposium. 1991 White and WhiteKarstSurfaces Denudation Rates700 600(P E)500 01C.._---lL-_---l. __ --l.. __ -L. __ ...L __ .l....-_--..: 100 200300400P-EImm/alFigure 2: Denudation rate as a functionofprecipitation minus evapotranspiration.The"world average temperate karst" is the regression lineofSmith and Atkinson (1976) for many experimental measurements.The"equilibrium" line is a first principles calculation based on the modelofWhite(1984). Points show experimental measurements from the Appalachians:WV(Ogden, 1982); VA (calcu lated from Harmon and Hess, 1982 and Davis and Hess, 1982); PAiscalculated from Thompson Springincentral Pennsy 1 vania.10 40 PA50Theequilibrium denudation equation is plotted in Fig ure2incomparison with the regression line for denudation rates intemperateclimatekarstscompiledfrom world sourcesbySmith and Atkinson (1976).Thetwo lines have similar slopesbut the regression line for the world data does not pass through the origin. There is an offsetofInanearlierinvestigation (White, 1984), the rateofloweringofa karst surface was calculated from rust princi ples using a model which assumed that infiltrating groundwatercameintoequilibriumwithcalciteat the soil bedrock contact.Theequation isThedenudation rate,Dn,is given in this equation in unitsofmm/ka. This is numerically equal to the unitsof preferredbythose whomakedirect microerosion measure ments and m/Ma often used for denudation rates calculated from clastic load in non-karstic river basins. P is bedrock density in g/cm 3 the various K'sareequilibrium constants for the usual carbonate reactions(seeWhite, 1988 and ma nyothersources),Pe02is CO2 partial pressure in unitsofatmospheres, and (P-E) is precipitation minus evapo transpiration (orrunoffin karstic basins with no surface streams) in unitsofmm/a. c.2 20 oEvolutionof ilkarstsurface Valley uplands With/fluvialsoils andgravels"'-............ _. "\ Floodplain V"lley,u_,.... .... widening Rejuvinaledvalley Evolutionof aIlon-k;:wsl surfac.e o,.i9in,IVvalley Figure 1 illustrates the essentialsofthe problem. In regionsofnon-carbonate rock, rapidly down-cutting val leys take on the characteristic V -shape. During extended periodsofstable base level, downcutting slows down and the valleys widen, forming awideflood plain. Renewed downcutting results in a rejuvenated V -shaped valleycutinto the old flood plain, remnantsofwhich remain behind as a terraceorvalley upland.Theusual assumption, not always stated specifically, is that the rateofloweringof the valley upland is much less than the rateofdowncuttingof the rejuvenated valley.PreservationofKarstSurfaces: Doline KarstandRiverTerraces. Manyofthebestpreserved remnantsof the Harrisburg Surface in the Appalachians are on carbonate rock. Be causeofinternaldrainagethrough sinkholes, carbonates remain as undissected plains with an aspectofan "erosion surface",atleasttocasualinspection.However,these "surfaces" are being continually lowered by dissolutionofcarbonate rock at the soil/bedrock contact and transportofdissolvedmaterialinto anunderlyingkarsticdrainagesystem. In contrast, thekarstplain neednotchangeits form.Tobesure, sinkholesbecomefilled, new sinkholes form, and there is a continual evolutionofthe detailofthe land scape. Viewed in the large, however, the visual appearanceof the landscape remains much the same. On the average, the sinkhole plain is simply loweredata rate determined by the rateofdissolutionofthe underlying carbonate rock. Unlike the valley uplandonclastic rock, it should notbeassumed that the sinkhole plain marks a fixed elevation.Figure1:Sketchshowingthecontrastinlandscapemorphology between an eroding valley and an internally drained karst surface. 3


White and White8 mm/ka at zero runoff. This may be merely an artifactofthe scatterinthe measured denudation rates.Inprinciple, the denudation rate in an actual karst drainage basin can be determinedifone knows the areaofthe basin, the basin runoff, and the total dissolved carbon ate load discharged by the springs which drain the basin.Inpractice, data are usually sparse and there are some pro blems with relating dissolved load to long term average loweringofthe land surface. The dissolved carbonate load derives in part from dissolution at the land surface which penultimately results in a loweringofthat surface, but al so includes contributions that are carried into the basin by sinking streams as well as contributions from the dissolut ionofbedrock at depthinthe subsurface. Bedrock removal in the subsurface includes dissolution along fractures and bedding planes and enlargementofconduit drainage sys tems. Furthermore, most limestone spring waters are knowntobe undersaturated with respect to calcite because of the relatively sluggish kineticsofcarbonate dissolution in open conduit systems. Thus, the measureddenudation rate should be less than the equilibrium denudation rate. However, the regression line based on observed denudation rates for world-wide datainFigure 2 is actually higher than the equilibrium line. Few denudation measurements have been made in Appalachian watersheds. Data from Monroe County, West Virginia and Burnsville Cove, Virginia are plotted on Figure 2. The West Virginia data were obtained from time series dataon three conduit flow springs receiving mostoftheir catchment from carbonate rocks with some contribu tion from surface streams (Ogden, 1982). The Burnsville Cove denudation rates were obtained by combining chemi cal analysis data for four springs (Harmon and Hess, 1982) with some low flow discharge data (Davis and Hess, 1982). Because of the useoflow flow runoff, these values are probably less than mean annual denudation rates. Data are rarely available to determine the flux in response to different runoff conditions. However, by a strokeofgood fortune, the small karstic basinsofThomp son Spring, a diffuse-flow system, and Rock Spring, an open-conduit-drainage system, both in Centre County, cen tral Pennsylvania, were instrumented for discharge and regularly sampled for chemical analysis during an eight month period that included the Hurricane Agnes storm that swept across Pennsylvania in Juneof1972. From these data (Jacobson, 1973; Jacobson and Langmuir, 1974) itispossibletocalculate the fluxofdissolved carbonate rock being removed from these two small basins (White, 1990). What was immediately obvious was that although the con centrationofdissolved carbonate is diluted during storms, the decrease in concentrationismore than compensated by increased discharge. The net result is a large increase in the rateofdenudation during high-runoff events. Becauseofsmaller fluctuations in discharge during storm events, diffuse-flow springs give a more reliable measureofkarst denudation, at least if one has only short periodsofrecord.4Appalachian Karst Symposium, 1991The rateofdenudation can be calculated by integrating the fluxofdissolved rock over the time periodofobserva tion. Because the Thompson Spring data are discrete ob servations taken at a few-day intervals, the denudation rate is expressed as a summation Fi=Qi Hdi/pA Fiisthe instantaneous fluxofcarbonate rock lost from the basin in unitsofcm 3 sec1 km2 ; Dn is the denudation rate in mm/ka. The time unit hereisindays. Qi is discharge in m 3 /sec, Hdiishardness asCaC03inglm3 ,p is rock density inglcm3 ,and Aisbasin area inkm2 Thompson Spring receives its entire recharge on a limestone and dolo stone valley upland which is the local expressionofthe Harrisburg Surface. The catchment areais17.6km2 (esti mated upward from the 11.1 km 2 given by Jacobson) giv ing a denudation rateof45 mm/ka. Average precipitation in the Centre County area is 980 mm according to the University Weather Station. Depending on estimatesofevapotranspiration losses which vary strongly with season, plant cover; and topographic location, runoff is about 650 mm, allowing the Thompson Spring data to be plotted on Figure 2. Based on Figure 2, a denudation rateof30 mm/ka has been selected for the Appalachians. The uncertainties are large and a finer tuning does not seem to be warranted. The number isinthe same range as denudation rates based on other basin measurements and calculations summarized by Sevon (1985) and with the valueof27 mm/ka deter mined by Sevon (1989)forthe entire Juniata River Basin.AgeDatingoftheKarstSurfacesOne approach to the datingoferosion surfaces is to examine residual soils.Ifthese can be demonstrated to be residual, let-down soils, rather than alluviumorcollu vium, then a comparisonofsoil thickness with insoluble residue in the underlying bedrock gives a minimum rock mass that must have been removed to produce the soil.Anexample from the Harrisburg Surface in Pennsyl vania was described by Parizek and White (1985). The Harrisburg Surface cuts across the broad, multiply-folded archofthe Nittany Anticlinorium and exposes the Cam brian Gatesburg Dolomite at its center. This is an inter fluve area with no evidence for transported soils. There occurs on the Gatesburg a residual soil with a thicknessof30 to 100 meters for which a representative thickness was estimated to be 50 m (165 feet). To get an ideaofthe amountofcarbonate rock removed to produce the residual soil, we take the measured average insoluble residueof7.27 percent, a bulk densityofthe soilof1.76 gcm3 and


Appalachian Karst Symposium, 1991 White and White 3:Sketch showing amountofcarbonate rock losttoproduce residual soil In central Pennsylvania. Thickness scale at rightisinfeel. Using 30 m/Ma as the denudation rate gives14mil hon years as the time required for the bedrock columntobedissolved. Some predecessorofthe present surface with enough stability to preserve the residual soil must have existed since mid-Miocene time. This is consistent with someofthe younger estimates for the ageofthe Harris burg Surface, but it immediately poses a different problem. The present day elevationsofthe moun tain ridges that border the Nittany Valley are 2000 to2500feet (610 to 760 meters) with a reliefof820 to 1310 feet (250to400 meters). Oneofthe assump tionsinthe interpretationofthe Appala chian landscapes is that the resistant quartzites thatsupportthe mountain ridges have lowered relatively little dur ing the Cenozoic. Adding 430 meterstothe valley floor moves it above the ridge tops, which seems unlikely. Thereiscarbonate rock available as shown on Figure 3, but becauseofthe curvatureofthe anticlinal arch, the mid-Miocene ridges must have stood farther out into the valley and they must also have eroded substantially. This begins to look like the dynamic-equilibrium model proposed by Hack (1960). One additional result can be deduced from this number-juggling. The present dayHarrisburgSurfaceincentralPennsylvaniaisdissected by present-day streams producing a local relief on the surface, particularly evident on shale floored valleys,ofabout 100 meters. Taken as a proportionofthe14Ma that the valley uplands have been lowering, about 3 Ma has been required for the re cent dissections. That also gives a late Pliocene or Pleistocene age for the oldest cave fragments that have formedintheremnantsoftheuplandsurfacein response to the downcutting.o600 800 400 200 1400 1000 1200Theother pieceofground truth comes from the Mammoth Cave area. The Mammoth Cave System occurs be neath the sandstone caprockofthe Mam moth Cave Plateau. The present-day relief between the Sinkhole Plaintothe south and east and the sandstone contactis al:lout 90m.Recharge into the sinkholes and sinking streams drains beneath the Plateau to some slightly back flooded springs on the Green River. As the cave system deepened leaving abandoned high-level trunks, the Sink hole Plain should have been lowering to keep pace with the active drainsatthe baseofthe cave system. There have been three different attempts to estimate the ageofMammoth Cave from geomorphic evidence (Deike, 1967; Miotke, 1975; Palmer, 1989) allofwhich more or less agree that the highest abandoned cave levels date from late Pliocene or early Pleistocene. The second lineofapproachisSchmidt's (1982) measurementsofmagnetic reversals. Downinthe floodwater zoneofthe cave, the magnetic mineral grains in the sediments are=14Ma165FeetResidualSoil SoilThicknessx SoilDensityInsolubleResiduexRockDensityIDenudationnRateRockThickness ...; -----Time=\\\\\\"\\\\\ \. BedrockRemoved=\ ., \\\\.Mid-MioceneSurface/I//// -..: 7/ .F -III (I/I II,1// II -;I//(a densityofthe dolomite bedrockof2.85 gcm-3 .Simple mass-balance considerations demand that roughly 430 m (1400 feet)ofcarbonate rock must have been dissolvedtoproduce the 50 metersofresidual soil (Figure 3). Thisisa minimum figure since removalofthe soil by dissolutionofsomeofthe componentsorby piping into solution cavitiesinthe bedrock would demand an even larger bed rock thickness. 7 -/ 7 / "77"7 //I//// 7 // 7/o/I <5/7 oI lJ // I_---L./ __I 1--'/'----7-..!.I:.-(--l // lJ/ ( / 5


White and White Appalachian Karst Symposium. 1991Figure 4: The Harrisburg Surface in Burnsville Cove, Bath County, Virginia.4001000 E NE3500fee'1100 Q) 900zaBULLPASTURE IMOUNTAIN800 <1 >2500-2600'eelW -1 700w WC> a:o C>W a: Jf(f)""n. JACK MOUNTAINCOWPASTURE RIVERIt is difflcult Losee how the karst erosion surface on the Greenbrier Limestone can be much older than 2 to 5 The southwestern boundaryofthe Little Levelsisthe sandstone-capped Droop Mountain, which separates it from the limestone-floored Hills and Bruffey Creek valleys to the northwest. Muchofthe Little Levels is backed up against the Yew Mountains where the limestone dips away beneath the Allegheny Plateau. The topofDroop Moun tain is 3000 feet (915 m), or a little higher, but the topofthe limestone is lower on its flanks.Ifthe karst surfaceisreconstructed by replacing the lost limestone, it requires only a few million years for the western edge to butt up against the Droop sandstone and runoutoflimestone. Meanwhile the eastern edgeofthe karst surface slides up the dip slopeofthe western flankofthe Browns Mountain anticline. However, if the Greenbrier River has not shifted its course, the eastward migrating surface willbecutoffby the river as itistoday.TheGreenbrierKarstofWest VirginiaIn the Greenbrier River drainageofWest Virginia there's another remnant, called the Little Levels, around the townofHillsboroinPocahontas County (Figure 5). The Little Levels are at an elevationof2300 to 2500 feet (700 to 760 m), about the same as in Burnsville Cove. The up land karst surface is a doline plain with large closed depres sions. The present-day surface drainage, the Greenbrier River, is cut down to aboutanelevationof2100 feet (640 m). Another fragmentofthis representationofthe Harris burg Surface occursinthe Swago Creek Basintothe north and a larger fragment appears as the Big Levels, also called the Great Savanna, a major karst surface in Greenbrier Countytothe south (Jones, 1973) (Figure 6).DRYRUNTOWERHILLMOUNTAIN3200fee' W W ..J ..J..J..J DRYRUN CHESTNUT RIDGE '"'"r2500-2600fee'''\ 1200 1000'-------------------------' SWz a I<1 > j 2000wo Q)enQ) >o ..c 3000 Q).:'! =4000 3800fee' .!!! SurfaceReconstructionsNow what happens if this surface is pushed back in time? Thereisa karst erosion surface whichisequal to sandstone-capped mountain ridges and things are stayinginplace. It really does look like an erosion surface because it cuts across bedrockofdifferent lithologies,butsome mechanism must be found for keeping the carbonates and the clasticsinaccordance. On the drainage divide at the headofBurnsville Coveisa little remnantofthe upland surface whichischaracter ized by large closed depressions (White and Hess, 1982). There are fairly steep valleys falling away on both sides. There is something rather curious (Figure 4). The Burns ville karst isatan elevationof2500 feet (760 m), the same as the topofBullpasture Mountain and Chestnut Ridge which are remnantsofthe erosion surface capped with clastics. Although the Harrisburg here is a ridge-top surface, it is much lower than Tower Hill Mountain or Jack Mountain which would represent the Schooley sur face. Present-day base level is down at the Bullpasture River.Ifwe consider the reliefofa passage like Collins Avenue, which is right up against the sandstone90meters above present-day base level, and ask when it was an active drain carrying water to the paleo-Green River, it was presumably getting its water from the paleo sinkhole plain when it was90meters above the present-day sinkhole plain. Dividing 90 meters by the denudation rateof30 m/Ma again gives an ageof3 million years.TheMammoth Cave area checks out. Three independent assessments: a geochemical argument, a paleomagnetic argument, and a geomorphic drainage basin developmentargument allcomeout with the same answer to plus or minus something in the fIrst decimal place. Thisisalso the best evidence that scaling karst surfaces backwardintimewith the denudation rate as a scale factor is not a completely erroneous exer cise.TheValleyandRidgeofVirginia: Burnsville Covepointing to the present-day north pole. Near the middleofthe system, around the elevationofCleaveland A venue, someofthe sediments are reversed, thenthey be come normal again, then reversed again. Comparing these results with the known paleomagnetic time scale also gives an ageofabout twotothree million years.6


Appalachian Karst Symposium. 1991 White and White4500 1400 e 01300III Q) III e 4000 Q) 0 I.-Q).... 1200 CLro III Q) III l::Je III e III eroI.-....Q)ro III >III1100 Q) Q) 3500 ro > ....I.-....Q)....roeQ)l::JQ) 0 Q)ro E l+-e U > I.> e 0 Q) 0U010000 2 ...J I.-....0::::Ue2ro 0 UI.0 e 0 Q)..02Q)I.-a.. 0 .... 3000 3: III ..0Q)+'"ro....Q)'900 roQ)....Q)I.> >...J I(f)..o > Q)eQ)wQ) 800 wQ) 2500 I.-lJ Greenbrier700Limestone2000 0 24 68101214 16(km)Figure 5: Cross-section drawneast-westthrough LobeliaandHillsboro 7.5-minute quadrangles, Pocahontas County, West Virginia. Profile is along N 38' latitude. Ma, unless theassumedvalue for the denudation rate haschangeddramaticallyoverthistimespan.Obviously,there was a precursor landscapebutit is very questionableiftheearlierlandscape had theappearanceofthe present surface.TheHighlandRimA thirdandfinalexampleis the western marginoftheCumberlandPlateauintheCumberlandRiverdrainage(Figure 7)(see alsoWhiteandWhite, 1983).Therethe HarrisburgSurfaceisrepresentedby theHighlandRimwhich isatabout1000 feet (305 m) elevation.TheCumberland Plateau itself variessomewherebetween 1800 and2000feet (550 to610m).Theelevationatthe topofthe Bangor limestone marks the stratigraphic horizon at which developmentofakarstplain could have started. That is as far up as therewasany limestone to dissolve. Dissolutionofthe limestone surface down to the present-day Highland Rim requiresremovalofabout150 meters through the whole section. Againwearriveatabout5 mega yearsasthe timeatwhich the Highland Rim butted up against the baseofthe clastics thatmakeup the topoftheCumberlandPlateau,andthis defines amaximumageatwhich anything resembling the present karst surface could exist.Conclusionsevidence,mostofthe cavesofthe Appalachians have ages thatextendmaybebackthrough thePleistocenebutnot much more than that. Many are related to valley systems thatfonnedby the dissectionofthe valley upland (Harris burg) surface.TheAppalachianshavebeen eroding for a long timebutthe recognizable fonnoftheHarrisburg Sur face appears to be much younger than the traditional geo morphic literature has claimed. Rather than Eocene, a very late Tertiary(maybelateMioceneorPliocene) age seems to fit the data best for this guidingsurface for the Appala chian karst.Theoldest record described in this investigation was14Ma, derived from residual soilsonthe Harrisburg Sur face in Pennsylvania.Thesesoils required, however, re movalofsuch a thicknessofrock thatonemustquestionwhatchanges took place in theothertopography, particu larly thequartzite-cappedmountainridges,duringthis periodoftime. Interpretationofdenudation rates on otherkarstsurfacesrevealsseverestratigraphic constraintsifthese surfaces are reconstructed farther back than a few mil lion years.Ifonemeans by "ageofthe surface", the timeatwhich the surface evolved into something recognizably similar to what is seen today, the Harrisburg Surface can not beolderthan a few million years. However, there isalsothe possibility that inspiteofrecentneo-Davisian interestinerosion surfaces, a Hackian modelofcontinuous evolution may beller fit the Appalachian landscape.Itseemsthat thisbriefexcursion into the relationship between karst and the Harrisburg Surface has raised more questions than it has answered. Judging from the available Returning to the two quotes cited earlier, can these be reconciled?TheAppalachians have been eroding since the 7


White and WhiteoSCALE(MILES)Appalachian Karst Symposium. 1991THEGREATSAVANNAAQUIFER,(ADAPTEDFROMPRICEAND HECK,19:59)::i' MAUCHCHUNKSERIES GREENBRIERSERIES McCRADYAND POCONOSERIES &iiifI ALLUVIUMFigure6:Map showing the "Great Savanna" karst on the Greenbrier Limestone, Greenbrier County, West Virginia. closeofthe Paleozoic but there has also been a huge quan tityofrock removed. The conflict appears to concern the landscape itself and to what extent present-day topography can be followed backwardintime. Evidence from karst investigations supports Thornbury's pointofview. The recognizable detailsofthe landscape are relatively young.8ReferencesDavis, N.W. and Hess,l.W.,1982, Hydrogeologyofthe drainage basin, Burnsville Cove, Virginia: National Speleological Society Bulletin,44,p.78-83.


Appalachian Karst Symposium, 1991 White and White300 350 250 800 500 400 550 450 1000 1200 1600 1400 1800FeetMeters HartsellePenningtonFormationPottsviIIIe(Sewaneess)MonteagleLimestoneBangorLimestone..' -..... ".."....","..' ;;:" ..": :"::: .;:.::::,;.=:.::".:. :.'":.' ; Escarpment IIICIII01 01C 0 01 0If)III > (Dolin' karstT150m1 St.LouisLimestone .........---======-:7 WarsawFm.IFt.PayneFmCumberlandPlateauProfile200 600Figure7:Schematic cross-section through the western marginofthe Cumberland Mountains, north-central Tennessee in the Cumberland River drainage. Davis, W.M., 1889,Therivers and valleysofPennsylva nia:National Geographic Magazine,v.1,p. 183-253. Deike, G.H., III 1967,The DevelopmentofCavernsoftheMammothCaveRegion:Ph.D.dissertation(unpublished), Pennsylvania State University, 235 p..Miotke, F-D., 1975, Der Karstimzentralen Kentucky beiMammothCave:JahrbuchderGeographischenGesellschaft zu Hannoverfur1973, 360p.Morisawa, M., 1989, Rivers and valleysofPennsylvania, revisited:Geomorphology,v.2, p. 1-22. Fenneman, N.M., 1938,PhysiographyofEastern United States,McGraw Hill,NewYork, 714 p. Hack, J .T., 1960, Interpretationoferosional topographyinhumid temperate regions:American JournalofScience,v.258-A, p. 80-97. Harmon, R.S. and Hess,l.W.,1982, Ground water geo chemistryofthe BurnsvilleCovearea, Virginia:National Speleological Society Bulletin,v.44, p. 84-89. Jacobson, R.L., 1973,Controls on the QualityofSome Carbonate Ground Waters: Dissociation ConstantsofCalciteandCaHCO]+from0 to50C:Ph.D. disserta tion (unpublished), Pennsylvania State University,131p. Jacobson, R.L. and Langmuir, D., 1974, Controls on the quality variationsofsomecarbonatespringwaters:JournalofHydrology,v.23, p. 247-265. Jones, W.K., 1973, Hydrologyoflimestone karst:West Virginia Geological Survey Bulletin,no.36,49p. Ogden, A.E., 1982,Karstdenudation rates for selected spring basins in West Virginia:National Speleological Society Bulletin,v.44,p.6-10.Palmer, A.N., 1989,Geomorphichistoryofthe Mam moth Cave System (Chapter 12) in White, W.B. and White, E.L. (editors),Karst Hydrology: Concepts from the Mammoth Cave Area:Van Nostrand Reinhold, New York, p. 317-337. Parizek, R.R. and White,W.B.,1985,ApplicationofQuaternary and Tertiary Geological FactorstoEnviron mental Problems in Central Pennsylvania:Guidebook to 50th Annual Field ConferenceofPennsylvania Geol ogists, p. 63-119. Price, P.H. and Reger, D.B., 1929,Pocahontas County:WestVirginia Geological Survey, 531 p.(seeplate 16B). Price, P.H. and Heck, E.T., 1939,Greenbrier County:West Virginia Geological Survey, 846 p. 9


White and WhiteSchmidt, V.A., 1982, Magnetostratigraphyofsediments in Mammoth Cave, Kentucky:Science,v.217, p. 827 829. Sevon, W.D., 1985,Pennsylvania's Polygenetic Land scape: Guidebookfor the 4th Annual Field Tripofthe Harrisburg Area Geological Society,55 p. Sevon, W.D., 1989, Erosion in the Juniata River Drainage Basin, Pennsylvania:Geomorphology,v. 2, p. 303 318. Smith, D.I. and Atkinson, T.C., 1976, Process, landforms and climate in limestone regions (Chapter13)inDerby shire, E. (editor),GeomorphologyandClimate,John Wiley, London, p. 367-409. Thornbury, W.D., 1965,Regional geomorphologyofthe United States:John Wiley, New York,609p. White, E.L. and White, W.B., 1983, Karst landforms and drainage basin evolution in the Obey River Basin, north central Tennessee:JournalofHydrology,v.61, p. 69 82. White, W.B. and White, E.L., 1974, Base level controlof10Appalachian Karst Symposium. 1991underground drainageinthe Potomac River Basin,inRauch, H.W. and Werner, E. (editors),Proceedingsofthe Fourth Conference on Karst Geology and Hydrology:West Virginia Geological Survey, p. 41-53. White, W.E. and Hess, J.W., 1982, GeomorphologyofBurnsville Cove and the geologyofthe Butler Cave Sinking Creek System:National Speleological Society Bulletin,v.44, p.67-77. White, W.E., 1984, Rate processes: chemical kinetics and karst landform development (Chapter 10)inLaFleur,RG.(editor),Groundwaterasa Geomorphic Agent:Allen & Unwin, Boston, p. 227-248. White, W.B., 1988,GeomorphologyandHydrologyofKarst Terrains:Oxford University Press, New York, 464 p. White, W.B., 1990, Surfaceandnear-surface karst land forms (Chapter 7)inHiggins, C.G. and Coates,DR(editors),Groundwater Geomorphology: The RoleofSubsurface Water in Earth Surface Processes and Land forms:Geological SocietyofAmerica Special Paper 252, p.157-175.


Appalachian Karst Symposium. 1991 Medville and MedvilleStructural Controls on Drainage beneath Droop Mountain, Pocahontas County, West VirginiaDouglasM.Medville and HazelE.MedvilleWest Virginia Speleological Survey 11762 Indian Ridge Road Reston, VA 22091ABSTRACTOver 8.7 miles (14 kilometers)ofsurveyed cave passages are found in a 590-foot (180-meter) thick sequenceofMississippian-age carbonates around the northern endofDroop Mountain in southern Poca hontas County, West Virginia. Regional dip is 4 to 5 degreestothe northwest. Whereas primary subsur face flow paths along both the western and eastern sidesofDroop Mountain are generally aligned along strike (N 30 E), other flow routes, passing beneath the mountain and crossing the dip, are used as well. This paper compares flow-path gradients, describes relationships among caves, and discusses the influence that faults subparallel to the dip have on flow-path direction. Inconsistencies among measured dip eleva tions and the topofthe Union Limestone on both sidesofDroop Mountain are also noted.IntroductionDroop Mountain isa north-south trending ridgeinsouthern Pocahontas County, West Virginia. Local reliefisabout 1000 feet with a maximum elevationof3100 feet (Figure 1). The ridgeiscapped with Mississippian-age sandstones and shales. Itisbounded on the east by the valleyofLocust Creek and on the west by the Hills and Bruffey Creek valleys, flowing from higher elevations to the northwest. The northeast flankofDroop Mountain is bounded by north-trending valleys that open to the southern marginofa sinkhole plain to the north (Little Levels). Middle Mississippian-age carbonatesofthe Greenbrier Group crop out on the ridge's western, northern, and east ern sides. These limestones, along with interbedded shales and sandstones, are about 650 feet thick. The main lime stone sequence in the Greenbrier extends from the Union Limestone to the basal Hillsdale Limestone. The thick nessofthis sequenceis600 feet, as measured by the Sun Oil No. 1 Droop Mountain well drilled at the northern endofDroop Mountain and describedinLeonard (1968). Regional dipisN 60 W, strike N 30 E, and measured dips on both sidesofthe ridge are 4 to 5 degrees (Worthington, 1984). The entire limestone sequenceisexposed along the mountain's eastern side with the baseofthe Hillsdale Limestone seenatan elevationof2100, feet at a spring forming the headofLocust Creek (Locust Spring). To the11western sideofDroop Mountain, in the Hills/Bruffey Creek valleys, only the upper 50 or so feetofthe Unionisexposed. Streams sinkingonthe western flankofDroop Moun tain, e.g.Hills Creek, Bruffey Creek, and Rush Run, flow entirelyorinpart, depending on flow conditions,tothe southeast, beneath the Droop Mountain ridge, crossing the dipofthe limestones, and risingatcave entrances and springs along the northeastern and eastern sidesofthe mountain. The area has been well studied since the late 1950's with the basic hydrogeologic selling summarized by White and Schmidt (1966) and early stream tracing car ried out by Zotter (1963,1965). Additional stream trac ing has been carried out by Coward (1970), Williams and Jones (1983), and Jones (1991). Speleogenesisofthe major caveinthe area, the Friars Hole System justtothe south, was studied by Worthington (1984) and the struc tural geologyofthe Droop Mountain area has been sum marized by Jameson (1985) and Kulander and Dean (1978). Within the past ten years all caves at the northern endofDroop Mountain that have any hydrological significance have been resurveyed. These surveys, totalling nine miles, have been carried outineight cavestothe north and eastofthe 43-mile Friars Hole System. The areal extent and rela tive locationsofthese caves are showninFigure 2. Pro files from the survey data demonstrate relationships among cave-stream gradients and can be used as indicatorsofin ferred flow routes where such routes have yettobe demon strated through stream tracing.


MedvilleandMedville Appalachian Karst Symposium. 1991PENNSYLVANIANWISSISSI PPIAN '."'-LETf ...111111\ Sillt---r--'-zrl--sri........ Greenbrl.rGroup CIIm.,' ... ) Group "d.I....) POlIIII.Group('d,I,c ;' .....r1 [EJ"'" ." ".::I!'I :":"...,. .....,...COllE IIUIl ......KIIOI... "' ...., ".ccr.d,Formatla.(,hal.)Figure1:Geologic mapofDroop Mountain and vicinity. From Jameson (1985).12


Appalachian Karst Symposium. 1991Key1.Locust Spring2.Locust Creek Cave3.GeneralAverellPit 4.Martha's Caves5.HughesCreek Caves6.Hills-Bruffey Cave M edville and M edville Figure2:13


Medville and MedvilleStructural SettingThe structural setting was summarized by Jameson as follows: "Droop Mountain is near the southern marginofthe central Appalachian fold and thrust belt... In this re gion, first order folds, major thrust faults, and many small er structures trend about N 30 E. The area is at the south ern marginofthe Webster Springs block; bounded on the eastbythe Browns Mountain anticline, on the south by theModoclineament, and on the west by the Webster Springs anticline ...TheModoc lineament, considered to be somewhat illusorybyDeanet al.(1979), is a3 to 7 mile-widezonewhich trendseastacross the regional structural trendofN30 E" (Jameson, 1985, p. 106, 108). Faults beneath the northernendofDroop Mountain that strike east-west and may be associated with the "illu sory" Modoc lineament are described in this paper. Their influence on both cave-passage termination and on allow ing flow to occur across the regional dip are noted. These faults are distinct from the thrusts that strike at N 25-30 E and are observedinmany locations in the Friars Hole Cave Systemjustto the south. An inconsistency between the observed elevationsofthe topofthe Union Limestone on the eastern and western flanksofDroop Mountain and measured dips should be noted; this inconsistency supports Worthington's hypo thesis that an anticlinal axis exists beneath the northern endofthe mountain (Worthington, 1984,p.24-26). Mea sured dips to the northwest, on both sidesofthe mountain, are 4 to 5 degrees. On the eastern sideofDroop Moun tain, the bottomofthe basal limestone member, the Hills dale Limestone, isat2100 feetatLocust Spring. Directly downdip and on the western sideofDroop Mountain, the Union/Greenville contact isatan elevationof2510 feet, directly above the Bruffey Creek Cave entrance.Therela tive elevationsofthese contacts are consistent with a uni-Appalachian Karst Symposium. 1991form dipof1.5 to 2.0 degrees rather than the 4 to 5 degree dips that are observed.Fora consistent 4 degree dip, the topofthe Union Limestone should be 200-250 feet lower than the floorofthe Hills and Bruffey Creek valleys rather than 50 feet above these valley floors asisthe case.Theinconsistency is illustrated in a profile beneath Droop Mountain looking along strike (Figure 3). Profilesoftwo major caves in the area Hills/Bruffey Cave on the left and Locust Creek Cave on the right are shown, along with the600-footthicknessoftheGreenbrierGroupbetween the Union/Greenville contact and the baseofthe Hillsdale Limestone. A4 degree dip is assumed and, on the eastern sideofDroop Mountain, the locationoftheTaggardFormation, aprominentshale-limestone-shale marker bed, is shown. This bed is observed in Locust Creek Caveinthe location shown on the figure. Deep-seated thrust faulting and an associated anticline beneath Droop Mountain has been suggested by Worthing ton (1984)asan explanation for the observed elevationsofthe Union/Greenville contact. The hypothetical fault plane parallels regional strike with the down thrown side to the west.Ifthis is the case, however, the topofthe Union, across an anticlinal axis westofthe fault, should be even lower in elevation than is thecasefor a uniform 4 to 5 degree northwest dip. Whereas westerly dipping, strike oriented thrusts are observed in cavesinthe areajustwestofDroop Mountain, these may be back thrusts associated with the hypothetical fault beneath Droop Mountain. Al though displacement along these is minimal, such faults,en echelonand beneath Droop Mountain, may account in part for the observed elevationsofthe Greenbrier Group in the Hills/Bruffey valleys.Subsurface DrainageKnown flow paths beneath the northern endofDroop3000 2000 ----.>...-:::.r-::= HB LC LSo40008000Figure3:ProfileofDroop Mountain looking north-northeast along strike. Key: HB Hills-Bruffey Cave; LC Locust Creek Cave; LS Locust Spring; andTg Taggard Formation. Elevations and horizontal scale areinfeet.14


Appalachian Karst Symposium, 1991 Mountain are illustrated in Figure 2.Todate, three such paths have been by!ones ot.hers. The nor thernmostofthese begins with the sinking ofBruffey Creekinits bedatthe topofthe Union Limestone, about a half-mile northofthe entrance to Bruffey Creek Cave. The water flows almost due east for 7400 feet, drops 220 feet, andisseen againinUpper Hughes Creek Cave on the eastern sideofDroop Mountain. Although the observed gradientinUpperHughes Creek and the projected straight-line gradient between the smkofBruffey Creek and thiscaveshould permit a fairly rapid flowthrough time, Jones (1991) reports a flowthrough timeofabout a month. This suggests a circuitous route beneath Droop Mountain and/or a substantial amountofponding between sink and rise. From Upper Hughes Creek Cave, water has been traced to Locust Spring, probably via Lower Hughes Creek Cave,LowerMarthas Cave, and General Averell Pit, although specific cave-to-cave traces have not been carried out. Hills Creek is also a losing stream and under most flow conditions enters small cavesatthe baseofa ridge on the western sideofits valley. This water has been traced for 11.3 miles, more-or-Iess along strike to the southwest, and through a seriesofcaves(CutlipCave, Cochrane Sinks, and the Friars Hole System) beforensmgat springs along Spring Creek.Inmoderateto high-flow conditions, however, someofwaterinboth Bruffey Creek and Hills Creekoverflows the sinkpoints noted above and flows into cave entrances on the western sideofDroop Mountain These entrances are about 50 feet below the topofthe Union Limestone and are about 1200 feet apart. The streams can be followedtoa junction and the combined flow follows a circuitous routetothe northeast and then south. The stream is finally lostinrockfall along an east-west trending fault that termi nates the cave. This overflow route for Hills and Bruffey Creeks (presumably, an older route) passes beneath Dr.O?P Moun tain, crossing the dipofthe limestones. Thensmgofthe Hills-Bruffey overflow waters is at Locust Spring, 9000 feet to the southeast and 480 feet lowerinelevation. The flow path is direct; dye hasnotbeen detected at Upper Hughes Creek Cave. The third path beneath Droop Mountain isthat be tween the ultimate sinkofRush Run and Locust Spring. Rush Run, flowing from the northwest, normally sinksinits bed andiscaptured by the southwest-trending Cutlip Clyde Cochrane-Friars Hole flow path noted above. ,un?er high-flow conditions, someofRush Run passesthiSsmk point and flows into cave entrances at the endofits blind valley, 1/4 mile westofthe western flankofDroop tain. While mostofthis water still flows southwest via ClydeCochraneSinks,someofit flows east, at Locust Spring (Williams and Jones, 1983). As with the Hills-Bruffey Cave water, dye was not detected at Upper15Medville and MedvilleHughes Creek Cave.Probable Flow RoutesLocust CreekCavelies behind Locust Spring and contains about2milesofsurveyed passage. The cave's entrance is flooded and thus the cave is accessible only to divers. Muchofthe cave consistsoflarge, base-level passagesinthe lowerDenmarFormation.Thecave contains three streams representing the downstream endsofthe flowpaths described above. All traces conducted to date have been to the cave's entrance; no traces have been completed to the streams inside the cave. notes that the cave's high-flow drainage area, mcludmg the Hills Creek and Bruffey Creek basins, is about 28 square miles and that discharge from the cave has been measured as 2.5 to 350 cubic feet per second. The cave contains three streams, allofwhich join up streamofthe sumped entrance passage.Themajor sourceofwater in the cave is a stream entering from the north. This may be the Hills-Bruffy water; the volume is consis tent with that seeninHills-Bruffey Cave and the hypothe tical route is direct; a continuationofthe easterly trending flowofthe Hills-Bruffey water. The straight-line distance is 5400 feet and the vertical difference is 320 feet (gradient is 0.06). A smaller stream in Locust Creek Cave flows from the southwest and probably represents water sinkinginvalleys on the eastern sideofDroop Mountain. This stream, however, may also contain the water traced from Rush Run. Figure 4 is a profile looking N 25 W and shows the relationship between the sinkofRush Run and Locust Creek Cave. The viewing direction is perpendicu lar to a straight line between the sinkofRush Run and the upstreamendofLocust Creek Cave's southwest stream. Assuming a fairly direct flow path to the northeast, the hypothetical gradient (0.024) is consistent with that seen in the cave stream. Finally, the Bruffey Creek streambed water, rising at Upper Hughes Creek Cave, has also been Locust Spring. Whereas the probable flow pathIS Via Lower Marthas Cave and then General Averell Pit, this has not been confirmed by tracing. Also, the specific stream withinLocust Creek Cave where this water rises has not been identified through stream tracing, although it is probably the stream enteringfTomthe north. Figure 5 is a profileofthese caves, looking from the east.Therelative elevationsofthe streams in the caves and the gradientsofthese streams (0.02) are consistent with a single flow path. The locationofthe large northern stream entering Locust Creek Caveisindicated by an "N"inthis figure. OriginofFlow PathsThe existenceofthree distinct and more-or-Iess par allel flow paths beneath Droop Mountain, eachofwhich


Medville and Medville Appalachian Karst Symposium. 19912400 2100 RushRunSink*-----? ---------LocustCreekC.o10,000Figure 4: Profile from Rush Run Sink to Locust Creek Cave. View direction is N 25 W. Elevations and horizontal scale areinfeet. crosses the regional dip and drops through up to 600 feetofcarbonates and interbedded clastics, is striking. The pre dominant directional orientation for solutional conduitsinthe area is along strike (NNE-SSW), following enlarged bedding-plane partings. Cave passages trending normaltothe strike tend to be high-gradient inlets following joints with theflow direction downdip to the northwest.Theoccurrenceoftraversible conduits containing streams flowing across thedipis infrequent, although such conduits must exist, given the resultsofthe stream traces carried out.2500f\2300 2100oIn someofthe cavesatthe north endofDroop Moun tain, several passage segments trending nearly east-west and containing easterly flowing streams, are accessible.Inother caves, east-west passage segments, not necessarily containing streams, are also accessible. In both cases, these passage segments parallel high-angle thrust faults that strike east-west to east-southeast/west-northwest,i.e.,subparallel to the regional dip. It is hypothesized that these faults provide a structural mechanism for solution tobeinitiated and subsurface flow to occur beneath Droop Mountain. The dipofthe faults is usually to the northorfjGAPf\i 6000Figure5:Profile showing caves draining to Locust Spring. View direction is looking west. Key:LCLocust Creek Cave; GAP General Averell Pit; LM Lower Martha's Cave; HC Hughes Creek Caves; and fESE-WNW trending thrust fault. Elevations and horizontal scale areinfeet.16


Appalachian Karst Symposium, 1991 NNE, although in one case, the dip is to the SSW. The fault planes are almost perpendicular to thoseofthe previ ously observed strike-oriented faultingincaves in the area. Examplesofthese faults are given below. (a) Hi lls-BruffeyCave:Thestream in this cave meanders to the northeast and then south for over6000feet following an enlarged bedding-plane parting in the Union Limestone. It then turns abruptly to the east and is lost in breakdown along a fault. The cave passage, however, can be followed for another 425 feet to the east following this fault, which strikesN80-85Wanddips to the north at 35 to 45 degrees. (b) Upper Hughes Creek Cave: For over800linear feet, the cave's stream flows from west to east along a fault strikingN85Wanddipping to the north at40to 50 degrees. This is the longest traversible segmentofsuch passage in theDroopMountain caves. As previously noted, the cave stream has been traced from the bedofBruffey Creek, on the west sideofDroop Mountain. (c) Lower Marthas Cave, trending SW along strike for 3000 linear feet, endsata fault striking N60Wanddip ping northeast at 45 degrees. At this point, streamflow turns to the southeast along the fault although, as is the case with other passages along these faults, the stream is soon lostinrockfall. (d) The entrance to General Averell Pit, on the north eastern flankofDroop Mountain, follows the planeofa thrust fault downward for 360 vertical feet to a stream segment. The fault strikesN60Wanddips steeply(60to70 degrees). The dip direction, however,isto the SSW,incontrasttothe fault terminating the south endofLower Marthas Cave. A few hundred feetoftraversible passage, containing a stream (almost certainly that seen in Lower Marth as) parallels the planeofthe fault. (e)CutlipCave,on the western sideofDroopMountain and containing that partofthe underground Hills Creek that flows to the southwest, terminates in rockfall along a fault that strikes at N 60Wanddips to the NNE at 20 to 30 degrees. The cave stream is lost in rockfall and continues southwest toward Clyde Cochrane Sinks. In this case, east-west faulting only affects passage mor phology and does not alter the flow direction.(f)At the upstream terminationofthe stream flowing from SW to NE in Locust Creek Cave, an upper passage ascends very steeply over rockfall, climbing over 120 ver tical feet and breaching the Taggard shales. While this may be an exampleofa passage ascending along a northerlydipping fault, additional evidence is needed. The dipsofthe generally east-west trending faults des cribed above are apparent when viewing the profilesofthese caves from the east;i.e.,along the strikeofthe faults. This can be seen in Figure 5 where, for examples17Medville and Medville(b),(c), (d), and(f)above, the arrows labeled"f'indicate the locationsofthe faults. Whereas the faults exist and apparently provide routes for water to follow when passing beneath Droop Mountain, their origin and relationship to other regional structures is a problem for further study. It is worth noting that these faults parallel and perhaps coincidewith the northern partofthe postulated east-west trending Modoc lineament and indeed, couldbeusedas evi dence for the existenceofthis linear zone. Corroborating evidence such as surficial fault developmentorchanges in axial trends,assuggested by Dean and others (1979), or changes in the alignmentsofsurface streams has been neither observed nor soughtinthe area.SummaryCave development beneath northern Droop Mountain is a fairly complex process (Jones, 1983) involving stream piracy from one subsurface route to another, flow both along strike and across dip, and a degreeofcontrol over the directionof flow by thrust faults that both parallel the strike and are subparallel to the regional dip. Whereas the overall subsurface drainage pattern has been established, much work remains tobedone concerning cave-to-cave flow routes and time-discharge relationships. Also, until such time as traversible conduits beneath Droop Mountain are found, the extent and natureofthose structures beneath the mountain that influence cave development and flow direction canbehypothesized and discussed,butnot observed.ReferencesCoward, J.M.H., 1975, PaleohydrologyandStreamflow SimulationofThree Karst Basins in Southeastern West Virginia. U.S.A.: Ph.D. thesis (unpublished), McMas ter University, Hamilton, Ontario, 394 p. Dean, S.L.; Kulander, B.R.; and Williams, R.E., 1979, Regional Tectonics, Systematic Fractures, and Photo linears in Southeastern West Virginia: Proceedingsofthe Second International Conference on Basement Tec tonics, Newark, Delaware. 1976,p.10-53. Jameson, R. A., 1985, Structural Segments and the Anal ysisofFlow Paths in the North CanyonofSnedegars Cave. Friars Hole Cave System. West Virginia: M.S. thesis (unpublished), West Virginia University, 421 p. Jones, W.K., 1983, Evolutionofthe Hills Creek basin, West Virginia (abstract): Programofthe National Speleological Society Annual Convention. Elkins, West Virginia,p.69-70. Jones, W.K., 1991, Locust Creek Cave hydrology,NSSNews, v. 49,p.139. Kulander B.R. and Dean, S.L., 1978, Gravity, Magnetics, and Structureofthe Allegheny Plateau/Western Valley


Medville and Medvilleand RidgeinWest Virginia and Adjacent States:West Virginia GeologicalandEconomic Survey. ReportofInvestigations,no. 27,91p. Leonard, A.D., 1965,The PetrologyandStratigraphyofUpper Mississippian Greenbrier LimestonesofEastern West Virginia:Ph.D. dissertation (unpublished),WestVirginia University, 245 p. White, W.B. and Schmidt, V.A., 1966, Hydrologyofa karst area in east-centralWestVirginia:Water Resour ces Research,v. 2, no. 3, p. 549-560. Williams, C.F. and Jones, W.K., 1983, Karst drainage systemsofthe northern Spring Creek basin, West Vir-18Appalachian Karst Symposium. 1991ginia (abstract):Programofthe National Speleological Society Annual Convention, Elkins. West Virginia.p.92-93. Worthington, S.R.H., 1984,The Paleo Drainageofan Appalachian Fluviokarst:Friars Hole. West Virginia:M.S. thesis (unpublished), McMaster University, 218 p. Zotter, H., 1963, Stream tracing techniques and results: Pocahontas and Greenbrier Counties,WestVirginia:NSS News,v. 21, no. 10, p 136-142. Zotter, H., 1965, Stream tracing techniques and results: Pocahontas and Greenbrier Counties,WestVirginia,Part2:NSSNews,v.23,no. 12, p. 169-177.


Appalachian Karst Symposium, 1991 Saunders and BalfourKarst and CavesofMercer and Summers Counties, West VirginiaJosephW.Saunders1and WilliamM.Balfour213207 Melody Lane Lansing,MI48912 2Marshall Miller&Associates P.O. Box 848 Bluefield, VA 24605ABSTRACTCaves and karst drainage occur at four general stratigraphic levels in Mercer County:TheCambro-Or dovician dolomites, mid-Ordovician limestones, Mississippian Greenbrier limestones, and upper Mississip pian A vis Limestone. In Summers County only the latter two are exposed.Thedolomites run the lengthofEastRiver Mountain, dipping into the mountain a thousand feetormore below the lengthy mountain crest. Most mountainside drainage above the lower exposuresofthe dolomites is captured and drains to springs in those lower exposures; several caves less than 600 feet long are known there.Theoverlying mid-Ordovician limestones host plentiful mountainside dolines and ponors.Thelongest known cave by far is three-mile-long, strike-aligned Beacon Cave; there is a swing and branching in the St. ClairFaultand mountainside drainage discharges from limestone.TheGreenbrier limestones are exposed as part ofan over turned syncline along the lengthofthe footofEast River Mountain and also in the Abbs Valley Anticline. Here, two major features are a water-supply cave draining Interstate 77, and a 4000-foot, straightline piracyofBig Spring Branch.TheAvis Limestoneisup to 40 feet thick, mostly relatively low dip, doline-scarce, and exposed in steeply dissected terrain in eastern Mercer County and mostofSummers County. Cave passagesinthe Avis are highly joint-controlled, often mazy, and associated with piracyofnearby streams. JonesCave(5655 feet long), where a landslide was instrumental in the cave's development, has two insur gences, a flood-outlet entrance and a perennial-spring entrance. Brickyard Ridge Cave (2450 feet long) is a complexmaze developed in the plungeofthe Abbs Valley Anticline, and has a high-water outlet. Therearefew perennial springsinthe A vis, but Adair Run valley has threeofthem along its length, where the lime stone drops 700 feet in two miles. At this point, below the valleyside exposuresofthe Avis, several seg mentsofalmosthalfofthe two-mile course are often dry, and subsurface flow occurs in the Payne Branch Sandstone. An 820-foot maze in the Avis Limestone is the longest cave known in Summers County.IntroductionMercer and adjacent Summers County lieinsouthern West Virginia (Figure1)and are twoofthe state's lesser known karst counties. Neither county has been involvedinanyofthe publicationsoftheWestVirginia Speleolog ical Survey. Davies (1958) didnotcover Summers Countyatallinhis compendium on the state's caves, and someofthe information listed for Mercer County wasinerror. Even today partsofboth counties containing caves are unknowntokarst researchers and cavers.StructureMercer County contains the southernmost extension19ofthe Valley and Ridge Province inWestVirginia. The southern edgeofthe county comprises the outcrop slopeofthe 23-mile long East River Mountain that continues into Virginia,oneofthe Appalachian long mountains whose karst was described by Saunders and others (1977). The longest major fault in the county, the St. Clair, parallels the flankofthe mountainside for the entire lengthofEast River Mountain (Reger, 1926). Just northofthe fault line lies the Hurricane Ridge Syncline, which is overturned on its southern limb. Further north, the structure grades into the gentle foldsofthe Appalachian Plateau Province which predominates inSummersCounty.JustnorthwestofBluefield is the very pronounced Abbs Valley Anticline. This anticline rapidly plunges to the northeast andisthe structural extensionofthe Boissevain Fault that dies out


Saunders and Balfour Figure1:LocationofMercer and Summers counties, West Virginia. Summers County is northeastofMercer County. as it crosses into Mercer County from Virginia. The topo graphic expressionofthis anticline is not apparent northofPrinceton eventhough the fold continues at depth into Monroe County.StratigraphyCaves and karst drainage occur at four different strati graphic horizonsinMercer County. The Cambro-Ordovi cian dolomites, upto3500 feet thick, crop out along the lengthofEast River Mountain, dipping fairly uniformly into the mountain crest. The lower exposuresofthe Cam bro-Ordovician dolomites are located along the St. Clair fault line, which limits the proportionofthe dolomites involved. The overlying mid-Ordovician limestones are approximately 100 feet thick, with the pink and white, thin-bedded, sandy Moccasin Limestoneinthe upper parts. Above the limestones are calcareous shales as well as sand stones, with the Tuscarora Sandstone forming the crestofthe mountain. The Mississippian Greenbrier Limestone crops out in a narrow band on the southern limbofthe overturned Hurricane Ridge Syncline that parallels the St. Clair Fault, as well as in a narrow, six-mile long strip along the crestofthe Abbs Valley Anticline in the western partofthe county. The Greenbrier is thicker (upto1700 feet) and more argillaceousinMercerCounty thaninsev eral counties along its outcrop belt to the north. The Up per Mississippian Avis Limestone (Hinton Group) reaches a thicknessofupto forty feetinpartsofMercer and Sum mers counties. Some partsofthe Avis are highly argilla ceous, and even the beds most conducive to cave develop ment have characteristic irregular, thin shale partings apparent throughout. The A vis Limestone is bounded below by a sharp contact with fissile red or yellow shale, 20 to 30 feet thick, and above by the Upper A vis Shale thatisupto40 feet thick. 20 Appalachian Karst Symposium,1991SurfaceExpressionThe Cambro-Ordovician dolomites are foundindoline poor terrain. The mid-Ordovician limestones crop out in dolineand ponor-rich terrain, often with benches and small, strike-aligned valleysofgentler slope. In the Greenbrier Limestone, dolines are common and limestones form valleys. The A vis Limestone crops out very promi nentlyinsteep, dissected terrain, usually as "turrets"upto ten feet long. Sandstone rubble covers the outcropinmany places,especially on the gentler slopes. Dolines are rare in the Avis.CavesTwo caves are known in the Cambro-Ordovician dolo mites. Big Spring Cave (500 feet long) and Tank Wilson Cave (approximately 60 feet long) both endinsumps. Big Spring Cave has a high-gradient stream with water falls. Beacon Cave (Figure 2), more than three miles long, is by far the longest caveinthe mid-Ordovician limestones in Mercer County.Ithas a major, strike-aligned trunk over 300 feet long that carries a large stream that sumps several hundred feet shortofits final resurgencein200 foot long Beaver Pond Spring Cave. Large blocks at the entrance to the Beaver Pond Spring Cave appeartohave dammed the water and caused several feetofdepthinthe sump. Removalofthe blocks might permit another access to Beacon Cave. Cave Rat Cave (3200 feet long)isa complex, but shallow, cave farther east on the mountain, enteredata ponor. KFC Cave, located behind the Kentucky Fried Chicken restaurant in Bluefield, drains the parking lot and is a small maze (500 feet long) that follows the dip. A 37-foot drop with a waterfall is devel oped near the endofthe cave, beyond which the stream sumpsina rubble-filled joint. Several pits with depthsof60-150 feet are also found in the mid-Ordovician lime stones. Holiday Hole, located adjacent to the parking lotofthe Bluefield Holiday Inn, is a 115-foot deep breakout dome with a 70-foot entrance shaft into a large breakdown floored room. Whispering Falls Pit, near Oakvale, is an enlarged joint measuring 120 feet long by 30 feet wide and 60 feet deep. A stream flowingoffEast River Mountain cascades into the pit as a 50-foot waterfall and sinksinrubble at the bottom. The longest known cave in the A vis LimestoneisJones CaveinMercerCounty (Figure 3), with four entran ces and 5655 feetofpassage, over a straight line distanceof3500 feet. Complex passage areas are found primarilyinthe vicinityofentrances, where dripping along ceiling cracks has created joint-orientated loops and deadends. Pal mer (1975) identified this typeofdiffuse infiltration as a mechanismofmaze-cave formation. A landslide composedofsandstone rubble at the upstream endofthe cave appears to have blocked a valley and also been instrumentalinthe cave's development, as the pirated waters flow generally


NIo !!!!!!FeetE=EntranceBEACONCAVEMERCER COUNTY, WEST VIRGINIA is"C")i!::3"";::'.......'0 '0....... EHK91Figure2:Beacon Cave, Mercer County. This caveisaligned along bedrock strike. Original Survey by V.P.I. Grollo, 1971-1974. IV -Figure3:Jones Cave, Mercer County. The caveisthe longestinthe Avis Limestone. Map from Saunders and Koerschner, 1976. V:ll:::31}:3I:l..t:l:lS;(;)l:::... EJONESCAVEMERCER COUNTY, WEST VIRGINIA-==N Feet E=Entranceo ,!III, EEHK91




Appalachian Karst Symposium, 1991downdip through the hillside to emergeina parallel valley (Saunders and Koerschner, 1976). Brickyard Ridge Cave (2500 feet long) has a complex maze connected by a straight conduit to the only entrance known in the Avis that neither takes nor discharges flow (Figure 4). Brick yard Ridge Cave is formed along the plunging axisofthe Abbs Valley Anticline,justbefore the Avis dips below drainage.Ofall the Avis caves, only Jones and Brickyard Ridge have significant fossil passages(i.e.not presently carrying flow). All caves in the Avis are strongly joint controlled and all except Brickyard Ridge have a stereo typical vadose-spring entrance. In Summers County, the longest known cave is 820 foot long Neely's Nose Cave, a maze developed in the Avis Limestone. At Barger Springs, the along Greenbrier River, thereisa cave by the same name that is developedinthe Alderson Limestone, the uppermost memberofthe Greenbrier series. This cave is approximately 600 feet long. Two topographic features in the county refer to caves: Cave Ridge may be named for 11O-foot long Barker Hollow Caveinthe A vis Limestone, and Cave Branch appears to be named after a large shelter caveinthe Avis Sandstone. To date there are 50 recorded cavesinMercer County, 20inthe A vis Limestone, 7inthe Greenbrier Limestone, and 23 in Ordovician carbonates. In Summers County there are13recorded caves, two large shelter cavesinsandstone, 10 in the A vis Limestone, and one in the Greenbrier.Subsurface Drainage PatternsMost, if not all, streams flowingoffthe upper slopesofEastRiverMountain sink into the mid-Ordovician limestones during all but high-flow conditions. Most dis crete sink points do not permit human entry; many areinsinkholes choked with sandstone float from the overlying caprock. Many stream sinks are located stratigraphicallyinthe highest limestone unit, the Moccasin. The sinking streams do not occupy allofthe upper-slope ravines on East River Mountain and preliminary examination has found that this higherdrainageis confluent, exiting primarily from solitary springs in as yet unidentified stratigraphic units somewhat above the Moccasin. By number and greater proportionoftotal discharge, most significant springs in the carbonates along East River Mountain are found in the Cambro-Ordovician dolo mites. Most examined so far do not have associated caves. Some, like Big Spring Cave spring and Tank Wilson Cave spring are located immediately above the Sl. Clair Fault, at the lowest available exposureofthe dolomite. Many other springsinthe dolomite are located somewhat above the lowest possible exposures. As noted earlier from more preliminary generalizations (Saunders and others, 1977), failureofmanyofthe discharge points tobefound at the lowest carbonate levelismostly a reflectionofthe dolo-23Saunders and Balfourmite lithology in the long mountain setting. The only possibleoverflow outlet known along East River Mountain maybeanother exampleofthe low per meabilityofthe dolomites. Pigeon Creek, southofOak vale, risesata spring in dolomite that is well above the Sl. Clair Fault. A high-water outlet in dolomite exists a thousand feet up the valley. Yet another thousand feet far ther up the valley is an outletinthe limestone with inter mittent flow, suggesting that it is partially related to flow in the intermittent outletindolomite. The largest limestone spring along East River Moun tain is the Beaver Pond Spring, the outlet for the Beacon Cave stream. Nearby is the most significant irregularity in the St. Clair Fault along East River MountaininWest Virginia.Thefault branches and swings closer to the mountain crest. It may not be coincidence that the longest and most accessible limestone cave along the mountainislocated here. No water tracing has been performedinthe East River Mountain karst. Based on modelsoflong, underground drainage basins proposed by Saunders and others (1977), two quite different configurationsofunderground flow can be visualized and may co-existinsomeofthe underground drainage basinsinEast River Mountain. One scenario, seeninBeacon Cave, involves essentially strike-orientated flow along and beneath the mountainside in subsurface basinsofelongate shape. A contrasting situation involves main flow generally diagonalorperpendicular to the strike, against the dip, away from the mountain crest. Tributary flow in both types is often down componentsofthe dip and into the mountain. Limestones usually host strike aligned main flow in the long mountains, as exemplified by Beacon Cave. In contrast, main flow against the dip appears to occur primarily in the dolomite, creating shor ter, but wider, basins. Unfortunately, the actual conduit orientations, if conduits exist, are largely unknown due to limited access to passages. However, Big Spring Caveislike several other dolomite caves elsewhere in the long mountain setting; it has a steep gradient with waterfalls and cascades, in contrast to the typical, nearly flat, gradient seen in limestone cavesinthe long mountains. The Greenbrier Limestone is exposedintwo linear bends in Mercer County, oneofwhich runs the lengthofthe county along the flankofthe Hurricane Ridge Syn cline. Significant hydrological features are evidentattwo locations in the latter. The eastern Big Spring Branch sinks entirely for partofthe year into its bed southwestofKellysville and appears to resurge 4000 feet down strike to the east along the East River. Secondly, Ingleside Cave carries a domestic water supply stream that would appear to drain nearby Spangler Valley, a doline-rich area through which Interstate Highway 77 runs. Most known cavesinthe A vis Limestone are asso ciated with sinking streams and most cave entrancesinthe


Saunders and BalfourAvis are springs. Mostofthe sinking streams and springs are seasonalinnature. All but the two longest cavesinthe Avis canbedescribed as simple dendritic-flow systems with a single discharge point at the solitary cave entrance. Thesecond longest Avis cave,Brickyard Ridge Cave, has both a perennial and a high-water outlet. The longest cave, Jones, also appears to be the most hydrologically complex. There are two streams sinking into the cave, as well as a high-water outlet behind a passable collapse zone, well upstreamofthe perennial spring. All four locales provide entrance to the cave. It appears that increasing cave length is associated with increasing hydro logic complexity. Other than the associationofBrickyard Ridge Cave with a steeply plunging anticline, no predictive relation shipsofAvis Limestone cave location with structure or topography have been formulated. Given the low-dip char acteristicofmostofthe A vis outcrop area, it can notbeconcluded that dip is a factorindevelopmentofthe com monly short (less than 900 feet long) and complex A vis caves. The abilityofthe A vis Limestone to conduct flow over long distances would entail captureofdrainage on up dip sidesofmountains and downdip discharge on the other sideofthe same mountains. Prospective flow under Bent or Tallery Mountains, for example, would occur along flow pathsofat least three miles in length. Perhaps the most interesting, as well as unusual, areaofsubsurface drainageinthe two-county area isinAdair Run valley, northofKellysvilleinMercer County. Situ atedina structurally transitional area between the over turned syncline at the baseofEast River Mountain and the gentle dip to the north, the valley follows the dip, losing 700 feetintwo miles. Just southofElgood, the head watersofAdair Run cross the Avis Limestone, sinking and rising within a hundred feet. Shortly downstream, flowinthe dry month, April 1988, sank into the Payne Branch Sandstone underlying the lower Avis Shale. Flow finally appeared amidst cobbles above the confluence with Woodall Fork (approximately two miles distant), which itself flows a short distance in the Payne Branch Sandstone before discharging from cracksinthe sandstone bedrock 24Appalachian Karst Symposium, 1991floorofAdair Run. For the entire lengthofthe valley, the A vis limestone is 20 to 50 feet above the valley floor. Three small caves are knowninthe A vis Limestone on the northeast sideofAdair Run, plus a thousand-foot piracyofa tributary, an unexplored cave, and two water-supply springs on the southwest side near the lower endofthe 2.5-mile long valley.SummaryIn summary, karst exploration in Mercer and Sum mers counties has begun to locate and define caves and karst drainageinthe long mountain setting as well asinan isolated limestone confined primarily to these two counties. Four stratigraphic levels have been identified where caves and karst are found. Much information is yet to be learned because these two counties have long been neglected in favorofthe more traditional karst areasofWest Virginia.ReferencesDavies, W.E., 1958,CavernsofWest Virginia:West Vir ginia GeologicalSurvey,v.19,330p. Palmer, A.N., 1975, The originofmaze caves:National Speleological Society Bulletin,v.37,p.57-76. Reger, D.E., 1926, Mercer, Monroe and Summers Coun ties:West Virginia Geological Survey County Report,963 p. Saunders, J.W. and Koerschner, W.F., III, 1976, West Vir ginia's Jones Cave:NSS News,v.34, p. 6-10. Saunders, J.W.; Medville, D.M.; and Koerschner, W.F., III, 1977, Karst drainage patterns in the long mountainsofthe eastern United States,inFord, T.D. (editor),Pro ceedingsofthe 7th International Speleological Con gress, Sheffield.1977:British Cave Research Associa tion, Somerset, England,p.375-376.


Appalachian Karst Symposium. 1991Hydrochemical and Structural Controls on Speleogenesis in the AppalachianFoldbelfKass KastningP.O. Box 1048 Radford, VA 24141ABSTRACTKastning The formationofcaves is affected by composition and porosityofrock, structureofbedrock, and the relationofcaves to groundwater flow and surrounding topography. Caves in Cave Hill, Augusta County, Virginia were investigatedinorder to determine the relative effectsofeachofthese factors in the folded Appalachian Mountains. Partofthis study was a determinationofporosityofsamples taken from Grand Caverns, a cave that cuts through many vertically dipping bedsinCave Hill. Previously, these samples were analyzed for con tentofcalcite and insolubles. Results indicate that bedsofrock that are purest in calcite and microcrystal line are the least porous, and those that are low in calcite content and contain mostly insoluble materials are the most porous. That the most soluble rocks are the least porous or permeable implies that the ground water flow that dissolved them may have been carried through the rock sequence along adjacent highly porous and permeable bedsofinsoluble material, such as interbedded sandstones. This mechanism for speleogenesis in rocksofmixed lithologiesisproposedasa new hypothesis that may have application in many other cave regions. A secondpartofthe study wastoproduce a geologic map of Cave Hill. The resulting map, incorpora ting measurementsofbedrock attitude and position and elevationsofcaves, explains muchofthe variation seen among caves in the hill.In troductionThe originofcaves has beenofinterest for some time and manyofthe details have been studied in various geographic and geologic settings (White, 1988; Ford and Williams, 1989; Palmer, 1990, 1991). However, at some localities, additional information can cometolight with precise fieldwork. Recent visitstoGrand Caverns, located in Cave Hill in northeastern Augusta County, Virginia, indicated that this would be an excellent sitetoinvestigate the rolesofbedrock composition, porosity, and structure in cave development in the Appalachian Mountain region. Because Cave Hillisa miniature versionofthe ridgesofthe Appalachian Mountains that run northeast and southwest through western Virginia, processesofcave development here should be similartothose that have occurred throughout the mountain range. Therefore this study should add to the knowledge about cave developmentinfolded rocksofthis region.Thefirst phaseofthis study, completed in 1990, analyzed selected samplesofthe Conococheague Forma tionoflate Cambrian age that is host to at least eight caves on Cave Hill. A very strong correlation between calcite contentofthe rock and passage enlargement was established (Kastning, 1990a,b). Several cavesofvarious sizes have formed in Cave Hill (Figure 1). They differ significantly in shape and size even though they are located very close to each other along a 915-meter stretchofthe hill. There must be basic geologic reasons for these differences. This studyisanin vestigationoffactors that have given each cave its geome tric characteristics, including(1)the porosity and permea bilityofthe bedrock, (2) the structureofthe bedrock (folds, faults, joints), and (3) the relationshipsofcavestothe surrounding land (topography) and how the originofthe caves may have been affected by changesinthetopography during the geologicpastHighschool science fair project.AuLhorwas a finalist and award winner at the 1990 and 1991 International Science and Engineering Fairs, and recipientofa 1991 Naval National Science Award andLhe1990 James G. Mitchell AwardofLheNational Speleological Society.25


Kastning200 0 !wee Alluvium Terrae. depollta Lowerlimtoneof B""manlown Group Cooococh.ogue FormationSand,toneIon.In theConococheogue Formallon ................ Inferred fault ,(22'Slrl'"ond dipofbed'Appalachian Karst Symposium, 1991QolFigure1:Geologic mapofCave Hill showing caves, bedrock, and structure. Cross-sections are shown in Figures 4 and 9. Three hypothesesaredeveloped in this study: (1) Bedsofthe Conococheague Limestone that are purest in calcite and microcrystalline are the least porous and those that are high in insoluble minerals have generally higher porosity, (2) Distinct changes in the structureofthe bed rock from one place on Cave Hill to another are responsi ble for differences in the morphologyofcave passages at these locations, and (3) Caves have formedatelevations that correspond to former levelsofthe streams in the area (namely the South River, a tributary to the Shenandoah River). As streams eroded and lowered the surrounding land surface, water tables dropped, and lower and lower cave levels developedinsequence. The caves are developed in the Conococheague For mation that is easily recognizable throughout the moun tainous regionofwestern Virginia because it consistsofa dense crystalline gray-to-black limestone that contains numerous thin wavy bedsoflight brown-to-tan clay and sand. Sedimentary rocks in western Virginia have been severely folded during the Paleozoic Era when continental26plate collisions formed the Appalachian Mountain range.ThebedsofCave Hill exhibit considerable structural deformation.Methods Laboratory AnalysisofRock SamplesSamples collected from Grand Caverns were analyzed in a laboratory to determine the approximate specific grav ity, porosity, and permeabilityofvarious typesofbeds.Thefollowing procedure was followed in acquiring data: (1) weigh the samples while completely dry using an anal ytical balance, (2) soak the samples in a beakerofwater for about ten minutes to allow the sample to takeinall the water possible, (3) weigh the wet samples, (4) deter mine volumesofthe samples by measuring the volumeofwater displaced as the sample is submersed, (5) weigh soaked samples while suspended in a beaker, that is kept above the panofthe analytical balance, and (6) calculate


Appalachian Karst Symposium. 1991porosity and specific gravity. Porosities and specific gravitiesofthe samples are listed in Table I.MappingItwas necessary to determine the precise locationsofimportant features on Cave Hill so that they could be placed on mapsofthe surface orofcaves. This was done with accurate "compass-and-tape" surveying. The survey progresses from station to station and the data is then processedbycomputer and plotted.Table 1Porosity and Specific GravitySample Percent Content Percent Specific Number Calcite* Porosity Gravity 1 89.8 3.3 2.71 2 0.0 27.1 2.14 3 32.6 2.2 2.62 5A 9.9 72.0 1.835B6.25 13.0 2.40 6 6.4 28.4 2.19 7 76.3 1.5 2.71 8 55.3 1.6 2.68 9 64.6 5.2 2.57lOA1.1 21.1 2.22lOB2.8528.6 2.21110.0 68.01.8112 0.0 40.4 1.701389.6 7.0 2.711410.9 25.9 2.17150.4 47.4 1.731688.1 3.8 2.621792.2 28.4 2.39185.9 33.1 2.341973.5 3.3 2.67 20 76.05-2.9 2.762159.2 4.4 2.70 22 36.4 5.2 2.72230.0 5.5 1.99240.0 20.92.242568.5 4.8 2.67 26 61.5 6.4 2.60 27A 38.9 2.05280.0 6.5 2.43 *From Table 5, Kastning, 1990a.27KaslningItwas also important to precisely determine the eleva tionsofthe caves in Cave Hill andofmanyofthe surface features that are in the vicinity. This was accomplished using the V-tube method described by Palmer (1970). The V-tube is a lengthoftransparent plastic tubing thatisnearly filled with water. Two stadia rods are fashioned from wood on which a datum can be marked. When the tube is opentothe atmosphere, water in the tube willbeat the same level (elevation) at each endofthe tube. Dur ing surveying, a datumismaintained and is moved up or down with rising or lowering terrain. In this way precise elevationsoffeatures can be determined. Sedimentary rock strata that began as horizontal de posits are steeply inclined at Cave Hill. Someofthe bedsinthe caves are even vertical. The orientations, or atti tudes,ofbedsofrock were measured as strikes and dips (Suppe, 1985; Marshak and Mitra, 1988). This was done with a Brunton Compass. During this study, forty-six strike-and-dip measure ments were made on Cave Hill and in the caves. Strike and-dip symbols were plotted on the geologic map (Figure1)as it was constructed. Once a numberofmeasurementsofattitudeofbeds were taken, it was possibletostatistically plot these on a stereographic equal-area (Schmidt) net(Figure 2). This graphical representation clearly shows clusteringofdata (beds that have nearlyidentical attitudes) and can be used to work out many geometrical solutions to folds, such as the attitudeofa fold axis or the plungeoffolds( the entire fold may slope into the ground). These analyses were conducted using standard stereo-net techniques (Mar shak and Mitra, 1988). Itisobvious from looking at mapsofcaves on Cave Hill (Figure1)thatpassages typically appeartobe com posedofstraight segments (lineaments). Manyofthese are paralleltoone another; others maybeperpendiculartothese. This suggests that these passages are controlledbythe geometry of the strata or by fracturesinthe rocks. Lineaments may be plotted on polar graph papertoindicate how these directions cluster. Statistical plots, known as rose diagrams, were constructed for the cave passage data using the methodofSmith (1968). Rose dia grams for eight caves on Cave Hill and the nearby South River are showninFigure3.Oneofthe main objectivesofthis study wastocom pile a geologic mapofCave Hill in the vicinityofthe largest caves. Because a map shows only a two dimen sional viewofthe area,itwas also desirabletoconstruct cross sections through the hill that would show the sub surface geology and cavesinrelation to each other. Both the map and cross sections together provide a three dimensional perspectiveofthe geologyofCave Hill.


Kastning Appalachian Karst Symposium, 1991NORTH41Cave 32 Fountain Cave -North I\..33 3121GrandCaverns-712II.5Northeast and Central42046Attitudes(UngroupeddataarefrompathstoGrand Caverns and FountainCave)Figure2:Schmidt equal-area net showing polesofdipping beds. Note differencesinorientationofbeds in caves northofthe faultinFigure 1 and those southofthe fault. The geologic map (Figure1)and four cross sections (Figures 4 and 9) were constructed using a photographicallyenlarged topographic map, existing geologic maps (Gathright, Henika, and Sullivan, 1978a,b), horizontal survey data, leveling-survey data, structural fielddata(Figure 2), and cave maps. The geologic cross sectionsofFigure 4 were constructed roughly perpendicular tothetrendofCave Hill and crossing through key geologic features including major caves. These sections showthe.South River and itsfloodplain. Cross sections were28


Appalachian Karst Symposium, 1991ROSEDIAGRAMSOFPASSAGE ORIENTATIONSKastning / StegersFissureGrandCaverns-NortheastJeffersonCaveMadisonSaltpetreCaveGrandCaverns-SouthwestFountainCaveMad-As-HellCaveBabyGrandCavernsGrandCaverns-TotalFakeSnakeCave AllCavesSouthRiverFigure3:Rose diagrams for eight cavesinCave Hill and the South River.29


Kastning Appalachian Karst Symposium. 1991CROSSSECTIONSOFCAVEHILLNorthwestASoutheast A 1200feel _ Rive_r _--vi 0011100Madison SaltpetreCave Eco IIII II8 8'II1300I,JIIIIGrandCavernsI.,. ,.1 III,1200 E:co JI1/II/I Eco IOldSouthRiver/I ';:----/II,001cc'Vertical exaggeration=4x2000200400600feel,.I I'I //I/ Eco //I//FountainCave1400 13001200Figure 4: East-west cross sections through Cave Hill showing attitudeofrocks and vertical positionofcaves.SeeFigure 1forlocationsofcross sections. 30


Appalachian Karst Symposium. 1991 Kastning drawn with an expanded vertical scale in order to best il lustrate vertical relationships, including slopes. The map and cross sections put the topographic features, caves, and bedrock together in one package so that allofthese things can be compared and relationships determined. Measurementsofbedding attitude were used to con struct beds at their correct angles on cross sections. Using data from the U-tube leveling surveys, elevationsofmajor caves were placed in their correct vertical position (Figure4).This was done to show how caves relate to the sur rounding beds, bedrock structure, topographic features, andtoeach other. ordertodetermine any relationship between these variables (Figure 5). The graphofporosity versus limestone purity shows distinct differencesinporosity for samplesofvastly differ ent composition. There is a strong correlation between porosity and calcitecontentIt is easily seen from Figure 5 that high-purity limestone samples (30 percent and higher) have very low porosity and that samples with a high levelofinsoluble material show considerable poros ity. In fact, porosityisgenerally less than 10 percent for relatively pure limestone and greater than 20 percent for rocks high in insoluble content.ResultsCharacteristicsofRocksThe graphofspecific gravity (relative density) versus limestone purity also indicates a distinct relationship. Rock samplesofhigh-calcite content have a high specific gravity. The valuesofspecific gravity for these samples lies very closetothe valueof2.71, the known specific gravity for calcite (Klein and Hurlbut, 1985). On the other hand, the specific gravity for samplesofmostly in soluble materialissignificantly lower, around 2.0to2.4. The significanceofthese determinations is that the most soluble materials (highest purity limestone with least insoluble material) arc the least porous and that the least soluble materials (lowest purity limestone with most insoluble material) are the most porous.StructuralControlofCavesIn an earlier study, Kastning (l990a,b) determined that high-purity limestone was dense and microcrystalline. The dataofFigure 5 supports the hypothesis that dense materials would have considerably fewer pores and will thereby hold less water.Ingeneral, the rose diagramsofFigure 3 show that orientationsofpassages in each cave agree well with the known structureinthe immediate vicinityofthe cave. This is particularly true for twoofthe large caves, Madi son's Saltpetre Cave and Grand Caverns. In Grand Ca verns thereisa difference between average passage orienta tion in the southwest endofthe cave and thatinthe main partofthe cave. In this case the rose diagrams agree well with the sLrike-and-dip readings takeninthe cave or just outside the entrances (compare FiguresIand 3). The notable exception to good agreement between the rose diagram and structural dataisFountain Cave (Figures1and 3). The structural data indicates that the strikesofrocks taken in the cave lie between N77W and N55W and between N500E and N86E. But, the rose diagram shows that mostofthe cave is oriented between true north and N30oE. Insteadoffollowing the strike, Fountain Cave is roughly following the plungeofa small syncline.Therose diagrams for smaller caves and for allofthe caves combined (Figure 3) show a trend thatisconsistent with the overall structural trendofCave Hill (FigureI).8070 ! 60 (jrarufCaverns... en 50 0 30Samp[es ...0 40 I:l..... 30 s:: 20 to)... 10 0.0 -10 0102030405060708090100PercentCalcite3.02.8 to 2.6 i 2.4 "to) 2.2 ,!3 2.0 (jrarufCavernsto)81.8 I30Sampfesf/} 1.6Porosity and specific gravity for samples collected in Grand Caverns are listedinTableI.The percent calcite compositions determined in an earlier study (Kastning, 1990a,b) are given in the second columnofTable1.Porosity and specific gravity were each plotted against ,compositionofthe beds (expressedaspercent calcite)in0102030405060708090100PercentCalciteFigure5:Graphsofporosity and specific gravity versus (Percent calcite for 30 samples from Grand Caverns.31


KastningThe combined rose diagram for all caves was constructed by weighing each cave according to its length. Conse quently, the trendsofthe three larger caves dominate this diagram. Figure 3 also gives a rose diagram for linear segmentsofthe South River in the vicinityofthe caves. The trendofthe river is generally paralleltothe overall structural trendofthe area. This is as expected since the river lies between mountain ranges that follow the regionalstructural pattern. The 46 strike-and-dip measurements are plotted on an equal area (Schmidt) net (Figure 2). Poles clustering at or near the circumferenceofthe net represent vertical beds, such as those in mostofGrand Caverns (poles number 216on Figure 2). Poles 17-19 are the attitudesofrocks in the southern partofGrand Caverns. These have a slightly different position and represent a different structural trend. BedsinJefferson Cave (poles 44-46) are close to being horizontal (the poles are near the centerofthe Schmidt net) but they have about a 13to17 diptothe southwest. BedsinFountain Cave (poles 30 through 41) dip between 10 to 30 degrees to the south-southwestorsouth southeast.DiscussionandConclusionsAnExplanation for SpeleogenesisinMixed LithologiesInsome limestones, the porous rock is not necessarilythe one that is the most soluble. Crystalline lime stone, although highincalcite content, may be imperme able to water flow.Inthe Conococheague LimestoneofCave Hill, numerous thin bedsofinsoluble material, gen erally sand and clay, easily soakupwater and transmit it. However, where acidic water comesincontact with lime stone, the rock will dissolve, and the insoluble material will not.Anexplanation for cave developmentisproposed here that is generally different from other explanations normallyfoundinbooks and articles on cave origin. Thisisa direct resultofthe studies performedatCave Hill and in the laboratory with samples that came from Grand Caverns. Because water has a difficult time getting through dense microcrystalline rock, it will be unabletotravel within the limestone and dissolve openings into the rock. However,ifadjacent beds are highly porous and permea ble, they can bring water to the limestone (Figure 6, step I). The water must flow through openings to get to the rock andtoleave the rock carrying dissolved material.Aswater flows through a bedofporous rock, it comesincontact with the adjacent limestone (Figure 6, step1).Anopening begins to form next to the insoluble (but permeable) rock. The opening isinthe limestone 32Appalachian Karst Symposium. 1991(Figure 6, step 2). When the opening becomes large enough, flow within it will increase, allowing still larger flowstooccur. The flows become rapid enough to effici ently excavate a cave (Figure 6, step 3). The insoluble beds are left behind, perhaps jutting from the walls or ceil ings until they break off (Figure 7).StructuralControlsCave passages northofan inferred fault, namely Steger's Fissure, Madison's Saltpetre Cave, and Grand Figure 6: Sequenceofdevelopmentofcave passagesininterbedded porous sandstone and non-porous, microcrys talline limestone. Initially, groundwater flows only in the sandstone (step1).Once adjacent bedsoflimestone begin to dissolve (step 2) both flow and solutional excavation accelerate within the limestone (step 3).


Appalachian Karst Symposium. 1991Figure 7: Beds in the ceilingofa large roominGrand Caverns. Note bladesofinsoluble sandstone and clay (dark colored) protrude into the passage, whereas beds .of microcrystalline limeslOne (light colored) have been dls solvedand are recessed. Widthofview is about 3 feel. Caverns (Figure 1), have developed along strike within nearly vertical beds that are highincalcite content and very soluble. In contrast, cave passages southofthe in ferred faull, namely Jefferson Cave and Fountain Cave, are not strike oriented. PassagesinJefferson Cave and partsofFountain Cave are aligned along fractures. Additional ly, other passages in Fountain Cave trend parallel to the axisofa small plunging syncline (Figures 1 and 4). Statistical support for these relationships are graphically shown in rose diagrams (Figure 3) and on a Schmidt equal-area net (Figure 2). Passages in the caves may be classifiedas following the strike, oriented along fractures,orpositionedinother ways (such as following the dip). Distributionofpassage lengths within each cave with respect to these three cate gories are showninthe ternary diagramofFigure8.This plot concisely summarizes the differencesinstructure and its control on speleogenesis for caves on either sideofthe faull.33KastningSTRIKEORIENTED100%100%50100%FRACTUREORIENTED OTHERFigure8:Ternary graph showing structural controls.oncave development in Cave Hill. Caves northofthe In ferred fault are Steger's Fissure (SF), Madison's Saltpetre Cave (MSC), Baby Grand Caverns (BGC), Mad-As-Hell Cave (MHC), and Grand Caverns (GC). Those southofthe fault are Jefferson Cave (JC) and Fountain Cave (FC). Fake Snake Cave (FSC) is a small cave on the western flankofCave Hill.Topographic RelationsCave levelsinCave Hill are lower toward the north. The older passages, toward the south, are well drained and flow now occurs in the flooded lowest levels (Figure 9). The geologic cross sections suggest a correlation elevationsofthe drained caves and former floodplainS, marked by river terraces on the edgesofthe modem-day floodplain (Figures 4 and 9).AcknowledgmentsI thank the following people for their assistance in this study:Mr.DavidR.Leatherwood, Superintendentofthe Upper Valley Regional Park Authority, for allo.w!ng me to collect samplesinGrand Caverns and permlllIng access to Grand Caverns and Fountain Cave during the off season. Dr. RobertK.Boggessofthe Chemistry Depart ment at Radford University, for allowing access to his chemistry laboratory and the useofequipment for study ing porosity. My parents for transportation to Grand Ca verns and assistanceinfinding references, sample collect ing, photography, laboratory work, and wor? processing. Most importantly, they shared their expen ences and knowledgeofcaves and geology. Mr. Ron


Kastning Appalachian Karst Symposium, 1991oSOUTH1400fee.LONGITUDINAL SECTIONOFCAVE HILL0'NORTH1300JEFFERSON CAVE dJ 17JJTER=OEPOSIT FOUNTAIN GRAND CAVERNS120South River Floodplain ----__n h __ ;j2___ \FAULT' n -South _-=-_=_..::..:::=_____ 17TERRACEDEPOSIT .---=-==-=._...=.. RlverOMADISON'S SALTPETRECAVE.-.. '/-:., 1000 400400 800leel-."....! !IVerlical Exoggeralion=4x STEGERS_.FISSURE'Figure 9: ofCave Hill showing major caves, topographic profile, South River, and mapped terrace and floodplamdepOSIts.SeeFIgureIfor locationofthe section. Morton, Mr. Tom Spina, Mr. Phil Lucas, and Dr. John Holsinger for providing information on caves located on Cave Hill.ReferencesFord, D.C. and Williams, P., 1989,KarstGeomorphology and Hydrology:Unwin Hyman, Winchester, Massachusetts, 320p.Gathright II, T.M.; Henika, W.S.; and Sullivan III, J.L., 1978a, Geologyofthe Crimora quadrangle, Virginia:Virginia DivisionofMineral Resources Publication 13text and1:24,OOO-scalemap. GathrightII,T.M.; Henika, W.S.; and Sullivan III, J.L., 1978b, Geologyofthe Grottoes quadrangle, Virginia:Virginia DivisionofMineral Resources Publication 10,text and1:24,OOO-scalemap. KasLning, Kass, 1990a,Speleogenesis in the Appalachian Foldbelt: Phase I: EffectsofLimestone Composition on Cave Formation:Science Fair Project and Report, Radford High School, Radford, Virginia, 88p.Kastning, Kass, 1990b, Effectsoflimestone composition on cave formation (abstract):Abstracts, 41st Interna tional Science and Engineering Fair, Tulsa Oklahoma,Science Service, Washington, D.C.,p.125. Klein, C. and Hurlbut, C.S., Jr., 1985,ManualofMiner-34alogy (After James D. Dana),20th edition: John Wiley and Sons, New York, 596p.Marshak, S. and Mitra, G., 1988,Basic MethodsofStruc tural Geology:Prentice Hall, Englewood Cliffs, New Jersey, 446p.Palmer, A.N., 1970, Some recent innovations in cavesurveying:National Speleological Society News,v.28, no. 4, p. 44-46. Palmer, A.N., 1990, Groundwater processes in karst terranes,inHiggins,c.G.and Coates, D.R. (editors), Groundwater geomorphology: The roleofsubsurface water in Earth-surface processes and landforms:Geological SocietyofAmerica Special Paper252,p.177-209. Palmer, A.N., 1991, Origin and morphologyoflimestone caves:Geological SocietyofAmerica Bulletin,v.103,no.I,p.1-21. Smith, A.R., 1968, A roseisa roseisnot a flower:The Texas Caver,v.13, no. 10,p.113-115. Suppe, J., 1985,PrinciplesofStructural Geology:Pren tice Hall, Englewood Cliffs, New Jersey, 537p.White, W.B., 1988,GeomorphologyandHydrologyofKarst Terrains,Oxford University Press, New York,464p.


Appalachian Karst Symposium, 1991 JamesonConcept and ClassificationofCave Breakdown: An AnalysisofPatternsofCollapse in Friars Hole Cave System, West VirginiaRoy A. JamesonDepartmentofGeology and Geophysics UniversityofMinnesota Minneapolis, MN 55455ABSTRACTCave breakdown consistsoflocally derived, predominantly bedrock fragments. Featuresofbreakdown include individual fragments, their accumulations, and the morphologic modifications they impose upon passages.Thesurfacesofbreakdown range from dissolutional surfaces to fracture surfaces as endmember forms. Individual breakdown is classified into fracture forms, dissolutional forms, and mixed forms, accord ing to surface type. Breakdown is described by loose shape terms such as curved sheet, flake, beam, wedge, blade, chip, slab,andblock. Breakdown with a predominant origin is classified by genesis; examples include exfoliation and crystal-wedging fragments. BreakdownwiLha distinctive combinationofshape and origin is classified morphogenetically; examples include pendant cluster fragments and canyon trench blocks. Breakdownisclassified according to its modeofoccurrencewiLhina collapse continuum.Thecon tinuum describes breakdown withinLhecontextofthe modifications that fragmentation and particle accumu lation impose upon passage morphology. The continuum is arranged in orderofroughly increasing size or complexity.Thecollapse continuum includesinsitufragments, isolated collapse fragments, local feature collapse, passage-junction collapse, major passage collapse, and large chamber collapse. Featuresofthe continuum in Friars Hole Cave System include paleo-mud-crack fragments, exfoliation fragments, gypsum crystal wedging fragments, cross-bed blades, fault wedges, pendant clusters, floodwater-maze spans and blades, pothole fragments, flute fragments, shaft debris, trench blocks, disintegrating blocks, junction-room trench blocks, breakout domes and associated debris, terminal breakdowns, and large chambers.IntroductionManyaccountsofbreakdowninlimestonecaves(Davies, 1949, 1951; White and White, 1969; Sweeting, Jennings, 1971, 1985; Bagli, 1980; Ford, 1988; hite, 1988) identify collapse as the predominant,orat !fast proximate (Ford and Williams, 1989) originofbreak'Gown.Often, breakdown is equatedwiLhjagged or angular bedrock fragments that have fracturedsurfaces. Such a conception, however, can easily misdescribe breakdown. Breakdown need notbeangularorhave a significant per centage ofarea as fractured surface at the timeof(or imme diately following) fragmentationorcollapse. Breakdown need not later retain more than a few percentofsurface area as fractured surfaces and stillbeconsidered breakdown. Moreover,breakdownneed not originate by collapse: much material considered breakdown, such as gypsum crystal wedging and exfoliation fragments, has fragmented35in situ,and then been displaced only a few centimeters, slowly by sliding and rotating, even upward against the forceofgravity (Figure 1). Suchconsiderations suggest that a conceptionofbreakdown as collapsed angular frag mentsisinadequate: itissimply too narrow to character ize the bedrock fragmentsLhatappear in limestone caves and both scientists and explorers habitually identify as breakdown. Thispapertherefore offers a reinterpretationofthe conceptofcavern breakdown.Theinterpretationisbased on studyofFriars HoleCaveSystem, West Virginia, and other eastern caves.Itis supported by observationsofbreakdown incavesin Texas,SouthDakota, Montana, Minnesota, and Mexico. Not suprisingly, altered conceptsofbreakdown force an altered perspective on breakdown classification. Thus, an additional purpose is to offer a new classificationofcave breakdown and collapse features.


Jameson Figure1:Lifted gypsum-crystal wedging fragments. Framboidal pyrite in argillaceous micrites and dolomicrites is altered, leaving hematitic pseudomorphs. Released aqueous sulfur in the formofsulfate reacts with calcitetoform gypsum. The Highway, Rubber Chicken Cave.DefinitionofBreakdownBreakdownis produced by the fragmentation and local displacementofbedrock from cave walls, ceilings or floors. Breakdown is also produced by local fragmentation of pre-existing breakdown. Individual breakdown consists almost exclusivelyofbedrock fragments. Minor amountsofbreakdown consistsofbedrock with speleothem encrus tations or clastic sediments that became cemented before or even after fragmentation. Chert nodules, fossil fragments, quartz silt and sand,orother material weathers from bedrock, but these are best considered weathering detritus (White and White, 1969). Although speleothems collapse,itisnot useful to designate their fragments as breakdown.Incaves with paleokarst fills, or caves intersecting intra stratal karst (palmer and Palmer, 1989), bedrock fragments may appear in brecciasofvarying age and origin. It may then be difficult to distinguish between fragments formed within the present cave and fragments formed in earlier caves, orinintrastratal karst zones preceeding cave devel opment. However, cave breakdown can usefully be charac terized as locally derived, predominantly bedrock fragments that owe their origin to processes operatingincaves.ProcessesofBreakdownA numberofprocesses affect the form, size, and shape of breakdown during its residence time within a cave. Too many studies lose sightofthe fact that the processes oper ate before, during, and after initial fragmentation. Proces ses contributing to the formationofbreakdown fall into at least three groups:(1)dissolution, (2) chemical growth and alteration, and (3) mechanical overloading. The pro cesses have been discussed by Davies (1949, 1951), White and White (1969), Renault (1968), White (1988), Jennings (1971, 1985), Powell (1977), Jagnow (1978, 1979), Bogli (1980), and Jameson (1983, 1985). The following 36 Appalachian Karst Symposium. 1991 paragraphs emphasize processes that operate before or dur ing fragmentation, and identify a few processes that do not appearinthe literature. Dissolutional processes include(1)enlargementoffractures to liberate in situ fragments, (2) enlargement of fractures to weaken bedrock bydiminishing the surface areaofcohesive bonds across fractures, (3) undercuttingofbedrock by free surface streams, (4) mining or coring by drips, films, sprays, or waterfalls,toreduce the areaofattach mentofspeleogens such as flutes, blades, and projections on shaft walls, or to cut through bedrock spans, blades, and ledges, and (5) dissolution and transportofcements or other soluble materialoutofporous beds, to produce weakened, often friable bedrock. Processesofchemical growth and alteration include(1)weatheringofclay minerals by leachingofions anduptakeofwater to produce swelling, mechanical weakeningofbonds, and openingofcracks and partings and (2) crystal wedging by growthofice or salts, including gypsum and halite. Processes directly triggering mechanical overloading include: (1) growthofthe conduittoa critical unsupported size, under the prevailing overburden stresses, residual tectonic stresses, and bedrock characteristics, to cause mechanical rupturing and often collapse, (2) expansion toward a free surfacetoproduce stress-release rupturingandfragmentation as a typeofexfoliation, (3) lossofbuoyant support as phreatic conduits are drained or floodwaters rise and fall, (4) lossofsupport by removalofclastic sedi ments, (5) impactascollapsing bedrock shatters, or pre existing fragments are swept over vertical shafts and then impact, and (6) hydraulic toppling (the breakageofcoher ent bedrock or tippingofin situ fragments, by water flow ing against weakly attached bedrock,oragainst bedrock with a large surface area but a small areaofattachment).Sourceof BedrockFragmentsBedrock fragments in caves may have a wholly inter nal (endogenetic) or external (exogenetic) origin. Endo genetic fragments are produced in caves, or at surface/ subsurface intersections such as cave entrances, as a resultoflocally operating processes.Exogeneticfragments originate outside caves, or originate as fragments derived from solution, chemical alteration, and collapse, in intra stratal karst preceeding cave development. Exogenetic fragments are transported by fluvial processes to deposi tional sites in caves. Subsidence, collapse,orsubglacial processes are also possible transport mechanisms. Com mon exogenetic fragments include surface weathering debris and fragmentsofsurface karren. In alpine caves, glacialorperiglacial debris forms exogenetic fragments. Also included are pinnacle and culler fragments producedatthe soil/bedrock interface or deeper within the subcutane ous zone.Incavesofthe eastern United States, pinnacle and cutter fragments are found near entrances,interminal


Appalachian Karst Symposium. 1991breakdowns,andindebrisassociatedwith shaftsanddolines. Recognitionoffragment provenance as endogeneticorexogenetic, and the identificationoffragments accordingtogenesis, isnotmerelyacademic. It isimportantin the evaluationofbreakdown features and other deposits, such as paleontological and archeological deposits, in which bedrock fragments are included.TransportofBreakdownInsitufragmentationofbedrock produces trapped frag ments that may resist relocationoverlong periodsoftime. Many fragmentsoriginatein situ,but rotale orslide as a resultofforeign crystal growth(orbyfluvial removalofunderlying support) tobecomebreakdown.However,most breakdown has been transported farther; the frag ments are displaced a few meters to tensofmeters from their pointsoforigin,accumulatingaslocaldeposits.Theupper limit on the distance breakdown can move and stillbeidentified asbreakdownispoorlydefined. It depends on size andshapeofthe fragments, processesoftransport, and gradientofthecavepassage.Mostimpor tantly, it depends on theextentto which fragmentsaremodified and incorporated into fluvialorother deposits. Anexamplethattests the limitsofdistinction be tween breakdown and fluvial sediments, and the conceptofa local origin forbreakdown,is the following. In high relicf karsts such as thoseofH uautla, Mexico, where caves 'have long sequencesofvertical shafts separated by short, horizontal passages,bedrockfragments dctach in large number fromshaftwal1s, particularly in thinlybeddedlimitsofhigh dip.Thefragments fall down shafts,orslide downsteepchimneys,oftcnaidedbystreamaction,especially during floods. Manyofthe fragments probably .are detached during the floods.Shaftfloors and ledges exhibit extensive pilesofangular debris,ordissolutionally modified angular debris. Itisnot possible to distinguish the distanceoftransportofthe debris. Muchofit likely is transported down a numberofshafts, oflentodistancesofthreeorfour hundred meters vertically below the locationoforiginal fragmentation and a few hundred meters hori zontally. Even assuming that the debris has fragmented iSeveral timesbyshalleringduringthemovement,and given that fluvial action is involved in transport, it is not clear how to interpret the originofthe sediment. Is it a fluvial sediment? Is it breakdown? I favor the latter inter pretation, and suspectmostcavers exploring these systems would too.CollapseCollapseistherelatively rapiddownslopemove mentoffragmentsunderthe influenceofgravity. Col lapsemovementsinclude falling, sliding, rotating, and bouncing. Collapse can be into open cavitics fil1ed with air.Orcollapse may involve selliinginwater, assuming37Jamesonthat much breakdown forms when passages are first drainedofwalerandbuoyantsupportislost(Whiteand Whjte, 1969).Collapsemostoften resultsasmassive failureofbedrock, in which rock strength is exceeded and initial frag mentation is followed by displacement and additional frag mentation(for example,byshattering as fragments impactagainstoneanother,thusproducingshatteringfragments).However, collapse neednotinvolve fragmenta tion. Collapse canoccurwhen pre-existing fragments are underminedbythe erosional removalofsupport provided by underlying fluvial sediments.BreakdownFeaturesBreakdownfeaturesinclude individual fragments, their accumulations, and the morphologic modifications they impose upon passages. Where breakdown features are distinctiveandcommon, it is useful to describe and name them.Thus,breakdowntypesare defined on the basisof(1)predominantmodesoforigin (exfoliation, crystal wedging fragments)or(2) the characteristics and originofindividual fragments (pendant clusters, spongework debris,bedrockspansfromfloodwatermazes).Breakdownassociationsare defined on the basis of:(I)the charac teristicsofbreakdown accumulations (entrance talus cones, breakout-dome talus), (2) the characteristicsof modified passagemorphology(breakoutdomes,rockstumpsremaining from collapseofexfoliation fragments), (3) the passagecontext(terminal breakdowns, passage-junction breakdown),or(4) some combinationofthe above.IndividualBreakdown:PreviousClassificationsDavies (1949) classified breakdown on a morphogene tic basis.Hedistinguished ceiling and wall blocks, ceiling and wall slabs, scaling plates, and scaling chips. A re statementofthe classification (Davies, 1951) emphasized breakdownascollapse debris for the purposeofmathemati cal analysisofcollapse mechanics, but retained the descrip tive terms block, slab, plate, and chip for individual forms.Whiteand White (1969) modified Davies' classifica tion, explicitly for the descriptionofbreakdown,torecog nize blocks, slabs, and chips, based on the numberofin cluded beds. Division intoblock,slab, and chip forms was presumed to offer a strict morphologic basis for classi fication, it beingjudgedthat a morphogenetic classifica tion was difficulttoapply.Whiteand White (1969) and later White (1988) nonetheless recognized that the classes allowedsomeleeway in actual shape, because(I)frag ments are rarelyend-memberforms, and (2) shapes aresomewhatvariable, particularly for chips.Theyalso rc cognized a scale dependence based upon bed thickness, be cause a given size fragment could be a blockorslab de pendjngonthe thicknessofbeds. Despite these difficul ties, White and White (1969) considered the classification generally useful because it wasjudgedtoallow classifica tion without speculation on origin.


Jameson The White and White (1969) classification has been widely but uncritically used. It adequately distinguishes blocks and slabsifbedsare homogeneous depositional units bounded by prominent, subparallel bedding-plane partings (no cross beds), andifmuch fragmentation fol lows prominent bed partings and bed-perpendicular joints. These conditions are locally common in the eastern United States, particularly in Kentuckyand Tennessee, but eveninthese areas, bedding can be lithologically variable and structures complex. In complex structural settings (with faults, inclined joints, stylolites, fault breccias) more characteristicofthe folded Appalachians,orin casesofbreakdown produced by exfoliation (Jameson, 1983) and gypsum-crystal wedging, the classification falters. Attempts to rigorously apply it leadtoserious difficulties. The problems begin with the failure to operationally define the term bed. A bed can be either (1) a depositional unit or (2) a unit bounded by field-recognizable bedding plane partings. Recognitionofbed partings depends on their prominence, which itself depends on chemical wea theringofthe beds and on the presenceofassociated disso lutional features.Ifdefinition (1) is used, then one has to look carefullyatfragmentstomake the distinction, which may not be easy. Textural variations within beds, surface irregularitiesofthe fragment, weathering rinds, and clay or other clastic-sediment coatings easily complicate the analy sis. Also, it is necessary to define depositional unit for this context, which literally no one has done.Ifdefinition (2) is used, the task is somewhat easier, because surface irregularities and coatings do not so easily hide partings. There still maybedisagreement on parting recognition, particularly for thinly bedded or laminated limestones with easily weathered clay aligned along laminae. Such layers often weather in a distinct fashion (so that they readilyformtight but recognizible partings) but fail to have asso ciated dissolution features, or promote significant fragmen tation. Usage in the literature appears to favor definition (2), which appears most useful, but that usage is hardly explicit. The choiceofdefinition makes considerable difference, and highlights oneofthe five difficultiesofthe block/slab/ chip classification. The first difficultyisthat the classifica tion fails to unambiguously classify fragments. For example, fragments classified blocksbyone definition are chipsbythe other. Figure 2 shows a curved exfoliation sheet.Ifa bed is a depositional unit, then the fragment is a block.Ifa bed is a unit bounded by prominent bed part ings, then the fragment is a chip. Even if one settles on the definitionofa bedasa unit bounded by prominent bed partings, the classification failstousefully classify fragments. The curved exfoliation sheetofFigure 2 is not usefully described as a block or a chip, since there is at least one better description. Figure 1 shows lifted fragments that contain recognizable part ings: Is it useful to classify these small fragments as blocks? Most workers would call these chips, but there38Appalachian Karst Symposium. 1991 Figure 2: Curved exfoliation sheet. The contact in back ground separates overlying pure biosparites from argillace ous micrites that have exfoliation. Lithologic variations within the argillaceous unit suggest that more than one bedispresent, but prominent bed partings are absent. may be yet better descriptors. A third difficulty is that the classification misdescribes actual shapes. The block/slab/chip classification is not really a shape classification, as is apparentfromthe White and White (1969) statementofit. Shapes are often varia ble, and do not closely approach end-member shapes that one might want to ascribe to the terms block and slab. However, if common shapes directly contradict the denota tions or connotationsofthe terms, then it is legitimatetocriticize the classification for misdescribing breakdown. Figure 3 shows breakdown that could be described as slabs and as wedges. The slabs would be classifiedbyWhite and White (1969) as blocks, slabs, or chips; the wedges would be classified as blocksorchips, depending, as usual,onthe choiceofbeddefinition. A fourth difficulty, recognizedinpartbyWhite and White (1969) is the problemofscale. They minimize it, but the problem is serious. Far too many small fragments must be considered blocks. Far too many large fragments, weighing tens to hundredsofthousandsoftons, must be


Appalachian Karst Symposium, 1991 Jameson Figure 4: Pendant breakdown. Near the Otter Slide, Friars Hole Cave.(chip) n<1(block) n=3 /'./ V V ./ (block) n=3(chip) n<1(slab) n=1Slab-shaped fonns Wedge-shaped fonns Figure3:Slabs and wedges. n=numberofbeds. Top: slab-shaped forms have identical shape, but can be blocks(n=3), slabs (n=1), or chips(n<1)with the White and White (1969) classification. Bottom: wedge-shaped forms, common in fault zones, have identical shape, but can be blocks (n=3) or chips (n<1). The geometryofwedges precludes their classification as slabs, by defini tion, unless the left and right ends were the bed partings. tric shapes as blocks, slabs, beams, rods, wedges, flakes, sheets, arrowheads,orthe like. These could then be com pared with breakdown and individual fragments could be more objectively classified. The author,infact, has begun this process for the analysisofexfoliation fragments, and expects therebytogain some insight into the mechanicsofexfoliation jointing on walls and at bendsofpassages. considered chips, because they are in relatively homogeneousbedrock without prominent bed partings. One wonders whether the often cited breakdown blockofCarlsbad Cav erns, Iceberg Rock, estimatedtoweigh over 200,000 tons (Bullington,1968), is in fact a chip rather than a block. A final problem arises for those cases in which spe leogens such as pendants (Figure 4), bedrock blades, verti cal shaft flutes,orpothole walls have collapsed, or frag mentedinplace and have been locally moved. These bed rock features are common, but are simply ignored by pre vious classificationsofbreakdown.IndividualBreakdown:AlternativeClassificationsMorphogeneticClassificationThis study will instead develop a morphogenetic clas sification basedinpart upon the characteristicsoffragment surfaces. In doing so, itisconvenient to retain the loosely descriptive terms block, slab, and chip. The tenns are al lowed only minimal shape connotations and are not defined on the basisofthe numberofincluded beds. The terms are supplemented by other loosely descriptive terms (sheet, flake, blade, wedge, arrowhead-shaped, etc.) as needed. The shape terms are combined with genetic or locational modi fiers that help identify a particular morphogenetic classofbreakdown (fault wedge, exfoliation sheet, cupola span, floodwater-maze span).ClassificationofBreakdownSurfacesPossibleMorphologicClassificationsA descriptive classification needs to less ambiguously classify breakdown, retain essential shape or size data, andbeuseful within the cave.Itis dubious that a single, purely descriptive classification thatisuseful for all break down, today can be constructed: Breakdown develops intoomany ways and exhibits too many poorly studied forms. Careful description and measurementofsmaller breakdownofcertain types (exfoliation, crystal-wedging fragments) may be feasible and leadtorigorous descriptive classifications. Such classifications may employ shape terms based on dimensionsoflong axes, such as that dev eloped by Zingg (1935) for pebbles. Alternately, descrip tive classifications might better define endmember geomeInitially, the surfacesofbreakdown consistoffracture surfaces or dissolutional surfaces. In analogy with the crystal facesofthe mineralogist, itissometimes usefultodescribe the surfaces asfracturefaces ordissolutionalfaces. Over time, fracture faces are dissolutionally modified, and have dissolutional forms such as scallops su perimposed upon them. In many caves, weathering rinds uptoa few centimeters thick also modify breakdown and obscure fragmentational origin. The degreeofmodifica tionisa functionoftime, the lithologyofthe fragment, and the hydrologic setting. In some settings breakdown lasts up to millionsofyears with little modification;inothers, breakdown surfaces may be modified significantly within decades. Collapses that choke passages and pro mote floodwater-maze development favor rapid dissolution al modificationofbreakdown. Argillaceous units with39


Jameson Appalachian Karst Symposium. 1991IncreasingSIzeand ComplexllyBreakdownFormshigh clay contents, such as the illiteand kaolinite-bearing micrites anddolomicritesofthe Pickaway Limestone in Friars Hole Cave System, develop a thick weathering rind that modifies fracture surfaces beyond recognition and makes breakdown extremely slippery. Breakdownisclassified into dissolutional forms, frac ture forms, and mixed forms, according to the predominant surface characterofthe fragment. Dissolutionalformshave surfaces formed primarily by dissolution. Examples include speleogen fragments such as pendants and pendant clusters, spongework spans, and pothole wall fragments.Large Chamber Collapse MajorPassageCollapse Junellon Collapse LocalFeatureCollapseFigure5:Collapse continuum.Isolated CollapseFragmentsLocalFeatureCollapseWhere fragments collapse individually andfailtoformmultiple accumulations, it is appropriate to speakofisolatedcollapsefragments.Isolated collapse fragments need not, but often do, appear as distinctive forms. Local feature collapseisdistinguished for sites with a multipleaccumulationoffragmentsthatform a distinctive, well-characterizable accumulation.ILisalso defined for sites with a modificationofexisting passage morphology brought about by multiple fragmentation.Inmany cases, only fractured walls and ceilings testifytoformer collapse, due to fluvial removalofdebris. Local feature collapse may occur as relatively isolated features, or as separate accumulations scattered over extensive lengthsofpassage, often merging into one another. Small breakout domes and associated debris can appear as local features, such asinMystery Cave, Minnesota andinmany South Wales caves (Davies, 1977). Circular cross sections and associated exfoliation featuresinIndiana caves are local breakdown features (powell, 1977). Ewers (1969) describes local feature rooms produced by lateral enlargement and collapseinKentucky caves.In SituFragmentsclassesofbreakdown features. The classes subsume particular features and breakdown forms (depending on lithology, structure, and dissolutional patterns), manyofwhich are as yet only poorlydefined or' documented. At the lower endofthe spectrum areinsitufrag ments. These are common where closely spaced fissures and partings intersect, forming three-dimensional solids whose surfaces can be solutionally widened sufficiently to loosen bedrock. Such fragments are often trapped on wallsorceilings until undercuttingorlateral enlargement by vadose streams is sufficient to allow their collapse or dis placement by stream transport. In situ fragments are not collapse features per se, but over time they will be moved and become indistinguishable from breakdown proper, or be modified and incorporated into other deposits.Collapse ContinuumIsolaled Collapse Fragmenls InslluFragmenlallonCollapseContinuumBreakdown features are usefully classified within the contextofa collapse continuum (Figure 5), arrangedinrough orderofincreasing size and complexityofindividual fragments, their accumulations, or the modifications they impose upon pre-existing passage morphologies. The continuum includes in situ fragments, isolated collapse fragments, local feature collapse, junction collapse, major passage collapse, and large chamber collapse.Inpart, the collapse continuum can be thoughtofas themodeofoccurrenceofbreakdown features. This interpretation works well for in situ fragments, isolated collapse fragments, and localfeature collapse. The interpretation seems less satisfactory for junction collapse, major passage collapse, and large chamber collapse, becauseofa tendancy to thinkoftheseasbreakdown featuresintheir own right. However, itismore appropriatetoconsider them asMixedformshave surfaces formed by relatively unmodified fracture surfaces in addition to dissolutional surfaces. There is a gradation from fracture formstomixed forms and dissolutional forms; fracture and dissolutional forms are thus endmembersofa continuum. The continuum between dissolutional and fracture forms is further complicated by the fact that, over a sufficiently long timescale,fractureformsshouldbecomedissolutional forms as topographic surface lowering destroys caves.Fractureformshave surfaces formed by fractures. These may be tectonic/sedimentation fractures (bed partings, joints, faults, stylolite surfaces)orfractures formed during cave development. The latter type includes induced tension or shear fractures (new fractures formed by fragmentation across cohesive rock, caused by mechanical overloading), shattering fractures, crystal-wedging frac tures, exfoliation fractures, and fractures associated with chemical weatheringofbedrock, particularlyofclay minerals to promote minor surface flaking.40


Appalachian Karst Symposium. 1991 JamesonJunctionCollapseInmanycaves,junctionsarethe preferred sites for collapse to form rooms.Therooms attain widths several to manytimesgreaterthan meanpassagewidthsawayfrom the rooms.Themechanicsofroom collapse are well known(WhiteandWhite,1969). Detailsofjunctioncollapse featuresarenotas well documented. However, passagejunctionsdoexhibitdistinctivepatternsofcollapse and fragment accumulation. Collapse patterns are a functionoflithology,structure,hydrologicsetting,passage type, and geometry and locationofthe intersection relative to other passage features. An example from Friars HoleCaveSystemisdescribed in a later section.Major Passage Collapseings, N 60-75 Esetjoints, inclined joints, thrust faults, fault-subparallel joints, exfoliation joints, and fractures inducedbygypsum-crystalwedgingorshattering. Poly gonaljointsboundpaleo-mud-crackfragments insomeargillaceous unitsoftheUnionand Pickaway Formations. In Monster Cavern, the CrowsNestRoom, and other large chambers estimated by Worthington (1984)toaccount forabout30%ofthe knownvolumeofthe system,mostofthe fragments are true collapse debris.Thedebris has in duced tension and shear fractures except where breakage isalongbed partings, faults,orshattering fractures.Faultzonescontainanabundanceoffragments.Themostdistinctive are wedge shaped, with faces formed by joints, a fault, and abedparting. Inclined extension joints appear inmanyfault zones,orin crestsofoverlying anticlines.Muchofthe associated debris is slab-shaped, with inclined joint-fracture faces (Figure6).BPMixedForms:Themostcommondistinctive mixed forms are canyon-trench blocks and fault wedges. Figures7and8illustrate a single type for each form. VariationsInclined jointDissolutional Forms:Themostcommonformsarependant clusters. Blades modified from N 60-75 E seten echelonjoint-bounded bedrock are typicaloffloodwater settings. Cross-bed blades are locally abundantinoolitic cross-bedded unitsofthe Union Limestone, butifdissolu tional modification is minimal, the blades are more proper ly classified as fracture forms. Flutes, blades, and irregu larly shaped (but typically sharp) bedrock fragments may be found detached on wallsofshaftsoras debris below shafts, either as debris pilesorisolated fragments littering shallow bedrock pans filled with water. Pothole wallfragments and undercut ledges, often containing chert lenses, are foundinmany canyons in the Pickaway Limestone. Bed parting Collapse along long segmentsofpassage is character isticofpassages that form in zonesofdensely fractured bedrock,orthat underlie valleysorother locations, such as sinkholes, where aggresive waters canbediverted under ground. In many cases, major passage collapse terminates passages, forming terminal breakdowns beneath isolated dolines or,morecommonly, valley walls (Brucker, 1%6). The classificationschemeisselectively illustrated be low for Friars HoleCaveSystem.Thisbranch workcaveexhibits the following morphologies: tubes, canyons, fis sures, shafts, and large collapse chambers.Mostofthe surveyed passage length (63 km) is in canyons. Additional structuraldataand stratigraphic sectionsarein Jameson (1985). Worthington (1984) gives a broad accountofthe developmentofthe system; Jameson (1985) analyzes the growthofthe NorthCanyonofSnedegar's Cave, and des cribes some breakdown in more detail.Breakdown inFriarsHole Cave SystemLargechambercollapsestypicallyoccuratpassagejunctions, in zonesofextensivefracturing, and in loca tions where passageshavedevelopedatseveral levels and one crosses bencath another. Largechambercollapses are also associated withnearproximity to the surface. Fish (1977) describes large-chamber-collapse features forsomeofthe deep Mexican surface pits. Manyofthese are huge phreaticchambersintersected bysurfacelowering. In some, conduits to lower levels apparently provided loci for removalofthe collapse debris, by subsidence and dissolu tional removal where streams were available. Large cham ber collapse has been well documented for anthropogenical Iy-inducedcollapsebydrainingofchambersthrough groundwater mining (Beck, 1984).LargeChamberCollapse Classification BySurfaceFormFracture Forms:Fracture faces are formed bybedpartFigure6:Originofinclinedjointslabs. A fracture fonn is shown, but mixed forms also develop.41


Jameson Appalachian Karst Symposium. 1991 JointB A Cross sectionFault fracture faceoCBed parting TrenchBlockBed paning underneath( after collapse)Figure7:Originoftrench blocks in canyons. The block has surfaces fonned by a bed parting (top); a trench wallanda joint (sides); and an undercut (bottom). replace one or more fragment faces with undercut surfaces, bed partings, N 60-75 E set joints,orinclined joints.ClassificationbyPredominantOriginExfoliation: The exfoliation (Figure 2) occurs as chips, flakes, plates, curved sheets, and wedges, ranging from cm sized to 10 m-sized fragments, with many about 1m long and high and 10-30 cm thick, but tapering toward ends. The fragments have surfaces curved convexly out into the cave, imparting a tendancy toward hyperbolic partsofcross sectionsofcanyons, rounded passage bends, and a varietyofother features described by Jameson (1983). This typeofexfoliation is confinedtoargillaceous unitsand is most abundantincanyonsinthe Pickaway Fonnation.Classificationbythe CollapseContinuumInSituFragments:The most common are gypsum crystal wedging and exfoliation fragments. Others are: cross-bed blades, en echelon joint blades, inclined exten sion joint slabs, fault wedges, fault-subparallel joint frag ments, and pendant clusters.IsolatedCollapseFragments:These are foundinmany canyons and tubes that otherwise have seen littleCanyon Joint solution faceFigure8:Originoffault wedges in canyons. Cross sec tions:(A)Canyon with bed partings and fault. (B) Can yon after lateral undercutting.(C)Collapseoffault wedge. Fault wedge: (D). fragmentational degradation. The fragments litter bedrock floorsorfloors covered with fluvial deposits. Canyon trench blocksorslabs are the most common distinctive forms. Blades, spans, and other spe1eogen breakdowninfloodwater mazes nearly always occur as isolated frag ments. Pendant clusters that have collapsed usually occur as isolated fragments.LocalFeatureCollapse: Multiple accumulationsoffragmentstofonn local feature collapses occurs with near ly allofthe fragment typesjustlistedinthe section on in situ fragments.Inmany cases, particularly with gypsum crystal wedging and exfoliation fTagments, in situ frag ments and collapsed fragments occur together. Thus they can be described collectively as a local feature collapse fonning a single accumulation over a given passage reach. Two other local feature collapse types at Friars Hole were described by Jameson (1979) from Snedegar's Caveasdisintegrating blocks and retreating walls.Disintegrating blocks separate along closely spaced fractures (Fig ure 9).Retreatingwalls are walls with closely-spaced fractures, in situ fragments, nearby collapse fragments, of tenofdisintegrating blocks, and wall rock stumps (Figure 10). Retreating walls have a recessed or re-entrant geometry.Some retreating walls are developedinclay-rich ho mogeneous or laminated micrites and dolomicrites overlain by purer biosparites and oosparites. Thrust faults locally splay up into purer beds from contacts between these units. Jameson (1979) suggested that the fractures were associated with thrust faulting, but later (1985) noted that at many locations the fractured walls are formedinpurer units with or without associated faults. Most fractured walls are at entrances or in near-surface passages within about 500 mofpresent or suspected former entrances.Inthe purer units, major fracture surfaces often parallel existing passage walls as nearly straight planes; this con-42


Appalachian Karst Symposium. 1991 tributes to an unusual tunnel-like cross section in some reachesofSnedegar Trunk. The major fractures also curve around passage bends. The geometryof the major fracturesisunlike the convex-out geometryof the exfoliation fracturesof the argillaceous units. These obser vations nonetheless suggest an unloading typeoforigin by exfoliation. However, the issue is further complicated by abundant minor fractures. These have elliptical traces in sections cut parallel to bedding. Many parting fragmentsfTomthese zones have arrowhead and wedge shapes. Otherscanbe described as sliversorfragments with shapes derivedfromthese forms.Thefragments are vaguely reminiscent of the lenticular massesofbedrock between normal faultsintensional structural regimes.Itis at least plausible that minor fractures and parting fragments represent the re sponseofthese particular lithologiestoan expansional re gimeinbedrock adjacent to the cave. The proximityofthese features to cave entrances, and the presenceofsur ficial periglacial featuresinthis partof the Appalachians, suggests the possibility that their unusual prominence has been aided by frost wedging. Frost wedging is active each winter on nearby surface cliffs andincave entrance rooms (e.g., at Crookshank and Toothpick caves).JunctionCollapse:A distinctive example occursatthe junctionsofcanyonsinthe Union Limestone. Can yons are often tall and narrow, but widen locally by lateral enlargement. Lateral enlargement occurs where clastic sed iments armour the floor,oran argillaceous unit impedes downcutting.Ineither case, solutionismore efficient on lower walls, particularly at passage bends where flow from one wall is directedtothe opposite wall. At junctions, the undermining may be directed at the same wall from oppo site directions,onefrom each passage. Large canyon trench blocks form at these locations. Typical fragmentfacesare(I)an upper bed parting, (2) a lower undercut sur face, and (3) two walls consistingofsolutional forms characteristicofentrenchment (cusps, ledges, and undercut -/--Figure 9: Disintegrating blocks. Nearby walls are heavily fractured. Blocks collapse from walls and disintegrate a long fractures. Fragmentation is today aided by summer condensation corrosion. Trunk passageofSnedegar's Cave.43Jameson Figure 10: Retreating wall. Surface above the recessed wall has quartz crystals set on calcite-crystal fibre slicken sides along a thrust fault. The fault is parallel to bedding at the contact between the lower argillaceous unit (recessed wall with spalls) and the purer unit above. Quartz RoomofSnedegar's Cave. surfaces; see Jameson, 1985).Onthe downstream endsofjunctions, trench blocks also form; they usually have at least one fracture face aligned on N 60-75 E set joints.ConclusionBreakdown consistsoflocally derived, predominantly bedrock fragments. Breakdown canbedescribed and classi fiedina numberofways: Fragment surface (or face): fracture surface, dissolu tional surface Fragment type: dissolutional forms, fracture forms, mixed forms Fragment shape: flake, sheet, chip, block, slab, wedge,beam,arrrowhead-shape, etc. Predominant modeoforigin: exfoliation, gypsum crystal wedging fragment,. etc. Morphogenetic class: trench block, fault wedge, pen dant cluster, etc. By modeofoccurrence in a collapse continuum: In situ fragments, isolated collapse fragments, local feature collapse, junction collapse, major passage collapse, large chamber collapse Breakdown fragments range over at least four ordersofmagnitudeinlength, twelve ordersinvolume, and ten or dersinmass. The large rangeinsize and mass, combined with an abundanceoflarge fragments, effectively limits the kindsofquantitative studies to those that are feasible within the cave with a minimumofdisplacive sampling. Obviously, such standard techniques as sieving are imprac tical. For this reason, among others, the studyofcave breakdown has remained largely qualitative, with the ex ceptionofthe theoretical analysisofcollapse mechanics (Davies, 1951; White and White, 1969).


JamesonThis study isnoexception to the trend; the preceed ing is based on qualitative observations.Toadvance the analysisofbreakdown, it is necessary to undertake more detailed, quantitative analysesofbreakdown form and gene sis. The accurate assessmentofbreakdown characteristics (sizeandshape)maybea difficult task,butfor certain typesofbreakdown such as thatofexfoliation and crystal wedging it may hold the key to a better understandingoffragmentational origin,aswel1as understandingofthe mechanical stress history surrounding cave passages.ReferencesBeck, B.F. (editor), 1984,Sinkholes: Their Geology. En gineering. and Environmental Impact: Proceedingsofthe First Multidisciplinary Conference on Sinkholes. Orlando. Florida.15-17October1984:Rotterdam and A.A. Balkema, Boston,429p. Bagli, A., 1980,Karst HydrologyandPhysical Speleol ogy,Springer-Verlag, Berlin andNewYork, 284 p. Bullington, N.R., 1968, Geologyofthe Carlsbad Caverns, in Delaware Basin Exploration:WestTexas Geological Society, Guidebook 68-55, p 20-23. Brucker, R., 1966, Truncatedcave passages and terminal breakdown in the central Kentucky karst,National Spe leological Society Bulletin,v.28, no. 4,p.171-178. Davies,C.W.,1977,BreakoutdomesinSouthWalescaves,inFord, T.D. (editor),Proceedingsofthe SeventhInternationalCongressofSpeleology.Sheffield. England. September.1977:BritishCaveResearchAssociation, Somerset, England, p. 136-139. Davies, W.E., 1949, Featuresofcavern breakdown:Na tional Speleological Society Bulletin,v.II,p. 34-35. Davies, W.E., 1951, Mechanicsofcavern breakdown:Na tional Speleological Society Bulletin,v.13, p. 36-43. Ewers, R.O., 1969,AModelforthe DevelopmentofSub surface Drainage Routes Along Bedding Planes,M.S. thesis (unpublished), UniversityofCincinnati, 84 p. Fish, J.E., 1977,Karst Hydrogeology and Geomorphologyofthe Sierra de el Abra and the Valles-San Luis PotosiRegion.Mexico,PhD.dissertation(unpublished),McMaster University, Hamilton, Ontario.469p..Ford, D.C., 1988, Characteristicsofdissolutional systems in carbonate rocks, in James, N. and Choquette, P. (edi tors),Paleokarst,Springer-Verlag, New York, p. 25-57. Ford, D.C. andWilliams,P.W.,1989,Karst Geomor phology and Hydrology,Unwin Hyman, London, Bos ton, Sydney, and Wellington, 601 p. Jagnow, D.H., 1978, Geology and speleogenesisofOgle Cave:National Speleological Society Bulletin,v.40, 44Appalachian Karst Symposium. 1991no. 1, p. 7-16. Jagnow, D.H., 1979,Cavern Development in the Guadalu pe Mountains:Cave Research Foundation, 55 p. Jameson, R.A., 1979, Breakdown in fault zones: Snede gar's Cave, Friar'sHoleCaveSystem,WestVirginia (abstract):Programofthe National Speleological So ciety Annual Convention. Pittsfield:National Speleolo gical Society, Huntsville, p. 27.Jameson,R.A., 1983, Exfoliation inFriarsHoleCave system,WestVirginia, in Dougherty,P.H.(editor),Environmental Karst,GeoSpeleo Publications, Cincin nati, p. 119-129. Jameson, R.A., 1985,Structural Segments and the Analy sisofFlow Paths in the North CanyonofSnedegars Cave. Friars Hole Cave System. West Virginia,M.S. thesis (unpublished), West Virginia University, Morgan town, 421 p. Jennings, J.N., 1971,Karst:TheMITPress, Cambridge, MassachusselS,252p.Jennings,J.N.,1985,KarstGeomorphology,Basil Blackwell Ltd, Oxford and New York, 293 p. Palmer, M.V. and Palmer, A.N., 1989, Paleokarstofthe United States, in Bosak, P.; Ford, D.C.; and Horacek,I.,(editors),Paleokarst: A SystematicandRegional Review,Elsevier and Academia, Amsterdam and Prague,p.337-363.. Powell, R., 1977, Lateral unloadingofisotropic rock as a processofsolution channel enlargement,inTolson, J.S. and Doyle, F.L. (editors),Karst Hydrogeology,Inter national AssociationofHydrogeologists Memoir 12, p. 433-441.Renault,P., 1968,Contributional'etudedesactions mechaniquesetsedimentologiques dansIespeleogenese, duxieme partie,Leseffets mechaniques a l'echelledelacavite:AnnalesdeSpeleologie,v.23, p. 259-307.Sweeting,M.M.,1972,KarstLandforms,Macmillan,London,362p. White, E.L. and White, W.B., 1969, Processesofcavern breakdown:National Speleological Society Bulletin,v.22, no. 1, p. 43-53. White, W., 1988,GeomorphologyandHydrologyofKarst Terrains,Oxford University Press,NewYork, 464 p.Worthington,S.R.H., 1984,The PaleodrainageofanAppalachian Fluviokarst: Friars Hole, West Virginia:M.S.thesis (unpublished), McMaster University, 218p.Zingg,T.,1935,BeitraegezurScholteranalyse:Schweizerische MineralogischeundPetrographische Mitteilungen,v.IS, p. 39-140.


AppalachianKarSISymposium. 1991Mud Flow in a Karst SettingLundquisl and VarnedoeCharlesA.Lundquist1and WilliamW.Vamedoe Jr.21 Research Institute Building UniversityofAlabama in Huntsville Huntsville, AL 35899 2Huntsville Grotto, NSS 500 Ketova Way Huntsville, AL 35803ABSTRACTMud banks are presentinmost caves, and therefore are appropriate for investigation as partofa typical karst environment. In 1987, a peculiar mud bankinShine Cave (Alabama No. 210) caught the attentionofthe authors. Rather than drip pits, the mud has linear drip channels, with each channel terminated up-slope under an actively dripping stalactite. The mud is also impressed with prominent down-slope striations caused by irregularities at a roof offset against which the bank is squeezed as it enters the cave room. The sourceofthe mud is Fault Sink which lies above and behind the bank, although no open passage connects the sink on the surface and the cave. The slopeofthe bank is 30to40degrees. Its general appearance is thatofa "mud glacier". The authors reported a qualitative assessmentofthe site at the 1988 Friendsofthe Karst meeting, but quantitative data were lacking then. Therefore,inNovember 1987 and additionally in February 1988 andinFebruary 1989, we placed groupsofmarkers on the slope. A marker consistsofa plumb-bob strung from the cave roof and a vertical wire inserted into the mud directly below the bob. As the mud moves, the bob tip and the wire separate. Measurementsofthese separations on 28 dates now span more than three years. Analysesofthe motions give average ratesofa few millimeters per year. A correla tion with rainfall was particularly dramatic after a record floodinDecember 1990, when the motion was 4 mm between measurements separated by 43 days. With regard to rainfall dependence, the phenomenon resembles surface mud flows.IntroductionShine Cave, (Alabama Cave Survey No. 210) is a short, isolated sectionofan abandoned trunk passage that can be traced through muchofthe four-mile lengthofNewsome Sinks (Jones and Varnedoe, 1980; Varnedoe and Lundquist, 1986). At both ends, Shine Caveisterminatedbycollapsed segmentsofthe trunk (Figure 1). These collapsed sites form Fault Sink to the south and Shine Sinktothe North.Thecave entrance is through Shine Sink. Just northofShine Sink, another intact trunk sectionisfoundinChapel Cave, (Alabama No. 208). The southern endofShine Cave is a typical mud and rubble slope, below and to the northwestofFault Sink. Figure 2 illustrates a profileofthe slope along the lineA-AinFigure1.Mostofthe slope appears chaotic, but one area,10m horizontally and 4m downslopeisa rather uniform mud bank with a slopeof30 to 40 from hori zontal. This is identified as "study area" on Figures 1 and2.Onits upper side, the slope reaches the roof, where the 45 mud seemstohave been squeezed under a vertical roof off set as the mud enters the room. Irregularitiesinthe roof at that point impress striations in the mud surface that continue down slope (Figure 3). Its general appearanceisthatofa mud glacier. This situation was broughttothe attentionofthe authors by Michael Martin during a trip on September5,1987.Further examinationofthe slope found parallel, linear drip channels rather than drip pits, under each active stalactite (Figure 4). At its up-slope end, each channel begins directly underanactive drip. Qualitative evidence said that the channels were caused by the mud moving slowly under the drip (Varnedoe and Lundquist, 1988).QuantitativeMeasurementsThe rateofmud motion down slope was an obvious next questiontoinvestigate. To obtain quantitative data, a seriesofplumb bobs and surface markers were installed: two on November 14, 1987, one on February 27, 1988


Lundquist and Varnedoe Appalachian Karst Symposium. 1991Figure1:Shine Cave lies between Shine Sink and Fault Sink. Coordinates, in feet are relativetoa common origin selected for the cavesinNewsome Sinks. and two on February 25, 1989. Fourofthese consistofa plumb-bob strung from the cave roof such that its tip nearly touches the mud. A stiff vertical wire was inserted into the mud directly below the bob. As the mud moves, the tip and the wire separate (Figure 5). One marker was placed at the very apex where the mud and roof meet. It again has a vertical wire in themud, but has a scratch on the roof insteadofa plumb-bob. Measurementsofmarker separations on 28 dates now span more than three years. Each reading was made indepen dently by each author and the mean re corded. These measurements never disa greed by more than one mm or occasion ally two. Thus the precisionofthe mea surement is judgedtobe plus or minus one mm. The earliest set-ups used screw eyes as plumb bobs, but these are notassatis factory as the bicycle spoke construction. On a few occasions between observa tions, the screw eyes became configured so that the string loop through the eye was notatits top, opposite the stem.Perhapsformationofrust generated enough force to move the string posi tion,orperhaps bats hit the string. In these few instances, the string and eye were readjusted to the proper configura tion before making measurements. On one occasion, twoofthe mud markers seemed to have been slightly disturbed between visits. These were reinitialized and measurements for that date discarded. segment is sharpened to form the bob tip. Another spoke section is the vertical wire in the mud.Theupper endofthe string must be attached to the roofina way that excludes motion.800 900 1000FEET700 500 600 2000Tiltingofthe wireinthe mud was another concern.Itwas unclear initially whether a short wire segmentinthe sur faceofthe mud was desirable or whether a longer segment should be pushed into the mud to observe deeper bulk flow. The shallow wire could be more suscep tible to surfaces changes from seasonal drying and wetting.Thedeeper wire could tilt from differential flow rates with depth. At the position in Figure 5, a shallow wire (few cm) and a deeper wire (more than 10 cm) were emplaced side by side.Thebobwas over the shorter wire. No appreciable differential tilt was observed in this case.Ingeneral, while minor tilts did occur, their effects were sig nificantly smaller than the observed displacements. Thus tilt was a systematic effect influencing accuracy, but not seriously so.FAULT SINKN I T 1900 o \\ \\\", \ \--v, \ ...:\IJ., \ .. .:: 1800\ I1700SHINE SINK1600The useofrust resistant metal was found desirable.Aninexpensive set-up uses sectionsofbicycle spokes. The threaded end with its tightening head make an effective centered attachment for the string.Thestring is fed through the central holeofthe head and knotted so that it will not pull out. A segmentofspoke with the threaded endisscrewed into the head. The other endofthe spoke Whereas individual measurementshave the uncertain ties described (plusorminus a mmortwo), these are small relativetothe long-term motion. Also, consistency among the resultsofthe five markers supports the relia bility of the data.46


Appalachian Karst Symposium. 1991 Lundquist and VarnedoeResultsendofthe period, the channels had indeed extended beyond the lines byanamountapproximatelyconsistentwith the observed marker motions Figure 6 is a graph for each markerofthe displacements versus Julian date. Table 1 shows the mean rate for each marker over the interval it wasinplace. These data show that a flow rateofa few mm per yearisa consistent characteristicofthe top (outer) several cmofthe bank. This confirms the qualitative explanation offered above for the appearanceofthe bank and its drip channels.Oneofthe three years spanned by the measurements, 1988, was slightly drier than normaL The National Weather ServiceatHuntsville-Decatur Airport recorded 46.66 inchesofprecipitationrelativeto the normal 54.72. TheAirport is approximately15miles from Shine Cave. Years 1989 and 1990 werewetterthan normal, with 73.58 and 72.26 inches respectively (NOAA).750 800 700 650 FEET (MSL)FAULTSINK FAULT !r,-----'-SHINE CAVE ..//'.,'-:.:-0'" "' ::::::;: Ii',I:II -,.;..:=(--'\ ,'le----. .'AREAOFSTUDYFigure2:The areaofstudy is on a mud and rubble slope fed by Fault Sink. This profile follows the line A-AinFigure1.Altitude is in feet above mean sea level. Figure3:The study area is shown looking soulh from a point on the profileinFigure2.Asa less accurate confirmation, lines were scribed inthemud tangent to the up-slope endofa few drip channels. This was done midwayinthe measurement period. By the1850Aninteresting resultisillustratedinTable 2 where first the mean rate per year iscomparedwith thedisplacementbetween Julian dates 2,448, 218 and2,448,261(1990 Nov. 23 and 1991 Jan 5). During this interval, from December21through 23, norlh Alabama exper ienced a very heavy rain; Huntsville reported 12.0 inches. In Huntsville lhe monlh ofDecember had 18.68 inchesofprecipitation, which was a record. The previous maximum for a December was 11.74 inches and the previous maxi mum for any monlh was 17.00 inches. Flooding was ex tensive, including lhe Newsome Sinks caves. Debris from high water indicated that Shine Cave was flooded at least to lhe base elevationof lhe study area(seeFigure 2), and water may even have covered someof lhe measuremem sites. Thisisvery much higher lhan at any previous time since lhe measurement began.Thelast columnofTable 2 gives the mean rates if lhe final measurement imerval lhat spans lhe flood is not included. Thus itisevident lhat anepisodeofheavy rain can cause as much displacementasa yearofmore usual conditions. On the other hand, even an extreme flood did not produce catastrophic slidingofthe mud bank. Dependenceofmud-slope motion on water conditions is to be expected, based on conventional lheories for such circumstances. In lhese theories, soil movementona slope depends on lhe balance between lhe shear strengthof lhe slope materials and lhe down-slope componentof lhe 1800 1750FEET170047


Lundquist and Varnedoe Appalachian Karst Symposium, 1991Figure 4: Drip channels such as these in the mud slope correspond to active stalactites on the roof. gravitational force duetothe weightofthe material above a potential slope surface (Spangler and Handy, 1982). The shear strengthisa linear function involving material para meters such as effective cohesion, c', thesocalled effective-friction angleofthe material, 0' and the difference(puw), where p is the total stress normal to a potential slip surface andUwisthe pore-water pressure: s=c'+(p uw) tan 0'. Figure5:This hanging bob and short wireinthe mudaremarkerNo.4.The longer wire goes deeper into themudto detect any tilt due to differential movement with depth. That pore-water pressure is indeed presentintheShine Cave slope during wet seasons is indicated by another observed phenomenon. Early in the study, while listening and looking for other drip channels, drip-like noises were detected near the very baseofthe entire slope, well below the study area. No roof drips could be found. Instead sites on the slope were discovered from which small jets of water rose periodically, making a noise resembling a drip. Pore-water pressure seems to be the likely causeofthis phenomenon. Knowing no accepted name for these fea tures, the authors have adopted the namepird(drip back wards). Pird activity was observed regularly during moder ately wet conditions, but not during dry periods orinex tremely wet times when small streams replaced the pirds. Another aspectofTable 1 may support the simple theoretical picture sketched above. Markers No.I,2,and3 are nearly in a line, down-slope on the northward end of the study area. MarkersNo.4and 5arenear eachotheratthe bottomofthe study area toward its southern end. The slopeofthe bank increases slightly from north to south, Marker Interval, Lapsed Time Displacement Mean Rate No. Julian dates(days)(mm) mm/yr(lastfour digits) I7113 to 8261 1148 10.5 3.5 2 7113 to 8261 1148 14.5 4.6 3 7218 to 8261 1043 15.0 5.3 4 7582 to 8261 679 14.0 7.5 5 7582 to 8261 679 13.0 7.0 '-'--Total 4597 67.0 5.3 When heavy precipitation falls in the collec tion areaofFault S ink, water drains into the poresofthe mud bank below the sink, thus initiating motion. This same scenario and theory is the ba sis for real-time surface landslide warnings during heavy rainfall in California (Keefer and others, 1987). Here, the important featureofthis equationisthe negative sign on the pore-water pressure. As this pressure increases, the shear strength at points TableI:Mean Ratesinthe bank decreases. When the shear strength becomes smaller than the down-slope component of the weightofthe material above a slip surface, down-slope movement occurs.48


Appalachian Karst Symposium. 1991 Lundquist and Varnedoe1988 1989 1990MARKER5128 412MARKER48 I. 412MARKER38I.4 12MARKER2 8 4128MARKER14 mm7200 7400 7600 7800 8000 8200Figure6:The movementofthe five markers is showninmmrelative to the last four digitsofJulian Date (2, 44X,XXX).49


Lundquist and VarnedoeTable 2: Flood Conditions Mean Rate Displacement Mean Rate Markermm/ to 8261 (Excluding last (Total Data) (mm) (43 days) 43 day interval) 1 3.5 3 2.5 24.65 3.1 3 5.3 44.04 7.5 45.757.04 5.3 Mean from approximately 30 to 40. Thus markers 3, 4 and 5 are along the bottomofthe study area with the first near its northendand the latter two nearer the south. Other things equal, theory wouldpredictthatconditions for motion would occur sooner for the steeper slope condition at the south. The mean rates are higher for markers4and5than for marker3,which is consistent with that concept. However, the difference isnotgreat, and maynotbestatistically significant. Marker1isatthe apex where the slope reaches the roof. Its motion couldberestrained by the surface friction between the bank and the roof. This effect isnotin the theoretical model.Themeasured rate is indeed lower, but again the differenceisnot great and maynotbesignificant.ConclusionIn the aspects discussed above, movementsofthe mud bank studied inShineCavecorrespondwell with the 50Appalachian Karst Symposium. 1991conventional theoretical model accepted for' slopes on the surface. While the specific conditions in the study area may not reoccur often, the general situation illustratedinfigures 1 and 2 is quite commoninAppalachian karsts and elsewhere. Thus the results from Shine Cave may provide a guide to phenomenainsimilar situations.ReferencesJones, W.B. and Varnedoe, W.W., 1980, CavesofMorgan County, Alabama:Geological SurveyofAlabama. Bulletin112,205p Keefer, D.K and others, 1987, Real-time landslide warning during heavy rainfall,Science,238, pp. 921-925,13Nov.NOAA,NationalClimateDataCenter,LocalClimatologicalData.AnnualSummarywith Comparative Data. Huntsville. Alabama. Ashville, North Carolina.1987. 1988.1989.1990.Spangler, M.G. and Handy, R.L., 1982,Soil Engineering:Harper and Row, Publishers, New York.Varnedoe,W.W.,Jr.andLundquist,C.A.,1986, Speleogenesis in Newsome Sinks, Alabama, USA:9th Congreso Internacional de Espeleologia. Barcelona, Spain,v.1 pp. 258-262. Varnedoe, W.W., Jr. and Lundquist, C.A., 1988, Plastic flowofcavemud(abstract),inMylroie,I.E.(coordinator),Program. 10th FriendsofKarst Meeting, San Slavador Island. Feb.11-15: Mississippi State University, p. 25-26.


Appalachian Karst Symposium. 1991Mud Pot: ANewThennal WaterCavein Alleghany County, VirginiaKeithE.GogginInstitute for Geographical and Geological Sciences George Mason University Fairfax, Virginia 22030ABSTRACTAs a partofthe Warm River Cave Survey a new cave was discovered in Alleghany County, Virginia. The cave was mapped in February 1990, and was found to contain a thennal stream. This cave and nearby Warm River Cave are the only such thennal caves knowninthe eastern United States. Both caves are developedinlimestonesofMiddle Ordovician age and are located near the southern endofthe Warm Springs anticline. Falling Spring, from which the waters seeninWarm River Cave are known to surface, acts as a regional resurgence for mixed watersofboth deep-circulating (hot) and shallow-karst (cold) origin (Hobba and others, 1979). Many proposals have been advanced as to the originofthe thennal waters, and these are summarized below. In early 1991 dye-tracer tests were conducted in order to establish any hydrologic relationships among Mud Pot and nearby Warm River Cave and Falling Spring. Itisconcluded that the waters seen in Mud Pot resurge at Falling Spring, but do not pass through Warm River Cave.GogginIntroductionThe Warm River Cave Survey was initiated in March 1989 byDickGrahamandKeithGogginwith the intentionsofproducing a high quality mapofWarm River Cave, as well as to explore and survey other cavesinthe vicinity. This resulted in the locationofseveral new cavesandthe rediscoveryofone or two previously reported ones. IL was during this exploration that a new cave containing thermal waters was found. This new cave and nearbyWarmRiver Cave are the only known caves in the eastern United States containing thermal waters. The presenceofhot water, combined with the extremely muddy natureofLhecave's lower levels, resultedinthe name Mud Pot. The water temperature in MudPoton 3 February1990was 28.5 C. Same day readings show this tobe2C warmer than the water temperatureinthe "hot" streamofWarmRiver Cave. Mud Pot's proximity to Warm River Cave and the similarityinwater temperatures suggested a hydrologic connection, and a nuorescein dye trace was con ductedinorder to establish any relationship between thetwocaves as well as with ncarby Falling Spring, a large karst spring which acts as a regional resurgence for waters of both deep-circulating (hot) and shallow-karstic (cold) origin.51The entrancetoMud Pot is a 23-meter vertical shaft, locatedina small, brush-filled sinkhole approximately 550 m northofWarm River Cave. This shaft leadstoa short segmentofcanyon passage in which the thennal streamisfound. This horizontal passage terminates both upstream and downstreaminsiphons, with two separate infeeders near the upstream tenninus. Figures I and 2 are plan and profile mapsofMud Pot.LocationandGeologic SettingThe karst features mentionedinthis paper are located in the southern endofthe Warm Springs ValleyinAlle ghany County, Virginia. Figure 3 is a mapofthe area, showing locationsofsome caves and karst features found in the area, as well as the implied now palhsofthe walers seeninMud PotandWarm River Cave, en route to Falling Spring. To dale, little detailed geologic mapping has been done in the southern endofthe Warm Springs anlicline. The rocks exposed here range from Lower OrdovicianlODevonian in age. The karst fealures describedinlhis report are developed in limestonesofMiddle Ordovician age. Rader (1984) summarizes the stratigraphyofthe wes tern anticlines containing thennal springs, and Bick (1962,


Goggin Appalachian Karst Symposium, 1991MudPotAlleghany County, Virginia.3752'19"N.Lat.7955'49"W. Long. Suunto 8 Tap.Survey by:K.GogginD.GrahamK.RosenfeldR.Simmons 3February,1990.SheetI:Plan View Symbols areN.S,S.Standard. CrossSectionsaretoScale.1991Keith Goggin.Figure1:Plan viewofMud Pot.oNmfeet20 /p.13) includes a stratigraphic sectionofthe Big Valley Formation (a Middle Ordovician limestone) from about1.5kmnortheastofthe entrancetoMud Pot. Structural features observed both on the surface and within the caves in the area include joint "swarms" and many nearly vertical, normal faultsofsmall displacement. The entrance sectionofWarm River Cave is developed along oneofthese normal faults, whereas muchofthe 'hot stream' sectionofthe cave exhibits joint-controlled devel opment along strike in steeply dipping limestones. Mud Pot appears tobedeveloped along a vertical joint,innearly horizontal limestone. A slight synclinal flexureisvisible inside the cave. Hobba and others (1979) observed the coincidenceof52 nearly all the thermal springs found in western Virginia with lineaments or intersectionsoflineaments visibleonLANDSAT imagery and high-altitude photography. These lineaments were interpreted to be surfaceexpressions of extensive fracture zones. Figure 4 is a generalized geologic mapofthe Warm Springs anticline showing the above mentioned lineaments. A numberofpapers discuss the geochemical nature of thermal waters emerging from Falling Spring and the geologyofthe massive travertine deposits located a short distance downstream from the spring. Primary references include Hobba and others (1979), Dennen and Diecchio (1984), Herman and Lorah (1986, 1987), Lorah (1987), Lorah and Herman (1990), and Dennen and others (1990).


Appalachian Karst Symposium. 1991 Goggino204060MudPotAlleghany County,VirginiaSheet2:Profile(90View)1991KeithGoggin.8086Figure2:ProfileofMud Pot; view is from the east.OriginofThermal Watersinthe Warm Springs AnticlineThe originofthe thermal waters has been a subjectofdebate for almost 150 years.Thefollowing is a summary,inchronological order,ofthe major proposals concerning lhe originofthermal waters in the Warm Springs anti cline.Oneofthe first reports dealing with this problem was thatofRogers (1843),whobelieved the waters tobeofnormal meteoric origin basedontheir chemical characteris tics.Rogersalsoobservedtheapparentconnectionbetween thermal springs and anticlinal axes and faults. Reeves (1932) expanded on Rogers' hypothesis, agreeing that the thermal waters wereofmeteoric origin, and sug gesting that thewatersenterpermeable strata at highalLitudeexposures on the crest or limbofone anticline and nowinan artesian mannertoa similar anticlinal exposure 53 at a lower elevation where the waters res urge as thermal springs. Dennison and Johnson (1971) note that the structural settingofanticlines in the thermal springareaofVirginia (centered in BathCounty)isnotsignificantly different from elsewhere in the Valley and Ridge province extending from New York to Alabama. They also note the presenceofmiddle Tertiary igneous intrusionsjustto the northofthe thermal springs areaofVirginia. They suggest that the heat source for the thermal springs may be residual heat in a deep, solidified pluton. Helz and Sinex (1974) furthered Reeves' hypothesis byplouingelevationsofthermalspringsinthe Warm Springs Valley, findingallLObebelow 700m.They sug gest that the recharge area for the thermal watersisproba bly the Browns Mountain anticline located to the westinGreenbrier and Pocahontas counties,WestVirginia. They


(CI.:40ft.)GogginoFEET1000Appalachian Karst Symposium. 1991Figure 3: Mapofthe study area showing locationsofsomeofthe caves and karst features found there,as well as implied, straight-line, flow paths determined by dye tracing. add that "altitudes in the Browns Mountain area reach 1100 m, providing a hydrostatic headof2()()-400 m to drive circulation toward the thermal springs.Waterfrom Browns Mountain would have to travel underground a distanceofa-bout 25 km to reach the Warm Springs Valley". Helz and Sinex (1974) provide geochemical data that show that the thermal watersdonotrequire a magmatic heat source as suggested by Dennison and Johnson (1971),butthisdatadoes not ruleoutthe possibilityofresidual plutonic heat. Perry and others (1979) calculated the geothermal gradientinthe Warm Springs anticline using down-hole54heat flow measurements and found this to be similar to the gradient in other areasofVirginia.Owingto this similarity, along with structural analysisofthe anticline, they postulate that "meteoric water enters the Silurian quartzites (Tuscarora Formation) and possibly adjacent carbonate units along steep to vertical bedding planesonthe northwest limbofthe (Warm Springs) anticline, reaches depths sufficienttoheat the water, and then rises rapidly along essentially vertical east-west fracture zones which intersect the bedding plane permeabilityatdepth". Perry and others conclude that their findings do not support Dennison and Johnson's (1971) proposalofa residual magmatic heat source.


Appalachian Karst Symposium. 1991 GogginFigure 4:Generalizedgeologic mapoftheWarmSprings anticline (adapted from Dennen andDiecchio, 1984), The study area is located close to the inter sectionofthree lineaments visible near the southernterminusofthe Ordovician limestone exposure.oI,!BO" ""\" I, ,),,/Clifton Forge5 ml L..-...J....--l..._L...--L.----I' 5 kmIDevonian shale Cayugan-HeIderberg-Oriskany [::::}::::}:::::I Tuscarora-Clinton Martlnaburg-Oswego-JunlataI... /-Yi2'1 Ordovicianlimestone ,. LineamentsfromLANDSATImageryDye-Tracing MethodsTwodye traces were conducted to as certain the subsurface flow pathofwater seen inMudPot. Receptors containing activated charcoal were constructed from fine nylon screening in themannersug gested by AJey and Fletcher (1976). In order to enable the receptors to hang free ly in the water and ensure maximum dye recovery, modified "gumdrops" were con structed in the field using medium-gauge metal wire wrapped around suitable rocks found near each location. Before dye was injected into the groundwater, a receptor was placedatFalling Spring. This con trol receptor was used to establish backgroundlevelsofany pre-existing sub stances(i.e.stock waste) in the ground water that could interfere with visual ob servations used to determine test results. In both tests,dyereceptors wereplacedatstrategic locations inWarmRiverCave andatFalling Spring. Num bered receptor locations in Warm River Cave are shown in Figure 5 and described in TableI.It should be noted that no re ceptors were placedinthe extreme down stream reachesofWarm River Cave, but there existnosignificant infeeders, ther malorotherwise, in this sectionofthe cave.Oncethe receptorswerein place, approximately 0.9kgoffluorescein dye(AcidYellow73)wasplacedin the stream flowing throughMudPot. A 5% solutionofpotassium hydroxide (KOH) in 70% isopropyl alcohol was used as an elutriant for all tests. Receptor#Location Positive Negative1Mouthof"cold-stream"X2 Baseofclimb-down to water, mixed streamX3 Thermal in feeder near baseofrimstone damsX4 Thermal infecder on west sideofpassageX5"Hot" stream, in pool at mud slide roomX6 Final duck-under, 30 m from terminusX7 Mouthof"hot" streamX8 Fallin!! Sorinrr XTable1:Descnptlonofreceptor locations shown 10 Figure 5 and results (cumulative data from both tests). 55First Trace ResultsThefirst trace was executed on 5 January, 1991. Because water le vels inWarmRiverCavewere very high, receptors were placed only at locations 1, 2, and 7, as well as atFallingSpring(location 8). Afteroneweek,receptorsatthe springwererecovered,butthose inside Warm RiverCavewere inaccessibleowingtosevereflood conditions.Therecovered receptors were elutri ated in theKOHsolution and were found to be "very strongly positive"usingthecriteriaby AleyandFletcher (1976). This confirmed the


GogginColdStreami__ometers30120mtoFallingSpringAppalachian Karst Symposium, 1991Warm River CaveAlleghanyCo.,Va.Figure5:MapofWarm River Cave (modified from Lucas, 1971). The circled numbers indicate locationsofdye receptorsasdescribedinTable1.hydrologic connection between MudPotand Falling Spring,butthe flow routeofthe water with respect to Warm River Cave could not be determined. Given the extremely high water levels at the timeofthe recovery and the possibility that back flooding in Warm River Cave could contaminate the receptors located there, it was decidedtorepeat the trace.SecondTraceResultsandConclusionsThe second trace was conducted on 24 February, 1991. Water levels, while still high, were low enough to allow receptors to be placedatall locations including Falling Spring.Atthis time receptors remaininginthe cave from the first test were collected. Afteroneweek, all receptors were successfully retrieved and e1utriated. Once again the spring receptors tested strongly positive, but all the receptors from Warm River Cave tested negative, suggesting that the water from Mud Pot resurges at Falling Spring after following a flow path totally independent from known passages in Warm River Cave.Toremove the very slight possibility that56the spring receptors tested positive duetoresidual dye from the first trace and that the second dye pulse had not yet reached the spring, the in-cavereceptors from the first trace were also elutriated. These receptors were foundtobe neg ative as well. This confirms that the water seen in MudPotpasses through no known passages in Warm River Cave en route to Falling Spring.AcknowledgmentsThe author would like to express his thankstothe fol lowing people for their assistance and support during this project. Eric Lafferty, Dick Graham, Scott Blaha, Mark Richardson, Melissa Weakly, and Susan Zywokarte helped with the first dye trace, and Greg Duncan, Dave Hubbard and Butler Stringfield helped with the second. Rick Diecchio, Jeff Harrison and Steve Kline reviewed the man uscript and provided many useful comments. Special thanks are directedtoGustav Asboth and Ted dy Dressler, the landowners on whose land the caves and Falling Spring are located. Without their kind permission noneofthis work would have been possible.


Appalachian Karst Symposium. 1991ReferencesAley, T.andFletcher,M.W.,1976,WaterTracer'sCookbook:Missouri Speleology,v.16, no. 3, 32 p. Bick, K.F., 1962, Geologyofthe Williamsville Quadran gle, Virginia:Virginia DivisionofMineral Resources ReportofInvestigations.,no.2,40p.Dennen, K.O.andDiecchio,R.I.,1984,TheFalling Spring, Alleghany County, Virginia, in Rader, E.K. and Gathright, T.M., II (editors), Stratigraphy and Structureinthe Thermal Springs Areaofthe Western Anticlines:Virginia DivisionofMineral Resources Sixteenth An nual Virginia Geologic Field Conference,p.17-22. Dennen, K.O.; Diecchio, R.I.; and Stephenson, M.A., 1990, The geologyofthe Falling Spring travertine de posit, Alleghany County, Virginia, in Herman, J.S. and Hubbard, D.A.,Jr.(editors), Travertine-marl: Stream depositsinVirginia:Virginia DivisionofMineral Re sources Publication 101,p.79-91. Dennison,J.M.andJohnson,R.W.,1971,Tertiaryintrusions and associated phenomena near the Thirty Eighth Parallel Fracture ZoneinVirginia and West Vir ginia:Geological SocietyofAmerica Bulletin,v.82,p.501-508. Helz, G.R.and Sinex, S.A., 1974, Chemical equilibria in the thermal spring watersofVirginia:Geochemica et Cosmochimica Acta,v.38,p.1807-1820. Herman, 1.S. andLorah,M.M.,1986,GroundwatergeochemistryinWarm River Cave, Virginia:National Speleological Society Bulletin,v.48,p.54-61. Herman, J.S. and Lorah, M.M., 1987,C02Outgassing and calcite precipitationinFalling Spring Creek, Vir ginia, U.S.A.:Chemical Geology,v. 62, p. 251-262. 57GogginHobba, W.A., Jr.; Fisher, D.W.; Pearson,FJ.,Jr.; and Chemerys, J.C., 1979, Hydrology and geochemistryofthermal springsofthe Appalachians:U.S. Geological Survey Professional Paper 1044-E,36 p. Lorah, M.M., 1987,The Chemical Evolutionofa Tra vertine-Depositing Stream,M.S. thesis (unpublished): Charlottesville, UniversityofVirginia, 175p.Lorah, M.M. and Herman, J.S., 1990, Geochemical evo lution and calcite precipitation rates in Falling Spring Creek, Virginia, in Herman, J.S. and Hubbard, D.A.,Jr.(editors), Travertine-marl: Stream depositsinVirginia:Virginia DivisionofMineral Resources Publication 101,p.17-32. Lucas, P.C., 1971, MapofWarm River Cave, in Virginia Regionofthe National Speleological Society, 1971, National Speleological Society Convention Guidebook (Section 1):The Region Record,v.1, no. 4, p. 79. Perry, L.D.; Costain, J.K.; and Geiser, P.A., 1979, Heat flow in western Virginia and a model for the originofthermal springsinthe folded Appalachians:JournalofGeophysical Research,v.84,p.6875-6883. Rader, G.K., 1984, Stratigraphyofthe western anticlines A summary, in Rader, G.K. and Gathright, T.M., II (editors), Stratigraphy and Structure in the Thermal SpringsAreaofthe Western Anticlines:Virginia Divi sionofMineral Resources Sixteenth Annual Virginia Geologic Field Conference,p.1-12. Reeves,F.,1932, Thermal springsofVirginia:Virginia Geological Survey Bulletin36, 56p.Rogers,W.B., 1843,Onthe connectionofthermal springsinVirginia with anticlinal axes and faults:AssociationofAmerican Geologists and Naturalists Reportofthe1st, 2nd,and3rdMeeting, 1840-1842,p.323-347.


Plate Appalachian Karst Symposium. 1991PlateA:Travertine-marl deposit at entrance to Wildwood Park, Norwood Street, CityofRadford, Virginia. This deposit has formed on the wallofan artificial cut in the hillside where groundwater now seeps from bedding partingsinthe Elbrook limestone. Note the extensive accumulationofalgal material and the steady dripping.This site was a stop on the geologic fieldtripofthe Appalachian Karst Symposium. For a discussionoftravertine-marl depositsseeHubbard and Herman, this volume,p.59.Photograph by KarenM.Kastning.58


Appalachian Karst Symposium.1991Hubbard and HermanTravertine-Marl: The "Doughnut-Hole"ofKarstDavidA.Hubbard, Jr.1and Janet S. Hennan21 Virginia DivisionofMineral Resources, P.O. Box 3667, Charlottesville, VA 22903 2DepartmentofEnvironmental Sciences Clark Hall, UniversityofVirginia Charlottesville, VA 22903ABSTRACTKarst is a negative relief topography formed by the dissolutionofcarbonate bedrock. Emergent karstic water, as springsordiffuse stream-bank or -bed seeps, has deposited Quaternary age travertine-marl buildups. Mostofthe travertine-marl deposits in Virginia are associated with known faults in folded and fractured carbonate rock. Someofthese accretionary or positive karst features morphologically resemble spelean formations typicalofdepositional vadose cave environments. The ubiquitous presenceofalgae and moss in travertine-marl differentiate it from its spelean counterparts, but whether the roleofthe biota is only as a passive frameworkoris also as an active metabolic influence remains to be determined. Locally valued for their aesthetic waterfalls and as a sourceofagricultural lime, travertine-marl deposits are an environmental barometer. Land-use practices resulting in increased runoffofprecipitation and increased erosion have degraded these features: travertine-marl deposits have been smothered and partially buried by erosional debris; features have been destroyed by the corrasive actionoflarge magnitude floods carrying large volumesofabrasive sediment. Polluted karstic water may inhibit calcite nucleation and the growthofframework flora, bothofwhich influence the rateoftravertine-marl deposition.IntroductionTravertine-marl depositsofthe Valley and Ridge physiographic province are composedofcalcium carbonate precipitated from streams and springs. These accretionary fresh-water carbonate deposits are a commonfeatUI:ein karst areasofwestern Virginia (Figure1;Sweet and Hubbard, 1990). Although karst is characterized as a negative relief topography formed by the dissolutionofcarbonate bedrock, chemical conditions may permit calcium carbonate to precipitate from karstic water. Travertine deposits in caves are well known to cave visitors and karst researchers. The term travertine-marl isusedhere to denote complex depositsofcalcium carbonate, precipitated by surface streams and springs, comprisedoftravertine, tufa, and marl components. Some featuresoftravertine-marl deposits morphologically resemble their spelean counterparts (Love and Chafetz, 1990). In surface streams,travertine-marldepositscommonlyform impressive waterfalls. The buildupsoftravertine and tufa which create these waterfalls allow marl to accumulate upstream as the buildups simultaneously aggrade. InCopyright1991,CommonwealthofVirginia59addition to their aesthetic waterfalls, these freshwater carbonate deposits have been utilized as a sourceofagricultural lime (Sweet and Hubbard, 1990) and for building stone (Austin and Barker, 1990).Depositional ProcessesKarstic groundwater, enriched in carbon dioxide, aggressively dissolves carbonate minerals until it reaches equilibrium with respecttothese soluble minerals. Upon emergenceatdiscrete and diffuse springs, karstic water often becomes supersaturated with respect to carbon dioxide as the gas exsolves from the stream water. Outgassingofcarbon dioxide in response to the low atmospheric levelofcarbon dioxide drives the watertohigh degreesofsupersaturation with respect to the carbonate minerals, especially calcite. Hydrological agitation and the dispersionofwater over stream-bottom obstructions influence the rateofcarbon dioxide exchange. Travertine buildups contribute to higher levelsofagitation and dispersion and provide a sourceofpositive feedbacktothe outgassing process and subsequent precipitationoftravertine.


Hubbard and HermanDISTRIBUTION OFDEPOSITSTRAVERTINE-MARLDEPOSIT J\ COUNTYWITHABANDONEDWORKINGSAppalachian Karst Symposium.1991Figure1:Travertine-marl deposits found in Alleghany, Augusta, Bath, Botetourt, Clarke, Cra.ig, Giles, Montgomery, Page, Roanoke, Rockbridge, Rockingham, Shenandoah, Smith, Warren, and Washmgton counlles (ModIfIed from Hubbard and others, 1985). In Virginia, calcite precipitationiskinetically inhibi ted until stream waterisfive to six times supersaturatedinFalling Spring Run (Hoffer-French and Herman, 1990) and five to nine times supersaturated in Falling Spring Creek (Lorah and Herman, 1990). Factors which influence cal cite precipitation rates include the presenceoforganic materialorstrongly adsorbing ions. Algaeand mosses seasonally increase the surface area available for nucleation and precipitationofcalcite. Whether biological processes actively drive the depositionoftravertine-marl has not been resolved. A 24-hour studyofFalling Spring Creek failed to detect any evidenceofphotosynthetic activity affecting the partial pressureofC02orthe saturation indexofcalcite (Lorah and Herman, 1988). The physical proces-sesofhydrological agitation andC02outgassing in re-sponse to a chemical gradient predominated over the bio logical effects onC02uptake by stream biotainanother diurnal and seasonal studyofa travertine-depositing streaminVirginia (Hoffer-French and Herman, 1989). Studies at Plitvice National Park, Yugoslavia indicate that algal mu cous excretions trap particlesofcalcite and serve to initiate precipitationoftravertine (Emeis and others, 1987). Re searchers, such as Emeis and others (1987) and Pentecost (1990) suggest that the metabolic involvementofalgae may have a more significant roleinthe depositionoftra vertine. Chemical processes also influence ratesoftraver tine formation.Forexample, calcite precipitation is inhibited by concentrationsofstrongly adsorbed ions that block nucleation sites (Morse, 1983). In this complex natural system influenced by chemical, hydrological, and biological processes, the rateoftravertine deposition must60exceed the rateoftravertine erosion for these depositstoform and exist over time.Travertine-Marl DepositsTravertine, tufa, and marl have been depositedbystream and spring waterinat least18ofthe 26 karst-bear ing counties in the Valley and Ridge physiographic pro vinceofVirginia (Figure1;Sweet and Hubbard, 1990).Ofthe approximately 70 deposits, most are associatedwithknown faults in folded and fractured carbonate rock. A considerable rangeinmorphology is exhibited by thesetravertine-marl deposits. Extensive low-relief deposits domi nated by marl are typicalofthe sites in the northern part of Virginia. Travertine and tufa buildups (Figure 2), forming bluffs or falls in active stream channels, are notable components in most deposits throughout the State. Marl accumulations (Figure 3) typically are located upstream of buildups suggesting that carbonate sediments simultane ously accumulated behind aggrading buildups. Deposits are characterized by two morphological signatures that can be recognized in the field or remotely sensed. Buildups create waterfalls and cascadesindepositing streams, where as marl accumulations form broad flat fills or terraces. Additional keys to field recognition include fragments of travertine and tufa located downstreamofincised deposits and the presenceoftravertine and tufa debris or an abun danceofgastropod shells in stream banks; however, most marl deposits have beenmodified by soil development and may not be recognized in the absenceofcoarse carbonate debris without the aidofacid.


Appalachian Karst Symposium. 1991 Hubbard and HermanFigure2:Beaverdam Falls, a buildupoftravertine and tufa on Sweet Springs Creek, Alleghany County, Virginia (McDonaldandBird, 1986, cover). Close examinationofthe travertine componentsofthe deposits reveals that this material is highly porous. ?r ganic detritus, including twigs and leaves trappedonbUJl.dups,greatly increases the available surface area for calcite nucleationanddeposition in the latesummerandfall.Mossand algae are commonly observed on many buildupsandprovide a framework for nucleation and deposition (Figure 4). Decayofthe organic materials after entomb ment incalciteresults in a porous travertineortufa. Some travertine components are laminated. Bandsofden ser travertine are separated by porous partings. Each bandandparting is thought to represent an annual of vertine deposition fromsummerthrough sprmg. Ste.ldtmann(1934) characterized a travertine depoSit near Lexmglonwith "winter layers being thin, compact, and relativelyclasLic,whereas the summer layersare"mossy" and highly calcitic." Manyofthe featuresoftravertine buildups logically resemble travertine featuresincaves. BothdnpSLoneand flows tone cave features tendtobe dense, lacking61the porosity associated with the decayoforganic substrates thataretypicalofsurficial travertine and tufa.Environmental FactorsinDeposit DegradationMost travertine-marl deposits, currently, appear to be undergoing net erosion. Accumulationsofmarl are en trenched, travertine buildups are incised, and large frag mentsoftravertineandtufa are found downstreamfTombroken buildups. Thornton (1953) suggested that traver tine-marl depositsarerelicts and were deposited during past climates. Steidtmann (1934), however, argues that the present destructive erosionofthe is a consequenceofland cultivation. Muddy waters are less favorable for the growthofalgae and moss 1935a), which serve as "both a framework and a protecuve cover for the growing travertine" (Steidtmann, p. 334). Signifi cant damage to travertine-marl depoSitsIScaused by floods (Emig, 1917;Steidtmann,1936;Chafetzand 1984 ). Hurricane Camille, in 1969, had a effect on travertine buildups in Moores Creek and at Gibbs


Hubbard and Herman AppaLachian Karst Symposium, 1991Figure3:Extensive accumulationofmarl along Redbud Run. Frederick County. Virginia. Information on commercial miningofthis depositinGiannini. 1990. (photograph by S.O. Bird. 1984.) Falls in Rockbridge County, Virginia. One observer (phillip Lucas. 1984, personal communication) estimated only 20 percentofthe travertine still remains at Moores Creek Falls. Considerable flood damage to deposits in Rockbridge and Alleghany Counties has been observed by the authors. Steidtrnann's ideas that the destructionoftravertine marl deposits are a consequenceofland cultivation and flooding are better termed a consequenceofland-use modi fication. Siltation resulting from past land clearing and tillage practices have introduced tremendous quantitiesofclastic sediment into travertine-depositing streams. This sediment has buried some deposits (Thornton. 1953) and diluted the purityofthe marl in other deposits during flooding (Sweet and Hubbard. 1990). Examinationofthe sedimentsinthe streamsinthe vicinityoftravertine build ups reveals that the carbonate content is significantly lower than in the adjacententrenched marl. The noncar bonate fractionofthis sediment also serves as an abrasive. Analysesofactive and relict travertine from 10 deposits indicate that active travertine generally containsmore non-62carbonate material than relict travertive (Herman and Hub bard. 1990). This finding is the reverse caseofthe relative decreaseincarbonate content that one would expectofex posed weathered relict travertine. The clearing and modifi cationofland surfaces generally results in greater runoff of precipitation. In addition to carrying additional sediment, increased runoff generates higher magnitude flood pulses. The erosive potentialofthe higher magnitude floods of corrasive sediment-laden waterstothe travertine-marl depo sits is far greater than for the flood events predating the agricultural practicesofthe last two and one half centuries. Degradationoftravertine-marl deposits alsoresulLsfrom pollutionofgroundwater. The presenceofpollutants in emergent groundwaterorin a calcite-precipitating stream may adversely affect the growthofalgal and moss substratesorreduce the volumeofcalcite precipitated. Strongly adsorbing ionic species can inhibit calcite growth by bonding to calcite crystals and effectively poisoning growth sites. Magnesium. phosphate. and some organic compounds notably inhibit calcite precipitation (Morse, 1983). Pentecost (1990.p.125) remarks that "high levels


Hubbard and HermanFigure4:Algal drapes cemented by calcite in Falling Spring Run, Augusta County, Virginia. of dissolved phosphates may ... encourage plant growthbutdiscourage travertine fonnation."SummaryMost karst features are the resultofsolutional pro cesses wherein carbonate bedrock is dissolved, fonning a topography characterized by closed-contour depressions and internal drainage. Emergent karstic water containing high levelsofdissolved calcium carbonate precipitate calcite as anomalous positive karst features. Travertine-marl deposits,comprisedoftravertine and tufa buildups and associated upstream accumulationsofmarl, are widespreadinthe Valleyand Ridge physiographic provinceofVirginia and havebeeneconomically utilized as a sourceofagricultural lime. The commercial miningofthese deposits has ceasedinVirginia partiallyasa resultofconcerns for the environ mental consequencesofnatural resource utilization.Itis ironic thatata timeofenvironmental awareness, when the aestheticsofthese waterfall-generating deposits oversha dows their mineral worth, the environmental consequences of land-use modifications are destroying these natural re sources. Siltation from the clearing and tillageofland and higher magnitude floods fed by runoff from artificial and modified land surfaces have severely eroded manyofthese 63 freshwater carbonate deposits. Polluted waters adversely affect the growthofalgal and moss frameworks or inhibit travertine deposition by blocking crystal growth. The de generationoftravertine-marl deposits has been attributedtoclimatic change; more realistically, it heralds the declineofwater resources by siltation and pollution.ReferencesAustin, G.S. and Barker, 1.M., 1990, Commercial traver tineinNew Mexico:New Mexico Geology,v.12, no. 3, p. 49-58. Chafetz, H.S. and Folk, R.L., 1984, Travertines: deposi tional morphology and the bacterially constructed consti tuents:JournalofSedimentary Petrology,v.54, no.I,p.2289-2316.Emeis, K.C.; Richnow, H.H.; and Kempe, S., 1987, Travertine formationinPlitvice National Park, Yugo slavia: Chemical versus biological control:Sedimen tology,v.34, p. 595-609. Emig, W.H., 1917,TravertinedepositsofOklahoma:Oklahoma Geological Survey Bulletin29, 16 p.


Hubbard and HermanGiannini, W.F., 1990, A commercial marl deposit near Winchester, Virginia,inHerman, 1.S. and Hubbard, D.A., Jr. (editors), Travertine-marl: Stream depositsinVirginia:Virginia DivisionofMineral Resources Publication 101,p. 93-99.Herman,J.S.andHubbard,D.A.,Jr.,1990,A comparative studyoftravertine-marl-depositing streams in Virginia,inHerman, J.S.andHubbard, D.A., Jr. (editors), Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p. 43-64. Hoffer-French,KJ.andHerman, J.S., 1989, Evaluationofhydrologicalandbiological influences onC02fluxes from a karst stream:JournalofHydrology,v. 108, p. 189-212. Hoffer-French, K.J.andHerman, J.S., 1990, AC02outgassing modelforFallingSpringRun,Augusta County, Virginia,inHerman, J.S. and Hubbard, D.A., Jr. (editors), Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p. 5-15. Hubbard, D.A., Jr.; Giannini, W.F.; and Lorah, M.M., 1985, Travertine-marl depositsofthe Valley and Ridge provinceofVirginia -A preliminary report: Virginia DivisionofMineral Resources,Virginia Minerals,v. 31, no.I,p. 1-8. Lorah, M.M.andHerman, J.S., 1988,Thegeochemicalevolutionofatravertine-depositingstream:Geochemical processes and mass transfer reactions:Water Resources Research,v.24, no. 9, p. 1541-1552. Lorah, M.M. and Herman, J.S., 1990, Geochemical evolu tionandcalcite precipitation rates in Falling Spring Creek, Virginia,inHerman, J.S. and Hubbard, D.A., Jr. (editors), Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p. 17-32. Love, K.M. and Chafetz, H.S., 1990, PetrologyofQuater nary travertine deposits, Arbuckle Mountains, Oklaho ma,inHerman, J.S. and Hubbard, D.A.,Jr.(editors), 64Appalachian Karst Symposium.1991Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p.6578. McDonald, J.N. and Bird, S.O. (editors), 1986, The Qua ternaryofVirginia -A symposium volume:Virginia DivisionofMineral Resources Publication75, 137 p. Morse, J.W., 1983, The kineticsofcalcium carbonate dis solution and precipitation,inReeder, R.I. (editor), Car bonates: Mineralogyandchemistry: Washington, D.C., Mineralogical SocietyofAmerica,Reviews in Mineralogy,v. 11, p. 227-264. Pentecost, A., 1990,Thealgal floraoftravertine:Anoverview,inHerman, J.S. and Hubbard, D.A.,Jr.(edi tors), Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p. 117-127. Steidtrnann, E., 1934, Travertine deposits near Lexington, Virginia (abstract):Proceedingsofthe. Virginia Aca demyofSciences,p. 56. Steidtmann, E., 1935a, Travertine near Lexington, Vir ginia (abstract):The American Mineralogist,v. 20,no.3, p. 206. Steidtrnann, E., 1935b, Travertine depositing waters near Lexington, Virginia:Science,v. 82, no. 2127, p. 333 334. Steidtrnann, E., 1936, Travertine-depositing waters near Lexington, Virginia:JournalofGeology,v.44, no.2,p. 193-200. Sweet, P.C.andHubbard, D.A., Jr., 1990, Economic legacy and distributionofVirginia's Valley and Ridge province travertine-marl deposits,inHerman, J.S.andHubbard, D.A., Jr. (editors), Travertine-marl: Stream deposits in Virginia:Virginia DivisionofMineral Resources Publication 101,p. 129-138. Thornton, C.P., 1953,The Geologyofthe Mount Jackson Quadrangle. Virginia:Ph.D. dissertation (unpublished): Yale University, New Haven, Connecticut, 211 p.


Appalachian Karst Symposium. 1991Meteorologyofthe Butler Cave-Sinking Creek SystemFred L. WeferTheMITRE Corporation 4600 Silver Hill Road Washington, D.C. 20389ABSTRACTThe Butler Cave Conservation Society (BCCS), Inc.isconducting a meteorological studyofthe Butler Cave-Sinking Creek System in west-central Virginia. The study consistsofan Entrance Project dealing with data from the passages forming the route from The Entrance to theTrunk:Channel deep inside the cave, and a Trunk Channel Project dealing with data from this 9,000-foot-long passage. More than 550 pairsoftemperature measurements (wet-bulb and dry-bulb) have been made since the study began in April 1984. The variationsofthe temperature, partial pressureofwater vapor, and relative humidity are being studied as functionsoftime (season) and position (within the cave). This paper reports preliminary resultsofthe Trunk Channel Project. The sixteen measurement positions employed are all located more than 1,000 feet horizontally from any known air entry pointsofthe cave and at least 150 feet vertically below the surface. Far inside the cave, meteorological conditions are remarkably stable. For example, seventy-one measurements made at one position (Sand Canyon) yield the following means and standard deviations: temperature 51.40.4 F, partial pressureofwater vapor 0.3750.005 inchesofMercury, and relative humidity 99.01.0% Three major trends are visibleinthe Trunk Channel data: gradually increasing temperature as one goes downstream (51.1 F increasing to 52.2 F); gradually increasing partial pressureofwater vapor as one goes downstream (0.367 increasingto0.385inHg); and relative humidities greater than 97% at most places and times. Four minor occurrences in the data require further study: unusually low temperatures (-1 F lower) at one position; unusually low partial pressuresofwater vapor (-0.01 inHg lower) at two positions; relative humidities as low as 92%; and higher variabilityinrelative humidity at some positions (6% variability versus a more normal 3%). Explanations are suggested for the occurrences observedinthe data.WeferBackgroundThe Butler Cave Conservation Society (BCCS), Inc.isconducting a studyofthe meteorologyofthe Butler Cave-Sinking Creek System(seeFigure 1). Owned and managed by the BCCS, this non-commercial caveislocated in Burnsville Cove, Bath County, west-central, Virginia. It consistsof17+ miles (27+ kilometers)ofpassages with a single human-traversable entranceoflessthanten square feet (one square meter). Wind speeds observedatthe entrance and inside the cave are low, approximately 10 ftls (3 m/s)orless. Other smaller openings are known to exist where water and/or air enterstJ1ecave system.65The system consistsofa central main passage called the Trunk Channel lying along theaxisofa syncline, and several, fairly distinct sectionsofpassages feeding down dip from the slopesofthe syncline. Becauseofits large size and relatively simple geometry, the Butler Cave Sinking Creek System provides the opportunitytostudy cave meteorology both near the entrance and far insideofa large cave. The BCCS acquired an aspirated psychrometerinthe early 1980s at a time when the traditional exploration and mappingofthe cave system were almost complete(seeWefer and Nicholson, 1982) and follow-on projects wereneeded.


Weier Appalachian Karst Symposium.1991Figure1:Thesixteen measurement positionsofthe Trunk Channel Project are shown by filled circles in this plan-view traverse-lineplotofthe Butler Cave Sinking CreekSystem.Thescale is shown byl00-foottic marks along the magnetic north/scale. Since the BCCS study was begun in April 1984, a totalof556 pairsofmeasurements have been made (asof28 February 1991).Theaspirated psychrometer allows the determinationofthe three parameters without the thermometer breakage usually attendant with the useofa sling psychrometer in a cave. The study is limited to three meteorological parame ters determinable via standard wet-bulb dry-bulb methods (temperature, partial pressureofwater vapor, and relative humidity). Previous studiesofthese parameters in caves include: Davies (1960), Cropley(1%5),Bamberg (1973), and Nepstad and Pisarowicz (1989), but there is little pub lished information on meteorologyoflarge cave systems. The period 1985 through 1989saw:refinementsinmeasurement techniques (detailed in Wefer, 1989a), continued measurements concentrating on theEntrance Project and in the Trunk Channel upstream from Sand Canyon (reportedinWefer, 1985), and calibrationofthe ther mometers (described in Wefer, 1988). Resultsofthe Entrance Project were pre sented at the 1989 NSS Convention(seeWefer, 1989b,c). During 1984someinitial explora tory measurements were made to esta blish that temporal and spatial differen ces in the three parameters could actuallybeseen in the cave(seeWefer, 1984). Also during 1984 two projects weredefmed, the Entrance Project and the Trunk Channel Project. Figure 2 shows a sim plified schematic mapofthe partsofthecave system directly involved in these projects.TheEntrance Project involves studyofvariations in the parametersbetween the Entranceandthe Trunk Chan nelatSand Canyon, a passage distanceofapproximately 1,800 ft (550 m). The Trunk Channel Project involves the stu dyofvariations in the parameters along approximately 8,000 ft (2,400 m)oftheTrunk Channel from theupstream endatPenn StateLaketonearthe downstream end at the July 6th Room.combinedintotherelativehumidity parameter. This paper presents preliminary resultsofthe Trunk Channel Projectintermsoftemporal (seasonal) and spa tial (along the Trunk Channel) variations in the three para meters: temperature(F),partial pressureofwater vapor (in Hg,i.e.,inchesofMercury),andrelative humidity (%). These are the unitsinwhich the instrumentisgradu ated and in which the analyses are performed. Conver sions to51(metric) units have been provided for discrete numeric values in the text. In the figures below,100times the partial pressureofwater vapor is plotted, conve niently bringing all three parameters into the same magni tude range. Whenever the term "partial pressure" isusedbelow, the partial pressureofwater vapor is meant.LXYZ (1500 1600 2240.) T100. 5-1.5BUTLERCAVE-SINKINGCREEKSYSTEM BATH COUNTY.VIRGINIA (J1 JANJARY1991) AmDB(0..-90 0.) Because each parameter provides fundamentally differ ent information, it is important to consider all three. The partial pressureofwater vapor, often ignoredincave mete orological studies, is a direct indicationofhow much wa ter vapor exists in the air. Relative humidity is the ratio (expressed as a percentage)ofthe actual partial pressure to the partial pressure at saturation, both partial pressures be ing determined at the measured temperature. Hence, the two parametersoftemperature and partial pressure areMeasurement PositionsSixteen Trunk Channel measurement positions (here after referred to simply as positions) are shown by filled circles in Figures 1 and 2.Thepositions are roughly equally spaced along the passage and are all more than 1000 ft (300 m) from knownairentry points. Figure 3 shows the elevations and depthsofthe positions plottedas66


AppalachianKarst Symposium,1991WeferSample MeasurementsFigure 4 shows sample summer measurements made on15July 1989. Note the date, periodoftime spannedbythe measurements, and initialsofthe ca ver making the measurements, all shownatthe topofthe plot. These eleven measurements were made during a 4.3 hour period, yielding an average total time per measurementof24 minutes. It typically takes something like15to30 minutes to pack up the instrument, trav eltothe next position, and setuptheinstrument again, plus 5to10 minutestomake the actual measurements, yielding an average total time per measurementinby about 80 ft (25 m) and are each about13ft(4m)thick(seeWhite and Hess, 1982 for a discussionofthe geology).Thelowersandstone unit forms the ceilingofthe Trunk Channel from Penn StateLaketo apointabout 3500 ft (1070 m) downstream from Sand Can yon. Beyond that, the lower sandstone unit forms the floorofthe Trunk Chan nel. The thick lines in Figure 3 indicate approximate locationsofthe two sand stone units. Thus, the five downstream most positions are in a different layer of limestone than the restofthe Trunk Channel. '/26oor----,--------------,c o -300 100r----,..--------------, 2400 .r: -200 vCl -w ..J til ::; .(; 2200> v::: -100LXYZ-(1500 . 1600 .2240.) SINKINGCREEKRESURGENCE SANDCANYON T-100.51.5 PENNSTATE LAKE \ \\SINKINGCREEKSIPHON WARM SHOWERS JULY6thROOM BUTLERCAVE-SINKINGCREEKSYSTEM COUNTY. (,,1Jm.JI>RY 1991)ENTRANCESTARTOFTHEUPSTREAM MAZE"-'Figure2:The partsoftheBuLlerCave-Sinking Creek System involved in the meteorological studies are showninthis simplified schematic map. a functionofdistance along the passage from Sand Canyon.By convention,negative distances are upstream (southwest)ofSand Canyon whereas positive distances are downstream (northeast)ofSand Canyon. This upstream/ downstream terminology is determined by the general directionofflowofSinking Creek(seeFigure2)and is consistent with the plungeofthe axisofthe syncline. The elevationsofthe positions are seen to decreasefromapproximately 2300 ft msl (feet above mean sea level)(700 m msl)atPenn State Lake,ina nearly monotonicfashion down to approximately 2000 ft msl (610 mmsl)at the July 6th Room. The depths below the surfaceareseen to range from -150 to -300 ft (-45 to -90 m).Theshallower depths downstream are due to a dropinel evationofthe surface above the cave. The importanceofthese depths is that heat conduction effects from the surfaceare completely negligible' since yearly (seasonal) vari ations as well as higher frequency(e.g.,diurnal) variations penetrate less than 50ft(15m)into the limestone(seeCropley, 1965). The Butler Cave-Sinking Creek Systemisdevelopedinthelower partofthe 330 ft (100m)thick Keyser LimeSlone.This limestoneisdivided into three distinct subdi visions by two sandstone units. These units are separatedDistanceFromSandCanyon(ft)Figure3:Elevations and depthsasfunctionsofdistance along the Trunk Channel from Sand Canyon. The hatched area represents the rock layers above the cave. 67


Wefer Appalachian Karst Symposium.1991TRUNKCHANNELPROJECT,15JUL1989(1425-18:45) (FlY) TRUNKCHANNELPROJECT,26JAN1991(15:05-20:30) (FW) 100. (In ,40 Figure4:Sample summer meteorological measurements for the Trunk Channel. ,-or...''.W INO 0I RE.CT ION ,<)---,,<)---, 100.'PARTIAtPRESSURE (InHQ) , (F'") 100908070-'-I> 6050 L 40 30 o200040006000DistanceFromSandConyon(ft)Figure 5: Sample winter meteorological measurements for the Trunk Channel. .'...,"TWINO DIRECTION '---0''---0' .(F) o200040006000DistanceFromSandCanyon(ft) 70 30 908060100the rangeof20to 40 minutes. The exact time between measurements depends on which positions are visited and the exact route taken through the cave, which can vary duetowater conditionsinthe Trunk Channel. Contrast this with a total timeofless than 9 minutes per measurementinthe studyofthe tour partofWind Cave, South DakotabyNepstad and Pisarowicz (1989). The temperatureateach visited position is shown by a square in Figure 4. As detailed in Wefer (1989a), the wet-bulb and dry-bulb temperatures are recordedtothe nearest 0.1 F. The mercury thermometersinthe psychro meter are graduated in stepsof1.0 F, hence this precision is attainable only with considerable practice and skill. The uncertaintyinthe temperatures is thoughttobe about 0.15 F(0.08 C), which is much smaller than the heightofthe squaresinFigure 4. The partial pressureofwater vapor (actually 100 times the value)ateach visited position is shown by atriangle in Figure 4. The uncertainty in the partial pressures (resulting from the above uncertainty in the wet-bulb and dry-bulb temperatures) is approximately.OO2in HgC.05mm Hg), which is much smaller than the height of the trianglesinFigure 4. The relative humidityateach visited positIOn is shown by a circle in Figure 4. The uncertainty in the relative humidities (resulting from the aboveuncertaintyinthe wet-bulb and dry-bulb temperatures) is approxi mately1.0%(seeWefer, 1989a). This uncertainty is shownbya small error bar on each circleinFigure 4. Figure 5 shows sample winter measurements madeon26 January 1991 using the same format asinFigure 4.Inthese two figures all three parameters are nearly constant over the lengthofTrunk Channel visited on the two dates; however, variations considerably larger than the respective uncertainties are seentoexist. Note that both the temper ature and partial pressure gradually increase as oneproceeds downstream. Because these systematic variationsareslight, horizontal lines have been drawn at ordinate valuesof40 and 50tomake them easiertosee.SpatialVariationsPlots such as Figures 4 and 5 suffice to show that spatial variations existinthe three parameters, evenonasingle date.Ofequal interest are trends in the parameters averaged over time. One wayofdisplaying such trendsisvia a composite plot (Figure 6). On a composite plot,allcurves(i.e.,data sets for which measurements are avail ableatseveral consecutive positions along the Trunk Channel on a given date) for a single parameteraresuper-60 ,---.,----,------.,------,---.,------, DistanceFromSandCanyon(tt)Figure6:Composite plotoftemperature variations along the Trunk Channel. Straight lines connecting consecutive pointsofall three parameters indicate expected values in the Trunk Channel between measurement positions. Dashed arrows indicate the wind directioninthe Trunk Channel. o20004000600068


Appalachian Karst Symposium. 1991 Weierimposed. Because only one parameter is being plotted, a greatly expanded ordinate scale is practical. This simple technique has been found to work well, particularly when curves are available throughout the year,asisthe case in this study. Composite plots have appar ently been previously used only in studiesofthe varia tions of the parameters with distance from the cave en trance(see/orexample,Gaum, 1952; Benedict, 1974a,b; Wefer,1989c). Another methodofdetecting trendsinthe parametersistoplot histogramsofthe distributionsofthe measured values (Figure 7). Ideally one would plot a histogram foreachposition; however, at this stage in the Trunk ChannelProject only the position at Sand Canyon has a suffi cient numberofmeasurementstoprovide a smooth distri bution. The Sand Canyon position has many more mea surements than the other positions for two reasons aspart of the Entrance Project it was visited even when no other positions in the Trunk Channel were visited and itwasnormally visited multiple times ona single date. 25.--------------------,UPSTREAU 20 UEAN= 51.06 .70 15NUN =77 10 25.------------------Accordingly, data from the sixteen positions has been combined into four groups as follows: the five upstream positions, the positionatSand Canyon, the five down stream positions located below the lower sandstone unit(seeFigure 3), and the five downstream positions located above the lower sandstone unit. Whereas the composite plots show only data where consecutive measurements are availableata numberofpositions on each date, the distributions make useofall available data, even isolated measurements. The mean value (MEAN), standard deviation (SIGMA) and the num berofmeasurements (NUM)ineach group are listedinthe figures.Spatial VariationsinTemperatureFigure 6 shows a composite plotoftemperaturesinthe Trunk Channel. Becauseofthe expanded ordinate scale (compared with Figures 4 and 5), the uncertaintyinthe temperature measurementsishere equaltothe half-heightofthe squares. The temperature in the Trunk Channelisseentovary between approximately 49.0 and 53.5 F (9.4 and 11.9 C). Figure 7 shows the distributionsoftempera ture measurements and the statisticsofthese distributions. A major trend in both figuresisthe gradual increaseintemperature as one proceeds downstream. Whereas the increaseisonly from approximately 51.1 to 52.2 F (10.6 to 11.2 C), it is, nevertheless, obvious in both figures. Possible originsofthis increase are discussedina separate section below.20 15 -s 10SAtiD CANYONUEAN=51.0 SIGUA.39NUN, 7150 51 A minor feature apparentinFigure 6 is systematical ly lower temperatures at the central upstream position located at the StartOfThe Upstream Maze. The highest temperature recorded there is actually lower than the average temperature at Sand Canyon. Measurements at this position are largely responsible for the bimodality of the upstream distribution in Figure7. 25..--------------------, Figure7:Distributions and statistics for all temperature measurements in the Trunk Channel.Spatial Variations in Partial PressureTwo possible explanations for this minor feature have been considered. Firstly, it might have been caused byincomplete temporal coverageinthe data(e.g.,the measure ments might simply have been concentratedinthe winter months); however, the measurements are,infact, well distributed throughout the year. Secondly, it might have been causedbythe position being located at a low pointinthe passage where a coldairtrap could form; however, Figure 3 shows that the position is actually at a high pointinthe Trunk Channel. The actual causeofthis minor feature remains unknown. Figure 8 shows a composite plotofpartial pressuresinthe Trunk Channel. The uncertainty in the measure mentsisshown by error bars on the triangles. Partial pressure in the Trunk Channelisseen to vary between 51 Te.peralure50 50 51DOWNSTREAM ABOVESANDSTONE UEAH 52.20SIGUA HUN 75SM'OSTONEUEAN 51.65SIGUA NUN 855 20 .E 15 1069


Weier Appalachian Karst Symposium.1991 (j;45IC a Cl. DistanceFromSandCanyon(ft)Figure8:Composite plotofpartial-pressure variations along the Trunk Channel. approximately 0.345 and 0.400 in Hg (8.76 and 10.16mmHg). Figure 9 shows distributionofthe partial-pres sure measurements and statisticsofthese distributions. A major trendinboth figuresisthe gradual increase in partial pressure as one proceeds downstream. Whereas the increaseisonly from approximately 0.367 to 0.385inHg (9.32 to 9.78mmHg), it is, nevertheless, obviousinboth figures. Possible originsofthis increase are dis cussed in a separate section below. A minor feature apparentinFigure 8 is systematical. ly lower partial pressures at the StartOfThe Upstream Maze. Measurements at this position are largely responsi ble for the bimodalityofthe upstream distributioninFig. ure 9. The partial pressure is also unusually low attheadjacent position just downstream from the StartOfThe Upstream Maze, an effect likely caused simply by thedryair flowing down the passageinthat direction. The lower partial pressures probably share the same origin as the lower temperatures discussed above; however, that origin remains unknown.Spatial Variations in Relative HumidityFigure10shows a composite plotofrelative humidi ties in the Trunk Channel. The uncertaintyinthe mea surements is shown by error bars on the circles. Relative humidity in the Trunk Channel is seen to vary between approximately 92 and 100%. Figure11shows distribu tionofthe relative-humidity measurements and statisticsofthese distributions. UPSTREAI.IUEAN =36.59 =. nNUU =77 20 __S:I:1435J6:17:lB394-0 41 15 .0..:i1 1025..----------------------, .U E96 I94 > -0 920:: 90 0200040006000A major featureinboth figures is that the vast major ityofrelative-humidityvalues are greater than 97%,i.e.,the relative humidity is greater than 97% at most places and at most timesinthe Trunk Channel.102The variationinrelative humidity is approximately 3% at most positions; however, several show variations Two minor features are apparentinFigure 10.Anunusually low relative humidityofapproximately 92% was measured at Penn State Lake. This was during wimer when cold, dry air enters the cave system from nearby Boundless Cave, flows down the Trunk ChanneltoSand Canyon, then out the higher Butler Entrance.DistanceFromSandCanyon(tt)Figure10:Composite plotofrelative-humidity varia tions along the Trunk Channel. OOWNSTREALlBELOWSANDSTONEUHH J7.76 SIGUA.58NUU 85 SANOCANYONUEAN =J7.51 SIGUA. 47 NUU 71 25..---------------------, 20 25..-------------------,20 15 i 10 %3435J637:lB394-0 41 % 34 35J6 373B 394-0 41 53435 36 3941 100PartialPressure HzO (inHg)Figure9:Distributions and statistics for all partial pressure measurementsinthe Trunk Channel. OOWNSTRfAlAABOVE SANDSTONE20 UEAN 38.47SIGUA &415 NUU 75 .0i 10 ;; 15 :i1 1070


Appalachian Karst Symposium, 1991 Weier(see/orexample, Gaum, 1952; Cropley, 1965; Benedict, 1974a,b; Wigley and Brown, 1976; Wefer, 1989b,c). Oneofthe goalsofthe Trunk Channel Project is to deter mine whether systematic temporal variations can be observed far inside a cave, as contrasted with near the en trance. This involves looking for parameter variations in time at each position along the Trunk Channel. sa 96 94 92 10 4050,---------------------,LPSTREAWIAH = 98.Cl.3SIGUA = 1.113WU =77 30.

Weier Appalachian Karst Symposium.1991$8-01.SANDCANYON 105,----------------------, As cave air moves along the Trunk Channel, the changing elevation subjects it to changing pressure. In an adiabatic process(i.e.,no heat entersorleaves the system) the rateofchangeoftemperature with elevation isconstant(overthe limited rangeoftemperature and eleva tioninthe cave). This so called "adiaba tic lapse rate" is approximately -5.4F/I,OOOft (-9.8CIl,OOOm)inthe But ler Cave-Sinking Creek System. driven by the temperature rate. A simple calculation shows that the observed par tial-pressure rateofFigure 8 is consis tent with the observed temperature rateofFigure 13, henceweneed only be concerned with temperature observations. In theloweratmosphereofthe earth, the observed "atmospheric lapse rate" is approximately -3.5F/l,OOOft(6.4CIl,000 m). The atmospheric lapse rate is lower than the adiabatic lapse rate primarily becauseofthe effectsofwater changing phase. Decreasing temperature with increasing altitude causes the rela tive humiditytoincrease and water vapor to condense into clouds. The heatofvaporization released in this processincreases the local temperature, having the effectofdecreasing the lapse rate....TEMPERATURE(F) 100,PARTIALPRESSURE(inHg)90RELATIVE HUMIDITY(X) 6035 555045 40 30 0:--l--.l....--.L-.L--:-:--'------'------'------'----c-L---'--'---'---L----l.-----'-------'------1..J 100 200300DayNumberInTheYearFigure 12: Temperature, partial pressure, and relative humidity as functionsofdayofthe year for seventy-one measurementsatSand Canyon. In the Trunk Channel the relative humidity is nearly 100%, the walls are wet, streamscovermuchofthe floor, and air, walls, and water are at nearly the same temperature.Thenet amountofwater changing phase mustbesmall, hence weexpecttheTrunkChannel lapse ratetobeclose to adiabatic, as indeed it is. points from the colder months with "x" symbols. There appears tobea tendency for the"+"symbols to be at higher temperatures than the "x" symbolsatthe same position. Because the measurements are not evenly distri buted throughout the yearatall positions, only a single least squares fit has been made to the data. The resulting slope (SLOPE), standard deviationofthe slope (SIGSL), correlation coefficient (CORR), and numberofmeasure ments involved (NUM) are shown on the plot. One could,ofcourse,prepare a plot similar to Figure13for the partial pressure. Recall, however, that partial pressure at saturation is a functionoftemperature only. Water vapor is availableinsufficient quantities from Sinking Creek, Sneaky Creek, and the cave wallstokeep the relative humidity in the Trunk Channel at essentially100%,i.e.,atsaturation. Thus the partial-pressure rate is Although it is interesting that the measured lapse rate is close to adiabatic, the reader is reminded that these results are preliminary and that adiabatic compressionisnot the only processatwork in the Trunk Channel. Pos sible sources (or sinks)ofheat (that would make the pro cess non-adiabatic) include heat fromairand water entering (and exiting) the Trunk Channel, as well as heat from the geothermal gradient. Water-temperature measurements indicate, however, that neither Sinking Creek nor Sneaky Creek is a thermal stream (despite the intriguing nameofthe sourceofthe latter). Errorsinthe measurements must also be considered. Possible sourcesoferrors include: errors in elevationsofthe positions, errors due to heat from the observer's body, and errors duetoheat generated inside the psychrometer. 72


Appalachian Karst Symposium. 1991 Wefer56 r-------------------, Figure13:Temperatureas a functionofelevationforall temperature measurements in theTrunkChannel. Points from thewarmermonths are plOlled with "+"s, points from the colder months with "x"s. A least squares fit to the data is also shown. 2300 + .," t

Wefer Wefer,FL.,1984, Cave meteorology,BeeSNewsletter, v. 10, p. 33-49. We fer, F.L., 1985, More on cave meteorology,BeeSNewsletter, v.11,p. 25-56. Wefer, F.L., 1988, Still moreoncavemeteorology,BeeSNewsletter,v.14, p. 76-84. Wefer, F.L., 1989a, On the measurementofrelative humidity in cave meteorology projects, Nittany Grotto News, v. 36, p. 6-14; reprinted in Speleo Digest 1989,p.395-399.Wefer,EL.,1989b, The meteorologyofthe Butler Cave Sinking Creek System (abstract), in Eck, C. and Olsen, 74 Appalachian Karst Symposium. 1991 L. (editors),1989NSSConvention Program,31July 4 August1989.Sewanee. Tennessee: National Speleo logical Society, Huntsville, Alabama, p. 49. Wefer,EL.,1989c, Yet still more on cave meteorology,BeeSNewsletter, v. 15, p. 18-38. White, W.B. and Hess, J.W., 1982, Geomorphology of Burnsville Cove and the geologyofthe Butler Cave Sinking Creek System, National Speleological Society Bulletin,v.44,p. 67-77. Wigley, T.M.L. and Brown, M.C., 1976, The physics of caves (Chapter 9), in Ford, T.D.andCullingford, C.H.D. (editors), The ScienceofSpeleology: Academic Press, London, p. 329-358.


Appalachian Karst Symposium. 1991Emerged Sea Caves and Coastal FeaturesasEvidenceofGlacio-Isostatic Rebound, Mount Desert Island, MainePaulA.RubinR.D.1,Box 159 Feura Bush, NY 12067ABSTRACTAcadia National Park is situated on Mount Desert Island, Maine. A numberofactively forming sea caves are present within the park. Two similarly appearing granite caves, one on Gorham Mountain and one on Champlain Mountain, have been identified at elevations well above the presentsealeve!.Aninves tigation was conducted to determine if these caves developed at a former sea level and were in fact sea caves. Previous studies had drawn conflicting conclusions as to whether the Gorham Mountain cave was indeed a sea cave, whereas the cave on Champlain Mountain was apparently not identifiedinthe literature. A third feature, a pinnacled rock referred to as Pulpit Rock, situated on Day Mountain at the baseofa cliff, resem bles sea stacksofcurrent-day rocky coasts. Limited debateispresentinhistoric literature as to its origin. A water-tube leveling survey was conducted to determine elevationsofthe two emerged caves. Cadillac Cliffs Sea Cave, on Gorham Mountain, occursatanelevationof231 to 238 ft ms!. Champlain MountainSeaCave occurs at an elevationof218 to 227ftms!. Pulpit Rock extends from 204 to 230ftms!. Thompson and others (1989) documented the maximum isostatic crustal uplift following deglaciationofMount Desert Island as 231 ft msl, based on an ice-contact glaciomarine delta. This figure represents maximum glacio-isostatic rebound following the late-Wisconsinan retreatofthe Laurentide Ice Sheet (14to 12 ka).Thelimitsofemerged boulder beaches on Day and Gorham Mountains were definable through addi tional survey work. The bottom and topofthese beaches lie at elevationsofapproximately 204 to 245,and173 to 198ftmsl, respectively. It had been assumed that these emerged boulder beaches were at approximately the same elevation. Instead, a continuumofglacio-isostatic rebound is portrayed as the depressed land surface rose following glacial unloading. The two emerged sea caves, the Pulpit Rock sea stack, the emerged boulder beaches, and nearby wave-cut platforms represent partofthe geomorphic recon struction. In addition, a revised maximum postglacial rebound figureispresented for Mount Desert Island.RubinLocationMount Desert Island is located off the coastofMaine110miles northeastofPortland, Maine and 130 miles southwestofSaint John, New BrunswiCk, Canada.Itis the largest and highestofmany islandsinthis area. Acadia National Park, situated on Mount Desert Island,haslong been a favorite hauntoftourists and geologists.The Upper Marine LimitAs the Laurentide Ice Sheet retreated from Mount Desert Island and the Maine seaboard, the depressed land75mass rebounded rapidly (Stuiver and Borns, 1975). Belknap and others (1987) found that glacio-isostatic uplift (land rise duetoglacial unloading) occurred rapidly during deglaciation, continuing asymptotically thereafter. Even though sea level at the closeofthe Wisconsinan was lower than today's,sealevel relative to the depressed land surface was considerably higher.AsRaisz (1929) aptly states: "The heightofthis 'Upper Marine Limit' is oneoftheoutstandingproblemsofMountDesert Island The interestof... the question lies not onlyinits bearing upon the general...geological historyofthe


Rubinregion, but also in the fact that...nearly all geologists who have studied the Island have ... expressed different opinions on the subject." Shaler (1889) and Bascom (1919) identified numerous terraces on Mount Desert Island that they believedtobeofmarine origin. These terraces are reportedly found at various elevations up to and above 1000 feet, with a marine origin generally not given credence by geologists today. Stone (1899) placed the upper marine limit at 230 feet, while Fairchild (1919) placed itat250-260 feet. Johnson (1925) questioned the marine originofthe higher cliffsofMount Desert Island, including the Cadillac Cliffs. Other authors(e.g.,Smith, 1966, and Thompson and others, 1989) have conducted extensive studies that placed the upper marine limit for coastal Maine more in line with the workofStone (1899) and Fairchild (1919). Manyofthese authors used an altimeter or barometer in ordertodetermine the altitudeofvarious features.Inorder to provide a reasonable interpretationofthe elevationofthe upper marine limit, it is important to examine as many linesofevidence as possible, that together correlate the available data and support a unified theory. These datainclude sea caves, boulder beaches, sea stacks, benches, and marine deposits.Itis necessarytounderstand the processes involved in the formationordepositionofthese features. This paper attempts to correlate recent, still actively forming, coastal features with emerged coastal features.LevelingProcedureInorder to assess and interpret various emerged coastalfeatures, it was necessary to accurately determine their elevation. A systematic leveling survey was conducted utilizing a water-filled V-tube. A water-tube leveling survey is one meansofrapidly determining elevational differences between locations with very small errors. The mean vertical error in loops for water tube surveys is 0.003 percent (palmer, 1987). All survey elevations were determined to the nearest 0.01 foot. Accuracy is certainly valid to the nearest 0.1 foot, although elevations are reported here to the nearest foot. A V-tube is particularly usefulinhilly, heavily vegetated, and very rough terrains. A V-tube can be used very efficiently by two people,asit requires no timelyorcomplicated setup procedures. Colored water at raised endsofthe tube rests at an equal elevation. The rangeofthe instrument utilized was 5.8 feet vertically and 50 feet horizontally. During the courseofthe survey a small numberofpermanent stations (such as the metal rungofa ladder along a trail) were established, separated by many temporary stations. Temporary stations consistoftwo limestone slabs, approximately seveninches in diameter, each with a drilled, 1/4-inch-deep by 3/4-inch-long grooveintheir center. Temporary stations were set by either firmly stomping a drilled rock into the groundorby placing it on 76Appalachian Karst Symposium.1991a bedrock or other solid surfaceinsuch a manner astoguarantee no vertical or horizontal displacement. Wooden rulers were placedineachofthe drilled-rock grooves and manually clamped to each endofthe V-tube. After removing a threaded metal cap from both endsoftheV-tube, the fluid level in the tube was allowed to stabilize and the level above each station recorded. The difference in elevation between the two stations is the gainorloss in topographic elevation along the traverse. This opera tion was repeated twice,orifnecessary more times between stations until a maximum differenceof0.01 feet between readings was achieved. A numberofcalculations were performed after each measurement was made in order to detect any errors. Periodically, the two endsoftheUtube were brought togethertoverify that the fluid levels were equal, thus insuring that no air bubbles had been introduced into the V-tube. Occasional placement of permanent stations reduces repeat surveying should air bubbles be detected. The survey continuedina step wise manner until the levelofeach desired survey pointwasestablished.Boulder BeachesForeaseofthe upcoming discussions, boulder beaches are discussed priortosea caves. However, itwasthis author's firm belief that the Gorham Mountain cave wasanemerged sea cave that ledtothe leveling survey of the emerged Day Mountain boulder beach. Well-developed boulder beaches are often found atthebaseofsteep coastal cliffs and cliff-bounded coves, where abundant source materialispresent. Rocks fallingfromthese cliffs are constantly abraded, rounded, and polished through continuous wave action. A relatively shallow foreland gradient promotes boulder-beach development. Storm surges further round boulders through mechanical corrasion, often hurling boulders against a shoretoelevations higher than the high tide mark. A period of only tensofyearsisneededtoround quarried, rectangular blocksofroadbed materials, such as those with drill holesinMonument Cove. Numerous examplesofmodem boulder beaches can be found along the coastofMount Desert Island. The Monument Cove area exhibits many fine examples. Protected rock-wall bounded coves often host laterally continuous boulders across their shorelines. One example had a maximum vertical extentofapproximately 35 feet, again reinforcing the concept that boulder beaches may represent a greater vertical extent than the difference between the highand low-tidal range. Three emerged boulder beach areas were examined. One excellent example lies on Gorham Mountain, slightly lower elevation ally than the Cadillac Cliffs trail, directly eastofa small trailside cave referred to here as Cadillac Cliffs Sea Cave. Coffin and others (1990) reference this


Appalachian Karst Symposium, 1991 Rubinresultofconstant wave action against highly fractured and intensely shattered country rock. Although slickensides werenotobserved,bothcavesappearto lie along lowanglethrustfaults.Moremassiveorless fractured bedrock is present toward the baseofthe respective caves.Theundulating ceilingofGreatHeadCave, possiblyexhibitingwellerodedslickensides,further suggests localized thrusting.Thehighly fractured bedrock, whether faultedornot, has provided a preferential pathway for attackbysurging ocean waters. Close fracture spacing in both caves, and vein filling in Anemone Cave, permit the continued undercutting, erosion, and removalofrelatively smallblocksofrock.TheceilingofAnemone Cave supports activecliffswallow nests, suggesting that only storm tides result in complete inundationofthe cave.I, AnemoneCave AcadiaNationalPark.MeGRADE 4 Figure1:MapofAnemone Cave. The bestexampleofa high-energy emerged shoreline that is known inNewEngland is foundonDay Mountain belowThe Cleft (Coffinandothers, 1990). Coffin and others cite this former beachasbeingnearly1200feet in lateral extent. Stuiver and Borns (1975) state thatthe beach developedinless than 600 years. The upper boundary was surveyed at 245 feet mslandthe lower boundary at 204 feet ms!.Thesurveyof the Day Mountainboulderbeach,PulpitRock,anda third boulder beach at the baseofPulpitRockwastied to aU.S.G.S.benchmark along Route 3. cobbleand boulder-beach deposit as a 75-foot-square area. Further examination revealed an extensive areaofcobbleandbeach boulders in what was once a small cliff-bounded cove, similar in setting to many coastal covesoftoday. Shipp and others (1985) classify this typeofgeomorphic feature(e.g.,a small cove) as apocketbeach. Through careful observationof the heavily wooded beach, it was possible to delineateanupper and lower boundary. Boundaryselection criteria included the presenceofatleast one clusterofclearly rounded and polished boulders, physically within close proximityofothermore massive concentra tionsofboulders.Theboulder beach top and bottom were determined tobeat198and173 feet msl, respectively.Theboulder beach, perpendicular to the strandline, encom passes 160 feet, having a shore-normal gradientof0.15. The lateral extentof the beach boulders, between nearby rock cliffs,is approximately 225 feel. This emerged cobbleand boulder-beach, complete with a seacavein its upslope wave-cut cliff,maybethe only known exampleanywhereofaformerhigh wave-energy pocket beach.ActivelyFormingSeaCaves ..---.;,.,:,',,'1"-',;,GRA DE 4T Engel P Rubin Great HeadCave AcadiaNationalPark.MeFigure 2: MapofGreat Head Cave. \'.\ .... I \ Anumberofactively forming sea caves are present within the Park. Three of these caves were mapped with a Suuntocompass and calibrated tape. Ane moneCaveandGreatHead Cave (Figures1 and 2), located near the northwesternterminusofOakHillCliffandatthe baseofthe westerly facingseacliffsofGreat Head respectively, are enlarging atthepresent mean sea level. S tagCave(Figure 3), located along the southeasterlyfacing seacliffofGreat Head, is also still forming. The developmentofAnemone and Great Head caves along the basesoftheir respective cliffs isnotsimply a matter of chance.Bothhaveformed as the 77


Rubin Appalachian Karst Symposium.1991Stag Cave AcadiaNallonalPark Me the summitofGorham Mountain. The Seal Harbor 7.5-minute Quadrangle cites this as 522 feet msl. This is believedtobe accurate to within one tenthofthe published contour interval(2 feet). :1' < -. In addition to the fact that Cadillac Cliffs Sea Cave falls well within the former sea level suggested by the Day Mountain boulder beach, examinationofvarious other linesofgeological evidence also suggest that itisanancient sea cave. The relatively short cave abruptly ends in a solid bedrock wall with no large avenues for waterinfiltration that could have ledtocave development.Afewsmall, isolated openings in the ceiling and behind a zoneofrounded granite boulders appear to be incapableof pro viding an avenue for sufficient water discharge throughthecave to account for its development. The only likely sourceofwater capableofinfiltrating downward to these openings is from very localized surface runoff throughalarge, vertical fracture a short distance westofthe cave mouth. This vertical fracture extendsdownward from the topofCadillacCliffs, but does not appear to directly itersect the cave. In addition, the potential recharge areatoThecave is formedina relatively massive pink granite. The resistant nat ureofthe bedrock is apparent as the cave is situated near the baseofa high cliff. Granite is not a rock type typically asso ciated with cave development becauseilScomponent minerals are hard and weakly soluble. The physical presenceofthe cave so far above today's sea levelhasled geologiststospeculate on whetheritis an ancient sea cave. Johnson (1925) doubted that the Cadillac Cliffs were of marine origin. Bascom (1919) and Smith (1966) believed the cave to be of marine origin, whereas Johnson (1925) and Raisz (1929) found otherwise..=GRADE5TEngelPRubInEmerged Sea CavesCadillac Cliffs Sea Cave (Figure 4), located near the baseofan easterly facing cliff on Gorham Mountain, is interpreted as representing evidenceoflarge scale glacio isostatic rebound following Wisconsinan glaciation. This cave is found along the baseofthe Cadillac Cliffs trail approach to the summitofGorham Mountain. A leveling survey determined the elevationofthe cave's floortobe 234 feet msl, with a ceiling elevationof238 feet msl. The seaward-sloping flooratthe entrance to the cave is 231 feet mslata point6feet southeastofthe dripline. The most reasonable location to initiate the leveling surveytothis cave and the nearby boulder beach was from Stag Cave has developed preferentially along a thrust fault. Extension veins, slickensides, and fracturing are present. Hundreds,ifnot thousands,ofyearsofwave action and quarrying have produced the caves weseetoday. Figure3:MapofStag Cave.T EngelPRubinCadillac Cliffs Sea CaveAcadiaNationalPark.MeGRADE4NmagEle. 234ftoIfeet40iFigure 4: MapofCadillac Cliffs Sea Cave. 78


Appalachian Karst Symposium, 1991thisvertical fracture or to any other nearby fractures is ex tremely limited in areal extent because the fractures occur virtuallyatthe topofGorham Mountain's summit ridge. Cadillac CliffsSeaCave's high elevation above sealevelmight, at fIrst glance, suggest that it formed prior totheWisconsinan glaciation. Perhaps the surface topo graphy comprising the cave's potential recharge area was considerably larger prior to glacial abrasion. Kasycki and Shilts (1979) estimate Canadian shield erosion rates to beaslow as33feet for each major glaciation. Gilman and others (1988) estimate that abrasion on Mount Desert Islandmay have lowered the average bedrock surface by oneortwo yards during the last glaciation. Thus, it is unlikelythat Wisconsinan glacial abrasionorplucking removed a significantpartofa formerly larger recharge area. The physical characteristicsofthe cave itself do not support a fracture-infiltration/vadose-stream formationmode.Had a large enough recharge area been available to support an intermittent stream to the cave, the small-size openinginthe ceiling would be expected to more closely conform to the cross-sectional dimensionsofthe cave. Instead, the cave and the cave's mouth are laterally wideandlack any evidenceofchannel incision. The limited recharge potential to the cave, coupled with the cave's blind terminus and lackofany incised floor channel, suggests a formation made other than vadose (unsaturatedzone)groundwater discharge. Cadillac Cliffs Sea Cave bears geologic characteristics moreinkeeping with taday's enlarging sea caves. The physical appearanceofthe cave is thatofan undercut over hang, reducing in size vertically and laterally toward its terminus. The presence far back in the caveofwell rounded granite boulders, up to approximately eighteen inchesindiameter, suggests aggressive wavecut erosion along either bedding planes and jointsoralong a localizedRubinfracture zone. The degreeoferosion evident in the cave appears inconsistent with themorelimited degreeofnatural weathering observed in such locations as the summitofCadillac Mountain. A gentle gradientinthe cave's floor from its terminus downward to the entranceissimilar in nature to that observed in Great Head Cave. This cross-sectional rise in the elevationofthe cave's floor, from front to back, is consistent with an expected reduction in erosive capacity as wave force attenuates toward the shoreline.Theevidence indicates that this slope once graded eastward into the forelandofan ancient coastline. Additional evidence that the Cadillac Cliffs Sea Cave is indeed a sea cave is found along the baseofthe Cadillac Cliffs, where numerous joints and fractures havebeenopened by wave carving. Excellent examplesofwave enlarged fracturescanbe found in the formofsmall caves(171ft msl) near the Waldron Bates memorial plaque situated immediately northeastofthe Cadillac Cliffs trail junction. Nearby wave-cut bedrock is suggestiveoflimited wave erosion. Champlain Mountain Sea Cave (Figure5)is situated on the northern flankofChamplain Mountain within sightofthe Park Loop Road, a short distance westofthe Champlain Mountain trailhead. Its elevation was deter mined by a leveling survey from the Bear Brook Pond. The pond's elevation is cited as 137 feet msI on the 1983 Seal Harbor Quadrangle. A beaver damatthe pond's outlet may possibly have raised the pond's elevation by4feet. The cave was found to occur at an elevation between 218 and 227 feet ms!. This spectacular emerged sea cave is foundatthe back endofwhat was once a small rock bounded cove. Massively fractured granite, with vein filling, provided a favorabie milling surface for breaking waves. This site would make an excellent tourist attrac tion.ChamplainMtn SeaCave AcadiaNationalPark,MeFigure5:MapofChamplain Mountain Sea Cave.79GRADE5TEngelPRubinEle.218Itofeet N... 00


RubinPulpit or Tilted RockPulpit Rockislocatedatthe baseofa 19-foot cliff that is approximately 2000 feet southeastofThe Cleft on Day Mountain and approximately 2900 feet from the sum mitofDay Mountain, along a bearingofS 34 E. Pulpit or Tilted Rock looks remarkably like a coastal sea stack, such as the oneinMonument Cave. Various researchers assert that itissea stack (Shaler, 1889; Bascom, 1919); others believe that itis not(Raisz, 1929; Johnson, 1925), and others believe that itmay be(Coffin and others, 1990). Pulpit Rock is pinnacle in shape, rises 25.9 feet from its lowest exposed point,andis separated from an upslope, laterally extensive cliff by several feet. The cliff base exhibits numerous wave-eroded joints and fractures, both adjacent to Pulpit Rock andatother locations along the cliff line.ThetopofPulpit Rock was surveyed as 230 feet msl. Nearby clustersofrounded and polished beach cobbles and boulders were surveyed between185and 201 feet msl.Thehighest beach-cobble cluster found (201 feet msl) was only 3 feet below the baseofPulpit Rock. The nearby emerged beach boulders, the wave-cut cliff, the physical characterofPulpit Rock itself, the possible wave-turned upper stoneofPulpit Rock, and its elevation within known former shorelines argue that it is indeed an emergedseastack.Dating the DeglaciationofMount Desert IslandDuring the Wisconsin glacial period, the Laurentide Ice Sheet covered Canada and partsofthe United States to a maximumelevationofabout 2.2 miles (3500 m), with about 2.6 miles (4200 m)ofice (Eyles and others, 1983). The land surface was depressed duetoisostatic subsidence. Similarly, Quigley (1983) estimates that the Laurentide Ice Sheet may have been up to 3.1 miles (5000 m) thick at the timeofmaximum glacial advance, 20,000 to 18,000 years before present (20-18 ka). The time frameofmaximum glacial advance prior to the onsetofclimatic warming and glacial retreat is further corroborated by the workofMorgan (1987). Morgan studied fossil assem blagesofColeoptera(beetles) in North America in order to establish the nature and timingofdifferent paleoen vironments at the marginsofthe Laurentide Ice Sheet as it advanced and retreated. Morgan documents insect recoloni zation following retreatofthe ice from its maximumat18ka and places the retreatofthe ice front along the Maine coastline somewhere between14and13ka. Teller (1987) also places the southern marginofthe Laurentide Ice Sheet northofMount Desert Island by 13.5 ka. Gil man and others (1988) estimate that a continental glacier covered Mount Desert Island some 14,000 years ago, having receded by 12,500 years ago. Thus, Cadillac Cliffs and Champlain Mountain sea caves must be younger than approximately14ka. 80Appalachian Karst Symposium.1991Cadillac Cliffs and Champlain Mountain sea caves hadtohave formed at a time when either the elevation of the land surface was considerably lowerorwhen thesealevel was considerably higher. At the timeofthe glacial maximum, the glacio-eustatic (a change in the elevationofmean sea level caused by the growth and decayoficemasses) ocean-water levels wereatleast 394 feet (120m)below present-day levels (Quigley, 1983). Gilman and others (1988) similarly place the minimum eustatic ocean water levelat328 feet(l00m) below today's meansealevel. A slow rise in sea level, broken only by minor glacial re-advances, followed as continental ice meltedandreceded. As continental ice receded from Mount Desert Island, the depressed land surface began to rise, or rebound. Although rebound continued to occur for thousands of years following glacial recession, the greatest rebound occurred immediately after glacial recession. Gilmanandothers (1988) document an elevated delta (a shoreline feature) southofJordan Pondat230 feet (70 m) above today's sea level. Shells from nearby clay deposits, dated by radiocarbon methods, establish a dateofapproximately 12.25 ka for this former sea level. Gilman and others point out that the lackoferosionally modified rocksorclay deposits higher than the Jordan Pond delta suggest that the sea never rose above 230 feet (70 m). Stuiver and Borns (1975) dated shells and seaweed taken from marine sediment from 22 Maine locations interpreted as having been deposited near melting ice. They determined that between 12.2 and 12.6 ka, and more likely before 12.7ka, the ice sheet had retreated from its terminal position on the continental shelf and passed northward through central Maine. Andrews (1987) reports the oldest dated shell-remains from the deglaciationofthe Maine coasttobe 13.4 ka. The lack, at elevations above emerged marine sediments,oforganic remains associated with emergedseacaves and boulder beaches may indicate that these features are somewhat older. Gilman and others (1988) state that fossils from mar ine clays discussed above reveal that Mount Desert Island rose some 217 feet (66 m) above today's sea level between 12 and11ka.Itis possible that the only evidence forahigher post-glacial rebound on Mount Desert IslandisCadillac Cliffs Sea Caveandthe upper limitofthe emerged Day Mountain boulder beach. Quigley (1983) presents the worldwide eustatic sea level curvesofKenney (1964) and Morner (1971). Kenney (1964) portraysanoscillating worldwide sea level ranging between 230 and131feet (70 and 40 m) below today'ssealevel between15and12ka. Morner's (1971) eustaticsealevel curves show a sea levelfluctuating between 249 to 197 feet (76 and60m) below today's sea level between15and 12 ka. Thus, the glacio-isostatic reboundsuggestedby the Day Mountain boulder beach (245 feet above today's sea level) is actually more on the orderof377 to 492 feet (115to150m)above the glacio-eustatic sea levelofsome14ka. Cadillac Cliffs Sea Cave and the emerged Day Mountain boulder beach then formedatan elevationof40to76meters below today's sea level. It is likely that the cave


Appalachian Karst Symposium. 1991formed between 14 and 12 ka, coincident with the recessionofthe continental glacier.Late Wisconsinan ReconstructionThe suggested Day Mountain upper marine limit (245feetmsl) coincides with massive talus lying at the footofTheCleft, and may therefore not be truly representativeof Ihe maximum glacio-isostatic crustal uplift. It is interestingtonote that the distance between the defined upper boundaryofthe Cadillac Cliffs Sea Cave's emerged boulder beach and the roofofthe caveis40feet verticallyand89feet horizontally. As discussed previously, the Cadillac Cliffs Sea Cave falls seven feet below the upper marine limit established on Day Mountain, thus indicatingthat the topofa boulder beach may not be truly reflective of an upper marinelimitInstead, the documentedelevational continuumofwell-rounded beach cobblesandboulders, sometimes with associated sea caves, may represent a periodofrelatively uniform glacio-isostatic rebound with accompanying marine transgression. Itmightbe argued that the surveyedlOpofthe Day Moun tain cobble and boulder beach represents the uppermost elevationofa former storm berminthe supratidal region.Theapparent elevational and lateral consistencyofthis emerged shoreline supports a mean high-water interpretation.Further credence is given to this interpretationby Ihe 238-foot-msl elevationofCadillac Cliffs Sea Cavewhichmust have been exposedlOconstant wave attack fo; someperiodoftime. In lightofthe physical evidence available at this time, and in the absenceoffurther verifiahIeevidenceofa higher upper marine limit, the upper most surveyed elevationofDay Mountain's emerged shoreline (245 feet msl) will be assumed to be coincidentwiththe maximum glacio-isostatic reboundofMountDesertIsland.RubinDay Mountain Cave Evidenceofthe Upper Marine Limit?It is possible that Day Mountain Cave might provide the only direct evidence of higher late-Wisconsinan glacio isostatic rebound for coastal Maine. This small granite cave, 36 feet in length,ispresent virtually at the baseofThe Cleft on Day Mountain. Its elevation was recently surveyed, placing it at an elevationof443-4515 feet msl. Day Mountain Cave has a maximum ceiling heightof8.7 feet near its terminus, and it is floored with a dry saprolitic granite-rich sediment. The rear and ceilingofthe cave are uniformly freeofany significant infiltration pathways, along which dissolution may have occurred. Its length is beyond thatwhich might be expected from driving rain or freeze-thaw mechanisms. No viable Holocene mechanism appears availablelOaccount for soil and rock removal from the cave.Inaddition, no surficiaI runoff or tributary-watershed explanation can provide infiltration to the cave. Figure 6 reveals the convex upward profileatthe terminusofthe cave's longitudinal section, almost asifit was formed by waves crashing into it along athin fracture. Until a viable alternative explanation for the forma tionofDay Mountain Cave is forwarded, a milling or wave-attack origin cannot be discounted. Similarly, the ultimate placementofthe upper marine limit for coastal Maine may also havelOaccount for this cave. A brief li terature search indicates that cavern development in granite is predominantly limited to solution within fractures along soluble vein fill or grussified (Esch, 1991) material open to erosional stripping. Finlayson (1983), Hose (1991), and Esch (1991) discuss varioustypesofgranite caves, allofwhich require the removalofweathered Today, Mount Desert Island encom passes approximately 107.8 mi 2 By utilizing the 245-foot elevational contour,itispossible to reconstruct the earlypostglacial appearanceofthe Mount Desert Island area at -14 ka. Atthistime, only 25 percentoftoday's landmassprojected above sea level. Seven islands over 200 acres in size were pre sent. Sixofthe largestofthese islands encompassedCadillac,Dorr.Day, Pemetic,Sargent,Norumbega,and YoungsMountains(13,132acres); Western and Bernard Mountains (1,732 acres); Champlain and Gorham Mountains(967 acres); Beech Mountain(741acres); St. Sauveur Mountain (393acres);and Acadia Mountain (213 acres).Anumberofother smaller islands werealsopresent. ;: ..... ,., ....Figure6:MapofDay Mountain Cave.81Day MtnCaveAcadiaNationalPark,MeGRADE5NmT Engel P Rubin40ifeet


Rubingranite by flowing water. Alternately, Dragovich (1969) discusses the formationofsidewallandbasal tafoni in graniticandgneissic rocks. Sidewall tafoni, foundonverticalornear vertical rock faces, generally show forms ranging from a shallow ellipsetoa near semi-circle. They form largely as a resultofweatheringbymoisture and temperature variations, having been documentedtodepthsofnearly eightfeetThompson and others (1989) have documented iso static crustal-uplift increasing from coastal to interior Maine from 193 to422feet msl.Perhapsthe initial glacio-isostatic rebound concurrent with a thinningorretreating ice sheet was rapid and sufficiently short-livedinthe Mount Desert Island area so as to obviate the accumu lationofany sedimentary recordalongsmall, cliffed islands.Thecombinationofinitial rapid uplift and ex posed steep topography may account for the lackofsedi ment depositionorboulder-beach development (above 245ft.msl) concurrent with ice sheet thinningordeglaciation. Perhaps the developmentofthe Day Mountain boulder beach heralded a reductionincrustal-uplift rates from even higher initial rates.Noadditional speculation is warranted without further field work.ConclusionThompsonandothers(1989)haveassigneda maximum crustal-uplift valueof231 feet to Mount Desert Island, basedonthe elevationofan ice-contact glacio marine delta. This study has established a slightly higher maximum crustal-uplift value based on the surveyed eleva tionsofemerged boulder beaches and sea caves. Develop mentofthe Cadillac CliffsSeaCavemust follow the retreatofcontinental ice from Gorham Mountain prior to substantial rebound and precede the lower sea-level stand documentedinthe elevationally lower clay deposits (12.25ka).Because rapid crustal rebound occurred concurrently with ice thinningordeglaciation, it appears reasonable to attribute a dateofbetween 14 and 12.7 ka to the emerged Cadillac CliffsSeaCaveand the surveyed, upper limitofthe Day Mountain boulder beach (245 ft msl). Additional ly, this study has raised questions regarding whether a higher maximum glacio-isostatic rebound figure might be verifiedatDay Mountain Cave through further investiga tion.AcknowledgmentsThe author extends thankstoThorn Engel who, eveninpouring rain, ably manned the other endofthe leveling tube. Thornisalso responsible for the cave surveys and draftingofthe maps accompanying this article. Thorn providedgoodhumorandcompanyoverthe courseofmany fine days on Champlain's "Isle des Monts Deserts". 82Appalachian Karst Symposium.1991ReferencesAndrews, J.T., 1987, The late Wisconsin glaciation and deglaciationofthe Laurentide Ice Sheet,inRuddiman, W.F. and Wright, H.E. Jr. (editors),The GeologyofNorth America,v.K-3. North America and Adjacent Oceans During the Last Deglaciation:The Geological SocietyofAmerica, p. 13-37. Bascom, F. 1919.ThegeologyofMountDesertIsland.Geographical SocietyofPhiladelphia. Bulletin17,p.117-130. Belknap, D.F. and others, 1987, Late Quaternary sea-level changes in Maine,inNummedal, D.; Pilkey, O.H., Jr.; and Howard, J.D. (editors), Sea level fluctuationandcoastalevolution:SocietyofEconomicPaleontologists and Mineralogists. Special Publication No.41,p.71-85. Coffin, T.E.; Tyler, Jr., H.R.; and Broyer, M.W., 1990,AnEvaluationofSites Representing Emerged Marine Shoreline Features in the New England-Adirondack RegionforTheir Eligibility as National Natural Landmarks,Prepared for Maine State Planning Office. Dragovich, D., 1969,Theoriginofcavernous surfaces (tafoni) in granitic rocksofsouthern South Australia, ZeitscriftfUr Geomorphologie,v.13, p. 163-181. Esch, L., 1991, Speleogenesis in theLostCreek Wilder ness Area, Colorado,Ge02 ,v.18, p. 31-35. Eyles, N.; Dearman, W.R.; and Douglas, T.D., 1983,Thedistributionofglacial landsystems in Britain and North America,inEyles, N. (editor),Glacial Geology:Pergamon Press Inc., p. 213-228. Fairchild, H.L., 1919, Postglacial upliftofsouthern New England:Geological SocietyofAmerica Bulletin,v.30,p.597-636. Finlayson, B., 1983,Theformationofcaves in granite,in Proceedingsofthe Anglo-French Karst Symposium. September19-26.England and Wales. United Kingdom.Gilman, R.A.;Chapman,C.A.;Lowell,T.V.; and Borns,Jr.,H.W., 1988,The GeologyofMount Desert Island: A Visitor's Guide to the GeologyofAcadia National Park:Maine Geological Survey, DepartmentofConservation. Hose, L.D., 1991, Millerton Lake Caves: A new model for granite cave development: Ge02, v.18, p. 36-37. Johnson, D., 1925,New England Acadian Shoreline:John Wiley and Sons, N.Y.


Appalachian Karst Symposium. 1991Kaszycki, C.A. and Shilts, W.W., 1979, Average depthofglacial erosion, Canadian Shield:Geological SurveyofCanada Paper 79-1B,p.395-396. Kenney, T.C., 1964,Sealevel movements and the geologic historiesofthe post-glacial marine soilsatBoston, Nicolet, Ottawa and Oslo:Geotechnique,v.14,p.203-230. Morgan, A.V., 1987, Late Wisconsin and early Holocene paleoenvironmentsofeast-central North America basedonassemblagesoffossilColeoptera,inRuddiman, W.F. and Wright, H.E. Jr. (editors),The GeologyofNorth America.v.K-3, North AmericaandAdjacent Oceans During the Last Deglaciation:The Geological SocietyofAmerica, p. 353-370. Morner, N.A., 1971, Eustatic changes during the last 20,000 yearsand a methodofseparating the isostatic andeustaticfactorsin anupliftedarea:Palaeogeography, Palaeoclimatology, Palaeoecology,v.9,p.153-181. Palmer, A.N., 1987, Cave levels and their interpretation:National Speleological Society Bulletin,v.49, p. 5066.Quigley, R.M., 1983, Glaciolacustrine and glaciomarine clay deposition: A North American perspective,inEyles,N.(editor),Glacial Geology:Pergamon Press Inc.,p.140-167. Raisz, E.1., 1929, The sceneryofMl. Desert Island: Its originanddevelopment:AnnalsofNewYork AcademyofScience,v.21,p.121-186.83RubinShaler, N.S., 1889, The geologyofthe IslandofMount Desert, Maine:U.S.GeologicalSurvey, Eighth Annual Report,Pl. 2, p. 987-1061. Shipp, R.C.; Staples, S.A.; and Adey, W.H., 1985, Geomorphic trendsina glaciated coastal bay: A model for the Maine coast:Smithsonian Contributionstothe Marine Sciences,no. 25. Smith, D.A., 1966,Late-GlacialEmergedShoreline FeaturesofMountDesert1sland, Maine:A report submittedtoAcadia National Park,SSp.Stone, G.H., 1899, The glacial gravelsofMaine and theirassociateddeposits:U.S.GeologicalSurvey,Monograph No.34, 499p. Stuiver, M. and Borns, H.W., 1975, Late Quaternary marine invasion in Maine: Its chronology and associated crustal movement:Geological SocietyofAmerica Bulletin,v.86, p. 99-104. Teller, J.T., 1987, Proglacial lakes and the southern marginofthe Laurentide Ice SheetinRUddiman, W.F. and Wright, H.E., Jr. (editors),The GeologyofNorth America,v.K-3, North America and Adjacent Oceans During the Last Deglaciation:The Geological SocietyofAmerica,p.39-69. Thompson, W.B.; Crossen, K.1.; Borns, Jr., H.W.; and Andersen, B.G., 1989, Glaciomarine deltasofMaine and their relation to Late Pleistocene-Holocene crustal movements,inAnderson, B.G. and Borns, H.W., Jr. (editors), NeotectonicsofMaine:Maine Geological Society, Bulletin 40,p. 43-67.


Plate Appalachian Karst Symposium, 1991PlateB:The Sumpinthe northeastern sectionofMcFail's Cave, Schoharie County, New York. Penetrationofthis nearly water-filled segment in the early 1960's ledtothe discoveryofapproximately five milesofcave passages. This cave stream flows for a distanceofabout 2.5 miles through explored sectionsofthe cave. Scallops are visible on the wallsjustabove the water level. For a mapofthe cave,seeFigurelAofMylroie, this volume, page 87.Photograph by ErnstH.Kastning.84


Appalachian Karst Symposium. 1991 MylroieCave Development in the Glaciated Appalachian KarstofNewYork: Surface-CoupledorSaline-Freshwater Mixing Hydrology?John E. MylroieDepartmentofGeology and Geography Mississippi State University Mississippi State, MS 39762ABSTRACTRecently, Panno and Bourcier (1990) proposed a new hypothesis for the mechanismofcave formationinthe eastern midcontinental United States. They suggest that the dissolutionoflimestone occurredasa resultofthe mixingofsaline formation waters with shallow, meteoric-derived fresh groundwater. Pleisto cene glacial loading was suggested as the mechanism which caused dischargeofthe saline waters. Labora tory experiments and mixing-zone examples from the Yucatan were used to illustrate the dissolutional mechanism. Further, the occurrenceinthe midcontinentofcarbonate and sulfate rocks, MVT-mineraliza tion zones and basin margins, karst regions, and the limitofPleistocene glaciations wereallcited as field evidence that supports the hypothesis. Unfortunately, the hypothesis is flawed because the field evidence does not in fact support what other wise appears to be a viable model. The following are someofthe major concerns:1.Calculationsofthe amountofrock removed by the mixing processistrivial compared to the initial conditions.2.Sulfate rocks cannot be treated as analogsofcarbonate rocksina discussionofmixing-zone dissolution. 3. MVT mineralization in the Midwest had nothingtodo with Pleistocene glaciation. 4. The model does not explain passage levels or passage geochronology. 5. The relationshipofkarst developmenttothe southern marginofglaciation reflects the mantling or quarryingofkarst features by glacial action, not preferred karst development at the glacial margin. 6. The appearanceofsaline waters in modem caves is an effect, not a cause,ofcave formation. 7. The morphology, plan, and profileofcaves developedinor near glaciated areas does not differfTomthatofcaves developed outside the rangeofinfluenceofglaciations; caves developedinconditionsofmixed-water (either H2S/freshwater or saline/freshwater) are distinct from caves developed by meteoric water/CO2-driven dissolution. Comparisonofcave morphologies from glaciatedAppalachian karstofNew York with cave morpholo gies from mixed-water cavesofthe Bahamas and Guadalupe Mountains illustrates several significant differ ences between them. Cavesofmixed-water origin have globular, irregular chambers with numerous win dows, thin wall-partitions, and blind passages.Incomparison, New York caves resemble those from all over the world that have developed as a componentofC02-charged meteoric-water infiltration, underground transport, and discharge backtothe surface. These caves are dendritic, coupledtothe surface hydrology, and contain little evidenceofa mixed-water morphology.IntroductionThe dissolutionofcalcium carbonate on a scale necessarytoform caves requires that water flowing within the carbonate aquifer be chemically aggressive during cave genesis. Bogli (1964) demonstrated that the mixingofwaters saturated with calcium carbonate at different partial pressuresofC02results in renewed dissolutional aggrcs-85sivityofthe mixed water. Beginning in the mid-1970's, the importanceofmixing watersofdifferent salinityinthe formationofcaves was recognized (plummer, 1975; Palmer and others, 1977a,b; Back and others, 1979). The developmentofsecondary porosity, including caves,incarbonates as a resultofmixing watersofdifferent salinities is now a well-established phenomenon (Back and others, 1984, 1986; Mylroie, 1988; Proctor, 1988; Sanford and


MylroieKonikow, 1989; Stoessell and others, 1989; Mylroie and Carew, 1990; Vogel and others, 1990). Other investiga tions have studied other mixed-water phenomena that in volve thermal, sulfate, and sulfide-charged waters (Egemei er, 1981; Hill, 1987; Bakalowicz and others, 1987; Pal mer and Palmer, 1989; Ford, 1989). These are thehypo geniccavesofPalmer (1991). The useofmixed-water models has helped explain the unusual morphologiesofdissolutional caves found in such disparate settings as the Bahama Islands, Guadalupe MountainsofNew Mexico, and Black HillsofSouth Dakota. A novel new model that utilizes the effectsofmixed waters for cave development was proposed by Panno and Bourcier (1990). They suggested that the caves along the southern marginofthe Pleistocene glacial advancesinthe midcontinentofthe United States were the resultofglacial loading and meltwater discharge that flushed basinal forma tion brines to shallow depths in limestones. There, it was suggested, those brines combined with glacial meltwater and meteoric water to produce a mixed water with the aggressivitytodissolve caves.Iftheir hypothesisisvalid, then these caves should have a morphology common to thatofmixed-water caves elsewhere, whichisdifferent from that seen in classic dendritic caves that are coupledtothe surface hydrology. The caves cannot be older than the onsetofthe Pleistocene glaciations. The caves should be demonstrably different from caves formed in similar rocks well away from the influenceofglaciation and possible expelled brines. Examinationofthe arguments presented by Panno and Bourcier (1990), coupled with field observations, suggest that dischargeofformation brines from intracratonic basins as a resultofthe effectsofPleistocene glaciation did not have a significant influence on cave developmentinthe midcontinentofthe United States. The caves that occur near the glacial marginsofthe midcontinent appeartohave developed like most caves in the eastern United States. That is, as dendritic caves that function as partofthe meteoric hydrologic cycle. Caves from the glaciated Hel derberg PlateauofNew York will be comparedtofresh water/salt-water mixing cavesofthe Bahamas to demon strate the morphological differences between caves pro duced by meteoric freshwater acting alone and those pro ducedbymixed waters.Problems with the Panno-Bourcier ModelThe Panno and Bourcier (1990) model has a number of serious difficulties. These are listed and discussed below.1.Based on their calculations, theoptimalmixing ratioofbrines to fresh water produces an additional dissolutional capabilityof0.06 mmol/l. The lengthoftime during which this activity occurred (and must have been less than thatofthe Pleistocene glaciations) was not discussed. Us ing a slightly lower valueof0.04 mmol/l, and an initial limestone porosityof10% (high for Paleozoic limestones) 86Appalachian Karst Symposium, 1991they determined that 300 m 3oflimestone could be dis solved for each square kilometeroflimestone. Thatisequi valent to a2m diameter cave passage 100 m long. This mayatfirst seem to besignificant cave development, but when the possible volumeoflimestone involved is consi dered, it is trivial.Ifwe assume a mixing-zone thicknessof10 m, then a1 km2surface area yields 1x 10 7 m 3 of limestone. Removing, under optimal conditions, 300 m 3 results in an increaseofporosityof300 m 3 /1O,000,000 m 3 or 0.003%. That increase is negligible comparedtothe 10% porosity assumed as the initial conditions. With a 10% initial porosity, itishighly unlikely that the minor increased dissolutional potential they indicated wouldbeconcentratedtoform oneormore tubesofexplorable size. 2.Themodel presents data on the distributionofcar bonate and sulfate rocks in the midcontinentofthe United States (their Figure 2A). The figure failstodifferentiate between the locationsofsulfates (gypsum and anhydrite), limestones, and dolomites. Further, the mixing model used in their argumentsisnot specific with regardtosul fate rocks, and does not deal with the difference between dolomites and limestones, which is necessary in light of recent work (palmer and Palmer, 1989). Useoftheir Fig ure 2A to compare glacial margins, Mississippi Valley type (MVT) mineralization, intracratonic basins, and karst development by basinal brines is diminished by the lack of specific data about soluble rock types in the figure.3.The model uses MVT mineralization as a demonstra tion that brines can be expelled from intracratonic basins and interact with adjacent rocks. While that has merit, the model expands the idea and couples Pleistocene glaciation and cavern dissolution with MVT mineralization. This coupling seems contrived at best, and it does not match the distributionofglaciation andMVTdeposits as seenintheir Figure2.4. The paper describes cavesofthe midcontinent as "gen erally simple and consistofone or two tubular passages that have formed along joints and fractures" (pan no and Bourcier, 1990, p. 770). On the contrary, cave develop ment within and near the Pleistocene glacial margin is much more complex than this description would imply. For example, ColdWaterCave in northeastern Iowa, Mystery Cave in southeastern Minnesota, and McFail's Cave and Skull Cave in east-central New York (Figure1)are all from glaciated areas; all have long, large, and complex passage development. In addition, the Panno Bourcier model does not explain this passage complexity, nor does it explain cave levels or passage geochronology. 5.Thepaper describes karstic terrain as concentrated around the southernmost extentofPleistocene glaciation. This karst concentration is attributed to brine expulsion caused by the effectsofglacial loading. Comparisonoftheir Figure 2A with 2C and 2D demonstrates that karstic phenomena northofthe glacial margin have most likely been glacially quarried or mantled with glacial drift. With-


Appalachian Karst Symposium. 1991 Mylroie Nr o300 o II IIIIIIIMcFAlL'S CAVENEIl YORK N o300 meters1......-' MYSTERY CAVE MINNESOTAN o300 '----' meters N o r 300 >------' metersCOLDWATER CAVEIOWA Figure1:Mapsofcaves from glaciated regionsofNorth America. Wiggly arrows indicate stream flowinthe caves. AMcFail's Cave, New York, modified from Kastning, 1975. Dashed lines indicate recently discovered, unsurveyed passage. BColdwater Cave, Iowa, from Welch, 1989/90. CMystery Cave, Minnesota, from Alexander, 1989/90. The cave feeds water east across meander necks. River backflooding has produced maze development. DSkull Cave, New York, from Kastning, 1975. The southern endofthe cave is a backflooded maze causedbyglacially-occluded resurgences.87


Mylroie out demonstration that the glacial margin karst is different from mantled karst to the north or glacially uninfluenced karst to the south, the model and its conclusions are not supported.6.The occurrenceofsaline waters in glacial-margin karst systemsispresented as evidence that brine mixing prduced the caves.Itis equally probable, however, that the existing karst drainage systems have served as an escape path for brines, in which case the occurrenceofthe brines there is an effect and not a cause. Further, they did not discuss the abundanceofsaline water in karst springs in distant, unglaciated areas. Whereas the existenceofhalide rich fluid inclusionsin carbonate minerals in these caves does indicate exposure to halide-rich waters, it does not indicate that those waters were dissolutionally aggressive and helped promote cave development.7.The key factor in their model is the useofa mixed water mechanism to explain the formationofcaves within and along the marginofthe Pleistocene glaciations. Caves formed by mixed waters are generally uniqueinpat tern, having a configurationofnetwork, spongework, or ramiform mazes (palmer, 1991). They rarely form dendri tic systems. On the other hand, caves that developed in a groundwater regime coupled with the meteoric hydrologic cycle are usually dendritic. In those cases, maze configura tions form only as an overprint in high discharge or back flooded passages,orwhen surface recharge is extremely diffuse (palmer, 1991). Caves along the glacial marginofthe midcontinentaI United States are primarily dendritic and do not have the unusual patterns associated with cave systemsofmixed-water genesis (Figure 1).ComparisonofGlaciated Appalachian Caves from New York with Mixed-Water Caves from the BahamasGlaciation has a dramatic impact on any landscape it directly affects, and through additional sea-level and cli matic effects, can influence landscapes thousandsofkilo meters away from the ice margin. By exposing new lime stone outcrops, incising valleys, and backflooding existing cave systems, glaciation has been demonstrated to have its own unique wayofenhancing cave and karst development (Mylroie, 1984). Panno and Bourcier (1990) have pro posed another typeofglacially-enhanced karst development that involves glacial loading and subsequent brine expul sion into shallow limestone aquifers to produce mixing dissolution and cave formation. Support for their model hinges on whether caves from the areaofPleistocene gla ciations have the characteristicsofthoseofa mixed-water modeofinitiation or development. Mixed-water caves in the Bahamas have developed on a time scale (Late Quaternary) comparabletothatofthe proposed glacial-margin cavesofPanno and Bourcier (1990). They also developed with minimal overprintingbyhydrologic processes other than mixingofsaline and88Appalachian Karst Symposium. 1991 fresh waters, and therefore represent a "type" sample,orendmember, for this methodofcave development.Incon trast, cavesofthe Helderberg PlateauinNew York devel oped in an environment that has been repeatedly glaciated. They are formed in dense Paleozoic limestones with pro perties similar to limestones found along the areaofPleis tocene glaciation.Ifglaciation does produce mixed-water dissolution as proposed by Panno and Bourcier (1990),theHelderberg caves should have a mixed-water morphology. However,ifmixed-water dissolution was not active, then the Helderberg caves should possess morphologies similar to that found in dense Paleozoic limestonesofan area far from glaciation and possible brine expulsion, such asinTennessee and northern Alabama. The Bahama Islands contain a large numberofcaves that developed within a freshorbrackish water lens during timesofpast, higher sea levels. The islands are tectonical ly stable, and the limestones are Pleistoceneinage (Myl roie, 1988). Therefore, dissolution that occurred duringtheglacio-eustatic sea-level high standsofthe Pleistocene are responsible for the caves now found at 1 to 6m above cur rent sea level (Mylroie, 1988; Mylroie and Carew, 1990; Vogel and others, 1990). These caves have not been exposed to other hydrologic regimes, and therefore show minimal overprinting by other processes. These caves consistofa seriesoflarge chambers with numerous tubes that interconnect and end abruptly (Figure 2). The cham bers are globular and irregular in shape and often separated from one another by extremely thin bedrock partitions through which small windows may have dissolved. This pattern is consistent with the ramiform or spongework pat ternofmixed-water caves in high porosity,poorly jointed rock (palmer, 1991). These caves are smaller than those foundinthe Guadalupe MountainsofNew Mexico, anddonot contain the abundanceofsulfate mineralization found there, but in termsofpassage morphology, cross section, and pattern, the Bahama caves are an excellent matchforthe hypogenic Guadalupe caves. Cavesofthe Helderberg Plateau, on the other hand, exhibit a classic dendritic pattern (Figure 1). Glaciation has modified that initial pattern, primarily by shifting water input and output points (Mylroie, 1977). In some cases, backflood mazes have been superimposed upon a dendritic plan (Mystery Cave and Skull Cave, Figure1).There is no evidenceofthe typeofnetwork development produced by hypogenic cave formation, such as that seen in the dense, well-jointed Paleozoic limestonesofthe Black HillsofSouth Dakota. Glaciation has also intro duced sediment into pre-existing cave passages, and deposits that reflect ice advance, still-stand, and retreat are present (Mylroie, 1984). The arrangementofpassages and their sediments indicate that the cavesare or have been coupled to surface hydrology. Unusual ramiform or spongework passage development is not present.Network mazes are found, but their configuration and placement within cave passages indicate that they resulted from back floodingofthe cave systeminresponse to obstructions by


Appalachian Karst Symposium. 1991 Mylroie L.Ji mo5 ...m o5 -.. m 0---3 mo20 ----.. m o3 "'""--' m CD Figure2:Selected mapsofcaves formed by mixed water in the Bahama Islands. North is to the top in all cases. Hachured line represents marginofIithified dune in which the caves are found. A Dance Hall Cave, San Salvador Island. BGeorge Storrs' Cave, San Salvador Island. C South Deep Creek Cave, South Andros Island. D Beach Cave, San Salvador Island. EBug City Cave, San Salvador Island. F. Maroon HilI Caves, Great Inagua Island. GLighthouse Cave, San Salvador Island. H Reckly HilI Pond Cave, San Salvador Island. Note the varietyofscales and the consistent general patternofthe caves. Modified from Mylroie and Carew, 1990. glacial sediment. In the midcontinental United States, Mystery Cave, Minnesota, contains passages that are pri marily a network maze (FigureIC).Thecave's setting, within a meander loopofa surface river, indicates that it is a meander-cutoff cave and its maze patternisthe resultofriver-induced backflooding (Mylroie and Mylroie, in press). Whencomparedtocavesfound in similar rocks in AlabamaorTennessee, the Helderberg caves are seen to be direct analogs. Despite glaciation, theNewYork caves share the overall dendritic pattern and morphologyoftheir southern U.S. counterparts. In the Helderberg caves there arenopassage morphologies, configurations,ordepositsthatrelate to a mixed-water origin. There is no evidenceofbrines or the productsofbrines. AlI passage morpholo gies, configurations, and deposits are consistent with those produced bydevelopmentofa dendriticcavesystem coupled to the meteoric hydrologic cycle.TheHelderberg caves differ from their southern glacially-uninfluenced counterparts only by sediment and non-brine floodwater modifications produced by ice contact.Conclusionsshipofdifferencesincave and karst development to the lo cationofthe Pleistocene glacial margin reflects the quarry ing and mantling effectsofglaciation not locationofbasin marginsorMVTdeposits. Co-occurrenceofsaline waters and caves today probably reflects the escapeofformation brines through pre-existing conduits.Theconfiguration, morphology, and patternofcaves in the proposed brine expulsion area do not show the characteristicsofcave sys temsofmixed-water genesis. Rather, those caves possess a dendritic pattern (often glacially modified, but not brine modified) that is consistent with development as partofa meteorichydrologic pathway.Themorphologyofcaves in the proposed brine-expulsion area do not differ from den dritic caves developed in areas far removed from glaciation. Panno and Bourcier's (1990) model in which glacially expelled basinal brines are an important cave-forming me chanisminthe midcontinent is not supported by the field evidence. Glacially-induced expulsionofformation brines from intracratonic basins may have occurred, and may have added some minor component to the developmentoflime stoneporosity at the glacial margin. However, the effect is obscure and minimal comparedtoprocesses operatinginthe meteoric hydrologic cycle. The roleofglacially-expelIed basinal brines on devel opmentofcavesand karst in the midcol1tinental United Slates is minimal. The amountofextra rock that could be removed by that mechanismisinsignificant.Therelation-AcknowledgmentsTheauthor acknowledges the Bahamian Field Station,89


MylroieDr. Donald T. Gerace, Executive Director, for logistical support in collecting the Bahamian cave data. Discussions with A.N.PalmerandW.E.White helped focus my ideas. Suggestions from J.L. Carew substantially improved the manuscript.ReferencesAlexander, E.C., 1989/90,Karsthydrologyofsoutheast Minnesota:Ge0 2 ,v.17, no. 2,3, p. 32-53. Back, W.; Hanshaw, B.B.; Pyle, T.E.; Plummer, L.N.; and Weidie, A.E., 1979, Geochemical significanceofgroundwater dischargeandcarbonate dissolution to the formationofCaletaXel Ha,QuintanaRoo,Mexico:Water Resources Research,v.15, p. 1521-1535. Back, W.;Hanshaw,B.B.;andVan Driel, J.N., 1984, Roleofgroundwater in shaping the eastern coastlineoftheYucatanPeninsula,Mexico,inLaFleur,RA,Groundwaterasa Geomorphic Agent,Allen and Unwin, Boston, Massachusetts, p. 281-293. Back, W.; Hanshaw, B.B.; Herman, J.S.;andVan Driel, N.J., 1986, Differential dissolutionofa Pleistocenereefin the ground-water mixing zoneofYucatan, Mexico:Geology,v.14, p. 137-140. Bakalowicz, M.J.;Ford,D.C.; Miller, T.E.; Palmer, A.N.; andPalmer,M.V.,1987,Thermalgenesisofsolution caves in theBlackHills, South Dakota:Geo logical SocietyofAmerica Bulletin,v.99, p. 729-738.Bogli,A. 1964,Mischungkorrosion,einBeitragzurVerkarstungsproblem:Erdkunde,v. 18, p. 83-92. Egemeier, S.1., 1981, Cavern development by thermal wa ters:National Speleological Society Bulletin,v.43, p. 31-49. Ford, D.C., 1989, Featuresofthe genesisofJewelCaveandWindCave, Black Hills, South Dakota:National Speleological Society Bulletin,v. 51, p. 100-110. Hill, C.A., 1987, GeologyofCarlsbad Cavernandother caves in theGuadalupeMountains,NewMexico and Texas:New Mexico BureauofMinesandMineral Resources Bulletin117,150p.Kastning, E.H., 1975, Cavern development in the Helder berg Plateau, East-CentralNewYork:New York Cave Survey Bulletin1,194 p. Mylroie, J.E., 1977,Speleogenesisandkarstgeomor phologyofthe Helderberg Plateau, Schoharie County, New York:New York Cave Survey Bulletin2,336 p. Mylroie, J .E., 1981, Pleistocene climatic variation and cave development:Norsk Geografisk Tidsskrift, p.151-156.Mylroie,I.E.,1988,KarstofSan Salvador,inMylroie, J.E. (editor),Field guidetothe karst geologyofSan90Appalachian Karst Symposium. 1991 Salvador Island. Bahamas: 10th FriendsofKarst Meet ing:Mississippi State University, p. 17-44. Mylroie, J.andCarew, J., 1990,Theflank margin model for dissolution cave development in carbonate platforms:Earth Surface Processes and Landforms,v. 15,p. 41342A. Mylroie, J.E. and Mylroie,J.R,in press, Meander cutoff cavesandself piracy:Theconsequencesofmeander incision into soluble rock:NationalSpeleological Society Bulletin.Palmer,AN.,1991, Origin and morphologyoflimestone caves:Geological SocietyofAmerica Bulletin,v.103, p. 1-25. Palmer,A;Palmer, M.; and Queen, M., 1977a, Speleo genesis in theGuadalupeMountains,NewMexico: Gypsum replacementofcarbonate by brine mixing,inFord, T.D. (editor),Proceedingsofthe 7th International Speleological Congress. Sheffield.1977:British Cave Research Association, Somerset, England, p. 333-336. Palmer, A.N.; Palmer, M.V.; and Queen, M.V., 1977b, Geology and originofthecavesofBermuda,inFord, T.D. (editor),Proceedingsofthe 7th International Speleological Congress, Sheffield.1977:British Cave Research Association, Somerset, England, p. 336-339. Palmer,AN.andPalmer, M.V., 1989, Geologic historyofthe Black Hills, South Dakota:National Speleologi cal Society Bulletin,v. 51, p. 72-99. Panno, S.V. and Bourcier, W.L., 1990, Glaciation andsaline-freshwater mixing as a possiblecauseofcave for mation in the eastern midcontinent regionofthe United States:Aconceptual model:Geology,v.18, p. 769 772. Plummer, L.N., 1975, Mixingofseawater with calcium carbonate ground water:Geological SocietyofAmerica Memoir142,p. 219-236. Proctor, C.1., 1988, Sea-level related cavesonBerry Head, South Devon:Cave Science,v.15, p. 39-50. Sanford, W.E. andKonikow,L.F., 1989, Porosity dev elopment in coastal carbonate aquifers:Geology,v.17,p.249-252.Stoessell,RK.;Ward, W.C.; Ford, B.H.; and Schuffert, J.D., 1989, Water chemistryandCaC03dissolution in the saline partofan open-flow mixing zone, coastal YucatanPeninsula,Mexico:Geological SocietyofAmerica Bulletin,v.101, p. 150-169. Vogel, P.N.;Mylroie,J.E.;andCarew,J.L., 1990, Limestone petrology andcavemorphology on San Sal vador Island, Bahamas:Cave Science,v.17, p. 19-30. Welch, L., 1989/90, Coldwater Cave 1990:Ge0 2 ,v.17, no. 2,3, p. 90.


Appalachian Karst Symposium, 1991 RubinModificationofPreglacial CavesbyGlacial Meltwater Invasion in East-CentralNewYorkPaulA.RubinR.D.1,Box 159 Feura Bush, NY 12067ABSTRACTPeriodsofhigh glacial meltwater have altered some preglacial cave-passage configurations. Floodwater and fossil karst features, whose formation cannot be explained based on available water from the surround ing watershed, are found superposed on actively forming cave passages. These features may be recognized through correlationofwatershed boundaries, peak-runoff observations through a cave system, the presenceofanomalous in-cave and surface features, and the geomorphic interpretationofthe areainquestion. Know ledgeofminimum ratesofkarstification may be usedtoinfer climatic conditions, making possible the reconstructionofthe hydrology associated with deglaciation. Clarksville Cave, situated in the hamletofClarksville, New York, provides an excellent exampleofinvasion by Wisconsinan meltwater on a preglacial cave system. Vadose developmentofa major partofthe explored cave has occurred preferentially aslant a thrust-fault ramp, often along a calcite bed/limestone contact created by pressure solution. Other fault-related features include slickensides, extension veins, fault bend folds, stylolites and the repeated basal Onondaga Limestone and impermeable Schoharie Formation thrust below the Onondaga Limestone stratigraphic column. An imbricate thrust eastofthe cave has upthrown the Esopus Shale against the Onondaga Limestone, forcing the developmentofan inefficient resurgenceatthe baselevel Mill Pond. During the Wisconsinan glacial stage, subglacial meltwater formed a seriesofnow abandoned bedrock channels and paleogorges that, due in part to topographic controls, found outlets along and over the flankofthe Helderberg Escarpment. Someofthis meltwater was pirated into Clarksville Cave where inefficient outlets resulted in the formationofhigher in-cave "intermittent phreatic" levels not controlled by the thrust fault. These levels abruptly truncate and grade to lower vadose passages. The characterofthese upper levels, the paleogorge and related caves, and elevated paleo-insurgence points correlate with described alpine karst settings.Physical SettingClarksville Caveisnestled under the flankofa low wooded ridge virtually in the centerofthe hamletofClarksville, New York (Figure 1). Itisformedinthe lower subunitsofthe Devonian Onondaga Limestone that were deposited approximately 380 million years ago. Its large passage size, up to 15 feet high and 40 feet wide, complete with multiple levels, makes it unique among other, usually smaller, Onondaga caves. The Clarksville area lies at an elevationof600to800feetms!. Itissituated within the foothillsofthe Helder berg Plateau, a partofthe Appalachian Plateau physio graphic province. Meyerhoff (1972) attributed thepresentdaydrainage patternofthis region to the normal erosive91processesofstream adjustmenttostructure. The Helder berg Plateau has been modified by stream incision, physi cal weathering, glacial and postglacial erosion, and deposi tion during the Cenozoic era (Dineen, 1987). Dineen (1987)hasdetermined that present-day drainage trends in the Hudson Valley were established before the Wisconsinan glaciation, sometime prior to 70,000 years ago. Glacial striationsintwo locations near Clarksville further indicate that taday's drainage was in place priortoinundation by the Wisconsinan ice sheet. The directionofglacial movement was almost exactly north-south(S13W), with a maximum ice thickness on the orderofone mile about 22,000 years before present (Dineen, personal communication). Late preglacial drainage along Onesque thaw Creek was probably little different from whatitis

PAGE 100

Rubin Appalachian Karst Symposium. 1991A' Esopus ShaloIri 1000Ifeel ::!:JmeltwaterttOW'dlfec:uon slOpe or ellnoI )"hll POnd Lo ..Row Condull =======paleoqotge hypothetiCaliocaoon 01cave passages .\.. "-.. ..........",",,0 ... 1" ....,..."..p./'0'" \;7<1'.... .. .... ..:...Figure I: Configurationofdrainage in the vicinityofClarksville, New York. Shown are Clarksville Cave and present and past routesofflow. For a detailed mapofClarksville Cave, see Figure 3ofthe following paper. 92

PAGE 101

Appalachian Karst Symposium. 1991 today. Evidence presented in this paper argues for pre Wisconsinan cave development.Structural SettingThe hamletsofClarksville, Tarrytown and Feura Bushhaveall been subjected to extensive faulting. Marshak (1986), Marshak and Engelder (1987), and Cassie (1990) discuss structural deformation within partsofthe HudsonValley Fol?-Thrust Belt (HVB). The HVB extends roughlyfrom Kmgston to Albany, New York, extending to a maximumof20kilometers east and westofthe Hudson River (Marshak and others, 1986). The deformation mayhaveoccurred during the Acadian (Cassie, 1990) or ABe ghanian orogenies (Geiser and Engelder, 1983), or duringMountMarion deposition (Murphy and others, 1980). Faulting and deformationofthe Esopus Shale, SchoharieFormation, and Onondaga Limestone, throughout the Clarksville area, may represent the farthest northwestern exposureofthe Hudson Val1ey Fold-Thrust Belt. The ex tensive structural deformation present throughout Clarks ville and the previously documented southern partsoftheHVBare ofdeformationofsedimentary rock under relatively low pressure and temperature conditions (Marshak and Engelder, 1985). Mappingofthe structuralorbedrock geologyinthe area, both on the surface and inthecave, reveals that faulting in the Clarksville area is characteristicofeither an imbricate thrust zone or a duplex.Atleast six elongate ridges, trending north-south ap proximately along strikeofthe faults, are unevenly spaced throughout the Clarksville area. They often exhibit exten fault-bend folding, slickensides, andinplaces an anti clmal structure. These limestone ridges, which are underlainbyoneormore basal thrust faults, can be mapped for distancesofup to one mile. One such deformed ridge, situatedat the eastern endofClarksville, has been breachedbyOnesquethaw Creek. Perhaps the most prominent exampleisfoundinthe upper Onesquethaw Creek gorge. The upper Onesquethaw Creek gorge exhibits the best out-of-cave exposureofthe repeated basal Onondaga Lime the impermeable Schoharie Formation (a quartz nch limestone) thrust below the Onondaga Limestone stra tigraphic column. Here muchofthe bedofOnesquethaw Creekisguided by fault-zone features. The thrust-faultramp,associated thick calcite bed, and fault-bend folds arethesame as those along which Clarksville Cave has developed, except that they are farther south along strike.Structural Features Influencing GroundwaterFlowinthe Karst AquiferFaulting in and immediately eastofClarksvil1e Cavehasresulted in thrusting, deformation, and upward move ment impermeable bedrock units underlying the Onon daga Limestone (Schoharie Formation and Esopus Shale)93Rubin into a position that makes the eastern escapeofground water impossible.Thegentle southwesterly dip of the bedrockoftheMiBPond aquifer (Figure 2) fails to directallsubsurface flow in this direction. Instead, significantly higher surface topography to the southwest(e.g.,Wolf Hill and Cass Hill) retards dissolution in this direction in favorofthe 1.3-degree apparent dip between Wolf Hill Dam and the base-level discharge pointatMill Pond. Tra cer studies generally verify this predicted flow path,atleast during periodsoflow discharge. However, tracer studies also document an unexpected easterly diversion ofmoder ateto high-discharge waters through Pauley A venue in Clarksville Cave. Thisissignificantly farther north than the Mill Pond. This easterly deflectionoffloodwaters may occur in response to an inefficient outlet and conduit lead ing to the Mill Pond. Pauley Avenue floodwaters flow easterly until they become perched ona thin bedofimpermeable Schoharie Formation that has been thrust below the basal or lower non-cherty subunitofthe Onondaga diver John Schweyen (personal communication) reports the presenceofthe Schoharie Formation overlying the lower non-cherty subunitofthe Onondaga Limestone approxi 700 westofthe north-south trending ClarksVilleCave.ThISnumber reflects a minimum westerly dis placementofbeds above the fault ramp. Aoodwaters re main perched, flowing down the apparent dip of the Scho harie Formation, until they encounter a fractured zone along a more steeply inclined partofthe fault ramp. Here, subsurface water is deflected sharplytothe south and aslant the strike and dipofthe inclined fault plane, with the pos sible localized exceptionoffollowing a horse for 200 feet northofthe Lake Room. Pirated surface water must riseatthe Mill Pond be cause the ir!1permeable Esopus Shaleisthrust upward agamst the cavernous Onondaga Limestone. The leading edgeofthis upthrown shale formation, an imbri cate thrust sheet separate from the fault zone that Clarks ville Cave formed along, trends roughly north-south (Fig ure 1). The Esopus Shale and thrust-fault-induced fault bend folds in the Onondaga Limestone, present slightly westofthe upthrown Esopus, ultimately form a wall or barrier to easterly karstic groundwater flow for a distance thatisweBinexcessofone-half mile. However, this geologic barrier has only retarded east ern groundwater movementintwo locations:(1)eastofthe known partsofthe Waterfall Passage and (2) at the Mill Pond spring resurgence. Formationofmostofthe north-south oriented cave occurred preferentially aslantaninclined rampofa thrust fault. Deformation along this thrust plane has produced a fault zone with at least three easily discerned, slickensided surfaces. Although separate, they occur within a few feetofone another. In muchofthe cave, this fault rampisaccented by oneormore promi nent calcite beds, often accompaniedbya zone of stylolites and calcite-filled extension veins. This calcite bedis

PAGE 102

Rubin Appalachian Karst Symposium, 1991400.w450.350 NewSalem w400900.1100 a '\ I III 7", I\II \. ,1375w,151111 ,;"\ \\I, '0 I:10,Al )..,--.'" .1577--WolfIHill,'S.Ii.buryISpr.I:1170II1500.\ \\ Coss' Hill,\B ",y' II\\1462..,\\,\.I ..... c, ''1'0\.,fa22,, /'7104000FEET1560 --oJ \\\\1 HelderbergEscarpment hillssculptedby IiIWIfJljlmeltwater" 1200"" ,,,".., 1790.. J I,I,f1710.\IJIIrI JI / 1670., "Figure2:Topography, drainage basins, and selected features along the Helderberg Escarpment, Albany County, New York.94

PAGE 103

AppalachianKarst Symposium.1991podiform in shape, with a central thickness ranging up to eight inches. Whereas the calcite forms a continuous be? of variable thickness aslant the strikeofthe thrust fault,itismost pronounced in the Gregory Sectionofthe cave, where the vadose partofthe cave is steeply inclined alongthefault ramp. The ledge forming the outletofthe WaterfallPassage is thesamebuff brown to weathered Schoharie Formation, with underlying calcitebed,as seenintheupper Onesquethaw Creek.gorge. The zones ofthecalcite bed, where dissolutIOn and crystalhzauon are greatest, coincide with the more inclined segmentso,fthe thrust ramp, where the stress was highest. remnant calcite blocks, up to eight inchesinthickness, 10 theWard's Sectionofthe cave provide the only evidenceoftheformer presenceofthe thick calcite bed. Ramsay(1980)providesevidencethatsimilar"extension veins are formed by an accretionary process in volving the formationofa narrow fracture followed by the fillingofthe open space by crystalline material, a nismtermed crack-seal." Such stress-induced chemical transfer,orpressure solutionofmaterials, seems relatively common (Ramsay, 1980).Thecharactenstlc crack-seal mechanismofrepeated tectonic stress (Ramsay,1980)is best illustrated in Onesquethaw 2 miles southeastofClarksville, where calcIte 10fJihng aslant the rampofa thrust fault reaches a maximum thicknessof27inches. Here, insight into the fault style and repeatedactivationinthe area is suggested by the presenceofmultiple calcite-vein infilling events along the fault ramp. Successive cracks often occur along vein-matrix contactsofa previously sealed crack system, because this is mechanically the weakest surface in the rock (Ramsay,1980).Fractures have been found to increase towards ma jor faults. The higher the fracture frequency, the higher the percentageofcalcite-filled fractures (Carrio-Schaffhauserand GavigI:o, 1990). It is a combinationofthis mechanicallyweaker calcite bed/Onondaga limestone boundary and related fault partings, all present along this inclined thrust ramp, that have served to orient the north-south ofClarksville Cave. Further structural and hydrauhc controlofcave-forming waters may also be locally attributedtoperching on a fault-thinned Schoharie Formation. Sim ilarly, muchofOnesquethaw Cave has developed down and along the mechanically weaker vein-matrix contact. Both caves exhibit characteristic fault-bend folds, stylolites, and extension veins adjacent to the prominent thrust plane.Ofmajor importance to the developmentofboth caveswasCenozoic structural deformation which provided a preferential solutional pathway along the inclined surface of a fault ramp. A steep hydraulic gradient was thus set up between infiltrating watersand their resurgence points along fault ramps. These faults may then be considered asbothnegative and positive influences on groundwater flow andcavern development: negativeinthe sense that downwarddissolution did not readily penetrate far below the fault zone (Kastning, 1977 and 1984), and positiveinthe 95Rubinsense that almost the complete trendofthe caves follows a structurally weakened zoneofincreased permeability.Intermittent PhreaticPassagesMeltwater invasionofpreglacial passages in Clarks ville Cave occurred during glaciation, significantly enlarg ing the cave and its tributary conduits within the aquifer. Observationofthe degreeofflooding within the cave dur ing major storm and runoff events reveals that only the lowest levelsofthe cave carry water along the fault zone. Two abandoned upper-level passages, both with relatively consistent ceiling elevations, were identified via a leveling survey. The levelofthese passages is determined by the relative uniformityoftheir ceiling heights. The highestofthese two upper-level passages extends from the Lake Room, through the Big Room, until its truncationinthe Pixie Passages immediately above the Corkscrew. (Note that a detailed mapofClarksville Cave is showninFigure 3ofthe following paper). This 714-foot level can roughly be characterized as the meandering upper levelofPerry Avenue. In places, the lower ceiling elevation is con trolled by chert beds. The 698-foot level extends from the Bathtub Passage through Upper Cook A venue, where the passage is truncated by breakdown, and where flow had been diverted down a steeply dipping, fault-plane-controlled tube leading to Lower Cook A venue. These large upper-level passages are generally high and dry, and are sometimes tubular in cross-section, sug gesting a water-table formation; they undulate upward and downward as is typicalofphreatic passages, and have formed with only limited influence from the fault plane. However, they lack the low hydraulic gradient typicalofphreaticorwater-table origin, are discontinuousinsize and extent and truncate suddenlyorbecome much reducedin area, grading systematically to active pre glacialvadose passages. Relatively smaller drain at the down-gradient endofthese passages comparedWIththe cross-sectional areaofthese floodwater conduits resultedintemporary phreatic conditions within the cave, quite dis similar from the conditionsofnormal phreatic water. Pal mer (1991) documents similar floodwater formationofconduits behind local passage impediments such as col lapse debris, insoluble beds,orsediment fill, where aggres sive water resultsinrapid passage enlargement. Forma tionofthese "intermittent phreatic" floodwater conduits, coincident with the direct influxoflarge quantitiesofsub glacial meltwater, occurredunder alpine karst conditions. A third solutionally developed upper level is also iden tifiableinthe cave at 739 feet msl. This discontinuous level is represented in only a short segmentofthe cave proximal to the Root Room (northeastofthe Lost Rock hammerRoom) and some nearby domes. It clearly repre sents the maximum flood level attained in the cave. Solu tion domesatthe 739-foot level are characteristic flood water-injection features. Like the two levels, it is systematically graded to the actively formlOg

PAGE 104

Rubin lowest level. These three abandoned levels are interpretedasreflectiveofdifferent glacial discharges (perhaps seasonalfluctuations) coursing through the cave, rather than dif ferent time-based developmental stages. Variable dischar ges, perhaps influenced byvariable outlet efficiencies and climatic conditions during glaciation, are inferred for for mationofthe levels. Thereisno obvious mechanism present today that would explain both the elevationofthese upper-level passages and their configuration, that exhibits little fault control. Other than the passage alignment, fault control appearstobe a significant developmental factor only in the low-discharge lower-level passages. The strongest argu ment for a preglacial originofthe vadose-level passagesinthe cave stems from the recognition that the mean annual precipitation and the sizeofthe Mill Pond watershed has probably not changed significantly in the last 10,000 years. Funk (1989), through the interpretationofarchae ological sites, established that climatic conditions within the last 10,000 years were at times either dryer or similartothatoftoday. Thus, the availabilityofthe significant discharges necessarytoform the upper-level passages was not there postglacially. These upper-level passages and related higher solution features on ceilings (upward to 739 feet) indicate that they are younger than many lower passages, having formed in response to aggressive glacial floodwaters behind an ineffi cient outlet. Itishypothesized that the lower-level Clarks ville Cave passages served as a natural in situ drainage systemfor glacial meltwaters underneath a warm-based Wis consinan ice sheet. Higher-level cave passages(e.g.,the 698and 714-foot levels) formedinresponse to the mas sive influxofsubglacial meltwaters behind inefficient, per haps partially ice-blocked outlets. Similarly, the forma tionofhigh-ceiling solution domes, anastomoses, pen dants, spongework-like dissolution, and diversion passages may be attributedtofloodwater invasion. The gradationofupper-level conduits tributary to the lower levelsofthe cave, from relict meltwater infiltration points, also lends supportive evidence for a preglacial originofthe linear(N12E) Clarksville Cave passage.RelictKarstA numberofrelict karst features are present both proximal to Clarksville Cave and to the northwest within the same watershed. These include a numberofsmall shafts and caves(e.g.,Trap Cave, North and Thook entran ces) that receive only minor amountsofdirect meteoric or snowmelt infiltration today. The most important relict karst feature is the Stove Pipe Paleogorge (Figure 1). This abandoned rockcut gorge grades directly into the pre glacial Clarksville Cave via the North Entrance, Brown's Depression area, Trap Cave, and the Thook Entrance. Its channeliswell defined for mostofits course. The mor phologyofupper reachesofthe gorgeischaracteristicofan ice-marginal meltwater channel with small-scale 96 Appalachian Karst Symposium, 1991 hanging valleys, rather than a well-graded stream bed which would be expectedofa former channelofOnesque thaw Creek. Sugden and John (1976) describethe ice-flow dynamics which cause favorable formationofdrainage routes in bedrock versus ice. In some places, the channel configuration is such that only large quantitiesofwater would have been capableoffilling the channel sufficiently high enoughtooverflow into sub-parallel channels. This paleogorge is sharply truncated to the north by a down wardly sloping limestone cliff. A negligible catchment area is present (Figure 2), certainly too small to carry any significant quantitiesofwater or sediment into the caveassuggested by thick sediment banks and upper-level phreatic passages. A second paleogorge, the Clarksville Paleo gorge, proximal to Osborn Cave, may represent either a preglacial drainage routeofthe Onesquethaw Creek prior to glacia tion or a channel carved around the Appleby Berm by gla cial meltwaters. Gorgesofthis nature can form in a rela tively short timeifsufficient abrasive material is carried through it. Von Engeln (1911) documents the rapid for mationofa rockcut marginal gorgeatthe outletofthe Hidden glacierinthe Yakutat Bay RegionofAlaska. Brown's Depression (739ft.msi) is an important lo cationinthat it received significant paleo-streamflowfromthe northwest (Figure1).Stove Pipe Paleogorge stream flow, originating from a vast subglacial watershed to the north, incised a channel through the Onondaga Limestone from tile northeast until itreached the Hunter's Fissure Cave and Diddly Cave area. Here, this paleostreamflow was responsible for the formationofthese caves. From Hunter's Fissure Cave the paleo-streamflow spread outtothe southeastoverthe gently undulating topography. However, its course was partially constrained by the eleva tionally higher surface topography to the west, north and east. Thus, muchofthe Stove Pipe Paleogorge stream flow was funneled southeast into the Brown's Depression/ North Entrance (above Lake Room) area, where it entered Clarksville Cave. During periodsoflow-tomoderate-glacial discharge, meltwaters converged proximal to Brown's Depression where muchofthe southeastern discharge was retarded from flowing east by a low north-south trending limestone ridge (747ft.msl). These meltwaters were pirated into Clarksville Cave through the Diddly Cave and Brown's Depression/North Entrance areas. Diddly Cave was recent ly dug open, increasing the length from5to550plusfeeLA dive push, a short distance into the cave, ledtoa master conduit which may soon be linked to Clarksville Cave. The exposed limestone pavement, coupled with the steep hydraulic gradient present between insurgence and resur gence points, provided a favorable avenue for subsurface piracy. High-discharge meltwater, with a glacial hydrostatic head, encountered the Clarksville Cave ridge and sought

PAGE 105

Appalachian Karst Symposium. 1991themost efficient outlet, flowing southward and over the ridge barrier. However, the even higher surveyed elevationsofa numberofabandoned surface stream infiltration points and their conduits leading to Clarksville Cave provideinformation on the large magnitudeofsubglacial dis charge necessary for their formation. Someofthese featuresinclude Trap Cave (753 ft. msl), the Thook Entrance(766ft.mst), and a deep solutionally enlarged joint neartheWard's Entrance (761 ft. mst). During periodsofhigh glacial discharge, meltwaters probably flowed both withinandoutside the Stove Pipe Paleogorge channel. This meltwater splayed outward around, and possibly over, the Appleby Berm. The alignmentofthe Thook Entrance pas sages and thePixiePassages suggests that meltwater coursing around both sidesofthe Appleby Berm soughttoenter the preglacial Clarksville Cave through the most direct pathway. Meltwater thus entered the Clarksville and Slove Pipe pal eogorges encountering the Gregory En trance to the cave, the Ward's Entrance, the Thook En trance, TrapCave, the North Entrance, Brown's Depres sion, and the jointed pavement above the cave.Sediment FillThick depositsofsedimentinthe cave provide direct evidenceofthe quantityofmaterial carried down the Stove Pipe Paleogorge by glacial meltwaters. The pointofentry of this material was largely through the Brown'sDepres sion/North Entrance area. The findingofsedimentsinthenewlydiscovered northwestern segmentofthe cave (pauley Avenue) also argues for input via Hunter's Fissure and Diddly Cave. Thick remnant sediments reveal that at leasttheWard's Sectionofthe cave was once sediment filled. Because the physical openingofthese sediment input pointsisbelieved to have been formed by subglacial melt waters, a sedimentation, passage infilling, and re-excavationhistory may be constructed. The thicknessofthe sedimentary column in contact withbedrock suggests that significant cave enlargementhadoccurred prior to sediment infilling, possibly during lIlinoian and/or Kansan glaciations. The basal deposits on bedrock include imbricate shale-clast-rich sediments with small cobbles, indicativeofrapid infilling. Once muchofthecave became filled, floodwaters stagnated, leaving their signature in finely laminated sand, silt, and clay layers. These layers are seen near the ceilinginthe Ward's Sectionandare interlayered with courser sediments in the Gregory Section (to at least the 714-foot level). At this time, bulk depositionofsedimentisinterpreted as occurring during earlyWisconsinan time, with re-excavation during late Wisconsinan or post-glacial times. Lackofsignificant sediment cover and fill in paleo-channels and abandoned in surgence points supports this hypothesis. Limited deposi tionofsedimentprobably also occurredinthe late Wisconsinan. Partial plugging by sedimentofClarksville Cave's overflow outlets may have contributed to the degreeof97 Rubin backflooding and upward passage developmentinthe cave. Evidenceisfound for this in large glacial cobbles cemented in a clay matrix now terminating The Hidden Room. Itispossible that sediments washing down the Clarksville Paleogorge partially or totally blocked the Gregory and Osborn Entrance overflow outlets for a periodoftime prior to being washed free again. Alternately, these outlets may have formed as a floodwater modification behind the ineffi cient Clarksville Cave low-flow outlet.Glacial GeologyTwo and possibly three glaciations are documentedasfar south as Corinth, New York(Lafleur,1991, un published report). Approximately 14,700 years bp, the Wisconsinan ice sheet receded from the Helderberg Plateau (DeSimone and LaFleur, 1985). Dineen (1986) gives an extrapolated bog-bottom dateof15,060,OOOyear bp for theGreatBearSwampsituatedsomewhatwestofClarksville. This date further confirms the timingofthe deglaciationofthe Clarksville area. DeSimone andLafleur(1985) provide a dateofapproximately 14,700 years bp for the recessionofthe Pine Swamp ice front from the Clarksville area. They depict the ice front as a lobe or tongue projecting southwardtoStuyvesant, New York, with Clarksville situated along the southwestern flankofthe ice margin. Dineen (1986) documents ice thinning during glacial stagnation over the Helderberg Escarpment. Large quanti tiesofmeltwater flowed southward proximal to the south western flankofthe Hudson Champlain Lobeofthe Scho harie ice margin. Dineen describes depositionofsedimentsinmultiple meltwater tunnels under stagnant ice. It thus appears that deglaciation from the Stove Pipe Road area was characterized by a southward thinning ice cover, with a southward meltwater flow direction. Free-surface flow was probably present in the paleogorge prior to the final retreatofthe Wisconsinan ice sheet from the Clarksville area.ImplicationsofWisconsinan ClimatesOfeven greater importance than the physical presenceofthe Stove Pipe Paleo gorgeisits relationshiptoHun ter's Fissure and Diddly Cave and the implication this has on interpretationofWisconsinan, and possibly Illinoian and Kansan climate. Hunter's Fissure and Diddly Cave formed along the abandoned Stove Pipe Paleogorge. The presenceofsmall scallop wavelengths in joint-controlled Diddly Cave indicates rapid streamflow along the baseofthe Wisconsinan ice sheet. Rounded stream cobblesinwalking-sized passagesinDiddly Cave provide clear evi dence that a large stream once flowed through the paleo gorge. The recent findingofbonesofa varying hare and the extinct passenger pigeon within clay depositsinthe cave may provide important scientific information on these species' recolonization duringorfollowing deglaciation, as well as datingofregional deglaciation. Additional field

PAGE 106

Rubinwork and radiocarbon dating are plannedinthis rare preser vationallocation. The Clarksville Paleogorge has probably not had a stream in it for the last 14,700 years, coincident with re treatofthe Wisconsinan ice in the mid-Hudson Valley (DeSimone and LaFleur, 1985). Furthermore, Wisconsinanice had retreated from the lower Hudson Valley 15,000 or 16,000 years ago (Connally and Sirkin, 1986; Dineen, 1986), with the ice front retreating to the St. Lawrence Valley by 13,000 years ago. Therefore, the maximum time frame for possible ice front stagnation in the Clarks ville area during active deglaciation is on the orderof1,000 years. Solutional cave formation will occur only where a pre-existing networkofintegrated openings connects the recharge and discharge areas (palmer, 1991). This is a process that requires a minimumof10,000 years (palmer, 1984; Dreybrodt, 1987, 1990; Palmer, 1991) before pas sageways obtain sufficient size for human entry. Addi tional passage cross-sectional size requires additional time. A shorter time period, on the orderof5,000 years, may be possible depending on joint widths present in the bedrock prior to infiltration by glacial meltwaters. Thus, subgla cial meltwaters apparently were not only responsible for the formationofDiddly Cave, but must have flowed for a minimumof5,000to10,000 yearsinorder for the cave to form. Because the maximum amountoftime the retreat ing ice front could possibly have stayed in the Clarksville area was on the orderof1,000 years, it follows that Diddly Cave, Hunter's Fissure Cave, the Brown's Depression area and other southern infiltration points were receiving melt water from below warm-based glacial ice for at least 5,000to10,000 years. Itislikely that temperate climatic condi tions were present during the early and later Wisconsinan.Meltwater Features inOtherNew York State CavesManyNew York State caves need to be re-examined for evidenceofglacial meltwater modification. Several caves in east-central New York exhibit features characteris ticofmeltwater invasion. Examples include Skull, Knox, Ella Armstrong, McFail's, Howe Caverns, SingleX,Schoharie, Gage, Onesquethaw, and Surprise (Mystery) caves, allofwhich have one or more passage segments su perposedabove passages receiving Holocene peak flood waters. For example, the upper levelsofSkull Cave have aragonite speleothems that are incapableofsurviving floodwater invasion. Similarly, caves such as Skull, Knox, and Ella Armstrong have watershed sizestoosmalltoaccount for the volumeofwater necessarytoform their observed vertical and areal extent. Other caves, such as McFail's and Surprise, carry underfit streams, yet exhibit anomalously large passage sizes. An artificially enlarged subglacial watershed would have been capableofproviding the necessary recharge. Related fossil karst includes aban doned and sometimes glacial-debris-covered sinkholes and 98Appalachian Karst Symposium, 1991shafts which once served as infiltration points. Anomal ous in-cave features such as abandoned pits and multiple level, ungraded passages may also reflect meltwater inva sion(e.g.,Surprise Cave). Similarly, significant cave development proximal to the headwatersofa drainage ba sin(e.g.,Gage Caverns and Phoebe Pit) may also reflect meltwater invasion from an expanded ice-sheet watershed. Other relict caves such as Knox, Salamander, several Sau gerties-area caves, and Joralemon's (Engel, personal com munication) are now abandoned and largely waterfree. Their derangement from active drainage patterns may por tray development during a previous interglacial period or, more likely, may be a resultofmodification by glacial meltwaters. The characterizationofpreglacial cave modificationbyglacial meltwater invasion poses many exciting geomor phic questions for researchers in New York State. Speleo themand sediment-dating techniques may shed lightonkarstic evolution and modification through threeormore glacialperiods. A complete geomorphic interpretation must include an assessmentofflow conditions and geolo gic featuresina defined watershed, both on the surface andinthe subsurface.ReferencesCarrio-Schaffhauser, E. and Gaviglio, P., 1990, Pressure solution and cementation stimulated by faultinginlimestones:JournalofStructural Geology,v.12,p.987-994. Cassie, R.M., 1990, Fault-related folding near Catskill, New York,Northeastern Geology,v.12, no.1,p.1927. Connally, G.G. and Sirkin, L.A., 1986, Woodfordian ice margins, recessional events, and pollen stratigraphy of the mid-Hudson Valley,inCadwell, D.H. (editor), The Wisconsinan Stageofthe First Geological District, Eastern New York:New York State Museum Bulletin455,p.50-72. DeSimone,DJ.and LaFleur, R.G., 1985, Glacial geology and historyofthe northern Hudson Basin, New York andVermont,inFieldTrip Guidebook:New York Geological Association, p. 82-115. Dineen, R.I., 1986, Deglaciationofthe Hudson Valley between Hyde Park and Albany, New York,inCadwell, D.H. (editor), The Wisconsinan Stageofthe First Geological District, Eastern New York:New York Slale Museum Bulletin455,p.89-108. Dineen, R.I., 1987, Preglacial and postglacial drainageofthe central Hudson Valley,in Field Trip Guidebook,New York State Geological Association,p.B-1B-30.

PAGE 107

Appalachian Karst Symposium. 1991Dreybrodt, W., 1987,Thekineticsofcalcite dissolution and its consequences to karst evolution from the initialtothe mature state:National Speleological Society Bulletin,v. 49, p. 31-49. Dreybrodt, W., 1990,Theroleofdissolution kinetics in the developmentofkarst aquifersinlimestone: A model simulationofkarst evolution:JournalofGeology,v. 98, no. 5, p. 639-655. Funk, R.E., 1989, Some contributionsofarchaeology to the studyofcave and rockshelter sediments: Examplesfromeastern New York:Man in the Northeast,no. 37,p.35-112.Geiser, P.A. and Engelder, T., 1983,Thedistributionoflayer-parallel shortening fabrics in the Appalachian forelandofNew York and Pennsylvania,inHatcher, R.D., Jr.;Williams,H.; and Zietz,1.(editors),contributionstothetectonicsandgeophysicsofmountain chains:GeologicalSocietyofAmericaMemoir158, p. 161-176. Kastning, E.H., 1977, Faults as positive and negativeinfluencesonground-waterflowandconduitenlargement,inDilamarter, R.R. and Csallany, S.C. (editors),Hydrologic Problems in Karst Regions:WesternKentuckyUniversity,BowlingGreen,Kentucky, 481 p. Kastning,E.H.,1984, Hydrogeomorphic evolutionofkarsted plateaus in response to regional tectonisminLaFleur, R.G. (editor),Groundwater as a Geomorphic Agent:Allen and Unwin, Inc., London, Boston, p. 351 382. Marshak, S., 1986, Structure and tectonicsofthe Hudson Valley fold-thrust belt, New York:Geological SocietyofAmerica Bulletin,v.97, p. 354-368. Marshak, S. and Engelder, T., 1985,Developmentofcleavageinlimestonesofa fold-thrust beltineastern 99RubinNew York,JournalofStructural Geology,v.7,p.345 360. Marshak, S., and Engelder, T., 1987, Exposuresofthe Hudson Valley Fold-Thrust Belt, westofCatskill, New York,in Geological SocietyofAmerica Centennial Field Guide,Northestern Section, p. 123-128. Marshak, S., and others, 1986, Structureofthe Hudson Valley Fold-Thrust belt between Catskill and Kingston, New York,in A Field Guide: PreparedforGeological SocietyofAmerica. Northeast Section, 21st meeting. Kiamesha Lake.N.Y.,70 p. Meyerhoff, E.A., 1972, Postorogenic developmentofthe Appalachians:Geological SocietyofAmerica Bulletin,v.83, p. 1709-1728. Murphy,PJ.;Bruno, T.L.; and Lanney, N.A., 1980, Decollement in the Hudson River Valley:Geological SocietyofAmerica Bulletin,v.91, PartI,p.258-262, Part II,p.1394-1415. Palmer, A.N., 1984, Geomorphic interpretationofkarst features,inLaFleur, R.G. (editor),Groundwater as a Geomorphic Agent:Allen and Unwin. Inc., London, Boston, p. 173-209. Palmer, A.N., 1991, Origin and morphologyoflimestone caves:Geological SocietyofAmerica Bulletin;v.103,p.1-21. Ramsay, J.G., 1980,Thecrack-seal mechanismofrock defonnation,Nature,v.284, p. 135-139.Sugden,D.E. andJohn,B.S., 1976,GlaciersandLandscape:Edward Arnold, London. Von Engeln, O.D., 1911,Phenomenaassociated with glacier drainage and wastage, with especial reference toobservationsin theYakutatBayRegion, Alaska,ZeitschriftfurGletscherkunde,v. 6,p.104-150.

PAGE 108

Plate Appalachian Karst Symposium. 1991PlateC:Passage leading from the Bathtubtothe Slickenside BlockinClarksville Cave, Albany County, New York(seeFigure 1ofRubin, this volume, page 103). The near-vertical fractureisajointThe inclined fracture is a major thrust fault along which muchofthe cave has formed. Slickensides on the fault plane are readily visible nearby. Note that this passage has formed along the lineofintersectionofthe fault and joint. These structures provided the initial avenues for groundwater flow resultinginlater enlargementofthe conduit to the dimensions visible here. View istothe northeastPhotograph by ErnstIi.Kastning.100

PAGE 109

Appalachian Karst Symposium, 1991Flow Characteristics and Scallop-Forming Hydraulics within the Mill Pond Karst Basin, East-Central New YorkPaulA.RubinR.D.1,Box 159 Feura Bush, NY 12067ABSTRACTThis is a studyofthe hydrologyofClarksville Cave and the headwatersofOnesquethaw Creek, situatedinthe hamletofClarksville, New York, specifically the Mill Pond karst basin. During mostofthe hydro logic year, water entering that partofthe watershed thatisdownstreamofthe Wolf Hill Dam is pirated into the Onondaga Limestone. Tracer tests and in-cave stream gaging indicate that extreme conduit conditions are present in the aquifer, with a maximum water velocity on the orderof5.3km/hr.It has been hypothesized that a submerged conduit must lie covered by breakdown blocks at the cave's northern terminus.Having established a knownpeakflow, a modified versionofthe Darcy-Weisbach equa tion was used to accurately calculate the minimum diameterofthis conduit.Knowledgeofthe structural geology throughout the watershed, coupled with a detailed leveling survey in the cave, permitted reasonable estimatestobe made for the two unknownsinthe equation. A submerged conduit was subsequently opened and explored. Scalloped cave walls are presentinPerry A venue at a key stream gaging location. BackOooding occurs behind inefficient passage constrictions a short distance downstream of, but not up to, this station. Evi dence exists that documents that only long return-interval flood stages cause backfloodingtothis station. This situation permits a reasonable estimateofthe maximum discharge and flow velocity responsible for scallop formation. Scallop wavelengths were measured below the elevationofpeak floodwaters. By input ting measured values for discharge and flow velocity into published equations, it was possibletoback calculate scallop Reynold's numbers that favorably correlate with measured flow velocities and discharges. A possible revisionofthe scallop Reynold's numberissuggested when itisutilizedinthe determinationofpaleo flow velocities. It also appears that scallop wavelengthispartially determined by the propertiesofthe rock comprising the wallsofthe conduit. RubinLocationandWatershedBoundariesA broad karst aquiferispresent in the Clarksville area.Itsboundaries extend north and northwestofMill Pond, situated less than 120 meters southofthe restaurant, June's Place(seeFigure 2ofthe preceding article). The farthest boundaryofthe Mill Pond karst basin lies about3.9kilometers to the northwest, proximaltothe Wolf HillDamon Onesquethaw Creek. The elevationofthe basin ranges from 1822 feet msl atop the Helderberg escarpmenttoapproximately 645 feet mslatthe Mill Pond. The boundariesofthe catchment basin are depictedinbold dashed lines. These boundaries were defined through theuseoflow-altitude stereo aerial-photography, U.S.G.S.101topographic maps, tracer studies and,inplaces, detailed structural geologic mapping. The Mill Pond watershed may be subdivided into two parts: A) that partofthe watershed located upstreamofWolf Hill Dam (1,245 hectares), and B) that partofthe watershed located downstream of Wolf Hill Dam (829 hec tares). The downstream partofthe Mill Pond watershed exhibits features characteristicofkarst terranes. These include sinking streams, limited surface drainage, solution ally enlarged joints, sinkholes, and the Clarksville/Diddly cave system. Structuraldeformation throughout the region has resulted in extensive jointing andfaulting, providing solutional pathways for infiltrating waters.

PAGE 110

RubinPiracyofOnesquethaw Creek WatersMostofOnesquethaw Creek downstream fromWolfHill Dam and upstreamofMill Pond is a losing stream, with a substantial amountofsurface flow lost to solution ally enlarged joints in the stream bed. During mostof the hydrologic year littleorno surface flow occurs in the area downstreamofWolfHill Dam, located nearly on the Marcellus Shale/Onondaga Limestone contact. Subsurface piracyofwater into the Onondaga Limestone belowWolfHill Dam occursvianumerous joints in the stream bed.Thewater briefly surfacesatSalisbury Spring, only to again sink into joints in the stream bed.Thevolumeofwater flowing in the stream bed and the relativeefficiencyofthe often partially sediment-choked joints governs the distance water may be found flowing on the surface down stream fromWolfHillDamand the Salisbury Spring. The greater the dischargeof the stream, the farther its flowiscapableoftraveling prior to complete subsurface piracy. During periodsoflowormoderate discharge, all Onesque thaw Creek surface flow is pirated into the karst network prior to where the bedof the Onesquethaw Creek passes beneath Rt. 443(seeFigure 1ofthe preceding article). Only large storm and snowmelt events generateenough surface flow in the watershed to cause Onesquethaw Creek to flow throughout its course. This represents a very small partof the hydrologic year. Surface-stream flow is short-lived even after major storm events.TracerTestsA seriesofuranine-tracer tests have permitted partial delineation under varying conditionsofdischargeof the subsurface flow paths throughout the Mill Pond drainage basin. Uranine is a non-toxic tracer frequently used in karst investigations (Smart, 1984).Itwasinjected into various joints in the Onesquethaw Creek stream bed that were pirating water. Activated-carbon detection bugs were placed at all likely resurgence points, collected later, and chemically elutriated with Smart solution (Quinlan, 1986). Tracer testing has revealed that dye injections from 3.2 km upstreamof the Mill Pond resurgence remain perched above the Upper Cherty Subunit foratleast 1.6 km before breaching chert beds that overlie the lower, more massive, non-cherty subunits. Water pirated into the Onesquethaw Creek stream bed immediately downstreamofWolfHill Dam and upstreamofRt. 85 remains inoneormoresubsurface conduits, until surfacing brieflyatSalisbury Spring, only to again sink intojointsin the stream bed downstream. Salisbury Spring is locatedon the western sideofRt.443,approximately 0.6 km southeastofRt. 85.Itisset back some distance from the road.Itis likely that piracyofthe Salisbury Spring discharge into the bedofOnesquethaw Creek is roughly coincident with the point at which this water breaches the Upper Cherty Subunitof the Onondaga Limestone. Thus, one major tributary con duit to the systemis likely to become physically impassa ble within 1.4 km northwestof the Lake RoominClarks102Appalachian Karst Symposium. 1991ville Cave (Figure 1). However, stream gaging and tracer studies indicate the presenceofa second low-flow conduit entering the known partsofClarksvilleCavefrom the large, heavily jointed watershed to the north-northwest. All subsurface flow resurgesat the Mill Pond. The relative inefficiencyof the outletof the Mill Pond conduitmaybedue to structuralproblemsresulting from the upward thrustingofthe impermeable Esopus Shale against the cave-bearing Onondaga Limestone(seeprevious paper).Thepresenceofimpermeable Esopus Shale in the bed of OnesquethawCreekatand immediately downstreamofMill Pond forces all subsurface flow from the karst aquifer to surfaceatMill Pond. This author established a gaging station downstreamofthis point(seeFigure 2ofthe pre ceding article). Tracer tests and discharge measurements throughout the watershed indicate that during periodsoflow discharge, pirated Onesquethaw Creek waters do not travel through Clarksville Cave. Surface and subsurface stream gagingandtracer tests establish the intersectionof the pirated Onesquethaw Creek low-flow conduit with the Clarksville Cave low-flow drainage conduit to be located between the southern endof the cave and Mill Pond(seeFigure 1ofthe preceding article). Hereafter, the conduit that resurgesatMill Pond and is physically separate from the Clarks ville Cave conduit north ofOsborn Cave, is referred toasthe MillPondlow-flow conduit. Although the exact elevationofthe lowest drainpointin Clarksville Cave remains to be surveyed, it lies slightly below an elevationof660 ft msl. The hypothesized flow routesofunentered partsof the network are portrayed in Figure 1of the pre ceding article. Tracer tests verify thatafter a certain critical subsur face discharge is reached, coincident with piracyofincreas ingly greater amountsofsurface flow into the subsurface conduit system, the efficiencyof the MillPondlow-flow conduit is exceeded and surplus water is shunted to the Lake Room in Clarksville Cave. The Mill Pond low-flow conduit utilized today, which bypasses Clarksville Cave, maybe the original flow route, with the flow route leading to the Lake Room (via Pauley Avenue) forming as a flood water-overflow route. Alternatively, the flow path to the LakeRoom may represent the original subsurface flow route that was later abandoned due to further stream piracy, possibly coincident with loweringof the regional base le vel. Under this genetic interpretation, diversionofwaters from the MillPondlow-flow conduit to Pauley Avenue would occur behind an immature drain. During periodsofbase flow, it appears that only water from the north-north western partof the Mill Pond watershed rises in the Lake Room. Moderate and high discharge in the subsurface causes a significant backupofwaterbehind the Mill Pond low-flow conduit, resultinginlarge overflows to the Lake Room.Thegreater the flow in Onesquethaw Creek, the more water is lost through joints in the stream bed, and the greater is the discharge that appearsin the Lake Room.

PAGE 111

E;'"("');:s-:os.g""j;:'......\Q \Q...... NorthEn!.r Peny Nt -6..,(1 HAlFSCAlEOFMAJNMAP 250 leel '""'X'" ,,,..,,"',, [.0"""""""0.',:', ,l ," \..0_/(/:/)"".Q I r.0,,C1."-""J0""" c_ W ""'"""..... ". .."""'-'=::\,1," 0"'"'.:'"",,,,"'U......(:?5hCitdr". . ,-'\.."""""...,.,,,"=\' S SI ... UllipIlUI CLARKSVILLE CAVE (Ward-Gregory Cave) Osborn to the Lake Room Albany Co., NYJBruwn.T. W FI'll:"C.1\.Flmsf.:Ilm
PAGE 112

RubinFrom the Lake Room the water flows south through the cave where someofit joins, in a tributary manner, the Mill Pond low-flow conduit somewhere between the down stream endofthe cave and Mill Pond(seeFigure 1ofthe preceding article). As the dischargeoffloodwaters within Clarksville Cave increases, the hydraulic efficiencyofthe branched conduit leading to Mill Pond is exceeded. The remaining water thatcannotbe handled by Clarksville Cave's low-flow subsurface conduitandthe Mill Pond low-flow conduit backs up within the cave as temporary storage. After a critical flow on the orderof2.7 cfs is reached, excess floodwaters are discharged along the Osborn Cave overflow route (677 ft msl) to the surface. Osborn Cave is situated directly southofthe Gregory entrance and is physically connected to Clarksville Cave by a water filledconduitFigure 1ofthe preceding article shows the Osborn overflow channel, that sometimes carries large quantitiesofwater.KarstBasinCharacterizationTracer tests conductedinpartsofthe Mill Pond aquifer reveal that all subsurface waters reappearorres urge at Mill Pond. During periodsoflow flow, all surface and ground water downstreamofSalisbury Spring and upstreamofthe bridge crossing Rt. 443 discharge through conduitsinthe karstic aquiferatMill Pond. During periodsofmoderatetohigh-subsurfacedischarge, partofthe subsurface flow is shunted through Clarksville Cave. All flow throughout the Mill Pond watershed thus surfaces eitherinMill Pond or Osborn Cave, where, for muchofthe year, it comprises the headwatersofcontinuous surface flowofOnesquethaw Creek. At times this flow is supplemented by water from the Clarksville South Road and western Bennett Hill Road subwatersheds. Water in Onesquethaw Creek, from that partofthe watershed upstreamofWolfHill Dam that is not artificiallydiverted to the Vly Creek Reservoir, also sinks into the subsurface downstreamofWolfHill Dam. Muchofthe flowinthe karst network originates as diffuse infiltration outside the Onesquethaw Creek corridor. Virtually all meteoric water and snowmelt contacting the heavily jointed, generally thin-soil-mantled limestone pavement within the Mill Pond watershedispirated into subterranean limestone conduits. Geologically, water entering the solu ble Onondaga Limestone must stay within it, becauseitis underlain by approximately 1mofthe Schoharie Forma tion (a quartzitic limestone) and approximately 30 mofimpermeable Esopus Shale. Physically unentered segmentsofthe conduit network may be envisioned as being similar to a tree, where all branches coalesce downstream toward the trunk. Palmer (1991) describes such branch work caves as the most com mon type. Water infiltrating from different segmentsofthe aquifer's recharge area converges as higher-order pas sages that decreaseinnumber and generally increaseinsize in the downstream direction. It is likely that the large104Appalachian Karst Symposium, 1991northwestern partofthe Mill Pond aquifer is branchwork in nature, with many tributaries coalescing downstreamtoward larger, master passages. It is also likely that seg mentsofthe conduit system directly underlie the bed of Onesquethaw Creek, whereas others extend far to the north west. Still other segments must enter from the northwest where runoff from the Marcellus and Hamilton beds ofWolfand Cass hills sinks near the Onondaga Limestone contact and is rapidly pirated into the system. Recently, exploration via the newly opened Diddly Cave entrance yielded approximately 0.5 kmoflarge stream passages extending north into the Mill Pond karst basin. These passages area branchwork in character, andifconnectedtoClarksville Cave would bring the cave's length to greater than 2 km. The dashed lines on Figure 2ofthe preceding article portray a simplified versionofthe hypothesized configurationofconduitsinthe eastern endofthe system.SubsurfaceTravelTimesThe combined flow from stream losses and diffuse fracture infiltrationisdocumented as moving very rapidly through the karst system. Although effort has not been made to absolutelyquantify the rateofsubsurface flowinthe aquifer, the timingoftwo tracer tests provides some insight on the situation. Under moderate flow conditions present on February 23, 1990, uranine tracer was injected into ajointin the bedofOnesquethaw Creek, 3.2kmnorthwestofMill Pond. At this time, all surface flowinthe upper reachesofOnesquethaw Creek was being pirated into this joint. Tracer-detection bugs were collectedfromClarksville Caveat4:00p.m. on February 24, 1990, about 22 hours after the tracer injection. All were positive for uranine. Thus, a subsurface groundwater transit time in excessof150 meters per hour was documented. A similar trace was conducted in October 1988 under low-flow conditions. In this instance, the tracer injection and sinkingofthe stream occurred farther northwest than during the above trace. In this second test the tracer-detec tion bug was removed from a location proximal to Mill Pond 27 hours after tracer injection. After elutriation,thedetection bug was positive for uranine. A subsurface groundwater transit timeinexcessof120 meters per hour was documented for low-flow conditions. In contrast, during a timeofpeak flow within the aquifer (March 15, 1986at1:45 a.m.), the discharge and velocityofflow within Clarksville Cave were measured. The velocity was recorded as 1.48 meters per second. This equates to 5,328 meters per hour (5.3 kms/hr) and may be considered as indicativeofthe peak velocityofpotential groundwater movement within the aquifer andofextreme conduit conditions. At timesofpeak flow, groundwater may move from end to end through the karst aquifer, a dis tanceofapproximately 3.9 km,inless than one hour. The rapid hydraulic response to significant precipita tion or snowmelt within the watershed has been repeatedly

PAGE 113

Appalachian Karst Symposium. 1991documented with stream hydrographs both in ClarksvilleCaveandinOnesquethaw Creek. Subsurface conduit flowinthe Mill Pond aquifer is roughly analogous to open channel flow in a surface stream. A thin soil-moisturebankover muchofthe watershed's limestone pavement further permits rapid infiltrationofmeteoric waters andsnowmelL,thus bolstering subsurface transit times. Flood pulses throughout the karstified system are flashy, providingevidenceofmature conduit development. Rapid flow characteristics present within the 2,074-hectare Mill Pond watershed, especially that part downstreamofWolfHillDam,make it and Onesquethaw Creek extremely sensitivetoinfiltrationofcontaminants. During muchofthe hydrologic year, discharge fromMillPond acts as the sole sourceofwater to the upper reachesofOnesquethaw Creek. During periodsofbasenowthis discharge has been gaged at less than 0.1 cfs.Therecent zoningofland centraltothe karst aquifer asruralcommercial may have severe effects on both the aquiferand Onesquethaw Creekifuntreated waste streams or septic infiltration are permitted (Rubin, 1990b).In-CaveandOnesquethawCreekFlowCalculationsMeasurementsofdischarge and streamflow velocityhavebeen made periodically in Clarksville Cave since1983.Over 99%ofthe water flowing through ClarksvilleCaverisesinthe Lake Room. This water has been gaged during both low and high flowatdischarges ranging between 0.002 andIIIcfs. A maximum water depthof63.5cm was measured during the stormofMarch 15,1986.Discussions with Ed Gregory revealed that the flood discharge componentinthe cave, associated with the 1938 failureofthe Helderberg Lake Dam, was significantly greater than the above maximum-gaged amount. Gregory reported that floodwaters were ponded toanelevationofapproximately 719feet msl, a short distance down the en trance slope inside the Ward's Entrance. The elevationofthecave passageinupstream Perry A venue, approximately11meters southofthe Lake Room, lies between 715 and708feet msl, thus indicating that allofPerry A venue was nooded during this event. Confirmationofthis flood level,and possibly anotherin1903, is manifestedina thickmudfilm covering historic names and dates chiseled nearthepassage ceiling. A gaging station was establishedinOnesquethaw Creek(seeFigure 2ofthe preceding article) in ordertoexamine the relationship between in-cave discharge and surface-watershed discharge. This was monitored twice daily for15months, more frequently during flood events,andperiodically for 4 years thereafter duringmajor runoff events. Stream discharge was gagedat13different stages. Curvilinear regression was then utilized to establish a seriesofmulti-order equations that could be usedtocorre late stage height with discharge.Thegreatest discharge recorded for Onesquethaw Creek during the courseofthis 105Rubinstudy was approximately 1337 cfs. This occurred on March 15, 1986at3:00 am following heavy rains(-7.0em) ona 38-centimeter snow pack. Temperaturesupto 40 F accompanied the coastal stormofMarch 13-14, 1986. Daily monitoringofstream stage in Clarksville Cave for the same IS-month period revealed that a direct correlation exists between this discharge and thatinOnesquethaw Creek. Approximately 8 percentofflood-peak dischargeinOnesquethaw Creek flows through Clarksville Cave. Knowledgeofexpected flood-return intervals and their magnitude in the Mill Pond karst basin was foundtobe essential to both the interpretationofhow abandoned upper-level passages in Clarksville Cave formed and anunderstandingofthe dynamics controlling scallop formation. The limited data for statistical comparison among hydro logic years in the Mill Pond karst basin necessitated exam inationofanother roughly comparable basininordertoas sess flood-return intervals. The farthest headwater gaging station on Schoharie CreekatPrattsville was selected. Many inherent differences occur between the basins, nota bly elevation, geology, regolith thickness, size, and loca tion. The Prattsville and Onesquethaw Creek gaging sta tions are approximately 48 kilometers apart. However, the Prattsville and Mill Pond watersheds are comparable under conditionsofa saturated soil-moisture bank, high runoff, and similar storm systems. Eighty-two yearsofdata were examinedatthe Prattsville, New York station. A Log-PearsonType-III and Gumbel-distribution statistical comparisonofhistoricpeakflowofSchoharie Creek gaging data with this study's hydrograph informa tion for Onesquethaw Creek indicates that the largest Onesquethaw Creekpeakofrecord (March 15, 1986) has a return interval on the orderof30to47 years. This corre sponds to a Prattsville hydrologic-year peak discharge of 54,900 cfs. Thus,if40 years was the expected flood return interval, 25 floods ofthis magnitude could be expec ted every 1000 years. Reconstructionofthe 1903peakdischargeatPrattsville (approximately 63,000 cfs), the highest on record, further reveals that a cave discharge well in excessofIIIcfs may also have occurred in 1903. The 1903 discharge, three standard deviations greater than the mean-annual Prattsville peak flow, has a predicted flood return interval on the orderof47 to 90 years. Although the magnitudeofthis flood was largerthanthe 1986 flood, it probably was not as greatasduring the dam-failure floodin1938. These infrequent storm or runoff events reasona bly represent a near-maximum quantityofwater available in the watershed under ideal, thin-soil-mantled, rapid-infIl tration conditions. Therefore, itisdifficulL to explain the "intermittent phreatic" upper-level passages in Clarksville Cave without a substantially greater quantityofwater. A subglacially enlarged watershed,asdiscussed in the preced ing article, appearstobe the only viable explanation.LakeRoomSubmergedConduitMeasured, statistically predicted, and inferred(e.g.

PAGE 114

Rubin mud-covered historic names and dates) high discharges rising from the Lake Room indicated that an conduit was present that must be capableoftransmlttmg large discharges. It was thus hypothesized (Rubin, 1989) that a submerged conduit must lie covered by breakdown blocks below the water surface in the Lake Room. By making a numberofreasonable assumptions, it was possible to calculate the minimum diameterofan assumed circular conduit capableofdischarging a given flow. A modified versionofthe DarcyWeisbach equation was successfully utilized to examine the sizeofthe, until recently, unentered upstream segmentsofcave conduit, northofknown partsofthe cave. Calculations were con finedtoa circularconduit capableofdischarging betweenIIIand 222 cfs(Q).The latter value was considered a rea sonable approximation for the 1938 dam-failure discharge. A friction factor(f)of0.1 was used. The two unknownsinthe equation were the change in head (elevationofwater upstreamofthe lake versus the elevationofthe lake, and the lengthofflooded passage upstreamofthe lake(L)during flood events. A wide rangeofvaluesof1.5 to 30 meters, and 6to1219 meters were tested, respectively, for these unknowns. Although someofthe values tested were likelytobe extreme in nature, they were selected based on knowledgeofthe structural geology within the watershed, coupled with a detailed leveling survey throughout the cave.Itwasbelievedpossiblethatsignificantback flooding might be occurring behind the Lake Room breakdown. Insertion in the modified Darcy-Weisbach equationofa reasonable rangeofvalues for the changeinhead and the lengthofflooded passage suggested that the minimum meterofa circular-conduit tributary to the Lake RoomISbetween 0.6 and 2.4 meters. Recent excavation and pene trationofa formerly blocked and water-filled conduit ex tending north and westofthe Lake Room verified the,cal culations (Rubin, 1990a). The lengthofthe water-filled passage was found to be61meters. The actual valueisprobably no more than 3 meters. The smallest dIameter foundinthese newly discovered passages was 1.4 meters. Maximum cross-sectional area found in the approximately 366 metersofconduit beyond the Lake Room that have been entered thus far is on the orderof8 square meters. These passage dimensions attest to mature conduit developmentinthe carbonate aquifer within the catchment basin. The assumed friction factorof0.1was found to accur ately reflect the flow conditions through the Lake Room breakdown. Approximately two vertical metersofclean washed angular breakdown, interspersed with minor quanti tiesofrounded glacial cobbles, were excavated. The heter ogeneous mixtureofbreakdown blocks rangedinsize from106Appalachian Karst Symposium. 1991 several centimeters in length, width and heighttoapproximately one meter. The water's approach angle, toward the lake surface, rises at approximately 30 degrees for the last 3 meters before reaching an irregular constriction (1.2 meters by 0.5 meters) in the breakdown. Prior to the last 3 meters, the submerged conduitisgenerally horizontal. The maximum conduit depth below the surfaceofthe lake was found to be approximately4.3meters.Scallop-Forming HydraulicsPhasesofthis study focused on defining the Mill Pond karst basin, the relationship between flow in the karst aquifer versus that in Clarksville Cave, the expected return intervalofpeakdischarge inside andoutside Clarks ville Cave, and flow conditions peculiar to Clarksville Cave. Specifically, a rangeofstream discharges and velo cities were measured in an air-filled segmentoflinear pass age and rectangular cross section. Water depth was record ed, as well as scallop wavelengths within the zoneofthe30to 47-year flood-return interval. Paleoflow information that researchers hopetorecon struct, based on scallop wavelengths and dimensionsofanabandoned passage,iseither empirically measured or rea sonably constrained. Measurements in Clarksville Cave permitted a cave-specific evaluationofBlumberg and Curl's (1974) scallop Reynold's number. One potential problem with characterizationofthephysical conditions under which scallops form is defining the discharge, or range thereof, responsible for scallop development. It was possible to define minimu":l and maximum discharge limits leadingtoscallop formation at a key stream-gaging location in Perry A venue. Here, cave wallsina fossiliferous sparite are scalloped. Streamflowacross the widthofthe cobble floor does not become deep enough, orofsufficient dischargetoform scallops until the waterisapproximately18centimeters deep. The 30 to 47-year flood (111 cfs)ofMarch 15, 1986 resultedina stream depthof63 centimeters, but decayed in 34 hoursto10 cfs with a stream depthofless than13centimeters. Backflooding occurs behindinefficient passage constrictions at the Big Room, approximately 120 meters downstreamofthis same key stream-gaging location. The levelofbackflooded waters, as measured on March15,1986 at the stream'ssurface, was only48centimeters lower in elevation than ponded water at the Perry A venue gaging station. A small additional discharge amount, such as that probable in 1903,orcertainly in the 1938 flood, would have substantially reduced the stream velocity here and its ability to form scallops. Thus, theIIIcfs measured on March 15, 1986 represents a value thatisclose to the maximum possible for discharge capable of forming scallops at the gaging station. This situation allows for a reasonable estimateofthe maximum discharge and flow velocity responsible for scallop formation.

PAGE 115

Appalachian Karst Symposium. 1991Statistical analysisofClarksvilleCaveflood-return intervals indicates thatcavedischarges in excessoftwo standard deviations about the mean-annual peak discharge maybesufficiently short-lived andofinfrequent recurrencetoform the observed scallops.Thefrequencyofshorter term floodintervalsisgreater, andperhapsitis these events which are recorded as scallops rather than very short duration, high-dischargelong-retum-interval floods. Al though the shorter-retum-interval floods are also relatively short-lived, itmaybethe combinedcontacttimeofwater with bedrockofmanysimilar magnitude floods that isofimportance.Therelationshipamongstream depth, dis charge, and flood-return interval in the watershed, as par tially indicated in the tablebelow,suggeststhatscallop formation in ClarksvilleCavemayoccurduring flood in tervals that range betweenoneand two standard deviations of the mean-annual peak discharge. Scallop wavelengths were measured below the elevationofpeak floodwaters. By inputting measured dischargeandflow velocity numbers into published equations, it was possible to back-calculate scallop Reynold's numbers that favorably correlated with measured flow velocities and discharges.Thisprocedure involved measuring scallops, stream flow, and stream velocity and examining the likely rangeofscallop-formingconditionsutilizing published equations.Forthe rectangular Perry Avenue conduit: The Sautermeanwas used tocalculatemeanscallopwavelengthsofscallop groups within 63.5cmofthecavefloor:(Curl, 1974)Rubinsionless Reynolds number: where: v=mean velocityoffluid flowing past scallop in em/sec L=mean scallop length incmp=densityoffluid == 1.0gm/cmfor 5C and 10 C Tl = fluid viscosity == 0.015gm/cm/secfor 5C and == 0.013 gm/cm/sec for 10 CThus,examiningthe specific flow conditions in a), b), and c) above(seeTable1)using L32 = 7.49 cm and Tl = 0.015gm/cm/sec.,arangeofsite specific Reynold's numbers was obtained:a)NR = 34,254 b) NR = 48,186c)NR = 73,801 Curl (1974) provides the limiting geometry for a rec tangular cave passage:Byinserting therangeofNR'S above into Curl's equation,wecan examine N*R,the scallop Reynold's number based on a rangeofactual flow velocities:a)N*R = 2,341b)N*R= 3,293c)N*R=4,337Blumberg and Curl (1974) derived a universal constant for the scallop Reynold's number, based on plaster model studies,of2200. Based on this study, it appears that N*R may actuallynotbea constant,butinstead maybestbe characterized by a rangeofvalues. These values, based on this cave-specific studyofa rectangular conduit, appear to be from 1 to 2 times the acceptedconstantEmpirical ob servationofthe flow dynamics in Clarksville Cave, cou pled with a characterizationofflood-return intervals within thecatchmentbasin,suggestthat ascallopReynold'snumberon theorderof3300 might fit the cave-specific conditions. It is possible that constants in accepted equa tions may lead to an underestimateofpaleoorrecent flow velocities and discharges. Further studiesofthe actual hy drologic conditions in which scallops form are warranted.ItshouldbenotedthatBL,anotherconstantin the Reynold's number equation (which deals with wall rough-Arangeofin-cave flow conditions was examined.Thethree flowconditionspresentedinTable1brackettheminimum andmaximumstream discharge and velocity believed to be responsible for scallop formationatthe key Perry Avenue gaging station. "The scallop Reynold's number,N*R based on friction velocity, is a universal constant for scallop formation andwasdetermined from model experiments (Blumberg and Curl, 1974) tohavethe numericalvalueN* R= 2200" (White, 1988). Scallop formation is controlled, inpart,by a dimenChannel Stream Discharge Velocity Width (m) Depth (em) (cfs) (em/sec)a)3.3 18.3 1468.6b)3.327.23096.5c)3.363.4111147.8 Table1:Threeflow conditions in Perry Avenue, Clarks ville Cave. 107*Thewall material subject to scallop formation may influence thevalueofthe scallop Reynold's number. Different typesofsurfaces, like limestone, ice, and plaster, may respond di(ferent-Iytowater scour.

PAGE 116

Rubin ness) was accepted on face value.AcknowledgmentsHeartfelt thanks are extendedtothe many northeastern cavers who have contributed to the various activities asso ciated with the studyofthe Clarksville/Diddly Cave sys tem. Different aspectsofthe project have included stream gaging, tracer tests, photography, leveling, digging, div ing, surveying, drafting, and wall scrubbing. Kevin Downey and Kevin Harris deserve special thanks for their many hoursofappreciated cave photography. Clayton Pauley, now deceased, was a true friend with whomIspent many fine evenings leveling through Clarksville Cave's labyrinth. Special thanks go to Thorn Engel, the unsung hero whoisalways there to help survey, stream gage, draft maps and formulate ideas. The studyoftheMillPond karst basin is northeastern caving at its finest.ReferencesBlumberg, P.N. and Curl, R.L., 1974, Experimental and theoretical studiesofdissolution roughness: JournalofFluid Mechanics,v.75, p. 735-742. Curl, R.L., 1974, Deducing flow velocity in cave conduits from scallops: National Speleological Society Bulletin,v.36, no. 2,p.1-5.108Appalachian Karst Symposium. 1991 Kastning, E.H., 1975, Cavern development in the Helder berg Plateau, East-Central New York: New YorkCaveSurvey Bulletin 1, 194 p. Palmer,AN.,1991, Origin and morphologyoflimestone caves: Geological SocietyofAmerica Bulletin,v.103,p. 1-21. Quinlan, I.F., 1986, Water-tracing techniques in karstterranes,revised extract from Practical Karst Hydrogeology, with Emphasis on Groundwater Monitoring: National Water Well Association, Dublin, Ohio. Rubin, P.A., 1989,ClarksvilleCave1990:TheNortheastern Caver,v.20, no. 4, p. 122-125. Rubin, P.A., 1990a, Breakthrough in Clarksville Cave: The Northeastern Caver, v. 21, no. 1, p. 3-8. Rubin, P.A., 1990b, Clarksville CaveandMill Pond Watershed Critical Area Definition: Report to theTownofNew Scotland, 25 p. Smart, P.L., 1984, A reviewofthe toxicityoftwelve fluorescent dyes used for water tracing: National Speleo logical Society Bulletin, v. 46, p. 21-33. White, W.B., 1988, GeomorphologyandHydrologyofKarst Terrains: Oxford University Press, New York.464p.

PAGE 117

Appalachian Karst Symposium, 1991 Palmer and PalmerReplacementMechanisms among Carbonates, Sulfates,andSilica in Karst Regions: Some Appalachian ExamplesArthurN.Palmer and MargaretV.PalmerDepartmentofEarth Sciences State UniversityofNew York College at Oneonta, NY 13820-4015ABSTRACTEvery carbonate rock formation contains examplesofreplacement among carbonate minerals, sulfates, and silica. Although the net geomorphic effect is rarely significant, porosity and permeability canbegreatly modified in this way. Recognitionofthese processes is also a powerful tool for interpretingpastgeochemical conditions. Examples canbeobserved in many areasofthe Appalachians: in caves, paleo karst zones, and carbonate oil reservoirs. Some mechanisms are well known, but their recognition in the geologic record is not. Replacementofsulfates by carbonates usually involves common-ion effects,e.g.where dissolutionofgypsum by calcite-saturated water causes precipitationofcalcite. Evidence for former evaporites includes sutured, lath-shaped, double-terminated and lozenge-shaped calcite crystals, scimitar and anastomotic veins, multi-stage breccias,boxwork, nodular or cauliform textures,andauthigenic carbonate sediment. Dolomitization and dedolomitization dependon the relative solubilitiesofcalcite vs. dolomite. Below 23 C, dolomite is more soluble than calcite, but the relationship is reversed at higher temperatures.Thesluggish kineticsofdolomite near saturation make it unlikely for dedolomitization tobea major karst process, except in sulfate-rich solutions, which greatly boost the solubilityofdolomite. Dedolomitization, recognized by scattered rhombsofcalcite,istherefore another hintofformer sulfates. Silica easily replaces carbonates and sulfates. Carbonates and silica dissolve and precipitate under the opposite conditions: risingpHand temperatures increase the solubilityofsilica but decrease the solubilityoflimestone and dolomite. Silica replacementofeither carbonateorsulfateminerals is usually very selective, so itisunlikely that such a broad-scale process as coolingofhigh-temperature fluids couldbethe main mechanism. Closed-system dissolutionofcarbonates isolated from carbon dioxide sources can raise the pH well above 9, allowing much silica to dissolve. Aeration, evaporation, and exposuretolocal acidity cause silica to precipitate.Theseconditions are common in aerated caves and in zones rich in sulfates or organics.IntrOductionCertain karst-related porosity and rock textures cannotbeaccounted for bysimpledissolutionofcarbonates. Replacementofonemineral by another is responsible formanyofthese features, but evidence canbeobscure. Only through recognitionofdiagnostic textures and mineral associations can these processes be reconstructed. Porosityisusually generated becauseofdifferential solubilityofthe various mineralsorby changes in volume caused by con trastsinmineral density. Thispaperoutlines thecommonreplacementpro cessesinkarst regions, describes some Appalachian exam109 pies, and summarizes the petrographic evidence for mineral replacement. The processes described here include:(I)do lomitization and dedolomitization; (2) replacementofgypsumoranhydrite by calcite,orvice versa; and (3) silicifi cationofcarbonate or sulfate minerals. All three are com monly associated with breccia and vuggy porosity caused by the recrystallization, dissolution, and movementofevaporites.Field DataMuchofour recent field work has been in the western United States, where late Paleozoic carbonate rocks have

PAGE 118

Palmer and Palmer Appalachian Karst Symposium. 1991Camp'sGulfCave, TennesseeFigure I: Location mapoffield areas described in this paper: A=Camp'sGulfCave, B=Jefferson City, C=Elmwood, D=gas fieldofeastern Ohio. Zinc minesofeastern Tennessee( JeffersonCity) are located in early Ordovician limestonesoftheKnoxGroup in the vicinityofa lower Ordovician paleokarst. Similar mines in central Tennessee(e.g.atElmwood), though notinthe Appalachians,arein the same strati graphie horizon and provide additional information.Thelimestone has been brecciated, dolomitized, and rendercd highly porous. Sulfide minerals line manyofthe voids. Although sulfates are absent, breccias and bedrock contain many textures associated with vanished evaporites (M.Palmer and A. Palmer, 1989). In some respects they resem ble thoseofthe sub-paleokarst zones in the Madison Limestoneofthe northern Rockies. There are two stagesofbreccia, only the later of which has been mineralized. Breccia bodies are intercon nected, irregular, and discordant to the beddingofthecarbonate rocks, with diameters on a scaleofmeters tomanytensofmeters. They resemble evaporite-induced breccia bodiesinsub-paleokarst zonesofthe Madisonofthe nor thern Rockies (A. Palmer and M. Palmer, 1989). The later ore breccia is more extensive than the earlier oneandfollows roughly the same trends. Sphalerite and other sul fidesinthe later breccia are Mississippi Valley-type hydro thermal ores (Kyle, 1983; Ohle, 1985).Themineralized breccias post-date the paleokarst surface, extendingbothbelow and above it. Although the breccia texture resem bles thatofevaporite breccia (with bedrock clasts "float ing" in a matrixofdolomite and other minerals), thereisno concrete evidence that the late breccias were causedbyevaporite-related processes. The mineralizing fluids caused solution and dolomitizationoflimestone and disaggrega tion ("sanding")ofdolomite. Sulfide ore both replacesthcbedrock and occupies pre-ore void spaces. Limestoneinthe early breccia has been partly replaced by quartz, someofwhich contains shrinkage cracks, that suggest an origi nal amorphous phase. In contrast, certain euhedral quartz crystals are doubly terminated Herkimer "diamonds",andothers are pseudomorphs after anhydrite needles and areoriented in flow patterns typicalofformer evaporites. Someofthe early breccias have been dolomitized, especially those in the Valley and Ridge Province. Chert containing criss-crossing silica lathslines the bottomsofsome orc zones in the mines. Allofthesecharacteristics are typicalofformer evaporite zones, suggesting that the breccias themselves are at least partly the resultofevaporite disrup tionofcarbonates by expansion, flow, and replacement. Unmineralized, bedded brecciainthe Knox form extensive aquifers throughout many eastern and east-central states.Sulfide Zones inCarbonatesof EasternTennesseeand calcite. It appears that exposure to freely circulating groundwater has favored the replacement. We had previ ously observed nodular quartz veins in Mammoth Cave, Kentucky, but until the replacement process was actually caught in theactin Camp'sGulfCave, their originwasjustspeculation.N.C.DOHIOKY.IND.cTENN.been modified by presentorformer evaporites. Concur rently we have encountered several examplesofporosity and mineralization in the Appalachians, someofwhich clarify the mechanisms observed in the western examples, and some that would not be understood without the more extensive evidence from the West. Partofthe western work is summarized in an article on the Black Hills cavesofSouth Dakota (A. Palmer and M. Palmer, 1989). Field sitesinthe Appalachians are located in Figure 1 and are described below. Becauseofthe consistent relationships between the replacement features and the geologic setting, it is possible to extend the ideas presented here to many other similar locations. Camp'sGulfCave, in Middle-Upper Mississippian carbonate rocksofthe Cumberland Plateau, Van Buren County, Tennessee, contains unusually extensive cave breakdown.Thecollapse zones provided viewsofcar bonate rockatdistances from theoriginal solutional cave wall ranging up to several tensofmeters. Veins and no dulesofwhite material have the appearanceofevaporites, but instead they consistoffibrous quartz and calcite. Ma nyofthe veins are thick and either discordant to the lime stone bedding along fracturesorconcordant to the bedding. Other veins are thin, with an interconnected "chicken-wire" pattern. Needle-shaped anhydrite inclusions are abundantinallofthem, suggesting that they were once evaporite bodies that have been replaced. In places the veins and no dules consist entirelyorpartlyofgypsum and/or anhydriteinvarious stagesofreplacement by the less soluble quartz110

PAGE 119

AppalachianKarst Symposium.1991Palmer and PalmerGasReservoirs in OrdovicianCarbonatesofEastern OhioIn 1987wewere contractedbyStoneResourceandEnergy Corp. (Worthington, Ohio) to investigate the originand distributionofporosity in the Ordovician gas field of the AppalachianPlateauin eastern Ohio.Theexactlocation and geological setting are proprietary.Tosupple ment available information, a10-cm-diametercorewasdrilled through thepayzoneata depthof2053-2071 m. Porosity is limited to specific stratigraphic horizons andhastextures diagnosticofformer sulfates. Porosity con sistsofvugs, intercrystalline voids, and small-scale brec cias. The vugsarecauliform, representing former evapo rite nodules,andarepartlyfilled with sanded dolomite. White nebulousmassesofdolomitearesurroundedbyporous dolomite. Anhydrite, locally replaced by quartz, occludessomeofthe pores. Chips from other wells in the most productive zones consistofdolomiteandauthigenic quartz crystals containing tiny dolomite rhombs with raggeddissolved edges.Thechips also contain large amounts of unoxidized pyrite. miteisalso volumetricallymoresoluble than calcite be low40C.Thesaturation values for dolomite are ratherapproximatebecauseofthe difficultyofachieving dolo mite equilibrium in the lab.Theions liberated by dissolutionofdolomite include thoseofcalcite. Therefore, calcite approaches saturation asdolomitedissolves,evenintheabsenceofsolid calcite.Below20C,calcitebecomessupersaturated as dolomite approaches saturation.Puresolutionsofdolomite are rare. Naturalwateralmostalways encounters a combinationoflimestone and dolomite,sothe degreeofcalcite saturationisgenerallyhigherthanineitherrockalone. Karst groundwater is typically less saturated with dolomite than with calcite (Thrailkill, 1977).Astemperature increases, bothdolomiteand calcitebecomelesssoluble. Dolomite solubility decreases at a faster rate,sotheopportunityfor dolomitization is en hancedathigh temperatures, particularly since the normal ly sluggish kineticsofdolomiteprecipitation would be speeded. Dolomitization and dedolomitization involve the following reaction:Geochemistry2CaC03+ Mg++ <=::::) CaMg(C03h+ Ca++(1)The porosity and rock textures in these field examplesarethe resultofmineral replacements and selective removaland precipitationofmineralsofcontrasting solubility. Several mechanisms are responsible, allofwhich relateto the original geologic setting.Byunderstanding the reac tions involved, it is possibletointerpret former geochemicalenvironments. with an equilibrium constant, equivalent to (Ca-rty(Mg-H), ofroughly 1.67at25 C.Thisequation doesnotapplyifthe waterisundersaturated with the replacing mineral. Replacementofonemineralbytheotheris often achieved simply through dissolutionofoneand later precipitationofthe other. In such a case the process isnotconsidered true dolomitizationordedolomitization. Although dolomite and calcite dissolveatroughly thesamerateatlow saturation ratios, the dissolution rateofdolomite drops dramatically below thatofcalcite as satura tion is approached (Herman and White, 1985). Because dolomite dissolvessoslowly near saturation, dedolomiti zationmustbevery slowornegligible in carbonate wa ters, and absent entirelyattemperatures above20C. Table I: Relative solubilityofcalcite and dolomiteatPe02=0.01 atm. Each mineral is considered independently, with no mixtureofthe two. The values for dolomite are less accurate than those for calcite. The saturation indexofcalcite is shownatdolomite saturation.Thesaturationconcentra tionsofcalcite and dolomite increase greatly withPe02'butthe calcite51values do not change significantly.Calcite DolomiteTemp(C)mmol/LImg/LIceILmmol/LImg/LTceIL SIcalcile 0 2.34 234 0.0864 1.6 2910.10+0.1010 1.94 194 0.0717 1.32300.081+0.0620 1.611610.0595 0.99 180 0.064 +0.01401.111110.04080.66120 0.043 -0.08Relative SolubilityofCalciteandDolomiteInthe following sections, geochemical relationshipswereobtained by a home-made iterative computer program (SOLEQUIL) designed to calculate carbonate equilibria in the presenceofother solutes. It uses the equilibrium con stants and their temperature dependence recommendedbyPlummer and Busenberg (1982) for calcium-carbonate so lutions andbyLangmuir(1971) for dolomite, and thermo dynamic values givenbyWagmanand others (1982) for other chemical species.TheextendedDebye-Huckel equation is usedtocorrectforactivity-concentration relationships. All relevant ion pairsareincluded in the calcula tions. Calcite and dolomite have approximately equal solubility in groundwater, but the exact relationship depends on the way in which sol ubility is measured.Asshown in Table 1 if calcite and dolomite are dissolved Iyat temperatures less than60C,dolomitehasa lower molar solubilitybuta larger solu bilityintermsofmasspervolume. Dolo-111

PAGE 120

Palmer and Palmer Appalachian Karst Symposium,1991 0.6 Dissolved Gypsum,moles/L w...U.J U ---wto_-------------------:001ATM--Peaz:.01 6 o PCOz =.001ATM_---+0.2 +0.4; -0.2 t<'"-0 CWhen waters at or near saturation with respecttocal cite encounter gypsumoranhydrite, the saturation ratioofcalcite increases becauseofthe presenceofthe common ion Ca++ (Figure 2). Calcite usually precipitates as gyp sum dissolves. From Table 1, it appears that dedolomitization (simul taneous replacementofdolomite by calcite)isfavored by low temperatures, with dolomite dissolving incongruently as calcite precipitates(c/.Evamy, 1967). However,inthe presenceofcalcium sulfate, dolomite is much more solu ble than calciteatall temperatures (Figure 3). Mg++ is depleted becauseofits strong affinity for sulfate ions, forming the ion pair MgS040. Ca++ is also depleted by forming CaS040, but not enough to offset the increase in Ca++ from gypsum or anhydrite. Therefore the saturation indexofcalcite rises much more rapidly than thatofdolo mite as gypsumoranhydrite is addedtothe solution.DissolvedGypsum,moles/L0.60'----.0....<0-1--.0.J..02--.-00L3--.0....l0-4--.0.J..0-5 --......J006Open-Systemvs.Closed-System Carbonate DissolutionIfthe water then reaches an aerated zone such as acaveor emerges at the surface, it encounters aPea2higherthanthe negligible values shown in Figure 4. CO2 israpidly absorbed, and there is a sudden drop in both pH and(perhaps surprisingly) C03=. The effectsofthese changesaredescribed in the next two sections.ReplacementofCalciteand GypsumbySilicaThesolubilityofsilica (either crystallineoramor phous) accelerates greatly as the pH rises above 9(seeDrever, 1982, p. 92). The high pH values and availabilityofsilica where water infiltrates through alternating sandstones and carbonates is favorable to dissolutionofmuch Si02 When the water encounters an aerated zone such as acave,or a zoneoflocally low pH, such as in an organic orsulfate/sulfide-rich zone, silica precipitates (Figure 4). This processismost effective at low initialPea2'which would be typicalofthe loose sandy soils that develop on sand stone.M.Palmer (1986) has observed replacementofcalsandstone before it encounters the carbonate rock, thewateris effectively cut off from itsoriginalC02source anddissolves the carbonate rockina system nearly closed to C02 (A. Palmer, 1987). Equilibrium pH can rise above9,whilePea2drops nearly to zero (Figure 4). The C03= concentrationisan orderofmagnitude greater thanintheopen system. This effect is reduced if the water becomes partly saturated with carbonate before it reaches thesandstone. Figure3:Contrast between calcite and dolomite saturation causedbydissolved gypsum (or anhydrite)ina solution initially 70% saturated with dolomite. Ionic ingredientsofcalcite are contributed by the dissolutionofdolomite.Saturation index=log (IAP/K.) for calcite and log(lAP/I<)O.5for dolomite(tomake the two SI values comparable).3050UNOERSATURATIONSUPERSATURATION Figure2:Increase in calcite saturation caused by dissolved gypsum (or anhydrite). Initial solutionis70% saturated with calciteatzero sulfate concentration. Saturation index=log (IAP/K). Where infiltrating water passes through soil directly into carbonate rocks, the water approaches saturation with respecttothe bedrock atornear the highPea2valuesofthe soil. Equilibrium pH is around 7.5-8.5,Pea2is high, andC03=activity is low.If,on the other hand, the water passes through a permeable but insoluble rock such as Dedolomitization is more likely to occur in the presenceofdissolved sulfates and at low temperatures, as shown by the contrast in solubilities between dolomite and calcite in Figure 3. The effectofcalcium sulfate on dedolomitization has been noted previously by Yanat'eva (1955) and DeGroodt (1967) but attributed to the high Ca/Mg ratio forcing calcite to replace dolomite, as in Equation1. t< +0.2 '"-0 C C00 0Ul-0.2 u0U-0.4112

PAGE 121

AppalachianKarst Symposium.1991PalmerandPalmer1.Ifcarbonate rock=pure limestone:sandstonecarbonaterock2.Ifcarbonate rock=pure dolomite: C 0.0005 atm 8.29 3.6mg/LlAx C0.0005 atm 8.17 3.6mg/LlAx B9x 107 9.95.9 0.0101B5.4x107 10.14 7.5 0.069A0.001 5.37 3.6 0.0066 A 0.001 5.373.60.0066 soiI:::':,' "'.':\.i'.:/ .: :..: I -L-1 I1__ L -1\-nBII I /--I\I t I/I caveI __ IIIIFigure4:Chemical evolutionofdiffuse vadose waterinquasi-closed conditions with respect to CO2 at 10 C. Water passesthroughsandstone, becoming isolated fromC02in the soil, then encounters carbonate rocks. Capillary and gravitational seepage into a dry cave causes a sudden increase inC02.Pe02ofsoil and cave is assumed tobe0.001 and 0.0005 atm (typical of caves in such settings). Saturation with carbonate bedrock is assumed by the time the water reaches pointC.C s = saturation concentration. Amorphous silica behavesina similar manner to quartz, but with a higher C s To achieve saturation at B, quartz must be available as insolubles within the carbonate,orcarbonates mustbeinterbedded with the sandstone belowA.citeinsmall amounts by silica in weathering rinds in thewallsofcertain caves overlain by quartzose sandstone. Silica solubility also increases with temperature and therefore, if present, silica may be precipitated by rising, cooling groundwater. Precipitation should be more widespreadthan by the mechanism described above and wouldnotnecessarily depend on local changes in pHoron re placementofother minerals.Carbonate/Sulfa te ReplacementGypsum and calciteareable to replace each other, depending on the (C03=)I(S04=) ratio: anhydrite by a carbonate-rich water would boost the Ca++ activity enough to precipitate calcite by the common-ion effect alone. The replacement would not take place on the molecule-by-molecule basis suggested by Equation 2. Where the water emerges into an aerated zone,C03=decreases rapidly, and the (C03=)I(S04=) ratio plummets. As a result, the ratio may easily drop below the threshold for calcite replacement, causing gypsum to replace calcite (A. Palmer, 1987). This is almost always valid only where evaporation is sufficient to drive gypsum to super saturation. Most gypsum crusts on cave walls represent an actual replacementofcalcite, rather than simple depos itsofgypsum on the limestone surface.PetrographyThisreaction has an equilibrium constantof103 .85at 25 C(seealsoWhite, 1976).If(C03=)/(S04=)> 103 .85, calcite will tend to replace gypsum. The opposite holdstrueat ratios less than 103 .85.A similar equilibrium canbeSlatedfor anhydrite and calcite, with an equilibrium conSlantof10-4 0 at 25 C. These statements are valid only ifthesupposed replacing mineral is supersaturated. Where water reacMs equilibrium with dissolved car bonatesinthe quasi-closed conditions describedinthe pre vious section,andthen encounters sulfates, the great C03= activity helps to keep the(C03=)/(S04=)ratiohigh,which favors the replacementofgypsumoranhydritebycalcite (A. Palmer, 1987). Dissolutionofgypsum or Recognitionofreplacement textures is best accom plished by microscopic examination. The following para graphs outline someofthe most diagnostic clues10the occurrenceofcarbonate-sulfate-silica replacement Dolomite bedrock is generally considered to be large scale replacementofpre-existing limestone. In the early stages, euhedral dolomite crystals replace calciteina scat tered manner and are sprinkled throughout the limestone. The crystal shapesofthe dolomite are independentofthe crystals being replaced and do not produce pseudomorphsofthe replaced mineral. Extensive dolomitization destroys mostofthe original sedimentary texturesofthe original limestone. Sharply defined, scattered rhombs are nearly always dolomite rather than calcite. Evaporites commonly113

PAGE 122

Palmer and Palmeroccur in dolomite bedrock because dolomitization also occurs in evaporitic marine environments. Dedolomitization leaves the edgesofdolomite crystals shredded and irregular, often with selective replacementofconcentric zones within the crystals. Common byproducts include dolomitesand(where the process has not pro gressed far), pink calcified clumps, and iron-oxide residues. Evidence for former gypsumoranhydrite is common, as dedolomitization is most vigorous in a sulfate-rich envir onment. In some cases dolomite is replaced by gypsum, which is later replaced by calcite. Quartz may replace gyp sum withinorbetween the original dolomite crystals, and when the carbonateorsulfate minerals are dissolved, the residue is very porous, forming a friable meshofquartz. This situation is typicalofthe bedrock in boxwork zones. Quartz is a great mimic, takingonpseudomorphic texturesofthe minerals it replaces. Replacementofanhy drite crystals often produces needle-shapedorlath-shaped quartz crystals. This provides themostconclusive evi dence for former evaporites. Early quartz replacementofbedrock is common in arid meteoric environments, and so quartz is often found replacing evaporite minerals.Inmeteoric water, calcite typically replaces evaporites. Euhedralpseudomorphsarerare,butmonoclinic and lozenge-shaped crystalsofcalcite appear to mimic former evaporite crystals. Calcite more commonly replaces large areasofformer evaporites, forming sheetsofsutured spar with banded inclusionsofinsoluble residue and wedgesofbedrock, floating dedolomite and dolomite sand. Clasts appear to "float," and calcite fills anastomotic veins that have been repeatedly fractured. Calcite eventually replaces evaporite-cemented breccias and former evaporite nodules. Gypsum and anhydrite are so soluble that they usually disappear entirely in near-surface humid environments. Their former presence must be inferred from relict textures such as thoseinFigure 5 (M. Palmer, 1988).Themost diagnostic textures are: pseudomorphsofevaporites after quartzorcalcite, lozenge-shaped calcite crystals, doubly terminated calcite and quartz crystals (Herkimer diamonds), breccia that includes upward-wedged clasts, crackle breccia (in which the fragments have barely moved relativetoeach other), fossil filaments ofiron-fixing bacteria, multi-stage calcite veining, scimitar-shaped and anastomotic veins, boxwork, dedolomitization, sutured-clast contacts, cauli form and nodular textures, and authigenic carbonate sedi ment.Nosingle feature is conclusive, but in combination their evidence is overwhelming.ConclusionsMuchofthe apparently karst-related porosity and min eralization at depth in the Appalachians is actually relatedtoformer evaporites. Observationsofmineral replacement and differential solubility help to interpret the origin and distributionofdeep porosity zones. It also allows more114Appalachian Karst Symposium.1991B C 0 0 EO A F t: H a Figure 5: Small-scale and microscopic textures associated with former evaporites now replacedbyotherminerals. Scalesareapproximate. A=Anastomotic calcite veins, wedge-shaped sliversofbedrock and semi-circular cracks (width=1 em); B=needles and lath-shaped pseudomorphsofcalciteorquartz after anhydrite (length=100microns); C=nodularorcauliform textures, (width=2 em); D=lozenge-shaped calcite crystals (length=500microns);E=doubly terminated quartz crystals with inclusions (lenglh=200microns); F=sutured stylolitic patterns (length=3em); G="chicken-wire" anhydrite texture (width=1em);H=dolomite clasts (shaded) assimilated by poikilotopic calcite spar after gypsum. Ragged bedrock remnantsanddolomite sand convertedtodedolomite and assimilatedbyspar (width=1mm). accurate projectionofpay zones, interpretationofthelocalgeochemical history, and stratigraphic correlation. Ifreplacement is not recognized, the original rock composition cannot be determined.ReferencesDeGroodt,K., 1967,Experimentaldedolomitization:JournalofSedimentary Petrology,v.37,p.1216-1220. Drever, J.I., 1982,GeochemistryofNatural Waters:Pren tice-Hall, Englewood Cliffs, NJ,399p. Evamy, B.D., 1967, Dedolomitization and the develop mentofrhombohedral pores in limestones:JournalofSedimentary Petrology:v.37,p.1204-1215. Herman, J.S. and White, W.B., 1985, Dissolution kineticsofdolomite: Effectsoflithology and fluid flow veloci ty:Geochimica et Cosmochimica Acta,v. 49, p. 2017 2026. Kyle, J.R., 1983, Economic aspectofsubaerial carbon ates,inScholle, P.A.; Bebout, D.G.; and Moore, C.B.

PAGE 123

Appalachian Karst Symposium, 1991 (editors.), CarbonateDepositionalEnvironments: American AssociationofPetroleum Geologists, Tulsa, p.73-92.Langmuir,D., 1971, The geochemistryofsome carbonate ground waters in central Pennsylvania: Geochimica et Cosmochimica Acta,v.35, p. 1023-1045.Ohle,E.L., 1985, Breccias in Mississippi Valley-type deposits: Economic Geology,v.80, p. 1736-1752.Palmer,A.N., 1987, Gypsum replacement by alternatingopenand closed systems in the vadose zone: Cave Re search Foundation.1986Annual Report, St. Louis, p. 27-28.Palmer,A.N. and Palmer, M.V., 1989, Geologic historyoftheBlack Hills caves, South Dakota: National Spe leological Society Bulletin,v.51,p.72-99.Palmer,M.V., 1986, Siliceous crusts on cave walls (ab stract): National Speleological Society Bulletin,v.48, p.39. Palmer,M.V., 1988, Influenceofformer evaporites onSouthDakota caves (abstract): National Speleological Society, Programof1988Annual Convention, Hot Springs, SD,p.31.Palmer,M.V. and Palmer, A.N., 1989, Paleokarstofthe 115 Palmer and Palmer United States, in Bosak, P.; Ford, D.C.; and Horacek,I.(editors), Paleokarst: A Systematic and Regional Re view: Prague, Elsevier and Academia, Amsterdam and Prague, p. 337-363. Plummer, L.N. and Busenberg, E., 1982, The solubilitiesofcalcite, aragonite, and vateritein CO2-H20 solutions between 00 and 900C and an evaluationofthe aqueous model for the system CaCO}-C02-H20: Geochimica et Cosmochimica Acta,v.46, p. 1011-1040. Thrailkill, J., 1977, Relative solubilitiesofcalcite and dolomite, in Tolson, J.S. and Doyle, F.L. (editors.), Karst Hydrogeology, International AssociationofHydrogeologists, 12th Memoirs, p. 491-500. Wagman, D.O. and 7 others, 1982, The NBS tablesofchemical thermodynamic properties: Selected values for inorganic andCIand C2 organic substances inSIunits: JournalofPhysical Chemistry. Reference Data11,Supplement2,p. 1-392. White, W.B., 1976, Cave minerals and speleothems, in Ford, T.D. and Cullingford, C.H.D. (editors),TheScienceofSpeleology: Academic Press, London,p.267-327. Yanat'eva,O.K.,1955, Effectofaqueous solutionsofgypsum on dolomite in the presenceofcarbon dioxide: Akad. Nauk. SSSR. Dok/ady,v.101, p. 911-912.

PAGE 124

Jameson Appalachian Karst Symposium,1991Fracture Controls on GroundwaterFlowand Cave Development in Northern Greenbrier and Southern Pocahontas Counties, West VirginiaRoy A. JamesonDepartmentofGeology and Geophysics UniversityofMinnesota Minneapolis, MN 55455ABSTRACTFractures fulfill a numberoffunctions as structural controls. They guide initial flow paths; promote the developmentofblind solution pockets, joint spurs, fissures, or anastomoses along the perimetersofexisting conduits; and promotein situdisaggregationofbedrock.Byproviding zonesofstructural weakness, fractures also promote collapse, which modifies passage morphology and leadstoenlargementofconduitsifbreakdown is removed. In northern Greenbrier and southern Pocahontas counties, West Virginia, bedding plane partings guide the majorityofearly flow paths. Most prominent guiding bed partings have anastomoses and are at contacts between argillaceous lower units and purer overlying units. N 60-75 E set joint", generally inen echelonzones, guide early flow paths in the Union Limestone, particularly in settings subjecttofloodwater influences, where the joints also guide blind joint spurs and promote distinctive breakdown features. N 30 45 E set joints locally guide early flow paths, mostly in the lower Greenbrier Group, but the N 60-75 E set joints are much more important where both are present. N 30-45 E set joints guide abundant blind pockets and joint spurs or bellsinthe lower Greenbrier Group, especiallyinbase level conduits near the Greenbrier River. Thrust-fault ramps from argillaceous units or prominentbedpartings, are common, and guide many early flow paths. Thrust faults provide fracture zones subject to extensive collapse, resulting in large collapse chambers and lengthy breakdown-filled canyons that readily transmit water but thwart cave exploration and mapping. Distinctive patternsofconduit growth associated with thrust faults include inclined anastomotic mazes, isolated vadose trenches, and isolated retreating canyons. Wedge-shaped breakdown is associated with many thrust fault zones. Crestsofanticlines have nonnal faults and high an gIe extension joints that locally guide early flow paths, have associated solution pockets, and produce slab shaped breakdown.116

PAGE 125

Appalachian Karst Symposium. 1991 SasowskySurface and Subsurface Drainage Basin Asymmetry: Ramifications for Karst Development in the Appalachian PlateausIra D. SasowskyDepartmentofGeosciences The Pennsylvania State University University Park, PA 16802ABSTRACTBasin asymmetry is a geomorphic parameter that quantifies disparate size and growthofa drainage basin.Forsurficial drainage, the Asymmetry Factor(AF)is determined by delineating the master stream, measuring the arealOthe rightofit (AR) and thelOtalbasin area (A1').AFis then given by: AF=100 (AR/ AT). AnAFvalue>50 indicates excessive areainthe sub-basin on the right. This study introduces two related geomorphic parameters useful in fluviokarst settings, Subsurface Drainage Density (Dsd) and Subsurface Asymmetry Factor (SAF). Dsd=Cave length / Basin area, for either the whole or a partofthe basin. SAF=l00(CLR/CL1')where CLR and CLT are the total lengthofknown caves for the right sideofthe basin and the total basin, respectively. A strongly karsted sub-basinofthe East Fork Obey River, Fentress County, Tennessee was chosen to explore the relationship between these parameters and karst development. This basin is ideally suited for this analysis, having a strong asymmetry and a large numberofknown caves (99).Thebasin is developed on the Cumberland Plateau, and is underlain by undeformed sedimentary rocksofmixed lithologies. Local relief is 340 meters, with sharply incised valleys dissecting the flat plateau surface. Regional dip is 0.5 degrees to the ESE. The sub-basin has a total surface areaof184km2 and an AFof29; strongly biased to the left. TheSAPis oppositeofthis,87,indicating that most cave passage occurs on the right. In this basin, AP is structurally determined. Dip slope (consequent?) streams on the left side initiated within larger sub-basins, and were ablelomaintain/expand them, while right-sided sub-basins descended the free face, and were limited in their catchment. Subsurface karstification, as indicated by the SAP, has been promoted on the right side by subsurface captureofthe master drainage, probably under structural control. AP andSAPmay serve as time-integrated proxies for surface and subsurface erosionrales.When adjustedloinclude only those caves relatedlotributary drainages, the SAP is 22, close to the AP (29) This suggests a possible relationship between the lengthofcaves developed and catchment area. Such a relationship would be expected from mass balance considerations.IntrOductionThe Appalachian Plateausare home to many wellknownfluviokarsts such as the Elk River and Greenbrier County karst terranesofWest Virginia, the Carter CavesandCaveCreekareasofKentucky, and RyeCove,Virginia. In theseareas, streams have incised through a mixtureoflithologies exposing speleoliferous carbonates. Abandoned caves are leftinthe valley walls, and the valley 117 bottom frequenLly has an active karst. There is usually some integration between the valley-wall and the valley bollom karst systems. Lithology and structureofthe Appalachian Plateaus are similar along the lengthofthe orogen. Cap rocks are generally Pennsylvanian conglomerates and sandstones. Beneath these are Lower Pennsylvanian andUpper Missis sippian clastics, underlain by Mississippian mixed lith-

PAGE 126

Sasowsky ologies, includingLhicklimestones. AILhough appearingOatand featureless (Figure 1), broad folds occurwiLhre gional dipsofup to several degrees. Faulting is presentinsome areas, in associationwiLhLhefolds andLheeastern boun-darywiLhLheValley and Ridge. Basin asymmetry is a geomorphic parameterLhatquantifies disparate size (and possiblygrowLh)ofcontribu ting areas in a drainage basin. It is simplyLhepercentageofLhetotal basin area which is present onLheright-hand sideofLhebasin (Figure 2).Forsurficial drainage,Lheasymmetry factor (AF) is determined by delineatingLhemaster streaminLhebasin, measuringLhearea toLherightofLhis(as viewed downstream),andLhetotal basin area. AFisLhengiven by:AF=100(AR/An,where AR isLhearea onLheright, and ATisLhetotal area. AF<50 indi cates excessive area onLheleft sideofLhebasin. An alter nativemeLhodis to measure total stream lengLhs oneiLherside, but this is more difficult and unnecessary, as area generally serves well as a proxy for streamlengLh.AF quantifies disparate growLh, and can be used to pinpointLhereasons forLhisdisparity. Hare and Gardner (1985), for example, used AF to discriminate actively up..... Appalachian Karst Symposium.1991lifting fault blocksinCosta Rica (Figure 3). Becauseitisdimensionless, it can be used to compare basinsofdiffer ent sizes. Two related geomorphic parametersLhatare applicable in fluviokarst settings are introducedinLhispaper:Subsurface Drainage Density (DscV, and SubsurfaceAsymmetry Factor(SAP;seeFigure 4). Dsd is theknownlengLhofcavesina basinorsub-basin, divided byLheareaofLhebasin. It is similar toDd(surficial drainagedensity), which is given byLhetotal streamlengLhdividedbythe surface area. Dsd isdesignedtoquantifythe"karstifica-tion"ofLhebasin inLhesamewayLhatDdquantifies surfi-cial drainage developmentSAPis essentially a comparisonofDsd 's, similartoAP inLhatwe normalize it toLhetotal length (total areainLhecaseof AF) inLhebasin.SAF=100 (CLR/CLT), where CLR (cave lengLh right) isLhetotal lengthofknown cave passage toLherightofLhemaster stream,andCLT(cavelengLhtotal) is the totallengLhofcavesknowninLhebasin. The formula used to calculateSAFcanbemodified for special purposes by normalizing for surface area, rock type, etc., as is demonstrated laterinthis paper. Figure1:Level upland surfaceofLheCumberland Plateau, Tennessee. Horizon and outcrop in foreground areLhePennsyl vanian RockcasLle Conglomerate.118

PAGE 127

AppalachianKarst Symposium.1991Hypothetical BasinSasowskyPlateaus, this wouldbesometime in the Late PermianorTriassic. In the present,ifwefind that onesideofthe basin is larger than the other, it implies oneof'twosituations:1)the larger side began with a larger catchment through fortui tous locationofthe master stream and has maintained this dominance; or2)the larger side has been more successfulinexpanding its catchment for some reason.Atthe basin scale, reason 2 is attributed to head ward growthoftributaries or pira cy, and the master stream is considered to be fixed in position. Primary controlling variablesofAFare:1)Structure Attitudeofthe rocks affects initial and subsequent drainagebasinenlargement.Streams flowingondipslopesaregiven more basin areaatthe outset.2)Stratigraphy In an areaofmixedlithology,sub-basinswith more erodable substrate may grow more rapidly than others.3)Climate / Vegetation Variable rain fall and typesofvegetation can sta bilize partsofthe basin, protecting them from erosion. Figure 2: DeterminationofAsymmetry Factor(AF)and Dd for a hypothetical basin.(AF=(Aright/Atotai)X 100 )(AF=(200/300)X 100=67 )(Od(right)=20/200=10)(Od(left)=10/100=10)Factors Controlling AFAt an instant in time,AFexpressesasone number aratioofbasin areas; what does this meanintermsofpro cesses? It can indicate many things, andany further inter pretationofthe phenomena must be made with additional geologic and hydrologic data.Inall but the largestofbasins, it is reasonable to assumethat erosionofthe "initial" surface began at the sameLimethroughout the basin. In the caseofthe Appalachian1194)Tectonism Currently active uplift can serve as a differential driving force for erosion and migrationofdivides. Variables3and4are not dominant fac tors in the Appalachian Plateaus.FactorsControllingDsdand SAFTobeatall meaningful, the areaofinterestmusthavebeen thoroughly searched for caves, and the caves must have been surveyed. Given this condi tion, for both Dsd andSAFthe assumption is that both the number and lengthofundiscovered caves on both sidesofthe basin are proportionately equal and evenly distribu ted(i.e.,ifwedo not have a complete sample, it is at least a representative sample). Primary controlling variablesofDsd and SAF are:1)Structure Controlling movementofgroundwater. Areas with more flow will develop more caves.

PAGE 128

SasowskyNWDrainage BasinAsymmetry72.570.349.8 c-:: Montana LineamentZoneLimones-CanasLinaamentFigure3:Asymmetryofseveral adjacent basins on the Nicoya Peninsula, Costa Rica. The researchers were able to delineate between uplifted fault blocks on this basis (from Hare and Gardner, 1985).2)Stratigraphy Locationofsoluble beds and their relation to surfaceand ground-water routesofflow.Appalachian Karst Symposium.1991Cumberland Plateau, in undeformed rocks with a regional dipof0.5 degrees ESE. The present tilt is associated with the uplift of the Nashville Dome, which occurred through out the Paleozoic Era. The Plateau surface and upper reachesofthe streams are underlain by Pennsylvanian and Upper Mississippian clastics. The lower reaches and val ley bottoms are floored by Mississippian carbonates. Lo cal relief is 340 meters, with sharply incised valleys dis secting the Plateau surface. The basin has a total area of 184 square kilometers, and an APof29, strongly biasedtothe left side. The gross SAP is 87, meaning that most cave-passage length is on the right side. Within this basin, AP is structurally determined (Figures 5 and 6). The sub-basin on the left (West) sideiscomposedofdip slope (consequent?) streams. Due to their initial large catchment, these streams have successfully eroded back into the Plateau. Right-side streams, descend ing the structural free-face, have been limited in their growth due to lackofinitial catchment.(SAF=(CLright / CLtotal) X100 )(SAF=(5300/7000)X100=76)3)Surficial erosion rates Caves may form but subsequently be destroyed or disjointed by surficial erosion (un roofing), thereby biasing the calculatedvalues.UtilityofAFandSAFThe utilityofAF's andSAP'sis two-fold. First, they provide a normal ized quantitative measure allowing com parisonoftwoormore basins. Second,inquantifying these discrepancies, we are forced to explain them, and so think about speleogenesis and basin growthinfluviokarsts.IfSAP exists, why does it? Destructionofcavesorlackofdevelop ment?Iflackofdevelopment, why do cavesformpreferentially on the one side? What does all this indicate about basin growth as a whole?An Example -EastForkObeyRiverA strongly karsted sub-basinofthe East Forkofthe Obey River (EFO), Fen tress County, Tennessee was chosen to explore the relationship between asym metry factors and karst development. This basin is well suited for such analy sis; its surface has been extensively walked by cavers, it possesses a strong asymmetry, and it has 99 known caves with a total lengthof296,021 feet.Subsurface Asymmetry Factor&Dsdypothetical BasinThe drainage is incised into the Figure4:DeterminationofSAP andDsdfor a hypothetical basin,seetextfordiscussion. 120

PAGE 129

Appalachian Karst Symposium. 1991HOrizontalPI-___ane---SasowskyFigure5:Schemefor developmentofsurficial asymmetry in the sub-basinoftheEastForkObeyRiveratFentress County, Tennessee. Broad arrows indicate consequent streams on plateau surface. Master stream is incised transverse to this. Sub basins on the left side have larger initial catchment, and rapidly incise. Sub-basins on the right have small initial catchment, and form only small, steep drainages. SAF, too, appears structurally determined, but closer analysisoftheasymmetryleadstosomeinteresting problems (Figure 6).Ofthe 99 caves in the basin, 58 areonthe left side (total length=39,624 feet), and41are on the right (total length=256,397 feet). On the right side, however, threecavesofgreat length (Xanadu, Zarathus tra's, and MountainEyecaves)aregenetically differentfromthose in the restinthe basin, and account for 96%(245,067 feet)ofthe length on the right side. These 3 caves are all found on the East sideofthe main stream, and appear to have been formed through captureofflowofthe EFO, whereas the othersinthe basin are related to tributarydrainages.Thesecaves skew the whole basin value.If they are excluded from the sample, leaving only 11,321ftofcaveonthe right, the adjustedSAFis 22.Theeffect of the shallow (0.5 degree) regional dip has had importancein the locationofthe three largest caves. Where does the adjustedSAFpoint? It brings theAF (29) andSAF(22) values quite close tooneanother, and points to a possible (and expected from mass-balance con siderations) relation between surface drainage and karst dev elopment that the lengthofcaves formed is proportionaltothe drainage area.121ConclusionsSAFandDsdareparametersthatare useful inconjunc-tionwithothermorphometricparametersin understanding a basin.AttheEFOsite, there is anapparentrelation be-tween surfacecatchmentsize and lengthofcaves. It wouldbeuseful to seeifthe relation holds true inotherareasin the Appalachian Plateaus. There is no lackofpotential test sites, but finding areas that meet the criteriaoftho-rough exploration and large size may be difficult.AcknowledgmentsTheauthorbenefited from discussions with T.W. Gardner and K.C. Sasowsky. Financial support from theNationalSpeleologicalSociety,theRichmondArea Speleological Society, and the Cave Research Foundation made this project possible.ReferencesHare, P.W.and Gardner, T.W., 1985, Geomorphic indica torsofvertical tectonism along converging plate mar gins, Nicoya Peninsula, Costa Rica,inHack, J.T. and Morisawa, M.E. (editors),Tectonic Geomorphology: Proceedingsofthe 15th Geomorphology Symposia Series, Binghamton,p. 76-104.

PAGE 130

SasowskyAF=29SAF=22Appalachian Karst Symposium, 1991+ 29% ofSurfaceArea 22% ofCaveLengthFigure 6: Morphometric relations for the karsted sub-basinofthe East Forkofthe Obey River. When the 3 longest caves are removed from the sample, surface area correlates well with known cave length. Values shown have been adjustedinthis manner. 122

PAGE 131

Appalachian Karst Symposium, 1991 Kastning and KastningEnvironmental Education RegardingKarstProcessesintheAppalachian RegionErnstH.Kastning and Karen M. KastningDepartmentofGeology Radford University Radford, Virginia 24142ABSTRACTTheintracaciesofkarstic systems make it difficult for a non-specialist to adequately understand what karst is or how it works. Yet, increasingly, it is important that individualsinindustry and commerce,inlocal and regional government, and in the private sector become awareofunique problems associated with karstic terranes. The Appalachian region contains someofthe finest karstlands in the United States. However, this region is experiencing oneofthe highest ratesofeconomic growth in the country. SectionsofAppalachian karst are being developed for housing, agriculture, industry, commerce, and transportation. Karst results from interactive geologic and hydrologic processes that operate both on the surface andinthe subsurface. A basic working knowledgeofthese processes needs to be providedtothose who must ad dress environmental problemsinthe Appalachian region. Todoso, the complex characterofkarst must be reduced to afewfundamental geologic and hydrologic concepts that can be easily visualized and understood. There are fivefundamental requisites for dissolutiontoproduce karst in any locale: (1) The bedrock must be relatively solubleinnatural water. (2) Karstic surfaceor groundwater must be chemically aggressive. (3) The bedrock must be both porous and permeableinorder for groundwater to move and transmit the solution and solute. (4) The region must possess topographic relieftoinduce and sustain water flow. (5) Sufficient time must elapsetoproduce karstic landforms. On a site-specific basis, five significant modifying factors govern the local characterofkarst:(1)Variations in lithostratigraphy determine which rock units are most susceptible to karstification. (2) Geologic structure governs the degree to which rocks are fractured and along what orientations flow systems will evolve. (3) Evolutionofthe regional topography ultimately es tablishes inputs, outputs, and pathsofflow through the bedrock sequence. (4) The hydrodynamic characterofthe flow system affects the rapidityofflow and rateofdissolution. (5) Local climate adjusts the rates and intensityofphysical and chemical processes that excavate the rock. Knowledgeofthese attributes leads to a better understandingoflocal karst settings and potential environmental problems. Specialistsinkarst must convey these concepts to the non-specialist in a clear manner using easily understood dialog and visu ally effective graphics. Educationinkarst processes could and should occur at all levels, including public schools, youth groups, institutionsofhigher education, civic groups, and governmental agencies. In recent years, various methods have been employed toward the goalofeducating the public about the sensitive natureofkarstic environments. These efforts have ranged from interaction between local cavers and the general publictocooperation between speleologists and governmental agencies. The cumulative effectofthese endeavors has been encouraging as some local communities striveto1)clean up from past abuses, 2) monitor their present use, and 3) protect the future resourcesoftheir karsllands. Caving groupsorother environmental organizations may be abletoincorporate into their educational outreach activities someofthe materials that were previously generated and are readily available as aids.IntroductionLandforms developed onorwithin carbonate rock (limestone and dolostone) through dissolving are collec tively known as karst (see Monroe, 1970, for definitions 123ofkarst terms). Muchofthe karstic landscapeofthe Ap palachian region lies within the Valley and Ridge and Ap palachian Plateau physiographic provinces andischaracter ized by sinkholes, caves, sinking streams, springs, and so lution valleys. This karst regionisamong the largest and

PAGE 132

Kastning and Kastning Appalachian Karst Symposium. 1991Figure1:Cause-and-effect relationships among geologic, biologic, geographic, and human factors in karst terranes. Arrows poru-ay directionsofeffects. From Kaslning (1989c).SoilMuchofthe karstic terraneofthe United States lies inrural regions where environmental impacts are generally limitedtothose imposed by agricultural practices and highways (Davies, 1970). In some cases, karst lies within the confinesofpublic land (parks, forests, and the like). However, urbanization is rapidly encroaching in many karst areas and economic development is resultinginsevere karst-related environmental problems.Karst as an Environmental LiabilityAnappreciation for the extent and complexityofen vironmental problems in karst is best gainedbyconsulting proceedings volumesofrecent conferences that have speci fically addressed these issues(e.g.Beck, 1984, 1989; Beck and Wilson, 1987; Dougherty, 1983; and this volume). The interaction among various natural elementsofthe karst setting and man's role in the system is illustrated conceptuallyinFigure1.The systemisfar from simplis tic and consistsofa seriesofnested loops with feedback that represent direct and indirect causes and effects. Mak ing changes inanyoneelementofthe system will have consequential impact on other elements. It is not theintent of this papertoaddress allofthe possible impacts that man and karst have on each other. Rather, itisinstructive to select a few situations where karst may be importantasan environmental concern in the Appalachian region andtoconsider methods to interpret pertinent geologic and envi ronmental principlestothe public at large. Mismanagementofkarstlands, whether through unsu pervised economic development, poor farming practices, improper waste disposal, or other means will often damage groundwater supplies, cave ecosystemsorman-made struc tures built on karst. Among the most severe and imme diate environmental problems associated with karst includeMineralandEconomic Resourcesfinest in North America (Herak and Stringfield, 1972; Kastning, 1986). Carbonate-rock terranes pose environ mental problems that are unique with respect to the widespectrumofbedrock types, and karstic landscapes are parti cularly sensitive to environmental degradation (LeGrand, 1973; White, 1988, p. 355-405). Stresses induced by mankind in karstic terrane result in environmental prob lems that are much more acute than those that would occur in terranes underlain by either crystalline (metamorphic or igneous) or clastic (other sedimentary) rock. Problems suchassupply and contaminationofgroundwater and land instability abound in the Appalachianregion, as they doinmost populated karst regions, worldwide. The studyofkarst is a relatively new science that draws largely on the principlesofgeology and geography. A thorough understandingofthe processes that occur both at the surface and underground and an appreciation for the total hydrologic system necessitates a global familiarity with scientific karst studies. The level and scopeofmod em karst studies is demonstrated by the recent proliferationoftextbooks on the subject (Kastning, 1989a). Recent texts in English include thoseof Bogli (1978), Dreybrodt (1988), Ford and Cullingford (1976), Ford and Williams (1989), Jennings (1985), Sweeting (1973), Trudgill (1985), and White (1988). Moreover, the numberofsci entific journal articles and graduate theses on karst is ex panding at a phenomenal rate(see for examplethe biblio graphiesofLamoreaux and others (1970, 1975, 1986), White and White (1984), and Huppert (1988)). Despite accelerating research on the subject, karst is not an easy concept for the lay person to understand at the outset. Generally speaking, most people know very littleofwhat exists and what happens below the surfaceofthe Earth. Yet, these same people may have a much greater perceptionofsurficial processes.Itis easytounderstand why. Surficial phenomena are easily observed and therefore understood; phenomena in the subsurface are not typically seen, much less com prehended ("outofsight, outofmind"). Perhaps most importantly, the average person views the aboveground and the underground astwo distinct zones,sep aratedbya boundary, namely the surfaceofthe Earth.Themost formidable challenge in successful karstIand managementisbreakingdownthisdichotomy. Accordingly, there are three fundamental concepts thatmustbestressed:(1)karst is asingleunifiedsystemthat integrates the surface andsubsurface,(2)karstlandsareenvironmentally very sensitive terranes, and(3)the physical and chemical processes that create karst are the veryprocessesthateasilyleadto environmental problems. 124

PAGE 133

Appalachian Karst Symposium. 1991 groundwater supply, groundwater quality, and land insta bility. Eachofthese problems is briefly discussed below(seealso succeeding papers in this volume).Groundwater SupplyIn karst regions, unlike other typesofterrane, ground water is channelized within a natural underground system of interconnected 'pipes' (see Plate B, page 84). These natural pipelines collectively transmit water from input points (recharge zones) to output points (discharge points). Recharge in karst terrane occurs as water percolates through the soil and into fractures in the carbonate rock over large areasofthe countryside (diffuse recharge)oras surface streams sink in their entirety where they flow into cavesorsinkholes (discrete recharge). In most karst regions, both mechanisms operate simultaneously. Discharge from karstic aquifers occurs where water seeps from the ground over a wide area (diffuse springs)orwhere underground riversorcave streams exit from large openings (discrete springs). Springs may issue anywherefroma few to thousandsofgallons per minute. A signifi cant quantityofwaterisalso discharged through man-made wells drilled to obtain water for domestic, commercial, agricultural,orindustrial use. Obtaining usable amountsofwater from karstic aquifersmay be a 'hit-or-miss" operation. Water is highly lo calized because itisflowing through solutionally enlarged fractures, within partings between bedsofrock (seeforexample, Zewe and Rauch, this volume), or along locally highly permeable beds (Kastning, this volume). In sand stone and other porous-media aquifers, flow is diffuse throughout. In karstlands, however,wells may not yield sufficient water unless a solutional conduit is intersected. Springs and wells in karst are also highly sensitive to variable weather patterns such as a draught or wet periods,andrespond rapidly to weather changes as waterisquickly conveyed along solutional conduits. Karstic groundwater supplies, therefore, are flashy and allowances mustbemade for this erratic behavior in the allocationofwater derived from springs or wells. Another common misconception by the general public relates to the relative rolesofsinkholes and groundwaternow. Water does not enter the subsurfaceatsinkholes just because they are there but, rather, sinkholes exist because water enters the subsurface at that locality and has done soforsufficient time to dissolve and remove the overlying materials. Infillingofa sinkhole will not alter that inher ent relationship to the subsurface. Man-made changes tothesurficial drainage and to sinkholes may, however, easilyalter the rateatwhich the underlying aquifer receives its normal recharge.Vegetation and soil cover slow runoffandabsorb some moisture, thereby providing less "flashy" recharge than would impermeable materials (e.g. cement 125 KastningandKastning drains, asphalt roadsorparking lots, and roofsofstruc tures). Sinkholes that have been infilled are less efficient inputs and may cause surficial water to pond or backflood, unless it is diverted away from its natural sinkpoint (therebyaltering the rechargeatyetanother sinkpoint) (see Mills and others, this volume). These activities, increas ing the rateofrunoff and/or blocking the input points, may cause castotrophic floodingorcollapse and drastically alter the quantityofgroundwater available for use in the immediate vicinity.Groundwater QualityIfthere is one single environmental issue that stands out in the karstofthe Appalachians, it would have to be the sensitivityofthe karsticaquifers to groundwater con tamination. Man's impact is most severe in cases where polluted surface watersenterkarstic aquifers. This problem is universal among all karst regions in the United States that underlie areasofeconomic growth. On the positive side, most karst in the Appalachians lies in rural areas.Onthe negative side, the region's karstic ground water problems are increasing with the adventof(1) ex panding urbanization, (2) increased productionofenviron mentally unacceptable artificial chemicals, (3) shortageofrepositories for hazardous wastes (both household and industrial), and (4) ineffective public education concerning waste disposal and the sensitivityofthe karstic ground water system. There is a general lackofpublic understandingofgroundwater behavior, particularly in karst. A common perception is that underground waters, such as those flow ing from springs, are fUtered and nearly pure. On the contrary,karst aquifers can notfilter contaminated groundwater sufficiently to render it potableatthe discharge sites; and because recharge points are directly connectedtodischarge points, conveyanceofcontamination is highly efficient Sinkholes are natural holes in the ground surface and to the general public appear to be natural sites for dumpingoftrash (see Plate K, page 204). The presenceofa sink hole "eliminates the need" to dig a pit into which refuse can be dumped. The numberofactive and inactive sink hole dumps in karst regions is staggering. For example, over 260 illegal dumps have been inventoried for Rock bridge and Botetourt counties, Virginia alone (Slifer, 1987; Slifer and Erchul, 1989; Erchul, this volume). It is esti mated that each county with karstinthe Valley and Ridge Province has hundredsofsinkhole dumps. The profusionofthese dumps is the resultof(1)a lackofa refuse-remov al service in rural areas and the expense and inconvenienceoftrash haulage on the partofthe landowner, (2) the con venient proximityofsinkholes, and (3) ignoranceofthe karstic groundwater system on the partofthe landowner. Sinkholes as natural funnels convey toxic substances directly into the karstic plumbing system (Kastning and Kastning, 1990). In many cases, chemicals may be trans-

PAGE 134

Kastning and Kastning mitted directly to domestic wells in a matter ofafewhours and without filtration. A farmer who places a carcassofa deceased farm animal into a sinkhole (an all too common procedure) may veryweI!be drinking water that passed through that sinkhole! Or, his neighbors may be. Unfortunately, sinkhole dumping is only one wayofcontaminating a karstic groundwater supply (Aley, 1972; Aley and others, 1972). Chemical fertilizers and pesticides appliedtofields overlying carbonate rock will enter the aquifer through diffuseinfiltration and contaminate springs and wells. Runoff from feed lots may also (see Brown and Ewers; Ogden, Hamilton, andothers, both in this volume). Improper sitingofmunicipal landfills onornear karst causes leakage or runoff from these landfillstoeasily contaminate karst waters. Corroded underground storage tanks, such as at active or abandoned service stations, may release hydrocarbons directly into karstic aquifers. Chemi cals introduced in these ways may include manyofthe most hazardous, including hydrocarbons, heavy metals, and others. Additionally, leaky septic systems, sewage lines, or effluent from faulty sewage-treatment facilities introduce coliforms and other disease-bearing organisms into the karst system. Many streams, including those in rural areas, are polluted and most surficial streamsinkarst terranes readily lose water into the ground through their stream beds. Contaminated waters from the surface that enter carbonate rocks introduce their toxic substances into subsurficial streams.Theonly difference between surface and underground waters in karst is that the latter is outofsight (andoutofmind)! Chemically they may be identical. Accidental spills from overturned tanker trucks and runoff from highways salted in winter toprevent freezing are just twoexamplesofpossiblecontaminationalongtransportation corridors (Werner, 1983; Lovegrove, 1988). Effluent from commercial and industrial operations along such corridors may also be a problem. However, industrial and governmental leaders are often unawareofthe sensitivy and inherent complexityofgroundwater flowinthe karstic subsurface. Caves contain fragile organisms that have evolved in the natural cave environment. Bats are the most common ly recognized creatureofcaves, but there are actually an amazing varietyofcave-dwelling organisms. Because these animals are highly adapted to their constant ecologi cal surroundings, they are particularly sensitive to distur bances. Foremostofthese is the introductionofforeign substances into the groundwater that flows through the caves. Even "clean" fill, such as brush, hay, sawdust, or dirt, may leadtochemical imbalancesinkarstic ground water as the vegetative matter decays rapidly and consumes oxygen. In order to conserve species endemic to caves, rare or otherwise, man must maintain clean groundwaterinkarst regions. Indeed, these vulnerable ecosystems and the organisms that resideinthem can serve as indicatorsofthe qualityofthe groundwaLer. The only effective means to 126 Appalachian Karst Symposium. 1991 monitor and protect these organisms is through detailed biogeographic inventories and educating the public. Many who visit caves on commercial tours are intri gued by the myriadofcave formations. There are,infact, many unusual, fragile, and often rare formations in caves (Hill and Forti, 1986). These formations, that take cen turiesormillenia to form, are highly susceptible to con tamination and derangementofgroundwater flow. The fra gility and rarityofthese deposits canbeeasily conveyedtovisitorsofcommercial caves. Important environmental messages can be provided in show caves by properly trained tour guides through entertaining, but educational, interpretation.Subsidence and Ground InstabilityThe potential for the surface in karst regions to give way in collapse is brought home from time to time in the media. Massive collapsesinwhich homes or businesses are swallowed by newly formed sinkholes make exciting news. In some states, such as Florida, Alabama, Texas, and Pennsylvania, such occurrences are somewhat frequent (see papers by Dougherty and Beck, this volume). Mostofthese events are triggered by man's intervention withthekarstic environment (Waltham, 1989). The most common cause for catastrophic sinkhole collapse is an overpumpingofgroundwater from karstic aquifers, resulting in a relatively sudden lossofbouyant forces that uphold roofsofcavernous openings. A second causeofcollapse occurs in response to changes in the positionofthe water table due to modifications to surficial runoff and infiltrationtothe karstic groundwater system. In areas undergoing development, sinkholes are often viewed as unwanted holes in the ground.Ifthey are filled in to produce levelland, the potential for ensuing environ mental problems is twofold: First, naturally developed pathsofinfiltration are often blocked, leading to ponding or flooding on the fill (see Plate G, page 158). Secondly, over the long run, fill materials willbesapped into the subsurface and settling may occur. These disturbances easily impact any structures built on the fill. Additional ly, the increased weightofwater, ftll, and structures upon the cavernous bedrock could cause catastrophic collapseinthe future (see PlateH,page 158).FundamentalsofKarst ProcessesThe originofkarstic landforms including caves and other avenuesofgroundwater flow in soluble rockisulti mately tiedtoseveral factors. To have a basic understand ingofwhat karst is and how it operates as a system re quires that the fundamental mechanisms responsible fortheoriginofkarst be clearly recognized. Basic educationinkarst processes should include knowledgeofthe requisites for karstification and local factors that control and modify these processes at specific sites.

PAGE 135

Appalachian Karst Symposium, 1991 The following discussion comprises an overviewofthenatureofkarst in the Appalachians. The humid-tem perate karst in the Appalachian foldbeltofthe eastern UnitedStates is characteristically different than karst regions elsewhere, for example in arid or tropical regionsorin areas underlain by relatively horizontal strata. Where ap propriate, karst in Virginia is used as an example because,inmany ways, the karstofVirginiaistypicalofkarst intheAppalachian region. Also, the authors have had recent experience with conveying the conceptofkarst to the lay public in Virginia.Requisites for KarstificationSolubleBedrock:The principal method by which caves and similar avenuesofwater flow are excavated fromrockisby dissolution, the chemical corrosionofthe bed rock by water flowing through it. Many rocks, such as sandstone, granite, shale, and gneiss, are relatively insolubleand very little materialisremoved by the dissolving actionofgroundwater. In contrast, calcite, the main ingre dientoflimestone and dolostone,ishighly soluble.Itfol lows that developmentofdissolutional openings will occur in sedimentary rocks that have a high contentofcal cite. Limestonesofthe Valley and Ridge and Appalachian Plateau provinces, for example, are relatively highincal cite content. In places, bedsofhigh solubility are inter bedded with those less soluble. Where this occurs, but where other factors remain relatively similar, the greatest developmentofsolutional voids (cave passages and thelike)conformstolimestone bedsofhigh purity (see Kast ning, this volume). Comparisonsofmaps showing the distributionofcaves and other karst features in Virginia (e.g. Hubbard, 1983, 1984, 1988; Miller and Hubbard,1986)and inventoriesofcaves (e.g. Douglas, 1964; Hol singer, 1975, 1985; Virginia Speleological Survey, 1987 present) with published geologic maps clearly show thattheareal distributionofcaves and sinkholes conforms to outcropsofcarbonate rock. PorosityandPermeability:Waterisunabletoflow through bedrock unless there are openings capableofholdingwater (porosity) and unless those pore spaces are inter connected. The abilityofa rock to transmit water along avenuesofporosity is referredtoas permeability. Well integrated and mature karstic drainage networks develop where porosity and permeability are high. Such conditions often include fractures such as joint sets and faults (see Plate C, page 1(0). Fractures form as a resultofdeformationofbedrock in response to crustal stresses (such as those that formed the Appalachian Mountain range). In tense folding, uplift, and other events resultina high den sityofjoints and faultsas brittle rocks break (Kastning,1977,1984). Careful perusalofmapsofcaves shows thatmanylinear passages and chambers closely follow the orientationofprincipal joint sets. Some significant cavesinVirginia are closely associated with faults (KrinilZsky,1947)and sinkholes are commonlyaligned along fracture traces, indicating that infiltrationofsurficial waters into127Kaslning and Kaslning karstic aquifers occurs along joints and faults (Kastning, 1989b).Inthe Appalachians, thisisbest illustrated where broad, low lying valleys are floored with nearly horizontal strata (Kastning, 1988, 1989b).ChemicallyAggressiveWater:In order for ground water to remove limestone by dissolution it must be acidic and not yet saturated with dissolved calcite.Inlimestone, freely circulating wateristypically charged with dissolved carbon dioxide (much like a carbonated beverage) thatwaspicked up as the water percolated through soil. This weak solutionofcarbonic acid seeps into fractures in the rock. Over time slow, but persistent, dissolvingoflimestonebythis solution can create voids in the bedrock ranging from inches in diameter to hundredsoffeet in width and height. Flow paths are often long, in many cases producing caves with tensofmilesofpassages. The extentofsoil cover over most areasoflimestone in the Appalachians ensures continual enlargementofflowpaths by chemically aggres sive karst waters.TopographicRelief: Groundwater will not flow un less water entering the ground at one elevation can exitata lower elevation. Differences in elevation, known as relief, are necessary to maintain an exchangeofwater along the flowpath. Without the removalofwater saturated with re spect to calcite (from the dissolvingofcarbonate rock) and the inputoffresh, chemically aggressive water continually coming into contact with the wallsoffractures and inci pient cave passages, no further dissolution will occur. The limestone areasofthe Valley and Ridge Province have considerable relief and the maintenance of groundwater flow is ensured. Moreover, flow rates are kept reasonably high by steep hydraulic gradients in partsofmany drainage networks. Runoff from intense rainstorms and snowmelt in the Appalachiansisefficiently conveyed within the karstic aquifers from the pointsofrecharge to springs.Asindicated above, this is an important factor in the rapid transmittalofpollutants through karstic terrane. Time: The evolutionofcave systems into well-integrated karstic drainage networksisslow by human standards.Ingeologic context, however, the process is quite rapid. Nearly all caves in the Appalachian region, regardlessoftheir ultimate size, developed during the Quaternary Period (the last two million or so years), even though the rocksinwhich they have formed are considerably older, typically dating from the earlytomiddle Paleozoic (525to320 mil lion years ago). Within our lifetimes, most flow networks will undergo very little structural change. On the other hand, the environmentofcaves and karsticDowsystems can be rapidly alteredbyhuman intervention.Local Modifying FactorsLithostratigraphy:As arule,carbonate rocks are not lithologically uniform. Sequencesofsedimentary depos its, such as thoseofthe Appalachian region, show consid erable stratigraphic variation. Layersofsandstone, shale,

PAGE 136

Kastning and Kastningor chert often separate bedsoflimestone. F10wpaths may be constrained by relatively impermeable beds, thereby channeling groundwater within carbonate beds or perching karstic groundwater above shale deposits(seepapers by Saunders and Balfour and Heller, this volume).BedrockStructure:Bedsofcarbonate rocks are not perfectly horizontal; in fact, most beds in the Appalachian region are steeply inclined due to intense Even the slightest tilt (dip)ofthe strata can exert influence on developing flowpaths. Flow withm dippmg carbonate beds primarily occurs either along strike (hori zontally and parallel to the trendofthe bed)or do,":n the dip (descending parallel to the slopeofthe bed).WhIChofthese may occur in any particular case depends on the con figurationofthe surface topography, the locationof level springs, and the hydraulic gradients imposed by lItho logic constraints and fracture porosity. Most inthe Appalachian regionofVirginia flows along stoke. Because the folded rocks and mountain ranges trend northeast-southwestinthis region, it is expected that karstic groundwater flow will generally be in these directions. (Of course, flow down the dip may occur locally; only careful mapping will determine this for sure.) As a result, extensive flowofkarstic groundwater typically occurs beneath and parallel to the lower flanksofmountains and hills. Flow may also be concentrated where thereisa higher densityoffractures and thus a greater permeability.TopographicEvolution: SlIbSllrficial karstic drainage systems develop in response to an evolving surficial land scape. Local baselevels are lowered as major surface streams cut downward through the landscape by erosion. Contemporaneously, the positionofthe potentiometric surface (water table) in the aquifer drops as baselevel springs migrate to lower elevations. Previously formed conduits and caves are abandoned by groundwater that now flows through the rock at lower elevations. Relict, water free caves are no longer a significant componentofthe drainage system (except during occasional high runoff). Younger, lower conduits have taken their place.Hydrodynamics:The rateofflow and levelofturbu lenceinkarstic groundwater flow is highly variable from place to place. High hydraulic gradients and constricted conduits can increase the pressure, forcing water through the system. This in turn increases turbulence and the rateofdissolution. Steep hydraulic gradients promote greater discharge and rapid transmittalofpollutants.Climate:Climate has been showntoseverely affect the rateofkarst development. Karstoftropical regions, for example, forms at a much greater rate than does thatofanarid, cold region. Karstificationinthe Appalachian regionhasbeen at a moderate rate. Because climates are relatively constant during the courseofhuman activity, this variable has little effect either spatially or temporally with respecttopresent-day environmental problems. 128Appalachian Karst Symposium. 1991Analyzing a Karstic SystemA complete analysisofa karst region would require careful documentationofeachofthe aforementioned ten variables (Kastning, 1990). An appreciationofenviron mental problems in the Appalachian karstlands requires a familiarity with the origin and configurationofsubsurface drainage. Accurate determinationofpathsofgroundwater flow necessitates detailed geologic mapping. The pathofflow from the pointsofrecharge (infiltration) to points of discharge (springs) is rarely a straight line and may in reality be quite convoluted. F1owpa.ths?f karst migration are often contrary to the directIOnofflow m nearby surface streams and water may be pirated through the subsurface from one surficial drainage basintoanother.Helping the Public UnderstandKarstAppropriate karst management must include an assess mentofthe vulnerabilityofintegrated karst systemstochanges incurred on the surface. The most effective waytoprotect the karst environmentisto develop an awareness and understandingofpotential problems on the partofthelocal residents. This translates into education; not simply limited to traditional schooling, but by any means where by the public is exposed to the problems and solutions. These include developing data for karstland management, promoting clean-up and restoration activities, working with civic organizations, youth groups, museums and science centers, developing displays and presentationsforlocal and regional events, and involving the papers, television, radio, etc.). ExamplesofaCtIVItIes.InVirginia and West Virginia are included in thediSCUSSIOnthat follows. However, the environmental problems are typicalofthe Appalachian region and the educational solu tions used in the Virginias may be easily applied in any of the other statesofthe Appalachians.InventoryingKarstMany karstic landforms, such as large sinkholes and sinking streams, are readily identifiable on standard U.S. Geological Survey 7.S-minute topographic maps (scale 1:24,(00). However, not all sinkholes, caves, and lesser karstic landforms appear on topographic maps; for example, many sinkholes are simply too shallow berepresented within the contour interval used on.a particular map,orin some cases karst features have been overlooked in the surveyingorcartographIC process. Precise inventoryofkarst necessitates additional work, including useoflow-altitude aerial photography and sur face reconnaisancebyvehicle or on foot. To date there are very few areasofthe country where such tories have been made. A noteable effort m accomplIshmg such a task includes recent surveying both on a state-wide and county-wide level in the CommonwealthofVirginia(seeHubbard, this volume).

PAGE 137

Appalachian Karst Symposium. 1991The karstofVirginia and West Virginiaischaracterizedby a high densityofsinkholes (Herak and Stringfield,1972;Hubbard, 1984; Kastning, 1986,1988, 1991). A seriesofthree maps showing exposuresofsoluble rock and the distributionofsinkholes and caves is being published for the stateofVirginia (Hubbard, 1983, 1988).Thedata is derived from topographic maps, aerial photo graphy, soil survey maps, and the speleological literature. Karst maps for twoofthe counties in this karst regionhaverecently been published (Miller and Hubbard, 1986; Hubbard, 1990). Karst terrane occasionally appears as a mapped environmental unit in local geologic mapping aswell(e.g.Schultz, 1981). Delineationofsinkholes on a map may readily indicate potential subsurficial flowpaths (Kastning, 1984, 1989b).Inmany situations, sinkholes are topographically aligned. This indicates a structural or stratigraphic controlinthe hydrogeologic setting wherein groundwater moves along linear flowpaths formed along bedding planes and fractures. The implication is that infiltration entering an aquifer through such sinkholes contributes water to an integrated flow system. The surface arrangementofsink holes thereby provides a hintofthe configurationofthe underground drainage. This concept is easily relatedtothe public through diagrams and mapsofthe areasinquestion. However, an experienced geologist or hydrologist must compile this data and reduce ittoan easily visualized graphical format showing how surficial and underground processes are related. The graphics canbeincorporatedinvirtually all typesofinformational material, published or otherwise, including pamphlets, posters, newspaper and magazine articles, handouts at meetings, classroom studyaids,and many others. Flowpathsofgroundwater through carbonate rock canbedeterminedbyinjectionofharmless fluorescent dyes intosubsurface streams and detecting where the dye emerges.Thetechniques of water tracinginkarst are well documentedinthe textbooks referred to earlier andinhandbooks. Water tracinginthe Appalachian karst has been underway since the 1950's. A few recent studies in the Virginias include thoseofJones (1973, 1983, and this volume, p. 217), Saunders and others (1981),Ogden(1976), Werner (1981), and Quinlan (this volume,p.168).Asflow net works are delineated through exploration and mappingofcaves, groundwater tracing, and other means, they becomepartofthe knowledgeofthe terrane, allowing assessmentofenvironmental problems that may occurifman modifiesthelandscape or water chemistry. This, in tum, allows local communitiestomake educated decisions on landuseandeconomic development. Inventoriesofcaves and other significant karst features are being maintained by privately operated speleological surveys in both states. These surveys have beeninoperationfor some time (Davies, 1958; Douglas, 1964; Hol singer, 1975; Virginia Speleological Survey, 1987 present). Recent efforts by the Virginia Cave Board and 129Kastning and Kastningthe Virginia Speleological Survey have identified caves considered to be highly significant based on geologic, bio logic, hydrologic, archeologic, and historic criteria (Hol singer, 1985; Gulden, 1989). Organisms inhabiting caves in West Virginia and Vir ginia have been investigated and documented(e.g.Holsin ger and others, 1976; Holsinger and Culver, 1988). Addi tionally, newly discovered species are being added at a regular rate. Research and publication on cave habitats and ecosystems should be partofthe inventory in the conser vation and managementofkarst regions. Caves servingashabitats for sensitive and rare organisms are increasingly brought to light, especially where developmentisthreaten ing their existence. The Virginia Cave Boardinconjunc tion with the Natural Heritage Program regularly contends with potential threats to cave ecosystems in the state.CaveandKarstProtection LawsFortunately steps are being taken to protect the karstic environment in the Appalachian region. For example, both West Virginia and Virginia have enacted state laws that protect caves and their natural contents from vandal ism and contamination. The public at large is not awareofthese laws and therefore their existence must be made known. Chaptersofthe National Speleological Societyinthe Virginias have placed special metallic signs inside many cave entrances informing the visitorofthe laws and penalties for violations. The CommonwealthofVirginia has established the Virginia Cave Board as partofthe DepartmentofConser vation and Recreationtotake up matters relatingtocaves and karstinthe Commonwealth, to advise other agencies, and to participate in education related to caves, cave science, and cave conservation. The board, composedofeleven members appointed by the Governor, meets three times a year and regularly considers environmental prob lems emerging at specific sites. Where necessary, itbecomes actively involved in mitigating potential threatstocaves, karst, and spelean biota.CaveandSinkhole RestorationCave restoration projects have become increasingly popular among concerned cavers and other volunteers. Restorationofsinkholes has also been attemptedinrecent years(seePlate D, page 146). Oneofthe largest and most successful efforts was the removaloftonsofrefusefromthe entrance areaofStillhouse Cave in Randolph County, West Virginia (Harter, 1991; Vlchek, 1991). Stillhouse Cave is within a highly scenic and popular karst region and lies just afewhundred feet from the well known SinksofGandy(seephotograph on the coverofthis volume). This clean-up was a cooperative effort between the speleo logical community and the West Virginia DepartmentofNatural Resources and DepartmentofHighways.Ilre sultedinthe first siteinthe official West Virginia "Adopt

PAGE 138

Kastning and Kastninga Dump" program for volunteer clean-upsofillegal dumps. Removaloftrash and restorationoforiginal contours around cave entrances have been very successful, but such efforts require considerable time and expense and in many states the shear numberofsinkhole dumpsisstaggering. However, clean-up and restoration projects catch the inter estoflocal newspapers and television stations. News sto ries about these projects are read by many people who live and work directly on the karst surface. Whether conciously or not, these readers note thatifvolunteers are expending time and energy in cleaning trash from a hole in the ground, there must be some reason.Ofcourse, it is useful if the reasons are giventothe reporters as partofthe story. Local chaptersofthe National Speleological Society regularly clean caves and sinkholes, leading to favorable publicity in the press. The New River Valley Grotto (Chapter)ofthe National Speleological Society in con junction with other grottos and a local Boy Scout troop has been cleaning a large trash-filled sinkhole in Pulaski County, Virginia. This was a former entrance to James Cave, the longest (over 7000 feet) cave in the county. The entrance has recently been reopened (after being buried for27years under tonsoffarm and household trash) and water quality in the cave streamislargely restored. About six clean-up events have taken place at the site to date and,insome cases, newspaper and video coverage has been a planned partofthe event (Farrar, 1989; Kittredge, 1989). Figure2:Cave Conservation Awareness Award earned by Boy Scouts in Virginia and West Virginia. Requirements for the award include (1) a short course about caves and cave conservation, (2) participation in a clean-up or restor ation project involving a cave, sinkhole,orother karst terrane, and (3) a caving trip led by experienced cavers. 130Appalachian Karst Symposium. 1991This clean-up was featured in a recent videotape produced by the Virginia Military Institute Research Laboratory and funded by the Virginia Environmental Endowment for use on public television and in classrooms throughout the karst regionofVirginia(seeErchul, this volume). Involvementofyouth groups in cleanups is both sen sible and desirable. Environmental awareness is currentlyata high level among teachers and studentsinpublic schools, and many younger people are willing to volunteer time and energy toward cleaning up their community. Some local Boy Scout troops in the New River ValleyofVirginia and West Virginia have established a continuing award program (Figure 2) incaveconservation that includes scheduled clean-ups at caves and sinkholes(seePlate E, page 146).Karst and Public EducationItis impossible for those concerned with preserving karst to confront allofthe environmental problems through remedial action, including cleanupsofcaves and sinkholes, legal action to prevent development ortoseek restitution from violatorsofenvironmental law,orother reactionary measures. Although these efforts will helpona case-by-case basis, they will not keep pace with the impactofprogress.Perhaps the single, most effective program to prevent the abuseofkarst and promote sound environmental aware ness is within the contextofprimary and secondary educa tion. How the characteristics and mechanismsofkarst dif fer from thoseofother terranes must be made graphically clear in the classroom, particularly in countiesorcities that lie within karst areas or are in close proximitytothem. Educational contact with this age group may also be made through youth programs including scouts, 4-H clubs, high school science clubs, and other outdoor oriented organizations. Organized cavers and karst researchers are often ap proached to lead workshops or fieldtrips for youth groups, science teachers, community leaders, symposia partici pants, or others. An understandingofkarst, including the subsurface, can be enhanced by viewing surficial features(e.g.sinking creeks, sinkholes, seeps, and springs) with out necessarily venturingunderground(seePlate A, page 58). Therefore, the publicatlarge, including the news media, could learn a great deal from a field trip on karst terrane. The news media can then effectively joinincarrying environmental messages to the general public. Graphic explanationsofactive karst processes in layman's terms can go a long way toward conveying the need to preserve fragile karst features, water supplies, and cave ecosystems. The useofphotography, video-recordings, graphic arts,andwriting, especially in conjunction with case histories, has been showntobe effective in reaching citizens living on

PAGE 139

Appalachian Karst Symposium. 1991karst. Distributionofthis information can be in various forms, including presentationsofpapers or multimedia programs at local, regional, or national meetings, posters, local clean-up and fund-raising events (Figure3)with allendant publicity in the media, exhibits at commercial caves, museums(seePlates I andI,page 186), scout shows, and other community events, and literature for distribution to the public andtolandowners. Some educational materials are readily available fromcaveconservation organizations, such as the National Spe leological Society and the American Cave Conservation Association. Likewise, the Virginia DivisionofMineral Resources has published materials on karst designed to educate the public and provide basic data for local com munities (Hubbard, 1988,1989, 1990, 1991; Miller and Hubbard, 1986). Additionally, the CommonwealthofVirginia, through rnianf(you!JorYour ContriGution!The Dixie Caverns Salamander(1'{lt!ioaonwe/in'i) The bind. Dixie CavernsSalamanderis a rdrc phase of Lhc Slimy Salaman(kr (Plethodong.glutinosus)andisfoundinDixieCaverns. New Dixie Cave. andBlankenship Cave. all inRoanoke County. This amphibian is oncofmanyspecies of animals.bothvcncbratc and invcncbrate.that inhabit C.:lves of this Inprotecting caves and the groundwater that circulates through them, we are ensuring tho.t sensitive crcaUlrcs such as this salamander continuetohave lhchabitat U1CYneedfor survival. (DrawingbyJohn Schoenherr. card designedbyErnsland KnrenKa."ning and Steve Lenhart. RadfordUniversily.)Yourvlsit to the AnnualDixiecavernsHauntedcaveIs contrlbuUngtothe conservationandprotectionofcavesandtheir ecosystems.TheNcwRiverValleyGrottoand BlueRidgeGrotto.chaptersofthe Nalional Speleological Socicty. usc proceeds from thcH3untcdCave for locaJ and slatewide cave conservation projccts and10suppOrt national caveprotcction progr:uns. Cavesarc a valuable.yet in the Appa13chian regionofVirginia and WestVirginia. Theyarean Integral pan of thegroundwi:ucr systeminlimestone terrarte (karsl). Conservationofcavcs. sinkholcs. and other karsl fealuresisimportamintheprotectionof TheNationalSpeleologicalSodelyisdedicaledtothe safeexploration. study,and conSCnt31ion ofcaves.Ifyou are imcrcslcdinC:lVCSand caving, please COnlJet eitherofthese chapters: New River Vnllcy Grotto.P.o. BOA 1027.Dublin. Virginillo24084Dlue RidgeGrotto.c/oAI StcWaIL2528Monq;orneryAvenue S.W.. Roanoke. VirGinia 24015 October 1990Figure3:"Thank-you" card presented to visitorstothe annual Haunted Cave fund-raising event at Dixie Caverns, Virginia. Proceeds are used for conservation projects.Thecard is a means to inform the public about caves, karst, and their environmental sensitivity.131Kastning and Kastningthe Virginia Cave Board, recently published a 22 inchby28-inch, full-color poster on karst groundwater protection(seePlate F, page 152). This poster, depicting the prob lems createdbypollutionof karst waters through sinkhole dumping, was distributed freeofcharge to ninth-grade Earth-Science teachers throughout the Commonwealth,toother educational facilities, and to selected governmental and environmental agencies and groups (Kastning and Kastning, 1990). This poster is also available by requesttosimilar groupsinother states.ConclusionsOneofthe best measuresofemerging environmental problemsinkarst regions is the incidence and frequencyofnews stories on such topics as sinkhole collapses(seepapers by Dougherty and Beck in this volume), contaminated springs and wells,oraccidental spills. This coverage alerts the populace to the extent and distributionofthe problem, as it should. Yet, positive environmental educa tion about karst processes is imperative in karst regions such as the Appalachians. Many potentially effective resources and methods are availabletoconcerned citizens who wish to further karst conservation. Among them are: cleanup and restoration projects preparation and disseminationofliterature on karst and speleology inventorying and mapping karst areas presentations and exhibits for the publicatcommu nityevents working with youth groups, civic organization, and governmental agencies and committees volunteer work with show caves, parks, museums, and outdoor organizations participation in public awareness activities(e.g.EarthDay, litterthons) involving the media wherever possible Educating the public about the environmental liabilityofkarst should be viewedasan opportunity whereby poten tial problems may be averted or minimized. Instilling an awarenessof karst, through remedial, monitoring, and pro tective actions, coupled with positive educational programs and media involvement,isthe most effective means toward managing this sensitive terrane.ReferencesAley,T.I.,1972, Groundwater contamination from sinkhole dumps:Caves and Karst,v.14, p. 17-23. Aley,TJ.;Williams,I.H.;and Massello, J.W., 1972, Groundwater contamination and sinkhole collapse

PAGE 140

Kastning and Kastning induced by leaky impoundments in soluble rock terrain: Missouri Geological SurveyandWater Resources. Engineering Geology SeriesNo.5,32 p. Beck, B.F. (editor), 1984, Sinkholes: Their Geology, Engineering and Environmental Impact: Proceedingsofthe First Multidisciplinary Conference on Sinkholes. Orlando. Florida, 15-17 October 1984: A.A. Balkema, Rotterdam and Boston, 429 p. Beck, B.P. (editor), 1989, Engineering and Environmental ImpactsofSinkholes and Karst: Proceedingsofthe Third Multidisciplinary Conference on Sinkholes and the Engineering and Environmental ImpactsofKarst. St. Petersburg Beach. Florida, 2-4 October 1989: A.A. Balkema, Rotterdam and Boston, 384p.Beck, B.F. and Wilson, W.L. (editors), 1987,KarstHydrogeology: Engineering and Environmental Appli cations: Proceedingsofthe Second Multidisciplinary Conference on Sinkholes and the Environmental ImpactsofKarst. Orlando, Florida. 9-11 February 1987: A.A. Balkema, Rotterdam and Boston, 467 p. Bogli, A., 1978, Karst Hydrology and Physical Speleo logy (translated from the Gennan by J.C. Schmid): Springer-Verlag, New York, 284p.Davies, W.E., 1958, CavernsofWest Virginia: West Vir ginia Geological and Economic Survey, volume 19,350 p. (reprinted in 1965 with 72-page supplement as volume 19A). Davies, W.E., 1970, Karstlands, in United States Geol ogical Survey, National Atlasofthe United StatesofAmerica. sheet 77. Dougherty, P.H. (editor), 1983, Environmental Karst (papers from karst symposium at the AssociationofAmerican Geographers meeting, Louisville, Kentucky, April 1980): GeoSpeleo Publications, Cincinnati, Ohio, 167 p. Douglas, H.H., 1964, CavesofVirginia: Virginia Cave Survey, Falls Church, Virginia,761p. Dreybrodt, W., 1988, Processes in Karst Systems Physics, Chemistry and Geology: (Springer SeriesinPhysical Environment 4): Springer-Verlag, New York, 288p.Farrar, B., 1989, Cave becomes an area time capsule: The Southwest Times (pulaski, Virginia), v. 86, no. 73 (27 March 1989), p. 1-2. Ford, T.D. and Cullingford C.H.D. (editors), 1976, The ScienceofSpeleology: Academic Press, New York, 593 p. Ford, D.C. and Williams, P., 1989, Karst Geomorphology and Hydrology: Unwin Hyman, Winchester, Massachu132 Appalachian Karst Symposium, 1991 setts, 320 p. Gulden, R., 1989, Listsoflong cavesofthe United States and the world (unpublished): National Speleological Society Committee on Long and Deep Caves. Harler, C., 1991, Stillhouse Cave: Cleveland Grotto clean-up: Nearby grottos help make project a success: NSS News (National Speleological Society),v.49,p.172-175. Herak,M.and Stringfield, V.T. (editors), 1972, Karst: Important Karst Regionsofthe Northern Hemisphere: Elsevier Publishing Company, New York,551p. Hill, C.A. and Forti, P., 1986, Cave Mineralsofthe World: National Speleological Society, Huntsville, Alabama, 238p.Holsinger, J.R., 1975, DescriptionsofVirginia caves: Virginia DivisionofMineral Resources Bulletin85,450 p. plus 7 plates. Holsinger, J.R., 1985, Annotated ListofSignificant CavesandKarstAreasin Virginia: Virginia Speleological Survey (limited distribution document, revised April 1985),251p.Holsinger, J.R. and Culver, D.C., 1988, The invertebrate cave faunaofVirginia and a partofeastern Tennessee: Zoogeography and ecology: Brimleyana (The Journalofthe North Carolina State MuseumofNatural Sciences), no.14(June 1988), p. 1-162. Holsinger, J.R.; Baroody, R.A.; and Culver, D.C., 1976, The invertebrate cave faunaofWest Virginia: West Virginia Speleological Survey Bulletin, no.7,82 p. Hubbard, D.A., Jr., 1983, Selected karst featuresofthenorthern Valley and Ridge province, Virginia: Virginia DivisionofMineral Resources Publication, no. 44, one sheet (scale 1:250,000). Hubbard, D.A., Jr., 1984, Sinkhole distribution in the central and northern Valley and Ridge province, Virginia, in Beck, B.F. (editor), Sinkholes: Their Geology, EngineeringandEnvironmental Impact: Proceedingsofthe First Multidisciplinary ConferenceonSinkholes, Orlando. Florida, 15-17 October 1984:A.A.Balkema, Rotterdam and Boston, p. 75-78. Hubbard, D.A., Jr., 1988, Selected karst featuresofthecentral Valley and Ridge province, Virginia: Virginia DivisionofMineral Resources Publication,no.83, one sheet (scale 1 :250,000). Hubbard, D.A., Jr., 1989, Sinkholes: Virginia Division of Mineral Resources brochure, 2p.Hubbard, D.A., Jr., 1990, Geologic mapofClarke Coun-

PAGE 141

Appalachian Karst Symposium, 1991 ty, Virginia Plate1;Mapofhydrogeologic compo nents for Clarke County, Virginia Plate2:Vir ginia DivisionofMineral Resources Publication. no.102,2map sheets (scale 1:50,(00). Hubbard, D.A., Jr., 1991, Caves and Caving in Virginia: Virginia DivisionofMineral Resources brochure, 2p.Huppert, G.N., 1988, Cave and karst related theses in United States and Canadian universities: 1899-1988: Cave Geology,v.2, no.1,82p. Jennings, J.N., 1985, Karst Geomorphology (revised and expanded editionofJennings, 1971): Basil Blackwell, Oxford and New York, 293 p. Jones, W.K., 1973, Hydrologyoflimestone karstinGreenbrier County, West Virginia: West Virginia Geol ogical and Economic Survey Bulletin, no. 36, 49 p. plus 2 plates. Jones, W.K., 1983, Karst hydrology in West Virginia(1983), in Medville, D.M.; Dasher, G.R.; and Werner,E.(editors),Anintroduction to the cavesofeast-central West Virginia: National Speleological Society Guide book Series, no. 23, p. 25-34. Kastning, E. H., 1977, Faul ts as positive and negative influences on groundwater flow and conduit enlargement,inDilamarter,RRand Csallany, S.C. (editors), Hydro logic Problems inKarst Regions: Western Kentucky University, Bowling Green, Kentucky,p.193-201. Kastning, E.H., 1984, Hydrogeomorphic evolutionofkarsted plateaus in response to regional tectonism, in LaFleur,RG.(editor), Ground Water as a Geomorphic Agent: International Series, no. 13, Allen and Unwin, Inc., Boston (proceedingsofthe 13th Annual ("Bing hamton") Geomorphology Symposium, Rennselaer Polytechnic Institute, Troy, New York, September, 1982),p.351-382. Kastning, E.H., 1986, Cave regionsofthe United StatesofAmerica, in Middleton, J. and Waltham, A., The Underground Atlas: A Gazetteerofthe World's Cave Regions: Robert Hale, Limited, London,p.203-220. Kastning, E.H., 1988, Karstofthe New River Drainage basin, in Kardos, A.R. (editor), Proceedings. Seventh New River Symposium. Oak Hill. West Virginia. April7-9.1988: New River Gorge National River, Oak Hill, West Virginia, p. 39-49. Kastning, E.H., 1989a, Karst Geomorphology and Hydro geology: A BibliographyofPrincipal References: Limited private printing, 8p (available from the author). Kastning, E.H., 1989b, Surficial karst patterns: Recogni tionandinterpretation,in Beck, B.F. (editor),133Kastning and Kastning Engineering and Environmental ImpactsofSinkholes and Karst: Proceedingsofthe Third Multidisciplinary Conferenceon Sinkholes and the Engineering and Envi ronmental ImpactsofKarst, St. Petersburg Beach. Florida. 2-4 October 1989: A.A. Balkema, Rotterdam and Boston,p.11-16. Kastning, E.H., 1989c, Environmental sesnsitivityofkarstinthe New River drainage basin, in Kardos,A.R(editor), Proceedings. Eighth New River Symposium. Radford. Virginia,20-221989: New River Gorge National River, Oak Hill, West Virginia, p. 103-112. Kastning, E.H., 1990, Virginia karst terrains: Unique problems associated with waste management and ground water protection, in Erchul, R.A. (editor), Proceedingsofthe SymposiumforVirginia Localities on Waste Management and Groundwater Protection. April3-4.1990 at the Virginia Military Institute. Lexington. Vir ginia: VMI Research Laboratories, Inc., Lexington,p.82-93. Kastning, E.H., 1991, Caves. Karst. and Environmental Impactinthe New River Drainage BasinofVirginia and West Virginia: Guidebookfora Geologic Fieldtrip. Appalachian Karst Symposium. Radford. Virginia.23March 1991: Radford University, DepartmentofGeol ogy, Radford, Virginia, 36 p. Kastning, K.M. and Kastning, E.H., 1990, In Karst lands ... What Goes Down Must Come Up!: Virginia Cave Board, DepartmentofConservation and Recrea tion, poster, 22" by 28". Kittredge, K., 1989, No place for trash: NewRiver Cur rent,v.1,no. 279, (13 January 1989), p. 1-2 (sectionofthe Roanoke (Virginia) Times and World-News). Krinilzsky, E.L., 1947, A fault-plane cavern: JournalofGeology,v.55,p.107-119. LaMoreaux, P.E. (editor-in-chief); Tanner, J.M.; and ShoreDavis, P., 1986, Hydrologyoflimestone terranes: Annotated bibliographyofcarbonate rocks (volume3):International AssociationofHydrogeologists. Inter national ContributionstoHydrogeology.v.2,341p.Lamoreaux, P.E., Warren, W.M., and others, 1970 and 1975, Annotated bibliographyofcarbonate rocks and hydrologyoflimestone terranes: Geological SurveyofAlabama Bulletin no. 94, parts A and E, 242p.and168p. LeGrand, H.E., 1973, Hydrological and ecological problemsofkarst regions: Science,v.179,no.4076(2March 1973), p. 859-864. Lovegrove, R., 1988, Tanker spills 3,000 gallonsoffuel: Roanoke (Virginia) Times and World-News,v.25,no.

PAGE 142

KastningandKastning21, (21 July 1988), p. B6. Miller, E.V. and Hubbard, D.A.,Jr.,1986, Selected slope categories and karst features mapofGiles County, Vir ginia:Virginia DivisionofMineral Resources Publica tion70,map, scale 1 :50,000, 1 sheet. Monroe, W.H., 1970, A glossaryofkarstterminology:United States Geological Survey Water-Supply Paper1899-K,26p. Ogden, A.E., 1976,The hydrogeologyofthe central Mon roe County karst, West Virginia:Ph.D. dissertation (unpublished),WestVirginia University, 263 p. Saunders, J.W.; Ortiz,R.K.;and KoerschnerIII,W.F., 1981, Major groundwater flow directionsinthe Sinking Creek and Meadow Creek drainage basinsofGiles and Craig Counties, Virginia, U.S.A.,inBeck, B.F. (editor),Proceedingsofthe Eighth International CongressofSpeleology, Bowling Green, Kentucky, July18-24,1981:National Speleological Society, Huntsville, Ala bama, v. 1, p. 398-400 (reprinted inThe Tech Troglo dyte,1985, v. 24, no. 2, p. 54-60). Schultz, A.P., 1981,Geologyofthe CityofRadford:un published map (scale1:7200),2sheets (copies available from the City Engineer's office, CityofRadford). Slifer, D.W., 1987, Rockbridge's illegal dumps:Focus on Water(Virginia Water Resources Research Center), no.2,8p. Slifer, D.W. and Erchul, R.A., 1989, Sinkhole dumps and the risk to ground water in Virginia's karst areas,inBeck, B.F. (editor),Engineering and Environmental Im pactsofSinkholes and Karst: Proceedingsofthe Third Multidisciplinary Conference on Sinkholesandthe134Appalachian Karst Symposium, 1991 Engineering and Environmental ImpactsofKarst, St. Petersburg Beach, Florida,2-4October1989:A.A. Balkema, Rotterdam and Boston, p. 207-212.Sweeting,M.M.,1973,Karst Landforms,ColumbiaUniversity Press, New York, 362 p. Trudgill,S.,1985,Limestone Geomorphology(Geomor phology Texts,No.8),Longman,Londonand New York, 196 p. Virginia Speleological Survey, 1987-present:Virginia Cellars,v.1-.Vlchek,F.,1991,Morethangarbage:NSSNews(National Speleological Society), v. 49, p. 176. Waltham, A.C., 1989,Ground Subsidence:Blackie and Son, London, 224 p. Werner, E., 1981,Guidebooktothe karstofthe central Appalachians: Preparedforthe Eighth International CongressofSpeleology, Bowling Green, Kentucky.U.SA..July18to24,1981:National Speleological Society, Huntsville, Alabama,51p. Werner, E., 1983, Effectsofhighways on karst springs An example from Pocahontas County,WestVirginia,inDoughterty,P.H.(editor),EnvironmentalKarst:GeoSpeleo Publications, Cincinnati, p. 3-13. White, W.B., 1988,GeomorphologyandHydrologyofKarst Terrains,Oxford University Press, New York, 464p.White, W.B. and White, E.L., 1984, Cave and karst-related papers in the mainstream scientific literature: A biblio graphy:Cave Geology,v.1 no. 9, p. 291-392.

PAGE 143

Appalachian Karst Symposium,1991HubbardRegional Karst Studies: WhoNeedsThem?DavidA.Hubbard, If.Virginia DivisionofMineral Resources P.O. Box 3667, Charlottesville, VA 22903ABSTRACTRegional karst studies delineating areasofcarbonate bedrock, sinkholes, caves, and permanent stream segments serve to indicate the degree to which areas, underlain by carbonate bedrock, have undergone karsti fication. There are three potential hazards associated with karst: subsidence, groundwater pollution, and flooding.Tothe inhabitantsofkarst areas, the lossoflife and propertytosinkhole collapseisprobably the most commonly perceived potential karst hazard. Karstisan unstable terrain typified by differential subsi dence with respect to sinkholesorthe highly irregular (pinnacled) soil-bedrock interface. Becauseofthe solutional originofkarst topography and its characteristic sinkhole, cave, and pinnacle features, the close relationship between karst and groundwater should be apparent The linkageofsinkholes and groundwater by caves or smaller conduits, the relatively rapid movementofgroundwater in conduit componentsofthe aquifers, and the directional, but commonly unpredictable, natureofkarst aquifers contribute to the high susceptibilityofkarst aquifers to pollution. Sinkhole-flooding hazards are most common in karst areas near base level; however, the useofsinkholes for drainage outfalls and poor managementofsiltation during construction can result in sinkhole ponding and flooding at any elevation. There is an increasing awarenessofa declineofgroundwater quality in Virginia, especiallyinkarst areas. Planners and local governmental officials are indicating concerns about the causesofgroundwater pollution and how it can be arrested. Regional karst studies, usedincombination with geologic and hydro logic information, provide the key to protecting groundwater resources from further degradationinthe Val ley and Ridge provinceofVirginia. Karst terrains are highly susceptible to pollution from waste disposal, leaks and spillsofchemicals and fuels, and the improper applicationofagricultural chemicals. The contain ment and transportationofchemicals, fuels, and wastes poses additional risk to karst aquifers becauseofthe instability inherent to karst terrains.IntroductionThick sequencesofvarious typesofcarbonate rock are cxposed to weathering in the Valley and Ridge physiographicprovinceofVirginia. Virtually all exposed carbonaterockinVirginia has undergone some degreeofkarstifica tion. The purposeofregional karst mappingisthe delineationofkarst areas and a relative characterizationofthe de greeofkarstificationineach area. Regional karst mappinghasbeen completed over approximately two thirdsofthe Valley and Ridge physiographic provinceata scaleof1:250,000.Thethree typesofhazards associated with karst are subsidence, flooding, and pollutionofgroundwater.Subsidence hazards have traditionally dominated thefearsofresidentsofkarst areas. Recently, concerns overthepollutionofgroundwater have prompted local govern mcnts to seek planning and land-use restriction as soluCopyright 1991, CommonwealthofVirginia 135 tionstoprotect their karstic groundwater resources.DistributionandMappingofKarstThe significant karst in Virginia is developed on the folded and faulted sedimentary carbonate rock types (Figure 1, L's) in partsof27 countiesinthe Valley and Ridge physiographic province. Sinkholes, caves, pinnacles, and subterranean drainage are indicativeofthe solutional originofthis karst topography; however, travertine-marl depos its are a common accretionary feature. Other karst areas include narrow beltsofmarble (Figure1,M's) in both the Blue Ridge and Piedmont physiographic provinces and par tially indurated shelly sands (Figure1,S's)ofthe York town Formation in the Coastal Plain physiographic pro vince. Karst features associated with the marbles include sinkholes, pinnacles, subsurface drainage, and possibly a few caves. Sinkhole development in the Coastal Plain

PAGE 144

Hubbard Appalachian Karst Symposium. 1991Figure1.Distributionofkarst in Virginia: L, M, and S indicate counties in which limestone and dolomite, marble, and shelly sands, respectively, have undergone some degreeofkarstification. Regional karst maps have been published for areas I (Hubbard, 1983) and 2 (Hubbard, 1988), and county karst maps published for Clarke (3; Hubbard, 1990) and Giles (4; Miller and Hubbard, 1986) counties. province isnotlimited to the outcrop extentofthe York town Formation; sinkholes are also found in the Bacons Castle, Windsor, and Charles City formations overlying the Yorktown carbonate-rich sands (Johnson and others, 1987). Subsurface drainage, pinnacles, andatleast one travertine deposit are also known, but no hard evidenceofextensive lateral solutional conduit development has been documented. Regional mappingofkarst in Virginia has character ized karst intensity by the location and distributionofsinkholes and caves in areas underlain by carbonate rocksintwo thirdsofthe Valley and Ridge provinceofVirginia (Figure1,areas 1 and 2). Two published 1 :250,OOO-scale maps (Hubbard, 1983, 1988) indicate the relative degreeofkarstification. Geologic information and cave locations were compiled from published and unpublished sources. Sinkholes were located by stereoscopic examinationofaerial photography previously used for photoinspection and revisionof1 :24,OOO-scale (7 .5-minute)topographic maps. This method yields significantly more sinkholes than sim ply using those indicated on the topographic maps because the contour intervals on the maps are too large to provide sufficient resolution. Thorough ground surveys, such as soil surveys, provide the most accurate representationofsinkholes, but accurate positioningofsinkholes onto base maps can be a problem, and pseudosinkholes such as old excavations and sag ponds maynotbe distinguishable from true sinkholes. A comparisonoftechniques used in Clarke and Giles counties reveals that only 19%ofthe sinkholes identified from photography are located on the topographic maps and only 19%ofthe sinkholes identifiedinthe soil survey are identifiableonthe photographyor136 only 3%ofthe soil-surveysinkholes canbefound on the topographic mapsofClarke County. Similarly, only 34%ofthe sinkholes identified from photography are locatedonthe topographic mapsofGiles County. County karst maps, depicting sinkholes and caves,forClarke (Figure1,area 3; Hubbard, 1990) and Giles (Fig ure1,area 4; Miller and Hubbard, 1986) counties have been published at a 1:50,000 scale. Sinkhole and caveinformation was determined by the same methods as for the regional karst maps. Additional work with sinkhole loca tions in Rockbridge (Ron Erchul, 1986, oral communica tion) and Montgomery (John Flynn, 1991, oral communi cation) counties has been conducted on 1:24,OOO-scale topographic maps. Both regional and county karst maps are usefulindetermining the relative degreeofkarst development for an area; however, these tools are for pre liminary assessment and are not intended to replace site specific evaluations.Potential Karst HazardsThere are three potential hazards associated with karst: subsidence, pollutionofgroundwater, and flooding. Karst terrains are inherently unstable and are typified by differen tial subsidence, expressed as sinkholes. Man-made struc tures also may undergo differential subsidence if their foun dations are not designed for the highly irregular (pinnacled) soil-bedrock interface common in the Valley and Ridge karst areas. The greatest perceived hazard to inhabitants of karst areas is subsidence. Although sinkhole formationisa natural process in karst, man isnotnecessarilyaninno cent victim. Man-induced hydrological alterations com-

PAGE 145

Appalachian Karst Symposium. 1991monly trigger the mechanismsofsinkhole development (Sowers, 1976; Hubbard,1989). Typically, surface water (runoff, drainage outfalls, or leaks) is introduced or reroutedandpercolates to solutionally enlarged bedrock fracturesandcreates voids that eventually stope to the surface, or fluctuationsin the groundwater levels (water well pumpingorextended droughts) drain water-filled voids that stopetothesurface to form sinkholes. Sinkhole flooding is not a major probleminthe karst areasofVirginia,butmay result from the constriction or pluggingofsinkhole drains or by the overwhelmingofthese natural drains by increases in runoff from artificial surfaces. Inadequate controloferosion during constructioncanpermit sediment-laden runoff to restrict or plug sink hole drains. The increased runoff from residential, com mercial, or industrial surfaces such as from roads, parking lots, and man-made structures is significant (Aley and Thomson, 1981) and can overwhelm the capacity of nearby sinkhole drains and connecting subsurface conduits. Sinkholeflooding occurred after the developmentofa shopping center and housinginsinkholes in the Fairlawn areaofPulaski County, Virginia in 1980 (R.B. West, 1980, oral communication). Litigious problems arise from flooding associated with housing developments (Quinlan, 1984). The most extensive hazard associated with the ValleyandRidge karst terrains is groundwater contamination.Thekarst aquifers are complex reservoirs containing diffuse(slow)and conduit (fast) flow componentsina "black box" arrangement. Contaminants introduced (by sinkholes, caves, or thin soil over fractured bedrock) into the karstic aquifcr may rapidly appearatsprings, cave streams,orwater wellsormay not appearatall. Contaminant inputsinkarst aquifers do not disappear, they just may not appear whcre they can be readily observedorare expected. The appropriate question in groundwater monitoring in karst tcrrains is not "can we detect the tracers or contaminants?",but"where are the tracers or contaminants going?" In karst terrains, "into the ground"issynonymous with "intothegroundwater." The three karst hazardsofsubsidence, flooding, and pollutionofgroundwater suggest that karst terrains are notwellsuited to human occupation. Karst areas are unstablewithrespect to topography as well as groundwater qualityandthe degreetowhich this instabilityisapparentisprobablygrossly proportional to the level to which man hasdevelopedthese areas.Governmental ConcernsinKarstInthe early 1980s, there was a subtle shiftin the requests by governmental authorities for geologic informationinthe sitingofwater wells for public water supply, especially from Augusta, Rockingham, Smyth, and WiseCounties.Not only were groundwater resources sought,butconcerns were expressed about the integrityofkarst aquifer recharge and if groundwater could be protected137Hubbardagainst contamination in karst areas. In 1986, county administrators and regional planners sought hydrogeologic information to formulate a groundwater protection plan for Clarke County. Environmental concerns about karstic groundwater quality have since been expressedinBotetourt County. Over the last year, concerns for the protectionofkarstic groundwater quality have been expressed by Mont gomery, Shenandoah, and Page counties. Repetitive themesofinquiry include: How are karstic groundwater resources polluted?; What can or cannot be putinsinkholes?; What should we do about sinkholes?; How can we protect and insure the qualityofour ground water resources?Utilizing Regional Karst MappingRegional karst maps in the Valley and Ridge physio graphic provinceofVirginia indicate the extentofthe carbonate rocks as well as the distributionofsinkholes and caves. A reasonable assumption is that the greater the densityofsinkholes and caves the greater the potential for subsidence and groundwater-pollution hazards; however, the absenceofsinkholes and caves does not indicate that no subsidence or groundwater-pollution hazardsexistRe gional karst maps depict a spectrumofthe sinkholes ob servable on aerial photography and the locationofknown caves. Unmapped sinkholes exist as well as unreported caves. The on-site absenceofsinkholes and caves does not indicate a lackofkarst. Herein lies the importanceofmapping the extentofcarbonate rock, as indicated on the regional karst maps, because all exposed carbonate rockinthe Valley and Ridge provinceofVirginia has undergone some degreeofkarstification. Land modification, contour smoothing, and filling in sinkholes and caves may alter the appearanceof the land, but the potential for subsidence, flooding, and pollutionofgroundwater still exists and mayinfact have been increased. Karst maps are most effective when utilized in con junction with geologic and soils maps. Correlations between patternsofkarst features and rockorsoil units may imply an increased riskofsubsidence or groundwater pollution hazards. Specific rock types may contain relatively higher or lower densitiesofsolutional features (sinkholes and caves); such trends may be especially man ifestedinfolded and faulted geologic structures. Rock units displaying relatively high densitiesofsolutional fea tures may represent important recharge areas for the groundwater aquifer; however, karstic aquifers are quite complex and even detailed hydrological studies may leadtoonly an elementary understandingofa specific karst area. Soils maps can provide information about the distributionofsoils too impermeable for septic-tank drainage fields; unfortunately soils that do not percolate slowly enoughtoadequately filter septic-tank effluents are not indicated. Karst areas are poor sites for waste disposal and stor age. Leachate and effluents from disposal and storage sites

PAGE 146

Hubbardpresent a potential groundwater-pollution hazard in karst terrains. Regional karst maps may aid in the preliminary sitingofwaste-disposalandwaste-storage facilities. Cer tain chemicals, includingsomedefoliants and pesticides should notbeused in karst areas; liquid hydrocarbons and other chemicals may pose substantial riskstogroundwater resources from improper storage, use,ordisposal in carbonateterrains. Storageandpipelinetransferofsomeliquid chemicals, fuels, and hazardous wastes may represent a considerable risktogroundwater resources in intensely karsted areas, especially where storage and pipelines are not designed with respecttosubsidenceorcontainmentofleak age.The fIrst key to understanding howtominimize karst hazards istodetermine where the karst is.ReferencesAley, T.andThomson, K.C., 1981,Hydrogeologic Map pingofUnincorporated Greene County, Missouri,toIdentify Areas Where Sinkhole Flooding and Serious Groundwater Contamination Could Result from Land Development:ProjectSummaryprepared for Greene County Sewer District,byOzarkUnderground Labora tory,Protem,Missouri,undercontract with Missouri DepartmentofNatural Resources,11p. Hubbard, D.A., Jr., 1983, Selected karst featuresofthe northern Valley and Ridge province, Virginia:Virginia DivisionofMineral Resources Publication44,onesheet. Hubbard, D.A.,Jr.,1988, Selected karst featuresofthe central ValleyandRidge province, Virginia:Virginia DivisionofMineral Resources Publication83, one 138Appalachian Karst Symposium, 1991sheet. Hubbard, D.A., Jr., 1989, Sinkholes:Virginia DivisionofMineral Resources Brochure,2p. Hubbard, D.A., Jr., 1990, Mapofhydrogeologic compo nents for Clarke County, Virginia:Virginia DivisionofMineral Resources Publication102,plate 2. Johnson, G.H.;Burmester,J.L.;McKay,W.A.; and Wallister, W.G., 1987, Distributionandoriginofsink holes in the Atlantic Coastal Plainofsoutheastern Vir ginia (abstract):Geological SocietyofAmerica, Ab stracts with Programs,v.19, no.2,p. 91. Miller, E.V. and Hubbard, D.A., Jr., 1986, Selected slope categories and karst featuresmapofGiles County, Vir ginia:Virginia DivisionofMineral Resources Publica tion70,one sheet. Quinlan, J.F., 1984, Litigiousproblemsassociated with sinkholes, emphasizingrecent Kentucky cases alleging liability when sinkholes were flooded,inBeck, B.F. (editor),Sinkholes: Their geology. Engineering and Environmental Impact: Proceedingsofthe First Multi disciplinary Conference on Sinkholes, Orlando. Florida, 15-17 October1984,A.A. Balkema, Rotterdam, and Boston, p. 293-296.Sowers,G.F., 1976,Mechanismsofsubsidenceduetounderground openings,inSubsidenceovermines and caverns, moisture and frost actions,andclassification:Transportation Research Board Record612, National AcademyofScience, Washington D.C., p. 2-8.

PAGE 147

Appalachian Karst Symposium. 1991 DoughertySpatial-Temporal CharacteristicsofKarst Subsidence in the Lehigh ValleyofPennsylvaniaPercyH.DoughertyDepartmentofGeography Kutztown University, Kutztown, PA 19530.ABSTRACTCausesofkarst subsidence and the geographical distributionofthis phenomena are analyzed for the Lehigh ValIey, the local name in eastern Pennsylvania for the Great Valleyofthe Appalachians. This is oneofthe most densely populatedareas in the Appalachians and it is oneofthe areas most prone to karst damage in the United States with over $1,000,000 damage occurring yearly. Four karst-related deaths have occurred this century, the latest as recently as 1990. This study is a necessary fIrst stepinunderstanding the processes responsible for karst subsidenceinorder that mitigation techniques canbeinitiated. An examinationofdata bases supplied by local utilities and municipal agencies, supplemented by field inspection and analysisofnewspaper reports,topographic maps, and air photos, has ascertained that the Richenback Formation has the highest densityofsinkhole development. The LeithsvilIe Formation has the lowest densityofsinkholes per square kilometer, owing to its more dolomitic nature. The most damaging subsidences also occur on the LeithsvilIe Formation because it has a deep colI uvial cover that, in places, exceeds 30 meters and allows the formationofsuffosion dolines. Causesoflocal subsidenceare highly correlated with human activity and include: road construction, general construction, faulty utility-line installation, groundwater drawdown from weIls, reactivationofcovered dolines, faulty designofdetention basins and runoff swales, and leaking water and sewer lines. There is a definite seasonal correlation between doline formation and the early "spring thaw"oflate January and February, folIowed by a secondary periodofsubsidenceinsummer due to the declineingroundwater level.IntrOductionNewspaper headlines and storiesinother news mediainthe Lehigh ValIey make numerous references to "sinkholes"and karst collapse. Over $1,000,000inkarst-related damage occurs yearlyinthe Lehigh Valley, that partofthe Great Valleyofthe Appalachians that cuts diagonally from northeasttosouthwest across Pennsylvania. In the United States, this amountofdamageissecond only to the much larger karst-prone areaofcentral Florida. Some collapse episodes within the past five years have resultedinlosesinexcessof$500,000each: theMacungiesinkhole(Dougherty and Perlow, 1988), the Vera Cruz road collapse (Bonaparte and Berg, 1987), and the Allentown church disaster (Clark and Rcaman, 1988). Becauseofthe high densityofpopulation, there is also a dangertohuman life. Three lives were lostina 1925 colIapse in the CityofAllentown (Wittman,1988), and another death and seven injuries resulted from the collapseoftwo townhouses andanaccompanying gas explosion on August 29, 1990 (Cassler, 1990).139It is not unusual to see headlines in local newspapers like "Residents flee street-gobbling Macungie sinkhole" (Buzgon, 1986), "Another day in the Valley, another sink hole" (Whelan, 1986), "30-foot sinkhole opensinshop ping center" (Morning Call, August4,1986), "Emergency workatUpper Saucon sinkhole complete" (Morning Call, November 4, 1986), "PennDOT says it's nottoblameinlatest sinkholeinUpper Saucon"(Darrah,1987), "Another U. Saucon road is affected by sinkhole" (Morning Call, March 7, 1987), "Sinkhole threatens to undermine North ampton Borough home" (Berton, 1987), "Muhlenberg dormitory gets that sinking feeling" (Youngwood, 1988), "City church collapses into sinkhole" (Clark and Reaman, 1988), "City firm hired to fill sinkholesatABEAirport"(Cowen, 1988), "Lower Nazareth woman files lawsuit over sinkholes" (Morning CalI, July 27, 1989), and many more. The news reports only indicate the largest and most disastrous sinkhole occurrences in the area because most colIapses are not reportedinthe news media. Perusalofroadmaster recordsinsuburban and rural townships show the problem tobemuch greater than indicated by the news

PAGE 148

Doughertymedia. Local residents are also eager to share accountsoftheir favorite neighborhood sinkhole collapse, telling sto ries about missing dogs and disappearing back yards, and even a humorous accountofa football coachata high school football game who, while pacing the sidelines, was engulfed waist deep in a sinkhole. The above referenced and personal accounts show that there are numerous col lapsesinthis region and that the problem bears investiga tion becauseofits economic and life threatening impact. In order to minimize the lossoflife and the destruc tionofproperty, local government officialsmustknow what causes subsidence-collapse. Clues to the formationofcollapses can be found in an analysisoftheir spatial and temporal distribution. It is important from a planning per spective to know what areas are the most prone to collapse so that zoning and subdivision ordinances can be writteninsuch a way as to minimize the danger from subsidence. The temporal aspect is also important so that emergency service organizationscanplan for the possibilityofa periodofgreater collapse activity. Therefore, it is the purposeofthe current research to investigate the causesofkarst collapse, the spatial distributionofoccurrences, and the temporal aspectsinthe Lehigh Valley. The Lehigh Valley is representativeofthe Great Valleyofthe Appala chians, and information from this study can be applied to the Lebanon Valley, East Penn Valley, Shenandoah Valley and other similar areasinthe Great Valleyofthe Appala chians. Thisisespecially trueofthe Reading, Harrisburg, and Hershey urban areasofPennsylvania where the strati graphic profile is similar.StudyAreaThe Lehigh Valley is generally considered to be that partofthe Great Valleyofthe Appalachians extending from the DelawareRiveron the east to the Schuylkill River on the west (Figure 1). This encompasses Lehigh and Northampton counties and includes the Allentown/ Bethiehem/Easton metropolitan area with a populationofnearly three-quartersofa million (Joint Planning Com mission, 1991).Ifnearby Reading is included, the urban area exceedsonemillion people, mostofwhom live on the limestoneofLehigh Valley. Contrary to the negative publicity the area received from the song "Allentown" by Billy Joel, the area is not depressed and withering on the vine.Itis a dynamic urban area that has been stimulatedbythe recent completionofInterstate 78. The area already contains theNortheastExtensionofthePennsylvaniaTurnpike andothermajor highways suchasroutes 22, 100, and 309. With easy access to New York City and Philadelphia, the region has experienced substantial growth as a warehousing center. Inexpensive office space has resultedinan influxoftertiary activities that have replaced jobs lost in the shrinking heavy industrial base. Expan sion has resulted in an 8.1%increase in population over thepastten years (Joint Planning Commission, 1991). That increase, along with movement from the core cities to the suburbs, is resulting in increased urban sprawl. 140Appalachian Karst Symposium. 1991Figure 2 shows the urban land use that is concentrated on the limestone lowlands, a use which may not be compati ble with the karst landscape.TheLehigh Valley is a distinct physiographic region located between South Mountainofthe Blue Ridge Pro vince, composedofPre-Cambrianand Cambrian-Ordo vician granitic gneiss, quartzite and sandstone, and Blue Mountain, the fIrst ridgeofthe Appalachians, composed of SCAlE,. ......100 ""un i! 1. iIi!ii! ,. .... loOOl.nOMn ...Figure1:Locationofthe Lehigh Valley study area, Penn sylvania (shown shaded).MapfromWoodand others, 1972. Figure 2: Urban land use in the Lehigh Valley, Lehigh and Northampton counties, Pennsylvania. Map from Joint Planning Commission, 1991.

PAGE 149

Appalachian Karst Symposium. 1991Silurian sandstone and conglomerate, partially metamor phosedtoquartzite. Between the two resistant ridges lies a 25-km-wide valley having over 400 minrelief. The val ley flooriscomposedofthe Martinsburg Formation shale forming a high structural benchofundulating topography on the northwestern sideofthe valley, and limestone on the southeastern sideofthe valley forming a flat agricul tural plain stretching to the baseofSouth Mountain (Mil ler and others, 1942). Figure 3 shows the geologic forma tionsofthe Lehigh Valley westofAllentown, extending from Kutztowninthe south to Slatingtoninthe north (Lash and others, 1984).SpatialAttributesofKarstCollapseKarst collapse in the Lehigh Valley is restrictedtotheDoughertylimestone belt on which mostofthe urban developmentislocated. There are six limestone formations within this zone. The shaly limestoneofthe Jacksonburg Formation in the northwest has several Portland Cement quarries. This is followed in sequencebythe progressively older Epler, Rickenbach, Allentown, and Leithsville formations. Table I shows the thickness and characteristicsofthe for mationsinthe Allentown area (Myers and Perlow, 1986). From a cursory examination, one should expect to find many sinkholes in the Allentown Formation becauseofits great thickness and large aerial extent. Fewer sinkholes should occur on the shaly Jacksonburg Formation because itisthinly bedded and impure. In addition, there should be few sinkholes on the LeithsvilIe Formation because itishighly dolomitized and is covered by an extensive South Mountain colluvium (exceeding 30 meters in places). The Cal CalCalOeCal Cnl .... ------CalFigure3:Geologic formationsofthe Lehigh Valley westofAllentown, PA. Shaded areastothe northwest are the shale hills composedofthe Hamburg sequence (Ohsg) and the Martinsburg Formation (Om). The limestone lowlandofthe Lehigh Valleyisin white and is composedofthe Jacksonburg Formation (Ojk), Ontelaunee Formation(00),Epler Formation(Oe),Rickenbach Formation (Ori), Stonehenge Formation(Os),Allentown Formation (Cal), and the Leithsville Formation (Clv). The shaded areastothe southeast are the Hardyston Formation (Cha) and the undifferentiated gneissesofSouth Mountain. (Source: Berg and Dodge, 1981.)141

PAGE 150

Dougherty Ontelaunee is a minor formationinaerial extent and there fore has few sinkholes. Table 2 shows the resultsofa studyofsinkhole occurrenceinthe Lehigh valley (Myers&Perlow, 1986). Data for the study was taken from topographic maps, aerial photographs, utility company records, and recordsoflocal government and engineering offices. A totalof1574 sink holes were identified.Itis not surprising to find the larg est numberofsinkholes on the Epler and Allentown formations, covering the greatest geographic area, and the least number on the Jacksonburg and Ontelaunee forma tions covering a small area. Note that the Rickenback Formation has the highest densityofsinkholes per unit area (9.3 per square mile), followed by the Allentown Formation with 8.6, and the Epler Formation with 6.9. An attempttoclassify the sinkholes into categories basedontheir origin shows the Richenback and Epler formations have the highest densityofnaturally occurring sinkholes, Appalachian Karst Symposium. 1991 whereas the Allentown Formation has a high utility-related component. Several recent sinkhole episodes, not reported by the media, have occurred on the Allentown Formation after it was stripped during constructionofa housing development. This means that human action aggravates sinkhole formation in an area that otherwise does not have a significant numberofnaturally occurring sinkholes. Therefore planners should pay more attention to the Rickenback, Allentown, and Epler formations because of the great likelihoodofsinkholes developinginthem. Investigationofnewspaper clippingsofsinkhole formation and field visits to sinkhole sites add further information to the previous study that is not apparentfromthe table. Although the Allentown Formation has a larger numberofsinkholes than most other formations, the indi vidual sinkholes are small becauseofthe thin overburden and the small sizeofthe joint-controlled points of recharge. Sinkhole "eyes" are close to the surface andareTable1:CharacteristicsofLehigh Valley Carbonate Rocks. Source: Myers and Perlow, 1986. Formation (Age) Thickness (m) Formation Description and Weathering Characteristics Jacksonburg For170-460 Dark-gray shaley limestone grading downward into crystalline, high-calcium limemation (M.Ord.) stone. Low to moderate porosity and permeability; thin soil mantle; relativelyfewsolution features. Ontelaunee For0-200 Medium-gray, finely crystalline dolomite; cherty at base; missingatmany locamation (L. Ord.) tions. Solution-enhanced porosity and bedrock pinnacles characteristic. Moderatetothick soil mantle. Epler Formation 270 Interbedded very fine grained, medium-gray limestone and gray dolomite. Solution(L.Ord.) enhanced porosity; few bedrock pinnacles; very thick soil mantle. Richenbach For220 Gray, fmetocoarse dolostones, thin beddedattoptothick bedded toward base. mation (L. Ord.) Solution-enhanced porosity and bedrock pinnacles characteristic; moderately thick soil mantle. Allentown Dolo575 Alternating bedoflightand dark-gray weathering dolomite; stromatolites and mite (U. Camb.) oolites common; some orthoquartzite beds. Solution-enhanced porosity and bedrock pinnacles characteristic; soil mantle generally thin. Leithsville ForInterbedded fine-tocoarse-grained dolostones andtanphyllite; few thin sandstone mation (Uppermost 350 L. M. Camb.) beds. Solution-enhanced porosity; bedrock and pinnacles common; commonly covered with thick colluvium near uplands. Table2:Average Sinkhole Density for various geologic formations and sinkhole types. Source: Myers and Perlow, 1986. TotalTotal Average Sinkhole Sinkhole Densit (sinks/square mile) FormationAreaNo.ofDensity (NoJmi 2 ) Naturally Construction Utility Structure (mi 2 ) Sinks(alloccurrences) Occurring Related Related Related Jacksonburg 24.0 054 2.2 1.8 0.3 0.1-Ontelaunee 06.4 028 4.2 1.4 1.4 1.4-Rickenbach 18.7 174 9.3 4.5 4.0 0.8 0.05 Epler 74.5 518 6.9 4.0 2.5 2.5-Allentown 85.07318.6 2.2 2.24.2-Leithsville 32.0 069 2.1 1.0 0.2 0.9-142

PAGE 151

Appalachian Karst Symposium. 1991easily repaired.TheLeithsvilleFonnationon the other hand, has the lowest densityofsinkholes (2.1persquare mile), but it is the site ofsomeofthe most disastrous collapsesin the area.TheMacungie sinkhole formed on June24,1986with a resulLing hole that was40meters acrossandnearly20meters deep andcostin excessof$700,000torepair (Dougherty&Perlow, 1988).Thevery large size of the sinkholeisrelated to the deep colluvial coverofthe formationallowingthedevelopmentofsuffosion sink holes. This is also the location where allogenic watersfrom the Hardyston Sandstone and undifferentiated gneisses of South Mountaincomein contact with the limestoneoftheLehigh Valley. Large, damaging sinkholes fonnatthislocation, although the overall densityofsinkholes is lowerinthe Leithsville than in any other fonnation in the Lehigh Valley.TemporalCharacteristicsofSubsidenceInthis context, temporal characteristics refer to the monthsofthe year when karst subsidence is most likelytooccur in the Lehigh Valley. A crude surrogate value to assess this property was needed because itisdifficulttoget access to records that list datesofsinkhole collapse. Arti cles about sinkholes in the AllentownMorning Call,the largest-circulation newspaper in the Lehigh Valley, were counted over the five year period from 1986-1990. Eigh ty-nine articles have sinkhole in their title or are directly related to subsidenceorsinkholes. Although many articlesarenot related to sinkhole subsidence specifically, theymayhave been prompted by a recent sinkhole collapse or a periodofsinkhole-collapse.Duringthesameperiod, another 22 subsidences were documentedbyfieldwork in Lower Macungie Township in suburban AIJentown. The months in which theIIIsinkholes fonned are shown onthegraph in Figure 4. There is a definite seasonal distributionofsinkhole occurrence with a peakinthe summer and early fall, and a secondary peak in January and February.Thesummerpeakcorrespondstothetimeoftheyearwhen the groundwater is at its lowest and there is little water in the 25r---------------------, JANFEBMARAPRMAYJUNJULAUGSEPOCTNOVDECFigure 4:Numberofnewspaper articles on karst subsidenceinthe Lehigh Valley, 1985-1990 by monthofoccurrence.Source:The Morning Call,Allentown, PA.143Doughertysoil orrockto helpbearthe weightofthe overburden. This is when suffosion sinkholes are most likely to occur (Dougherty and Perlow, 1988; Miller, 1987; Buzgon, 1986). When the ground is saturated there is lillIe chance for surficialrunoffto enterjointsandareasofincipient sinkholes; but,duringthesummer,surfacerunoffis directedtothese pointsofaccess and this preferential flow can reactivate a sinkhole.Theproblem is aggravated by excessive well pumpage due to a higher seasonal demand from municipalandagricultural wells. This results in a loweringofthe groundwater table and a lesseningofthe overburden support.Thereis also the possibility that excessive lawn watering can cause reactivationofsomeofthe smaller sinks that appear in domestic yards at this timeofthe year. Summer is thepeakconstruction period, and during this time large areasofearth are disturbed, resulting in newdrainagepallernsandthedevelopmentand/or reactivationofsinkholes. A January and February secondary maximumofsink hole activity is more difficult to explain. This follows aperiodofquiescence withDecemberbeing the lowest monthofsinkhole activity.Theground may be frozen and there is a higher likelihoodofsnowthan rain during December. Precipitation falling on frozen ground is more likelytoend up as surface runoff than as infiltration in the incipient sinkholes. By late January and February, an ear ly "spring thaw" occurs resulting in the liberationoflarge amountsofwater that flow into the sinkhole eye and can reacti vate it. Although not statistically significant becauseofthe small sizeofthe sample, the pallernofkarst collapses in thepastfive years indicates that themostprevalent mechanism is a typeofsubjacent urbankarstcollapse. Areas that have been developed are covered by buildings, streets, and parking lots, fonning an impenneable surface. Leaking water and sewer lines may provide someofthe water that flows under the urban caprock and seeks a drain. Under these conditions, loose unconsolidated sediments are washed below, leavinggapingvoidsunderthe urban landscape, and resulting in eventual collapse (Sanchez, 1988; Wittman, 1988; Clark and Reaman, 1988). Other collapses have occurred owing to indiscriminate fillingofsinkholes and later developmentofthe area (Harris, 1986; Lowry, 1987).Oncethe organic maller in the fill has decayed, the sinkhole may rejuvenate and cause a collapse. There are also several incidencesofsinkholes repeatedly forming on the same site, such as the Macungie sinkhole (Dougherty and Perlow, 1988) and the Vera Cruz sinkhole (Darrah, 1987).Otherepisodes have been caused by development (Myers and Perlow, 1986), runoff from high ways (Leffler, 1988), summer drawdownofthe water table (Miller, 1987; Dougherty and Perlow, 1988; Wittman, 1988),faulLyinstallationofwater and sewer lines (Nixon, 1988; Sanchez, 1988), andfaulLydesignofstonn-water detention basins and surface drainage routes (Lerner, 1988). Allmajorepisodesofcollapsementioned in the news media during the five-yc<'lf study period had a threshold that was most likely exceeded by human intervention.

PAGE 152

DoughertyConclusionsThis is only a preliminary analysis based on a small sample. Records from the roadmasters and highway depart ments in the area may eventually yield a larger sample on which to base a more detailed analysis. Spatially, the Al lentown and Leithsville formations appear tobethe most dangerous areasofthe Lehigh Valley; the former becauseofa higher densityofsinkhole development and the latter becauseofthe large suffosion dolines that form in the deep colluvial cover. Temporally, there is a summer maximumofsinkhole activity and a late winter/early spring secondary peak.Further work is needed to substantiate both the spatial and temporal patterns identified above.References3D-foot sinkhole opens in shopping center:The Morning Call.Allentown, PA, August 4, 1986. Another U. Saucon road is affected by sinkhole:TheMorning Call,Allentown, PA, March 7, 1987. Berg, T.M. and Dodge, C.M., 1981,AtlasofPreliminary Geologic Quadrangle MapsofPennsylvania:Pennsyl vania TopographicandGeologic Survey, Harrisburg, PA. Berton, V., 1987, Sinkhole threatens to undermine North ampton Borough home:The Morning Call,Allentown, PA, September 12, 1987. Bieber, S., 1986, Homes evacuated as hole opens in Macungie street:The Morning Call,Allentown, PA, June 24, 1986. Bonaparte,Rand Berg,RR,1987, The useofgeosyn thetics to support roadways over sinkhole prone areas,inBeck, B.F. and Wilson, W.L. (editors),Karst Hydro geology: Engineering and Environmental Applications: Proceedingsofthe Second Multidisciplinary Conference on Sinkholes and the Environmental ImpactsofKarst, Orlando. Florida.9-11February1987: A.A. Balkema, Rotterdam and Boston, p. 437-445.Buzgon,M.,1986,Residentsfleestreet-gobblingMacungie sinkhole:The Morning Call,Allentown, PA, June 24, 1986. Cassler, K., 1991, Gas explosion kills woman, levels Allentown row homes:The Morning Call,Allentown, PA, August 30, 1991. Clark, J. and Reaman, D., 1988, City church collapses into sinkhole:The Morning Call,Allentown,PA,February 18, 1988. Cowen, D., 1988, City firm awarded contract to fill sink holes at ABE (airport):The Morning Call,Allentown, 144Appalachian Karst Symposium.1991PA, July 27, 1988. Darrah, T., 1986, PennDOTsays its not to blameforlatest sinkhole in Upper Saucon:The Morning Call,Allentown, PA, September 25, 1986. Darrah, T., 1987,UpperSaucon horror: ReturnoftheSinkhole:The Morning Call,Allentown, PA, January 10, 1987. Dougherty, P.H. and Perlow, M., Jr., 1988, The Macun gie Sinkhole, Lehigh Valley, Pennsylvania: CauseandRepair:Environmental Geology and Water Science,v.12, no. 2, p.89-98.Dougherty,P.H., 1989,Landuse regulations in the Lehigh Valley: Zoning and subdivision ordinancesinanenvironmentally sensitive karst region,inBeck,B.F.(editor),EngineeringandEnvironmental ImpactsofSinkholes and Karst: Proceedingsofthe Third Multidis ciplinary Conference on Sinkholes and the EngineeringandEnvironmental ImpactsofKarst, St. Petersburg Beach, Florida,2-4October1989: A.A. Balkema, Rot terdam and Brookfield, p. 341-348. Emergency workatUpper Saucon sinkhole complete:TheMorning Call,Allentown, PA, November 4, 1986. Harris, K., 1986, Sinkhole site once a pond, geologist reports:The Morning Call,Allentown, PA, July1,1986. Joint Planning Commission, 1991,Population SummaryforLehigh and Northampton Counties:Allentown,PA.Kochanov, W., 1987, Sinkholes and Karst Related Fea turesofLehigh County, Pennsylvania:Pennsylvania Topographic and Geologic Survey,Open-File Report. L. Nazareth woman files lawsuit over sinkholes:The Morning Call,Allentown, PA, July 27, .1989. Lash, G.G.; Lyttle, P.T.; and Epstein, J.B., 1984, Geol ogyofan accreted terrain: The eastern Hamburg Klippe and surrounding rocks, eastern Pennsylvania:49thAnnual Field ConferenceofPennsylvania Geologists,Harrisburg, PA. Leffler, P., 1988, Farmers sayrunofffloods their fields:The Morning Call,Allentown, PA, June 20, 1988. Lowry, T., 1987,UpperSaucon sinkhole repair ques tioned:The Morning Call,Allentown, PA, January16,1987. Miller, B.L.andothers, 1942, Lehigh County, Pennsyl vania: Geology and Geography:PennsylvaniaTopographic and Geologic Survey. County Report39, 492p.Miller, T., 1987, Sinkhole prone Saucon Valley worries

PAGE 153

Appalachian Karst Symposium. 1991 planners: The Morning Call, Allentown, PA, December 28, 1987. Myers, P.B. and Perlow, M., Jr., 1986, Development, occurrence, and triggering mechanismsofsinkholes in the carbonate rocksofthe Lehigh Valley, eastern Pennsylvania, in Beck, B.F. (editor), Sinkholes: Their Geology, Engineering,andEnvironmental Impact: Proceedingsofthe First Multidisciplinary Conference on Sinkholes, Orlando. Florida. 15-17 October 1984: A.A. Balkema, Rotterdam and Boston,p.111-115.Nixon,J., 1988, Broken water main caused Easton sinkhole:The Morning Call, Allentown, PA, December IS,1988.Sanchez, L., 1988, City crews to repair washout below street: The Morning Call, Allentown, PA, February 20,1988.145 Dougherty Whelan, F., 1986, Another day in the Valley, another sinkhole: The Morning Call, Allentown, PA, June 26, 1986. Wittman,B., 1988, Deep historyofsinkholes in Allen town: The Morning Call, Allentown, PA, April 10, 1988. Wood, C.R.; Flippo,H.N.;Lescinsky, J.B.; and Barker, J.L., 1972, Water resourcesofLehigh County, Pennsyl vania: Pennsylvania Geological Survey Water Resour ces Report W31, 263 p. Youngwood,S.,1988, Muhlenberg dormitory gets that sinking feeling: The Morning Call, Allentown, PA, January 30, 1988. Zoning OrdinanceofLower Macungie Township: Lower Macungie. PA: Lower Macungie Township BoardofSupervisors, 1989.

PAGE 154

Plate Appalachian Karst Symposium, 1991Plate D (Above): Clean-upoflarge dump site at entrance to Stillhouse Cave, Randolph County, West Virginia, August 1990. This event was organized by the Cleveland Grottoofthe National Speleological Society with the cooperationoftheWest Virginia DepartmentofNatural ResourcesandDepartmentofHighways. Cavers from many grottos in the Appalachian region devoted two daystothe effort and removed15truckloads, or 22,500 pounds,ofdebris. The site is the first "Adopt-A Dump" Project in West Virginia.Seein this volume Kastningand Kastning,p.123, and Erchul, p. 147, for discussionsofclean-up projects andillegal dumping in sinkholes.Photograph by KarenM.Kastning.Plate E (Left): MembersofBoy Scout Troop46ofRadford, Virginia at entranceofNew River Cave, Giles County, Virginia, following a triptoclean trash from this heavily visited cave. Anewcave conservation program for scouting groups has been initiated in the New River ValleyofVirginia and West Virginia. Participationina cave or karst clean-up is a requirement fortheCave Conservation Awareness Award(seeFigure 2ofKastningand Kastning, this volume,p.130).Photograph by KarenM.Kastning.146

PAGE 155

Appalachian Karst Symposium, 1991Illegal Disposal in Sinkholes: The Threat and the SolutionRonaldA.ErchulDepartmentofCivil&Environmental Engineering Virginia Military Institute Lexington,VA24450ABSTRACTIn 1988 the Virginia Military Institute conducted a study supported by the Virginia EnvironmentalEndowmenton sinkhole dumping and the risk to groundwater in Virginia's karst areas.Thekarst areaofVirginia is located in the VaHey and Ridge Province and consistsof24 counties on the western edgeofthe Commonwealth.Dueto the karst/limestone geology that characterizes the region, these counties share acommonvulnerability to their groundwater resources. Karst aquifersareamong themostsensitive to disturbance and are readily contaminated.Oneofthe most visible and obvious sourcesofcontamination is the dumpingoftrash and waste into sinkholes. Botetourt and Rockbridge counties, the areas that were the focusofthe research, have thousandsofsinkholes within their boundaries.Thepossibilityofcontamina tion to local groundwater is great due to the presenceofhousehold hazardous waste in dumps. A totalof260 illegaldumpsites were documented in the study area, with 75 percentofthese existing in karst areas. Approximately 23 percentofthe illegal dumps wereinsinkholes. More than 90 percentofthe populationofthe study area take their water from wells and springs.Theresearch has concentrated on an effective methodologyoflocating and assessing potentially dangerous sinkhole-dump sites.Theproject has docu mented these techniques, allowing other localities in karst terrain to obtain similar information. A slide show and video have been produced to show local schools and citizen organizations the hazards associated with sinkhole dumping. Also, a survey cost analysis for identification, evaluation,anddocumentationofthese illegal disposal sites has been developed.Thetaskofcompiling a listingofall known open dumps within a regional boundary canbeeffectively realized using the methodology proposed in this study. In addition, the extent (numbers, nature, and sizes)ofeach site should be described to aidinproper remediation and elimination.ErchulIn trod uctionIllegal disposal in sinkholes hasbecomeso wide spread that it has resusted in oneofthe most comprehensivestudies ever conducted in the CommonwealthofVir ginia to identify, evaluate and document this threat to the environment.Thestudy was conducted by personneloftheVMI Research Laboratory (VMIRL) in 1988 and 1989andsupported by a grant from the Virginia Environmental Endowment. A previous independent study conducted in Rockbridge County, Virginia by Slifer highlighted the problem and alluded to the scope and magnitudeofsuch astudy.Slifer's study was published by the Virginia Water Resources Research Center (Slifer, 1987) and establishedtheguidelines and impetus to obtain more data to better define the threat to groundwater in the karst areasofVirginiacaused by illegal disposal in sinkholes. Fortunatelyforthe second study, Slifer was able to contribute as a 147 memberofthe VMIRL research team.Hewas active incoordinating, analyzing and documenting the data collected in this studyandcomparingitwith previous data he collected in Rockbridge County. This paper reviews the effectivenessandmethodologyofthe Rockbridge and VMIRL studies, provides a cost analysis, and discusses the draft and promulgated regulations relating to illegal disposalinlocal solid-waste management plans.The StudyAspreviously mentioned, a case studyofillegal dis posal sites in Rockbridge County, V irginia was conducted and documented by S lifer in 1986 and 1987.In1988 a similarbutmore comprehensive survey commenced in neighboring Botetourt County (Slifer and Erchul, 1989; Slifer, 1990). Botetourt and Rockbridgearesimilar in

PAGE 156

Erchulsize, geology, and population. Both counties are primarily rural and each operates a single sanitary landfill. Whereas Rockbridge maintains a solid-waste collection systemofdumpster boxes, Botetourt relies on private commercial haulers. These counties differ in their legal approach to the problem. Rockbridge lacks an ordinance that even recognizes or defines illegal disposal and has made little effort at enforcement. Botetourt has an adequate local ordi nance banning illegal disposal, but enforcement efforts have been random rather than from a systematic programtoidentify and eliminate illegal disposal sites. For both surveys on illegal disposal sites, information was collected by mailing a letter and county map to approximately two dozen persons who are knowledgeable about the counties. Reported sites were then field checked and evaluated.Inboth cases the rateofresponse exceeded 50 percent. This technique is inexpensive, simple, and produces reasonably good data for those sites that are the most visible and that are usually located along road sides.However,thisquestionnairetechniquedoesnotapproximate the actual numberofdisposal sites in a countyowing to the large numberofsites located on private property and not visible from roads. Consequently, in the Botetourt study, it was decidedtosupplement the question naire data by conducting aerial investigationsofthe county.A small plane was flown at altitudes low enough to identify most illegal disposal sites. Two flights were madeataltitudes ranging from 600 to 1,000 feet during periods when leaves wereoffthe trees. A total flight timeofapproximately 10 hours was required to cover and recheck the county. A setoftopographic quadrangle maps was used by a spotter to help guide the pilottofly a seriesofparallel flight lines across the county, and to record locationsofdisposal sites as they were observed.Itwas found that the most efficient way to record site locations was to note the latitude and longitude coordinates from a digital LORAN-C display as the plane was directly above a site. These coordinates, when plotted on topographic maps afterward, were usually accurate to within several hundred feet. A totalof106 disposal siles were recorded during these two flights, and manyofthem were photographed. The aerial data was then verified by field checking each site. The total numberofconfirmed sitesinBotetourt was 168.Anupdateofthe 1986 surveyofdis posal sites in Rockbridge County led to documentationof92 sites. Nearly twice as many sites are knowninBote tourt, primarily because that county was inventoried by ae rial survey. A similar aerial survey for Rockbridge would probably double the numberofdisposal sites known there. During the field check each site was evaluated and manywere photographed. A field-check form was developedtoensure a standardized evaluation. A score was developedforeach site by assigning numerical values to factors such as disposal-site size, site geology, topographic setting, public access, frequencyofuse, waste types, and proximitytowater. The ranked scores were assigned to categories that describe the relative threattogroundwater at each site. 148Appalachian Karst Symposium. 1991Forboth counties illegal disposal sites ranking inthemost serious categories comprise approximately 29 percentofthe total. These illegal sites are believed to be actualorimminent hazards in termsofgroundwater contamination. An averageofapproximately17percentofthe total numberofillegal disposal sites fell within the low-impact category. The numberofillegal disposal sites located in thetwokarst areas is nearly identical-73percent for Botetourtand74 percent for Rockbridge. Rockbridge has a greater shareofillegal-disposal sitesinsinkholes (29 percent) thandoesBotetourt (16 percent). This is probably due to a greater densityofsinkholes in the Rockbridge area. Because groundwaterisinherently at risk from illegal disposalinsinkholes, the weighingofranking factors tendstoplace sinkhole-disposal sites into the more serious categories. In Rockbridge 58 percentofthe illegal disposal siteshavepublic access, compared to only 40 percent in Botetourt. This is a reflectionofthe survey methodologies; because the Rockbridge survey was not aerial, it tendstofavorthemore visible roadside sites that, by definition, have public access.Animportant, apparent difference betweenRockbridge and Botetourt illegal disposal is the prevalenceof55-gallon drums observed in Botetourt sites. Drumsarerare in Rockbridge sites but were counted in21percentofthe Botetourt sites. Drums areofcourse often usedtocontain hazardous liquid chemicalsorwaste. Manyofthedrums are empty but some still contain unknownsubstances. Because personnel evaluating the Botetourtsiteswere not trained to sample for hazardous materials, dataisnotavailable to characterize contentsofthe drums.Perhaps Botetourt's relative proximity to the CityofRoanoke and its numerous industrial and commercial facilitiesmayexplain the presenceofdrums in at least 36 sites. Using the resultsofthe case studiesofRockbridgeandBotetourt counties, and extrapolating the data, thetotalnumberofillegal disposal sites within the ValleyandRidge Province (the karst regionofVirginia) is projected to be approximately 4032.Ofthat total, about2943would be located in karst and about 927 would beinsinkholes. Based on an averageof29 percent, approximately 1169 illegal disposal sites would be in the serious-threat. category.Ifillegal disposal is not limited to the karst regionofVirginia, the statistics for the entire Common wealth could exceed 15,000 illegal disposal sites through out the state. This alarming statistic would requireanactive educational program, stringent state regulations,andenforcement by localities if the numberofillegal disposal sites aretobe reduced and eliminated.Draft RegulationsEffective July1,1989, the Virginia GeneralAssembly directed the DepartmentofWaste Managementtoprepare regulations governing the preparationoflocal

PAGE 157

Appalachian Karst Symposium, 1991solid-wastemanagementplansandtospecifytheproceduresbywhich local governments must attain certain recycling goals.Anadvisory commitlee was formed and adrafldocument was prepared.Thedraft regulations were concep-tual innatureandwerebasedonnumerousperspectives.Thesedraftregulationswereto receive public comments and formal hearings and then wouldberevised tobecomethe promulgated regulations for the developmentofsolid-wastemanagementplansbylocalities in the Common-wealthofVirginia.Inthe draft regulations, specifically noteworthy to this study, was the requirement for all Virginia localities to provide a listing in their solid-waste management plans regarding all known open dump sites within the regional boundary, to name the current ownerofthe site, to des cribe the nature and sizeofthe site, and to determine the extentofillegal disposal and describe actions to eliminateit.The draft version defined an "open dump" as a site onwhichany solid waste is placed, discharged, deposited, injected, dumped, or spilled so astocreate a nuisanceorsoastopose, within the determinationofthe Executive Director, a substantial present or potential hazard to humanhealthor the environment, including the pollutionofair,land,surface water,orgroundwater. "Illegal disposal" wasdefinedas the disposal which is contrary to applicable laworregulations.Itappeared fortuitous that the regulation requirement just stated couldbeeffectively met by the methodology proposed inourcomprehensive study conducted on illegal disposalinBotetourt County. A letter was addressed to all countiesinthe Commonwealth informing themofthe studyandthat a cost analysis for the identification, evalua tion, and documentationofillegal disposal sites was being preparedtoallow them to understand the methodology andtheassociated costs. This would enable them to make necessarybudgetary adjustments if they desired to implementsucha study for their locality.SurveyCostAnalysisSome generalizationisrequiredincompiling an accuratesurvey cost analysis for the identification, evaluation,anddocumentationofillegal disposal sites for a specific locality. An assumption is that the cost would be basedonour research studyofa western Virginia county. Realizingthat flying, labor, and material costs will vary dependingon location, sizeoflocality, and extentofevalua tion required, the cost analysis presents rangesofcosts foreachitem. Various options are consideredinorder to provideanestimate for budgetary planning based on individual locality requirements. A survey cost analysisisdivided into three line items; identification, evaluation, and documentation. Table 1showsthe cost breakdownofeachofthese line items. The identification line-item is divided into maps, inquiry, aerial reconnaissance, and data recording. Mostofthe identifica149Erchultion costs are fixed, one-time purchases. However, even without a surveyofknowledgeable citizens, aerial recon naissance must be conductedifeffective and complete data is tobegenerated. Various options are offered in the evaluation line-itemofthe cost analysis that couldbeselected depending on the extentofdetailed information desired.Thegreater the detailed information desired the greater the cost, and con sequently the evaluation line-item will usually be the most influential factorinthe total survey cost. In addition, eachofthese options have subsets affecting cost. For example, each option could be conducted using available personnelorcontractingothersto dothework. VMIRL had proposed to use VMI cadets, home on summer furlough, toconductthe evaluation survey nearorin their home localities. This would providesummeremployment for cadets to conduct the evaluationifcounty personnel were not available. Cadets wouldbetrained in proper data collection techniques prior to conducting their survey. Finally, the documentation line-item provides the fi nalized data analysis, locationofsites on maps, and other detailed information desiredina final report and/or listing on a computerized spread sheet.Thecostofthe documen tation line-item is relatively fixed and the rangeofcost will only vary as a resultofrequiring contract supportorifthe work canbeconducted efficiently by available personnel. A survey for a county the sizeofBotetourt County (548 square miles) should cost between $4,500 to S8,400 depending on the detailofdata desired. This equates toS8persquare mile using available personnel andSISper square mileifcontracted personnel are required. These cost figures should be adequate for budgetary estimates for providing data to determine the location, size, nature and extentofsitesofillegal disposal and open dumping within the regional boundariesofany locality. By determining the locationofsites, the owner could be ascertained and a courseofaction couldbeprescribedtoeliminate the site. It is true that new illegal disposal sites could occur at any time and another survey should be plannedinthreetofive years to determine the effectsofthe current action to elim inate sites discovered on the first survey and to locate any new illegal disposal sitesifthey exist.PromulgatedRegulationsOnMay 15, 1990, after numerous public meetings and formal hearings on draft regulations, the Common wealthofVirginia DepartmentofWaste Management pro mulgated regulations for the developmentofsolid-waste management plans. These regulations made it mandatory for every city, county, and town in the Commonwealthtodevelop a solid-waste management plan and submit it to the state for approval. However, nothing was mentioned about requirements to provide a listingofall open dumps within the regional boundary,todetermine the extentof

PAGE 158

Erchul Appalachian Karst Symposium.1991In summary, this research demQnstrated a technically feasibleand cQst-effective approach and methQdologytoaccurately determine the number, location, size, andthreatto the environmentQfillegal disposal sites.Itprovided documentation in the formofa report, and video tape,andslidestQinform and educate peQpleofenvironmental problems associated with illegal disposal. ItalSQprQvidedcQmplete datafQrillegal dispQsal sites forQnelocality (BQtetQurt CQunty)inthe Commonwealth. A need existsConclusionsgenerateaplanwithQut infQrmationWQuldrequire data gathering, prohibit ing timely submission. b.ItWQuldbe difficult andCQstlytoenfQrceregulatiQns and determine actionstQeliminate illegal disposal sites. c. There are complex legal and financial aspects associated with clean upQfallillegal disposal sites within the Com monwealth.TheCommonwealthQfVirginia Depart ment QfWaste Management has excludedthedraft-regulationclausedealing with open dumps and illegal disposal in the promulga ted regulationsfQrthe develQpmentofsolid waste management plans for localitieswithin the Old Dominion.Itis this investiga tor'sQpiniQnthat the effectivenessofother formsQfwaste management, such as source reduction, recycling, landfilling, andcombustiQn, usedtQcQntrQl the waste stream will be questiQnableQrmisleadingifinfor matiQnQnillegal disposal isnQtdetermined.Theresearchpresentedin this paperonillegal dispQsal in sinkholes highlightedanenvirQnmental problem in the karst region of Virginia. However, it is this investigator'sQpiniQnthat illegal disposal in the rest of the Commonwealth, and probablyinmuchofthe natiQn, existsQnthe same scaleasinthetwocounties studied. BQtetQurtandRockbridge cQunties cQmprise approximatelyWOOsquare miles, and extrapolatiQnQfdatafrom a sampling areaofthis sizetQtherestQfthe state isnQtunrealistic. Even ifwecQnservatively estimate fromQurdatathat10,000 sitesofillegal dispQsal andopendumping existinthe CQmmQnwealth,thisis a considerable amQuntofthe wastestreamthat is tQtally unregulated. In addition,fromthe BQtetourt CQunty study, the fact that21percent ofthesites contained 55-gallQn drumsCQuldinfer that a partofthis unregulated waste stream maybehazardQUSwaste. a. InfQrmatiQn and dataQfsites isnQtreadily availabletQthe localitiesQrstate and thus the time needed to illegal disposal,QrtQprescribe actiQntQeliminate dumps. In fact, the Qnly pertinent items mentiQnedinthe final versionofthe regulations were the definitiQnsQfan"Qpendump" and "illegal disposal." The omissiQnQfany meanstocurtail open dumping and illegal disposal in the promul gated regulatiQns was very surprisingtQthis investigatQr in lightofthe data generatedinthetWQcQunties that were studied.Oninquiry with the DepartmentofWaste Man agement, the follQwingreaSQnswere given as to whythe illegal-disposal regulations were drQpped from any waste management plan. Table1:CQunty Survey CQst AnalySIS for IdentificatIOn, EvaluatiQn and DocumentatiQnQfIllegal Disposal Sites INDENTIFICATION: RangeQfCQstMaps (TQpographic and GeolQgic) $ 75-$ 100 Inquiry (Write letters, send questiQnnaires and maps, $ 370-$ 820 and process and plQt responses) Aerial RecQnnaissance (Aircraft rental flying time: 2 $1,500-$2,200 trips withonemake-up flight scheduled, 10 to 20 hours, film for aerial photographs) $ 500-$ 690 Data RecQrding (Data plQtting on maps and entryintQ$ 570-$ 870CQmnuter)IdentificatiQnTQtal$ 3,015-$4,680EVALVA TION: Option1:(Check all illegal disposal sites in detail. $ 1,500-$2,500 This includes measurements, analysisQfCQntents and photographic data) OptiQn 2: (Check Qnly the largestormQstaccessible $ 750-$1,250 illegal disposal sites) Option3:(Evaluate and prioritize accQrdingtQphotQ$ 550-$1,050 graphic data, DRASTIC mapsQrsimilar information on surface and grQundwater vulnerability) OptiQnlA,2A: (Meet with county representatives $ 300-$ 600 and train them in field checking and the datacQllectiQn routine and allow themtQcheck all sites) OptionIB,2B: (Allow VMI cadetstQcQllect data $1,000-$1,500 during the summer vacatiQn break:) EvaluatiQn Total $ 550-$2,500 DOCUMENTA TION: RepQrt andDataAnalysis (Assemble all data and $ 920-$1,220 analyses.Producereport and cQmputer spread sheets. Meet with county representativestQpresent results) DQcumentatiQn TQtal $ 920-$1.220 GRAND TOTAL $4,485-$8,400..150

PAGE 159

Appalachian Karst Symposium, 1991toconduct this typeofstudyinother localities. The data obtained would thenbeavailable for solid-waste manage ment plans, and a regulated, methodical eradicationofillegal-disposal sites could then progress throughout the Commonwealth.ReferencesSlifer, D.W., 1987, Rockbridge's illegal dumps:Focus on Water(Virginia Water Resources Research Center), no.2,8p.Slifer, D.W., 1990, Groundwater at riskinthe common dumpofVirginia,inErchul, R.A. (editor),Proceedings151Erchulofthe SymposiumforVirginia Localities on Waste Management and Groundwater Protection, April3-4,1990 at the Virginia Military Institute, Lexington, Virginia:VMI Research Laboratories, Lexington,p.96-101.Slifer, D.W. and Erchul, R.A., 1989, Sinkhole dumps and the risk to ground water in Virginia's karst areas,inBeck, B.F. (editor),Engineering and Environmental Im pacts on Sinkholes and Karst: Proceedingsofthe Third Multidisciplinary Conference on Sinkholesandthe Engineering and Environmental ImpactsofKarst,St.Petersburg Beach, Florida,2-4October1989: A.A. Balkema, Rotterdam and Brookfield, p. 207-212.

PAGE 160

Plate Appalachian Karst Symposium. 1991MustComeUp!What GoesDownPlateF:Environmental poster produced by the Virginia Cave Board, DepartmentofConservation and Recreation. This color poster. measuring 22 by 28 inches. has been distributedtoteachersofEarth Scienceinpublic schoolsofVirginia. The poster graphically emphasizes the sensitivityofkarst groundwater to contamination from dumping in sinkholes and other sources. Poster was funded by the VirginiaNatural Heritage Programofthe DepartmentofConservation and Recreation, Virginia DepartmentofEducation. and the Cave Conservancyofthe Virginias.SeeKastning and Kastning, this volume.p.123 for discussion.Photograph provided by Jack JeffersofRadford University.152

PAGE 161

Appalachian Karst Symposium. 1991 DeStephen and MilnerEvaluating a Landfill Expansion in Karst TerrainRaymondA.DeStephen, P.E. and Brian Milner, C.P.G.Schnabel Engineering Associates, Inc. One West Cary Street Richmond, VA 23220-5609ABSTRACTA landfill expansion is being proposed in Rockingham County, Virginia to meet future waste-disposal needs. The site is underlain by Ordovician-age dolostoneofthe Beekmantown Group. Unlike several other states, Virginia does not expressly prohibit constructionofa landfill inkarstHowever, the applicant must provide evidence that the landfillisnot sited in a geologically unstable area in which geologicorgeomor phic features may resultinsuddenornon-suddenevents that could cause subsequent failureofliners. Also, evidence must be giventoassert that a monitorable groundwater table exists beneath the site an often diffi cult task in karst terrain. This paper describes the geologic investigation and approach to risk assessment that was conducted to evaluate the site for landfill development. A numberoftechniques were utilized to collect subsurface data including test borings, air-track probes, air-rotary drilling, downhole geophysics, and slug tests. This data was evaluatedinconjunction with geologic reconnaissance data, fracture-trace data, air photos, and sinkhole maps to assess geologic risk factors at the site. Project-development risk factors that also could affect future sinkhole potential wereconsidered, such as dewatering, final depthofoverburden, and changesinsurface-water infiltration. Based on the assessmentofboth geologic and developmental risk factors, it was concluded that the site was favorable for landfill development.IntroductionSolid-waste regUlations for many states lying partiallywithinthe Appalachian Valley and Ridge, such as Penn sylvania and West Virginia, prohibit the developmentoflandfills above carbonate rock. Virginia allows consideralionofsuch sites, although the onus is clearly on the owner to demonstrate geologic stability and monitorable groundwater conditions. The major concern with stabilityisthe potential lossofliner integrity should a subsidence sinkhole occur. Subsidence sinkholes develop due to sub surface erosionofoverburden soil into pre-existing solutionvoidsorcavities in the rock. Because groundwaternowthrough fractured media such as carbonate rocks will follow preferential flow pathscreatedby solutional enlargement along joints, fractures, and bedding planes, itissometimes not possible to install monitoring wells at locationssothat the data obtained is representativeoftheaquiferflow beneath the landfill. Rockingham County currenLly operates a 48-acre land/illsouthwestofHarrisonburg, Virginia. A 40-acre expansionisplanned that will meet new regulatory requirements 153ofa doubly lined system. To limit potential negative en vironmental impactsofthe current site, "landfill mining" (removalofexisting wastes to newly lined cells) is being recommended. This is considered a desirable improvement over present conditions, because wastes might currenLly beinclose proximity to,orindirect contact with, karst-solu tion features, allowing unhindered avenuesofseepage into the groundwater table. Mined-landfill areas will laterbedevelopedinaccordanee with the double liner requirements.GeologicSettingThe landfill site is locatedinthe Valley and Ridge Province. Bedrock beneath the site consistsofthe upper dolostone memberofthe Beekmantown Group (Brent 1960), that is made upoffive, thickly-bedded Ordovician age dolostone and limestone formations containing numer ous hard, thin bedsofchert. The Upper Dolomite mem ber, as well as the limestone members, generally weather into deep, red, elastic silts and lean clays with chert frag ments scattered throughout The Beekmantown Group un derlies the entire site, extending well beyond and trending northeasterly.

PAGE 162

DeStephen and Milner The Staunton Thrust, a major thrust fault, strikes northeastwardly and dips southeastwardly from its outcrop two miles southeastofthe site (Gathright and Frishmann, 1986). Local structure is impacted by the presenceofthis thrust sheet, with several small northeast-trending folds paralleling the frontofthe thrust. At the western endofthe landfill site, mapped bedrock strikes range from N 25 W to N 63 E, with dips ranging from 2 to21degrees to the west. At the eastern endofthe site, strikes range from N 20 W to N 70 E with dips ranging from 5 degrees eastto10degreestothewestThese dips are influenced by the small folds, the Middlebrook Anticline to the southeast, and the Long Glade Synclinetothe northwest A fracture-trace analysis was usedtohelp define poten tial monitoring-well sites and identify lineaments along which solutional enlargement might be more prominent. Approximately 33 fracture traces were identified on aerial photos in and around the landfill site. The strongestofthe fracture trends are oriented N 0-10 E and N 50-90 W. The first trend parallels the axisofa small north-trending anticline located midway between the Middlebrook Anticline and the Long Glade Syncline. The second trendisnormal to the regional strikeofthe Beekmantown Group.Approach to Risk AssessmentIn order to define the levelofrisktothe site, various geologic risk factors were identified including sinkhole frequency, fracture trends, and groundwater conditions. The geologic risk factors were defined through a three-step pro cess which included: (1) a preliminary site evaluation, (2) site reconnaissance, and (3) a comprehensive subsurface ex ploration program. To assess the overall risks to the site, the geologic factors were considered together with project development risk factors including final overburden thick ness following waste-cell excavation, changes in surface water infiltration, and groundwater pumping (DeStephen and Wargo, 1990). This evaluation resulted in an opinion as to whether the site was either favorable or unfavorable for landfill development.Preliminary Site EvaluationThe first step in the preliminary site evaluation was identifying the potential for sinkhole development within limestones and dolostonesofthe Beekmantown Group. A comparisonofsinkholes on a regional basis (Hubbard, 1983) was made with respect to the Beekmantown Groupasshown on Figure1.Hubbard identified sinkholes using stereo examinationoflow-altitude aerial photography, and only sinkholes about 30ftin diameterorgreater are shown. From this data it was concluded that on a regional basis, sinkhole occurrence is not significant within the BeekmantownGroup. A higher sinkhole frequency in the Beekmantown Group around Harrisonburg is attributabletothe Upper Limestone Member, which predominatesinthat area.154Appalachian Karst Symposium. 1991 The region surrounding the site was further investi gated for sinkhole occurrence by identifying closed-end depressions from topographic maps and stereographic pairsofaerial photographs. Probable sinkholes were located, and their proximity to the site are indicated on Figure2.Based on this analysis, a low sinkhole frequencyofabout 3.2 sinks/square mile was identified for the Beekmantown Group limestones and dolomites. The proposed landfill disposal area is only 0.034 square miles. From this anal ysis, the probability for future sinkhole occurrence atthesiteisconsidered very low.Site ReconnaissanceA site reconnaissance was conducted and included a thorough observationofsurface features at the site. Inter views were also conducted with landfill personneltodevel op a historyofthe site. Site reconnaissance was aimedatidentificationofsubtle features such as wide bowl-likedepressions, small swales, changes in vegetation and similar features that might be a resultofkarstification. No indica tionsofsinkholes were observed within the landfill-expan sion area, although a sinkholeispresent adjacent to theexisting landfill.Subsurficial Exploration ProgramSeveral subsurficial exploration techniques wereemployedtoobtain information on the overburden soils,characterofthe underlying rock, groundwater occurrence,andkarst features. These techniques included15test borings, eight air-track probes, continuous rock coring at onedeepmonitoring-well location, air-rotary drilling at threemonitoring-well locations, borehole-geophysical logging,andgeologic loggingofa 55-foot-deep borrow trench.Ofthe five air-track probes drilled into rock,noneshowed evidenceofvoids, that is, lackofdust expelled from the holeorsudden drop in the drill stem. Allairtrack probes in the rock showed a fairly steady drillrate,with a continuous thick dust with rock chips, indicativeofcontinuous rock. Within the landfill-expansion area, overburdenthicknesses were substantial, varying from 20 to 60+ ft asindicated on Figure 3. General trends in the depth torockcould be estimated, although several rock pinnacles hadalready been encountered byearthmoving equipmentduringshallow excavationofborrow soils. Geophysical logging was performed in three ofthefour wells drilled on site to correlate geological strata encountered in the boreholes andtoidentify possiblewaterbearing fracturesinorder to place the well screens. A suite offour logs consistingofnatural gamma, density,resistivity, and caliper, was run in eachofthe three boreholes. No significant fracture zones were identified on thegeophysical logs. The data was used to select well-screen intervals, which ranged from depthsof70to140ft.

PAGE 163

Appalachian Karst Symposium. 1991 DeStephen and MilnerLEGENDSINKHOLESCAVEENTRANCEFAULTSBEEKMANTOWNGROUP., '. FROMSELECTEDKARSTFEATURES OFTHENORTHERNVALLEY h.'r AND RIDGE PROVINCE. VIRGINIA.DAVIDA.HUBBARD.JR.1983Figure I: Regional occurrenceofsinkholes in the vicinityofHarrisonburg, Virginia. Groundwater was not encountered within the soil over burdenatthe site. Allofthe monitoring wells drilled inrockproduced water,butnot immediately after drilling.Waterlevels rose above the screened depths, indicativeofa confined aquifer. From groundwater-elevation data, flowisgenerally westerlyata steep hydraulic gradientof4%.Lowtransmissivitiesof45 to 200 gpd/ft calculated from falling-headin situpermeability (slug) testsineach well suggested a lackofsolution features. All three down gradient wells were equally spaced across the site and equi distant from the upgradient well. The hydraulic headsin155 these wells were very similar, indicating that there were no major interconnected karst solutional openings that might give rise to unpredictable groundwater levels over short distances. This implied that groundwater flow beneath the site could be monitored, with water-quality results repre sentativeofgroundwater quality beneath the landfill.Developmental RiskFactorsDevelopmental risk factors were also consideredinthe evaluationofthe site. Developmentofa site can either be

PAGE 164

DeStephen and Milner Appalachian Karst Symposium, 1991 ObudLEGENDPROBABLESINKHOLELOCATIONSBEEKMANTOWNGROUP .oelh .; 2000 5 I/5 ObmlFigure2:Probable sinkhole locations in vicinityofproposed landfill expansion. a benefit or a detriment with regardtosinkhole potential. The majorityofinduced sinkholes are subsidence sink holes, formed by subsurface erosion into pre-existing voids, induced by allowing increased infiltrationofsurface water,orby intense pumpingofgroundwater. The pro posed landfill development was not consideredtobedetri mental with regard to sinkhole development for three rea sons:(1)the double-liner system would reduce (eliminate) surface infiltration, a major triggering mechanism in the inducementofsubsidence sinkholes, (2) the maximum 15 foot cuts necessary for development would leave a con siderable overburden thickness above the rock, and (3) future groundwater pumping adjacenttothe landfill will be limited. Where overburden soils are thick, the potential for soil raveling into any solutional voids is lessened (Newton, 1987). Surface failure usually does not occur, even under wetted conditions, unless overburden thicknessisless than about 7 feet (Williams and Vineyard, 1976). Typical over burden thickness remaining at the site following landfill construction will vary from10to 50 feet, with an average depth greater than25feet. 156 Subsidence sinkholes can also be triggered byintensegroundwater pumping, most notably where the ground water occurs in the soil overburden. Groundwaterpumping can result in a downward migrationofoverburdensoilby wayofthe following mechanisms: (1) lossofbuoyant support to residual soil arching above rock openings,(2)increase in velocity and seepage forces associatedwithgroundwater movement, and (3) movementofwatertobedrock openings. However, groundwater conditions allhe site are favorable with regard to water-table lowering,asthe occurrenceofgroundwaterisbelow the rock surface. Based on estimated drawdowns calculated from therockpermeability dala, singular domestic deep wellswithpumping rates less than 10 gpm were not consideredtoimpact the site. Any such wells willbeat least 500feetfrom the landfill as required by regulations.ConclusionsAllofthe geologic risk factors evaluated at thesitesupported a very low potential for future sinkholeactivityin the landfill development area. These include thefollowing:

PAGE 165

Appalachian Karst Symposium.1991LEGEND OVERBURDENTHICKNESS(FT) $ TESTBORINGLOCATIONAIRTRACKPROBEMONITORINGWELLLOCATIONROCKPINNACLES \\ EXISTINGLANDFILL5000500SCALEFigure3:Thicknessofoverburden at proposed landfill-expansion site priortoconstruction.DeStephen and MilnerInaddition, project-development risk factors related to the landfIll construction were consideredtobenon-detrimental with regardtosinkhole potential. Posi tive aspectsofthe design included(1)average final overburden thicknessesinexcessof25 feet, (2) useofa double liner system to eliminate surface-water inflltration, and (3) favorable groundwater conditions together with regulatory limi tations on future groundwater pumping. Basedona combinationofpositive geologic and developmental risk factors, the overall sinkhole potential at the site was estimated tobelow. The site was considered favorable for landfill design,anda Virginia DepartmentofWaste Management Part A landfill permit was applied for and approved. No sinkholes were observed within the landfill devel opment area based on site reconnaissance, and reviewoftopographic maps and aerial photos. The Upper Dolomite memberofthe Beekmantown Group is not particularly sinkhole prone, with an estimated sinkhole frequencyofonly about 3.2 sinks/ square mile in the vicinityofthe site. There are no major faults or other structural features occurring on site that would increase the likelihoodofsolution features. Fracture-trace analysis indicated no major lineaments except surficial drainagewaysatthe site. No cavitiesorvoids were indicatedinanyofthe test boringsorair-track probes that were drilled on site. No voidsormudseams were identifiedinthe 78 li near feetofrock coring performed. Likewise, no lossofdrilling water was observed, as might be indicativeofhighly fractured rockorsolutional voids. No solutional openingsormajor fracture zones were indicatedinover 450 linear feetofgeophysical log ging at the well locations. Transmissivities for the uppermost aquifer, as calcu lated for eachofthe wells, were very lowat45 to 200 gpd/ft. 157ReferencesBrent, W.B., 1960, Geology and mineral resourcesofRockingham County:Virginia DivisionofMineral Re sources Bulletin76,174p.DeStephen, R.A. and Wargo, R.H., 1990, Risk assess ment for foundation design in karst terrain:AnnualMeetingofthe AssociationofEngineering Geologists. Pittsburgh, Pennsylvania. October1990.Gathright, T.M., II and Frischmann, P.S., 1986, Geologyofthe Harrisonburg and Bridgewater Quadrangles, Vir ginia:Virginia DivisionofMineral Resources Publica tion 60.Hubbard, D.A., Jr., 1983, Selected Karst Featuresofthe Northern Valley and Ridge Province, Virginia:Virginia DivisionofMineral Resources Publication44, one sheet. Newton, J.G., 1987, Developmentofsinkholes resulting from man's activitiesinthe Eastern United States,U.S.Geological Survey Circular968, 54 p. Williams, J.H. and Vineyard, J.D., 1976, Geologic indica torsofcatastrophic collapse in karst terrain in Missouri:National AcademyofScience Transportation Research Record612,p.31-37.

PAGE 166

Plate Appalachian Karst Symposium. 1991Plate G: "Lake Fairlawn," Pulaski County, Virginia, early 1980s. Infilling and pavingofa sinkhole during constructionofcommercial establishments resulted in blockageofnatural drainage and caused periodic flooding following storms. This problem resultedinthe installationofa drainage system at considerable expense. Pulaski County has since established guidelines for new constructioninkarstic terrane. Bus is travelingofU.S Route11.SeeMills and others, this volume, p.159for a discussionofsinkhole flooding.Photograph by StephenD.Hale.PlateH:Sinkhole collapse within townofAustinville, Wythe County, Virginia, October 1989. Although sudden collapses such as this are infrequent, they may pose a problem for local residents. Note the porch to a homeinthe left foreground and the clothesline hanging into the sinkhole. A snow fence was erectedtokeep people back from the edgeofthe sinkhole.SeeDougherty, p. 139, and Beck,p.231, this volume for discussions on sinkhole collapse.Photograph by ErnstH.Kastning.158

PAGE 167

Appalachian Karst Symposium. 1991 Mills, George, Taylor. Ogden. Robinet-Clark. and FordePredicting Sinkhole Flooding in Cookeville, Tennessee, Using SWMMandGISH.H. Mills, D.B. George, H.N. Taylor, A.E. Ogden, Y.Robinet-Clark, andR.FordeCenter for Management, Utilization, and ProtectionofWater Resources Tennessee Technological University Cookeville, Tennessee, 38505ABSTRACTMore than 160 sinkholes occur within the city limitsofCookeville, Tennessee, manyofwhich are subject to flooding. To aid in managementoffuture development, the Stormwater Management Model (SWMM) and the ARC/INFO Geographic Information System (GIS) were used to predict the degreeofsinkhole flooding for specific rainfallevents. First,aUsinkholesandtheir tributary watersheds were delineated on special 1 :2400-scale maps with 5-foot contour intervals. Next, the following GIS coverages were prepared: topography, sinkholes, watersheds(ofindividual sinkholes), and land use (as determined from a 1987 city zoning map). The TIN moduleofARC/INFO was used to generate an additional cover age, a slope map. For each watershed, land-use coverage was used to estimate the average impervious area and slope-map coverage was usedtodetermine the average slope. Information provided by these coverages was entered into the SWMM and surface runoff then simulated for the following events: 50-year and 100 year, 3-hour events (3.0 and 3.9 inches, respectively) under dry and saturated antecedent-moisture conditions. The TIN module was used to calculate the volumeofeach closed-contour interval for each sinkhole and these volumes were then compared to the volumeofrunoff predicted by the SWMM. The GIS was pro grammed to shade the areas between closed contours within each sinkhole according to the incremental vol ume flooded. Results showed that for the lOO-year storm under saturated conditions, 27 sinkholes would overflow and an additional 33 would have their uppermost closed contour interval 50-100% fIlled. The gra phical output allows planners to rapidly determine the effectofproposed development on sinkhole flooding in a watershed.IntrOductionFloodinginsmall drainage basins poses a problem in densely populated areas. The problem isofparticular concernwhere sinkholes are abundant, for muchofthe runoffmaybe retained in the basins for hours or days rather than moving rapidly downstream as it does in nonkarst areas. Urbanization compounds the problem in several ways.First,the increased impervious area increases the proporlionofrainfall that runs off, filling sinkholes more rapidly.Second, after more desirable building sites are depleted, constructionofhomes and businesses often shiftstoless desirable lower areas,inand near sinkholes. Third, partial fillingofsinkholes often occursinconnection with this construction. This decreases the storage capacityofsinkholesthat serve as retention basins, and although it may prevent floodingofstructures built on the fill, it resultsinmorerapid risesofwaterinthe remaining partofthe sinkhole.Sinkhole capacity may even be exceeded so that 159 water spills over and floods other sinkholes and areas not previously subject to flooding.Forthe planningofdevelopment it is highly desirable to anticipate these effects. Blanket ordinances prohibiting building inornear sinkholes,orany fillingofsinkholes, may unnecessarily stifle development. City officials are likely to come under strong pressure to grant exceptions, someofwhich may be justified. A more discriminating and probably effective approach can be provided by the ability to predict the specific effectsofparticular proposed development projects. Demonstrating in a quantitative manner the deleterious effects that a proposed project is likely to have can provide a more convincing case against undesirable development than a blanket ordinance. This paper describes a computer-based methodofpredicting sinkhole flooding andofevaluating thepotential for flood ing for any proposed development. This study was partofa project that alsoinvolved the evaluationofwater quality

PAGE 168

Mills. George. Taylor. Ogden. Robinet-Clark. and Forde and the delineationofsubsurface flow paths within the city (George and others, 1990; Pride and others, 1988).Previous WorkWhite and Reich (1970) found that in drainage basins ranging from 2 to 200 square miles in area, the mean annual flood is much smaller in basins underlain by carbonate bedrock than in basins underlain by clastic rock, presumably because in the former basins muchofthe storm water enters groundwater storage and then res urges slowly over the next several days. Whether this result applies to basins on the orderofa square mile or less is not certain. Betson (1976) studied runoff yield from four urbanized basinsineast Tennesseeranging in area from 0.24 to 1.60 square miles, twoofwhich were underlain by insoluble or moderately soluble rock and two by highly soluble rock. Urbanization produced a much greater increaseinrunoff yield in the' latter two than in the former two. Apparently under rural conditions muchofthe runoffinbasins underlain by highly soluble rock is losttothe subsurface. Therefore, as the land surface becomes increasingly impervious with increasing urbanization, the relative increase in runoffinthese basins is much greater thaninthose underlain by less soluble rock. Kemmerly (1981) discussed problemsofsinkhole flooding and suggested a four-step approach for dealing with this hazard. First, all sinkholes and areas likelytobe affected by sinkhole flooding should be determined from maps and field inspection. Second, aerial and ground-based photography should be taken during rainy periods to help determine the volumeofrunoff collected in sinkholes. This should be supplemented with staff-gauge, water-Ievel recorder, and precipitation data. Sinkhole floodstage, storage capacity, and subsurface drainage data can then be evaluated using appropriate hydrologic models. The third step is calculating the volumeofrunoff thatissupplied to a sinkhole by its catchment basin during a 100-yr intensity rainfall. The elevationtowhich the sink holeisfilled by this volumeisthen calculated and shownonthe map as the "sinkhole floodplain" (100-yr flood-line elevation) analogous to the 100-yr floodplain along streams used for flood insurance purposes. Fourth, recom mendations are madebylocal planning commissionstoregulate land use within sinkholes that constitute flooding hazards. Faulkerson and others (1981) and Mills and others (1982) carried out partsofsteps 1 and 2 for the Cookeville area. Forty-two topographic maps (I :2400 scale with a contour intervalof5 feet) were used for the study. They were based on aerial photographsofthe city takeninJanuary, 1972. All sinkholes within this area, together with their drainage basins, were delineated on the topographic maps.Themorphometryofsinkholes and sinkhole catchments in the Cookeville area has been studiedbyMills and Starnes (1983). Extensive field 160 Appalachian Karst Symposium. 1991 observations were also madeofhydrologic conditions and sinkhole flooding. Evidence for flooding, such as the presenceofdried mudorflood debris on treesinthe sinkholes, was found in 33.9%ofthe sinkholes. Such evidence tells little about the frequencyofflooding but provided a first approximationofflooding hazard.Theonly extensive effort to collect sinkhole water level data in an urban karst area appears to be thatbyCrawford (1982) and Crawford and Groves(I984)forBowling Green, Kentucky. Little informationofthis sort has been obtained in Cookeville, although extensive groundwater tracing has been conducted by Faulkersonandothers (1981) and Hannah and others (1989).Physical SettingThe study areaofabout 35 square miles lies in partsofthe four U.S. GeologiCal Survey 7.5-minute quadrangles shown in Figure1.The area is located approximatelyinthe centerofthe eastern Highland Rim physiographicprovince between the Cumberland Plateau to the east andtheCentral Basin to the west. Outliersofthe western Cum berland Escarpment occur in the eastern partofthe study area in the formofmesas capped by sandstonesoftheHartselle Formation. These mesas provide the highest elevations, about 1430ft.The lowest point in the study area is 930 ft, so the maximum relief is about 500ft.Excluding the mesas, the relief is only about 160ft,withelevations over muchofthe area averaging about 1100ft.Allbut4%ofsinkholes occur on either the St. Louis Limestone or the Warsaw Formation. Both formations contain large quantitiesofinsoluble material, and as a result, many metersofresiduum commonly mantlethesurface. At Cookeville the mean annual temperatureisabout 57 F and the mean annual total precipitationis56infor the period 1931 to 1985 (NOAA, 1986; National Weather Service, 1988). With regardtohydrologic setting, Cookevilleissitua ted ona drainage divide between Roaring River andCaneyFork River, both tributariesofthe Cumberland River.Asa result, only small streams occur in the study area (Figure2).Most streams are only a kilometer or so long andterminate in sinkholes. Approximately 7 milesofcavepassage have been surveyed within the city limits. Metqods Geographic Information SystemLarge-scale maps delineating sinkholes and theircatchment areas prepared by Faulkerson and others (1981)(e.g.,Figure3)were digitized on the ARC/INFO Geographic Information System. ARC/INFO consistsoftwo software packages linkedtogether. ARC is the software thathandles coordinates and topology, whereas INFOisthesoftware that handles non-spatial attributesofthe map.When

PAGE 169

Appalachian Karst Symposium,1991Mills. George. Taylor.Ogden,Robinet-Clark. andFordeNI mile ;;?/K 36IS/NCOOKEVILLECOOKEVILLE WEST EAST BURGESSFALLSDRYVALLEY8537.S'W8522.S'WFigure1:Index mapofstudy area. Shaded areaiscoveredby42 1:2400-scale topographic maps used for this study. amapis digitized a numberoffilesarecreated. This setoffilesis called a coverage. The following coverages were prepared: sinkhole topography (represented by 5-ft contour lines), watershed topography (represented by 25-ft contourlines),watershed boundaries for each sinkhole, and land use(asdetermined from a 1987 city zoning map). The TIN moduleofARC/INFO was used to generate an additional coverage, a slope map. The basic approach in this study was to use a runoffmodelto calculate runoff generatedinsinkhole catchmentsbystorm eventsofparticular intensities. Using ARC/INFO,the predicted volumeofrunoff was then comparedwiththe volumeofthe sinkhole at various contours andtheareaofthe sinkhole inundated was then computed. For a first approximation, sinkhole floors were assumed to be impervious. The GIS was programmed to shade the area between closed contours within each sinkhole accordingtotheincremental volume flooded.Stormwater Hydrology ModelingTheStormwaterManagementModel (SWMM), developed by the U.S. Environmental Protection Agency (Huber andothers, 1987), was used to predict stormwater runoff quantity. A modelofthis sophistication was not really needed for the hydrologic simulations carried out herein, but SWMM also has the ability to model waterquality,desirable for future research. Simulationsofsingle-rainfall events by SWMM wereusedtodetermine the potential for flooding in eachofthe Figure2:Mapofstreams and ponds in Cookeville area. Heavy line shows Cookeville city limits. drainage basins within the smdyarea.The storm duration to use for modeling is a questionofsome importance.Indrainage basins a fractionofa square mile in area, thepeakdischarge is produced by rainfall durationsofonly a few minutes. Where sinkhole flooding is the main concern, however, obviously a much longer duration is called for. For a completely impervious sinkhole, the appropriate dur ation would be a dayorlonger. Because most sinkholes do drain to a degree, however, some duration between these two extremes seems most appropriate. As a reasonable compr'Jmise, a durationofthreehours was selected, rough ly corresponding to the durationofwinter storms in the study area. Thel-year,3-hour storm has been accepted by the U.S. DepartmentofHousing and Urban Develop ment as the determining storm for sinkhole floodplainsinBowling Green, Kenmcky (CrawfordandGroves, 1984). In the present study, simulations were made for both 50andl-year,3-hour storms.Asno National Weather Service hourly precipitation data base exists for Cooke ville, intensity-frequency duration curves for Nashville, Knox ville, and Chattanooga were used. These may under estimate the intensity by a small amount, as the mean annual precipitationofCookeville is several inches higher than the other cities. The precipitation intensity is about 1.0 in/hr (3.0 in total) for the 50-yr, 3-hr storm and about 1.3 in/hr (3.9 in total) for the l00-yr, 3-hr storm. Each simulation was performed under two different soil-moisture161

PAGE 170

Mills, George, Taylor, Ogden, Robinet-Clark, and Fordeconditions:1)saturated conditions prior to the rainfall event (zero initial moisture deficit), and 2) verydrycondi tions (initial moisture deficitof0.32 volume-of-air to volume-of-voids). The hydrologic input parameters to the SWMM that were needed to characterize the sinkhole watersheds were divided into climatic and morphologic types. Concerning the former, the duration and frequencyofthe storm event were usedtodevelop a hyetograph. The time step used for the hyetograph was five minutes. The resulting hyeto graphs for the 50and l00-year storm intensities were uniform block shapes and not representativeofa natural storm. They were, however, appropriate for planning the volumesofrunoff water to be transported through each watershed, which was the output datumofchief concernforsinkhole flooding. Morphologic hydrologic parameters included watershed area, average watershed slope, average percentageofimper vious area, roughness factors, depression storage, averageAppalachian Karst Symposium. 1991infiltration capacity, drainage network, and widthofover land flow. The watershed area was obtained from the GIS watershed coverage and the average slope was obtained from the slope-map coverage. The average percentage of impervious area was obtained from Cookeville zoning maps by the following procedure. First, zoning maps were superimposed on aerial photographsofthe city. Then, a grid was placed over representative areas on the photographs corresponding to each zoning category andthepercentageofgrid squares covering paved areas determined. This percentage was then used for other city areas falling into the same zoning category. For a given watershed, then, the percentage impervious area couldbecomputed based on the types and proportionsofzones in that water shed. These computations were performed by the GIS. The roughness coefficients used in the model were 0.014 for the impervious areas and 0.30 for pervious sur faces. These values are standard Manning's roughness coefficients for asphalt and short turf (Huber and others, 1987). Depression storage values used in the model Figure3:Exampleoflarge-scale maps delineating sinkholes and catchment areas. Contour interval is 5 feet. Dark hachured lines show closed depressions (sinkholes), whereas dark lines without hachures show catchment divides. Numbersidentify individual sinkholes.162

PAGE 171

Appalachian Karst Symposium. 1991 Mills. George, Taylor, Ogden, Robinet-Clark, and Forde 35,...----------------T ResultsFigure4:Comparisonofpredicted and measured hydro graphs for Sinkhole 62 for a OA-in rainfall event. c .Q : 2 OJa:'L'::J o .c'".c u c<.::Time(minutes) 25 10 20 '"2 Four sinkhole drainage basins were used to calibrate the runoff model.Asan exampleofthese results, Figure 4 shows the hyetograph/hydrograph calibration results for sinkhole 62.Themeasured hydrograph for August4,1987, was the resultofa short, intense summer storm. After adjusting the lengthofthe overland flow, the predicted-runoff volume was 2% less than the calculated volume. Using this calibration, additional simulations were then run for other observed rainfall events at this and other sinkholes. Table 1 shows the difference between measured-runoff volumes and volumes predicted by the SWMM. Although some simulations wereoffby as much as 58%, the simulationatleast gives the correct orderofmagnitude for runoff volumes, which is sufficient for the present study. Partofthe error maybecaused by difference between the actual impervious area within some basins and that predicted by the zoning map. TheSWMMwas used to simulate runoff in all 164 sinkhole basins for both 50and l00-year, 3-hr storms under both dry and wet antecedent-moisture conditions. Only the extremesofthese four simulations are discussed here: the 50-yr storm with dry-antecedent conditions and thelOO-yrstorm with wet-antecedent conditions. 30 compared the volumeofrunoff for a watershed withth.evolumeofthe sinkhole. The GIS was then used to graphI cally display the area inundated by the rainfall input. A simulated hydrograph was generated by entering the hyetograph into the SWMM. This hydrograph was super imposed on the measured hydrographs and the differencesinvolumes and peaksofthe two hydrographs compared.Thewidthofoverland flow was adjusted to calibrate the predicted hydrograph to the measured hydrograph. Having calibrated the SWMM, additional runoff events were simu lated using the calibrated model. The degreeofmatchof the predicted and measured hydrographs for eachofthese events was then evaluatedinorder to verify the calibration. Each sinkhole watershed was characterizedinthe mannerdescribed previously and the data entered into a com puter file for the particular sinkhole. The overland-flowwidthfor each watershed was entered as the lengthofthe lop-most closed contourofthe sinkhole. The SWMMwasaltered to produce an output file, generated by a specifichyetograph, that contained the total volumeofsurface runoff within each watershed. This output file was thenusedas an input file for the GIS, that mathematically The drainage network was not simulated. The width of overland flow normally is assumed tobetwice the lengthofthe main drainage channel. However, during calibrationofthe SWMM, it was found that this value produced excessively large flow rates. By experimentationitwas found that the best predictionofthe hydrograph was obtained when the widthofoverland flow was set equal to the lengthofthat partofthe top-most closed contourof the sinkhole that faces the major partofthe sinkhole drainagebasin. Calibrationofthe SWMM required measuring storm water runoff from runoff events. Four sites were selectedforthis calibration. They were representativeofwater shedsinthe study area and had well-defined, channelizedflowentering the sinkhole. This allowed adequate flow monitoring. A simple staff gauge was used to measure theriseand fallofwaterinthe channel over time. The waterlevelwas recorded at five-minute intervals and a velocity meter was used to determine stream dischargeatvariousLimesduring the rainfall event. These measurements, aswellas Manning's equation for open-channel flow, wereusedto develop a flow-rating curve for each location so that hydrographs could be constructed. Hyetographs for eachofthe four catchments were obtained by placing rain gauges within each basin and observing them during rainfall events. Rainfall intensitiesweregrossly estimated for live-minute intervals. simulations were 0.011 in for impervious surfaces and 0.1infor pervious surfaces. The average infiltration capacitywasdetermined from the Green-Ampt equation. Valuesusedfor particular parametersofthis equation were: capil lary suction, 12.0 inofwater; saturated hydraulic conduc tivity, 0.3 in/hr; and an initial moisture deficitof0.0 vo! umeofair/volumeofvoids (indicativeofsaturated condiLions).163

PAGE 172

Mills. George. Taylor. Ogden. Robinet-Clark. and Forde Appalachian Karst Symposium. 1991 \) 10J(t'1000ItIJ'8 \) 10J(A*Not filledtocapacity. example, would spill into sinkhole 63, and 63 into 64. Sinkhole 65 also flows to 64. Table 2 shows the volumeofrunoff predicted to overflow these sinkholes during both eventsofboth magnitudes. Figure6:Examplesofpredicted flooding in sinkholesina single-family residential area. A. 50-yr 3-hr rainfall with dry-antecedent conditions.B.100-yr 3-hr rainfall with wet-antecedent conditions. Solid shade indicates complete inundationtothat level. Crosshatch pattern indicates50100% inundationofthat contour interval, and diagonal line pattern, 25-50%. Contour intervalis5 feet. Lines with dots show catchment divides. Numbers identify particular sinkholes. TABLE2.Predicted Sinkhole Overflow Sinkhole Predicted ovecflow Predicted ovecflow number from 50-yr event, from 100-yr event,dryconditions wet conditions (cubic feet) (cubic feet) 62 368,100 2,126,100 63*490,900 64 131,300 483,600 65 76,200 380,300 888335004170800 rl" ,1000IIFigure5:Exampleofpredicted floodinginsinkholes in a commercial area resulting from 50-yr 3-hr rainfall with dry antecedent conditions. Solid shade indicates complete inundation to that level and crosshatch pattern indicates 50 100% inundationofthat contour interval. A 50-yr 3-hr rainfall with wet-antecedent conditions results in overflowofall sinkholes shown here. Lines with dots show catch ment divides. Numbers identify particular sinkholes. Note: Positive sign(+)indicates an underpredictionoftotal runoff volume. Negative sign (-) indicates an overprediction. Figures 5 and 6 show the levelsofsinkhole inunda tion that were produced by these simulations for two small areasofCookeville. Figure 5 shows a groupofrelatively small sinkholes with highly developed watersheds.Ascan be seen, all but oneofthese is filled beyond capacity by the 50-yr, dry-antecedent-condition event, not to mention the 100-yr, wet-antecedent-condition event (not shown). Severalofthese sinkholes, in fact, are known to flood to substantial levels even during storm events with recurrence intervalsofonly a few years. The effectofflood waters cascading from basin to basin was not evaluated, but thisisobviously a phenomenon to consider. Sinkhole 62, for TABLE1:Runoff Calibration Results Sinkhole Volume difference number Date between measuredand predicted (%)62 08/04/87+2.0 62 08/05/87 34.0 62 01/20/88+58.07109/29/87+16.0 36 09/29/87+0.5 88 10/10/87 7.0 164

PAGE 173

Appalachian Karst Symposium. 1991 Figure 6 shows a residential area with somewhat larger sinkholes. In the 50-yr, dry-antecedent-condition event, only the lowest partsofsome sinkholes are flooded (Figure 6A). In the lOO-yr, wet-antecedent-condition event, flooding is more extensive, with wide areas being inundated (Figure 6B). Only small sinkholes, however,arefilled to the highest closed contour. Simulations for all 164 sinkholes showed that the 50yr,dry-antecedent-condition event results in 5 sinkholes filled to overflowing and another 6 with their topmost conLOurinterval 50-l()()% filled. The lOO-yr, wet-anteeedent condition event results in 27 sinkholes filled to overOowingand another 33 with their topmost contour interval 50-100% filled. The main factors affecting the degreeofflooding in a given sinkhole should be the proportionofimpervious areainthe watershed and the ratioofthe areaofthe drainagebasin to the volumeofthe sinkhole. Figure 7 shows a scatter plotofthese two parameters for81sinkholes. Note that the area:volume ratioismuch more importantforpredicting the degreeofflooding thanisthe proportion o[ impervious area and,infact, appears to provide a good preliminary method [or assessing the likelihoodofflood mg. Other factors not quantified in this study also may0LEGEND0 0o Topcontourexceeded '" UppercontourInterval 50-100% filled oLeuerdegree'of flooding 4) 0E ...20 000 >-4) 0 -0.L 0 0 00 II> 0"0 0-4)dl 0 L.i'" 0 '" 0 '" '"'"0'" 0 4)!'" 0 0'"'".L 0 '"II> 0 L.II 0 4)0 000-0 0 0 0 0 0 II o0 0 0 0 O-t------r---,------.-------r------, 0.3 0.4 0.5ProportionimperviousareaFigure7.Plotofpercent impervious area against basin arca:sinkhole-volume ratio for81sinkholes. 165 Mills. George. Taylor, Ogden, Robinet-Clark. and Forde bear on whether a sinkholeislikely to represent a flooding problem.Forexample, the formofa sinkhole influences the likelihoodofdevelopment. Deep, steep-walled sink holes are obvious hazards and do not invite development. Shallow, gentle-walled sinkholes, however, often do not seem hazardous, and may well undergo development. Commonly, construction initially occurs within the upper contoursofthese sinkholes, which may pose no flooding hazard owing to adequate storage in the broad, flat bot toms. As the higher sites are taken, however, eventually development extends to the sinkhole bottoms, resultinginsuch hazards as ponding over roads, flooded lower levelsofbuildings, increased infiltration into sewer lines, and collapse duetopipingoffoundation material. Sinkholes 64 and 65 (Figure 5), for example, are highly developed. The amountofdevelopment in locations downgradient from sinkholes may also be important, particularly for sinkholes likelytooverflow. Sinkholes 62 and 88 (Figure 5), for example, have intensive commercial development downgradient.DiscussionThe procedure outlinedinthis paper providesanexcel lent way to screen a large numberofsinkholes for poten tial l100ding hazard. Once the high-priority sinkholes are determined, however, detailed hydrologic data are required for more precise modeling. Few such data have been gathered in Cookeville. The first need is installationofrecording rain gauges that can provide rainfall intensity data for five-minute or smaller time intervals. For high priority sinkholes fed chiefly by one stream, a water-level recorder ncar the stream mouth would allow the stream dis chargetobe monitored, thereby allowing an accurate recordofwater entering the sinkhole. Still more important, however, is the placementofwater-level recorders within sinkholes to measure leakage rates. Leakage ratesofsinkholes with regolith-covered floors can be approximated by measuring the hydraulic conductivityofthe regolith, but leakage ratesofsinkholes drained by swallets probably can only be determined em pirically. The rate at which the sinkhole drains,ofcourse, is important for modeling. The appropriate rainfall dura tion to useinthe model probably will be differentforfast draining and slow-draining sinkholes. In addition, the volumeofleakage must be subtracted when computing the degree of Ooodingina sinkhole. Another reason that water-level recordingisimportant is that, as Crawford (1982, p.8)has pointed out,insome cases even the assumption that a sinkhole flooriscom pletely impervious may not be conservative enough. Ma ny sinkholes are interconnected by a common subsurface conduit into which they normally drain.Ifthereisa con striction in this conduit downstream from the sinkholeofconcern andifthere are upstream sinkholes feeding the conduit, flood water may backupuntil it actually rises outofthe conduit into the sinkhole.

PAGE 174

Mills. George. Taylor. Ogden. Robinet-Clark. and FordeAnother effect that can confound modeling is that the groundwater basin that is supplying water to the sinkhole maybelarger than the surficial topographic basin, so that the amountofwater entering the sinkhole during a heavy rain maybemore than predicted. This difficulty canbecorrected by using additional dye tracing to better define areas contributing flow to sinkholes. Resource limitations allow only a small numberofsinkholes to be continuously monitored. For sinkholes with lower priorities, however, it is possible at least to measure the maximum depthofflooding with only a small amountofeffort. Crest-stage gauges (Leopold, 1968) can easily and inexpensivelybeinstalled in a large numberofsinkholes. Aerial and ground photographyofsinkholes during and after intense storms wouldbeanother quick method.Anadvantageofthemodeling approach discussed here is that, as leakage rates and other hydrologic data are ob tained from field measurements, they can readily be incor porated into the model, thereby increasing its sophistica tion with time.ConclusionsSurface runoff from 50-yr and 100-yr storms were simulated using the SWMM for 164 sinkholes and their catchment areas. Outofthe 164, only11sinkholes were predicted to flood beyond the top closed contourorhave greater than 50%oftheir highest contour interval inun dated during the 50-yr storm with dry-antecedent soil-mois ture conditions. For the l00-yr event with wet-antecedent conditions, however, a totalof60sinkholes were predicted to similarly flood. A priority list was compiledofthe 28 sinkholes most likely to produce flooding problems, based on the simulations, the amountofpresent development in the sinkholes, and the amountofdownstream develop ment. The modeling procedures developed in this project will assist city planners and developers in minimizing storm water flooding hazards.Tofurther enhance the proper managementofstorm water within urban karst areas, the following recommendations are made:1.The GIS database should be upgraded to allow fine discretizationofthe study area into subbasins, including storm drains. In addition, more dye tracingofsubterranean flow paths should be conductedinordertomore accurately delineate groundwater drainage basins andtobetter under stand the hydraulic connectionsofsinkholes.2.Recording precipitation gauges should be installedinthe study area to provide precise rainfall-intensity data needed for future modelingofrunoff. For the highest-pri ority sinkholes, water-level recorders should be installed both in inflowing streams (to aid in the calibrationofrun off) and in the sinkholes themselves (to establish hydraulic 1M Appalachian Karst Symposium. 1991conductivitiesofsinkhole floors and subsurface conduits). Crest-stage recorders shouldbeplaced in other sinkholes of possible concern in ordertodocument depthofflooding. 3. The study area should be modeled to determine the effectsofoverflowing sinkholes cascading into down-gra dient sinkholes. In addition, high-priority sinkhole drain age basins should be modeledfor various development and abatement scenarios.ReferencesBetson,RP.,1976, The hydrology of karst urban areas,inDilaniarter,RRand Csallany, S.C. (editors),Hydro logic Problems in Karst Areas:Western Kentucky Uni versity, Bowling Green, Kentucky, p. 162-175. Crawford, N.H., 1982, Hydrogeologic problems resulting from development upon karst terrain, Bowling Green. Kentucky:Guidebook PreparedforKarst Hydrology Workshop. August31-September3.1982,sponsoredbyGroundwater Section, Region IV, U.S. Environmental Protection Agency, 34 p. Crawford, N.H. and Groves, C.G., 1984,Storm Water Drainage Wells in_the Karst AreasofKentucky and Tennessee:Report prepared for the U.S. EnvironmentalProtectionAgency,UndergroundWaterSource Protection Program Grant No. G004358-83-0, 52 p. Faulkerson, J.; Burden, D.; Burden, K.; Edwards, c,; Kinley, T.;Lee,T.; Sparks, V.; Starnes, D.; Walls, E.; and Webster,S.;1981,Karst Hydrology. Morphol ogy. and Water Quality in the VicinityofCookeville. Tennessee:Report to the CityofCookeville: Tennes see Technological University, Cookeville, Tennessee. 67 p. George, D.B.; Taylor, H.N.; Pride, T.E.; Ogden, A.E.; Robinet-Clark, Y.; Forde, R.; and Mills, H.H., 1990,Assessmentofthe Cookeville. Tennessee. Stormwater Management System:Center for the Management, Utilization,andProtectionofWaterResources, Tennessee Technological University, Cookeville, Tennessee, 122 p. Hannah, E.D.; Pride, T.E.; Ogden, A.E.; and Paylor,R,1989, Assessing ground water paths from pollution sources in the karstofPutnam County, Tennessee,inBeck, B.F. (editor),Engineering and EnvironmentallmpactsofSinkholes and Karst: Proceedingsofthe Third Multidisciplinary Conference on Sinkholes and theEngineeringandEnvironmental ImpactsofKarst.St.Petersburg Beach. Florida.2-4October1989:A.A.Balkema, Rotterdam and Brookfield, p. 183-188. Huber, W.C.; Heaney, J.P.; Nix,S.1.;Dickinson, R.E.; and Polmann, D.1., 1987,Storm Water Management Model; Users' Manual. version III. EPA:600/2-84-109a.

PAGE 175

Appalachian Karst Symposium. 1991(tenth printing): Stonn and Combined Sewer Section, Systems and Engineering Evaluation Branch, Waste water Research Division, Municipal Environmental Research Laboratory, Cincinnati, Ohio. Kemmerly, P.R., 1981, The need for recognition and im plementationofa sinkhole-floodplain hazard designationinurban karst terrains:Environmental Geology,v.3, p.281-292. Leopold, L.B., 1968, Crest stage gageofU.S. Geological Survey,inTricart, J. (editor),StudyofSlope and Flu vial Processes:Paris, Jacques et Demontrond,p.181. Mills, H.H. and Starnes, D.D., 1983, Sinkhole morpho metry in a fluviokarst region: Eastern Highland Rim, Tennessee, U.S.A.:Zeitschrift fur Geomorphologie,v.27, p. 39-54. Mills, H.H.; Starnes, D.D.; and Burden, K.D., 1982, Predicting sinkhole floodinginCookeville:Tennessee167Mills. George. Taylor. Ogden. Robinet-Clark. and Forde Tech Journal,v.17, (Tennessee Technological Univer sity), Cookeville, Tennessee,p.1-20. National Oceanic and Atmospheric Administration (NOAA), 1986,Cookeville. Tennessee. Station Precip itation Datafor1931through1985period:NOAA, Cookeville, Tennessee Station, Cookeville, Tennessee. National Weather Service, 1988,Climatological data: Tennessee:National Oceanic and Atmospheric Adminis tration,v.93, no. 1-12. Pride, T.E.; Ogden, A.E.; Harvey,MJ.;and George, D.B., 1988, The effectofurban development on spring water qualityinCookeville, Tennessee:Proceedingsofthe National Water Well Association Karst Conference. Dublin. Ohio,p.97-120. White, E.L. and Reich, B.M., 1970, Behaviorofannual floods in limestone basinsinPennsylvania:JournalofHydrology,v.10,p.193-198.

PAGE 176

Quinlan. Ray. and Schindel Appalachian Karst Symposium. 1991ApplicationofDyeTracing to Evaluationofa Landfill Site in a Karst Terrane in the Tennessee AppalachiansJames F. Quinlan1 ,Joseph A. Rayl, and GearyM. Schinde12 lQuinlan & Associates Box 110539 Nashville, TN 37222 2Eckenfelder, Inc. 227 French Landing Dr. Nashville, TN 37228ABSTRACTExperience has shown that all landfills leak. Therefore, when a landfill is being designed, one must assume that it will leak, assess the consequences accordingly, and prepare to writeoffpartofthe underlying aquifer as a sourceofpotable water. Tracing, most easily and most economically done with fluorescent dyes, is theonlyreliable technique available for verifying one's interpretationofthe hydrogeologyofan area underlain by carbonateorother fractured rocks, determining the sites to which leakage will flow, designing a reliable monitoring system, and evaluating whether the system, once installed, is germane to the monitoring problem. Someofthese conclusions are illustrated by a site proposed for landfilling30miles eastofKnoxville, Tennessee, in Knox Group dolomites. A different dye was simultaneously put into eachoffour wells drilled to the soil-bedrock contact, as much as ninety feet deep. Flow was downdip and bi-directional along the strike. Dye was recovered at more than ten wells and thirty springs, manyofwhich are used as public and domestic water supplies, andatsome sites for more than nine months. Pre-test mappingofthe potentiometric surface made it possible to predict the probable flow directionsofdye and leachate. These predictions were subsequently verified. The Tennessee DivisionofSolid Waste Management maintained that tracing was irrelevant to evaluationofthe site; it recommended instead that hydrogeologic evaluation consistofgeophysical surveys that measure electromagnetic induction and earth resistivity. The Division then recommended that three additional piezometers be drilled so that a totalofseven on the site couldbemeasured once and used to establish a groundwater divide (theirs). This divide allegedly separated thatpartofthe property that drains to the water supplies from that part that (in one tracer test) did not. Such thinking is, at best, muddle-headed. The site is likely to be litigated. 168

PAGE 177

Appalachian Karst Symposium. 1991 Field and CritchleyComputer EnhancementofDownhole-Video Borehole LogsMalcolmS.Field1and Michael Critchley2 1 U.S. Environmental Protection Agency OfficeofResearch Development(RD-689)410M St.SWWashington, OC 20460 2pRC ,Inc. 1500 Planning Research Dr. McLean, VA 22102ABSTRACTVideo borehole loggingisa commonly used geophysical logging technique whereby direct observationofthe subsurface rockorwell casing is desired. Information typically obtained by video borehole logging wouldbeborehole shape, degree and extentoffracturing, evidenceoflarge cavities, and degreeofin-filling. However, recording conditions are often not conducive to the acquisitionofclear, distinct logs, thus result ing in insufficient detail regarding subsurficialgeologic conditions.Byutilizing a video-graphics computer workstation, it is possible to capture the entire video borehole log into a picture file. This picture file can thenbeviewed in either its original formorafter various com puter-aided enhancements have been made to the picture file.Theseenhancements canbeas simple as manipulating the color, lighting, and/or shadingofthe picture file to as complex as thatofrotating selected snapshotsofthe video log to the vertical and flattening the picture on the computer screen.TheHenderson Road Superfund site (Upper Merion Township, Pennsylvania) is located in a karst ter rane where the subsurface conditions are very poorly understood. Downhole video logging was conductedinan effort to gain more direct information on the subsurface. These video borehole logs were then enhanced using a video-graphics workstation. Enhancements included adjusting color, lighting, and shading.IntrOductionSubsurficial investigationsinfractured rock and karst tcrranes are generally much more difficult and complexthanthose in granular media. Although this problem is wellknownfromattemptstoexploitground-waterrcsources in fractured rock and karst terranes, the situationhasbecome especially evident recently with the adventof OlC Superfund program. Numerous examples can be citedinwhich instancesofknown releasesofcontaminants to olc subsurface could not be adequately monitored because SUbsurface transport was restricted to discrete pathways(c.g.fracturesandconduits) asopposedto dispersion Olroughout a porous granular medium.Asa resultofthe need to better define subsurficial conditions at Superfund sites, many researchers have begun investigating the useofalternative methods designedtoprovidemoredirectinformation from the subsurface. These methods are primarilyoftwo types, hydraulic and geophysical. New hydraulic techniques range from modifi cationofexisting hydraulics equations (Sen, 1987, 1988; Thrailkill, 1988)todeveloping better methods for hydraul ic testingofaquifers (Molz and others, 1989; Milanovic, 1981, p. 235-253) to ground-water tracing studies (Molz and others, 1986; Quinlan and Ewers, 1985). Geophysi cal methods range from attempts to apply existing meth ods in varying manners to glean additional information regarding the subsurfacetothe developmentofalternative techniques for investigating the subsurface (Lange andDisclaimer: The views expressed in this paper are solely thoseofthe authors and do not necessarily reOect the views or policiesoflheU.S.Environmental Protection AgencyorPRe.Inc. Mentionofeither trade names or commercial productsinno way constitutes official endorsement or recommendation for use.169

PAGE 178

Field and Critchley Kilty, 1988; Lange and Quinlan, 1988). Oneofthe most useful geophysical methods recently developed and now commonly usedatSuperfund sites is the downhole video borehole viewer. This allows a continuous video recordingtobe made on any standard video recorder for later repeated viewing in an office setting. Creationofvideo borehole logs provides a means for visualizing subsurficial conditions in boreholes. Shape, degree and extentoffracturing, evidenceoflarge cavities, and degreeofin-filling are someofthe more useful infor mation directly available from video recording a borehole. Unfortunately, the video recording often may notbeofsufficient quality to provide the detailed information still required for defining subsurficial transport conditions. By utilizationofa computer-graphics workstation, itispossi bletocapture the entire video borehole log into a compu ter picture file. Enhancements can be as simple as mani pulating color, lighting, and/or shadingofthe picture filetoas complexasrotating selected snapshotsofthe video logtothe vertical and flattening the picture on the compu ter screen.Site Description and Natureofthe ProblemDuring the 1970's, an existing water well at what has come to be known as the Henderson Road Superfund site (Figure 1) was converted from a water-supply well to a waste-injection well. Although itisbelievedtohave been operated as an injection well for several years, very little evidence existstoeither support or refute this information. Wastes injected into the well consistedofvarious indus trial solvents pumped from tanker trucks. Operationofthe injection well ceased after Pennsylvania State Police observed the ownersofthe site clandestinely injecting industrial-solvent waste into the well. The Henderson Road site consists of 7.6 acres located at the intersectionofHenderson and Church Roads. Itisoccupied by a sanitation company, several automobile shops, and a drilling contractor. Approximately 2000 feettothe north and downgradientofthe Henderson Road siteisthe Upper Merion Reservoir(UMR),an old rock quarry now used as a water supply for the TownshipofUpper Merion. Groundwaterispumped from the UMRata rateofapproximately 7.5 mgd, causing substantial draw downinthe local aquifer for several square miles in the area. The Henderson Road site is also approximately 350 feet upgradientofa local lumber company, where a water supply well serves15employees (USEPA, 1988). The Henderson Road site lies within the Piedmont Provinceofthe Appalachian Highlands andissituated on metamorphosed limestone and dolostone (Elbrook lime stone, Conestoga limestone, and Ledger dolostone). Graytoyellowish gray, the Elbrook limestoneisa commonly silty, sandy, siliceous limestone interbedded with dolomite whereas the Conestoga limestone, the main rock type 170 Appalachian Karst Symposium, 1991 underlying the site, is gray to bluish-gray and finelytocoarsely crystalline. Bothexhibitschistocity. The Conestoga Formation unconformably overlies the Elbrook and Ledger formations, andisthereby younger than both of them.TheLedger dolostone is light gray, massive, coarsely crystalline, sparkling, and contains siliceous layers. Allofthe rocks have beensubjected to intense deformation as can be seen by the complex seriesofanti clines, synclines, fractures, and faults in the general area(Nevius,1987, p. 9-12). Karstificationofthe limestones and dolostones has resulted in significant alterationofthe regional geology. Sinkholes observed throughout the area and solution voids detected in boreholes provide substantial evidence that dissolutionofthe carbonate rocks is extensive and that subsurface conduits probably play an important roleintheflow and transportofliquids and gases. However, these conditions are sufficiently complex so astoprevent a clear comprehensionofthe hydrogeologyofthe site regarding the rate, direction, and extentofsubsurface transport. Several efforts, from hydraulic testing to geophysical investigations, have been attempted with the intention of better defining the subsurficial geology. Vacuum testingofselected wells for possible vacuum extractionofsubsurficial volatile gases yielded highly ambiguous results. Two radically different interpretations from the same vacuum-extraction data have been offered with no resolution astowhichismore correcl. In both instances, intelpretations were directed toward establishing zonesofinterconnectedness. Unfortunately, this hasnotyet been possible. As a result, substantial confusion regarding pathways utilized for gaseous and liquid transport in the vadose zone still remains. A second hydraulic analysisofthe subsurface attheHenderson Road site consistedofconducting conventional aquifer tests on selected wells. Figure 2 is a semilog plotofthe late-time dimensionless time-drawdown dataofoneon-site pump test with the theoretical Jacob Straight line plotted on the same graph. However, useofthe Jacob methodinfractured and/or karstic terranes is generallynotvalid owing to the extreme limiting conditions associated with it. By plotting both the field data and the theoretical Jacob curve on the same graph, itispossible to determineifthe Jacob methodisin fact a valid methodofaquifer analysis (Sen, 1988). According to Sen,ifthe slopes of the two plots are parallel, then the useofthe Jacob method could be considered valid for analysisofthe site aquifer-test data. However, the slopesofthe two lines are not parallel in this case, so the Jacob methodofaquifer-test analysis can only be considered a first approximation. The flatter slopeofthe field data would indicate that the actual trans missivityisless than that calculated whereas the storativi tyisgreater than that calculated. Also, leakage fromsurrounding fractures may be inferred from the fieldplOLIntermsofthe pump-and-treat operation planned for thissite,accurate determinationofsubsurficial storage, flow,andtransport parametersisof paramount imporlance.

PAGE 179

Appalachian Karst Symposium, 1991 Field and CritchleyFigure1:Aerial photographofthe Henderson Road Superfund site and the Upper Merion Reservoir, downgradient from thesileoThe karst landscape is masked by the high levelofurbanizationinthe area, making field interpretations more difficull Borehole geophysical methods were also utilized at the sile to further define the zonesofgreatest importance re garding subsurficial contaminant transport. These consisledprimarilyofcaliper, resistivity, spontaneous-potential,171and natural-gamma logging. Downhole television loggingofthe boreholes onsite was also conducted. However, noneofthese methods have yet been able to provide th levelofdetail necessary for accurate determinationof

PAGE 180

Field and Critchley Appalachian Karst Symposium. 199116 141210 :J 8 6 4 2 0-510Jacob Straight Line HendersonRoadDataNote: All Units Dimensionless-110DescriptionofEquipmentVideo loggingofthe boreholes was initially conducted by suspending a tele vision camera equipped with a 4.5mmwide-angle lens and a manually adjusted f-stop, fixed-focus speed control. Light ing was provided by a low-light-level sil icon-target vidicon tube to allow image detection by the camera with minimum lighting power requirements. The rest of the video recording equipment consistedofa nine inch, studio-quality television monitor and VHS video recorder. Itisuseful to note here that useofa Super VHS would have probably greatly en hanced the overall qualityofthe record ing.I I10-610-5Uf10-410-310-2Figure2:Comparisonofthe dimensionless time-drawdown data for the Hender son Road site with the theoretical Jacob's Straight Line. Note that the lines are not parallel, indicating the inapplicabilityofthe Jacob method for aquifer-test analysis at this site. subsurficial zones most likely to provide pathways for contaminant transport. In particular, the downhole-video borehole logs were especially disappointing. Borehole video logsofthe monitoring wells for the Henderson Road site were collected with the intentionofproviding direct observationofsubsurficial geologic condi tions. Specific objectives included identificationoffrac tures and large voids, bothofwhich may be responsible for significant amountsofcontaminant transport. How ever, the qualityofthe video logs was significantly less than desired and/orrequired for gaining the necessary infor mation. Generally, the camera angle was not conducivetoa clear displayofimportant geologic features, the lighting was excessively bright, and the resolution was insufficient. Clear printsofselected features from each borehole were desired for inclusion in the official public record, but these could not be obtained becauseofthe relatively low quality of the video logs. Owing to the importanceofthe information needed, it was determined that either new video logs would be required for each well or some other method would havetobedevisedifthe information desired was ever to be ac quired. Objections to relogging each well forced considera tionofalternative methods for acquiring the information necessary for determining the important flow and transport zones at the Henderson Road site. An alternative that shows considerable promise is the capturingofthe video logs into a computer picture file that can then be manipu lated for enhanced clarity.Thecomputerequipmentusedtocapture and enhance the video frameswasaCommodoreAmiga2000 Personal Computer. Image capture was carried out on an A2000 equipped with a 68030 accelerator board, 8 megabytesof32-biLdynamic RAM and the recently developed Video Toaster by Newtek Inc. Imageenhancement was performed on an unacceI erated A2000 with 3 megabytesof16/32-bit RAM using off-the-shelf image-processing software. The Amiga com puter is manufactured by Commodore Business Machines as a high-quality graphics-and-sound capable personal com puter with proprietary audioand video-chips, multitasking operating system, and Graphical User Interface (GUI).Totalequipment costs, including both hardware and software, for an accelerated system with a large storage capacityareless than $10,000, making the technology very affordable. Image capture on the A2000 was carried out through useofthe Video Toaster, a new professional-level frame buffer/video-manipulation board and software. The Toaster is essentially a computer-on-a-card which providestheA2000 with full television color and resolution while furnishing it with the capabilityofperforming real-time video manipulation. Hence, output from Video Toaster approaches broadcast-quality video on a computer.TheVideo Toaster allows for the productionofnetwork-quality television on an A2000 by provjding several very powerful video tools such as a 4-Input Production Switcher, Digital Video Effects, Lightwave 3D Animation, Toaster Charac ter Generator, Toaster Paint, Dual Frame Buffers/Genlock, ChromaFX Color Processor, and Frame Grabber/Frame Store.DiscussionSelected borehole-log frames were captured intoNewtek's Video Toaster buffer from a standard VHS recorder. The video buffer was then written to a graphics fileina 172

PAGE 181

Appalachian Karst Symposium. 199124-bit color fonnat (24 color representations for each pixel).These files were then loaded into a graphics-conversionprogram calledTheArtDepartmentand manufacturedbyASDG, and converted to the Amiga's standard l6-bit Interleaved File Fonnat(IFF). Standard 24-bit graphics allows for millionsofcolors tobedisplayed on screen atonetime and has generally been available only at a profes sional level. The current Amiga displayand graphics software standards are 16 bit. This allows several on screen color displays depending on resolution and displaymode.The variations are 16, 32, 64, or 4096 colors on screen at one time. Resolutions range from a 320x200 toanOverscan modeofapproximately 800x500. The imagesusedinthis presentation were captured in the 24-bit, millionsofcolors mode, and then converted to an overscan nedHoldand Modify(HAM)mode that provided 4096 colorsonscreen at a resolutionof768x530 pixels. Various motion-removal algorithms were usedinthe captureofselected framestoremove flicker that occurswhena video tape is paused while in the "play" mode.Thiscouldbeaccomplished because each frame consistsoftwofields (every other scan line) that provide the Video Toaster with the abilitytograb four fieldsatonce, thus acquiring the full amountofresolution and color fidelity (McMahon, 1991). After flicker removal, the frames maybeloaded into Toaster Paint for manipulation. They mayalsobeloaded into Lightwave 3D and mapped onto various objects or they maybeloaded into the Video Toaster'sothermanipulation algorithms for additional enhancement.Field and Critchleyverting a 4096-color HAM image.Notethe excessive brightness on the right-hand sideofthe figure. Numerous efforts to modify this aspectofthe video frame have not yet been possible, providing evidence that greater careinthe original video-recordingiswarranted. There is a large void on the left sideofthe figure, the dimensionsofwhich are not readily discernible. To further enhance this image, an edge to the gray-scale image was overlaid onto the ftle producing more detail as shown in Figure 4. This was accomplished by applying a Laplacian Edge Detection OperatortoFigure 3. Improvements to the clarityofselected features in Figure 3 are readily apparent in Figure 4, although further improvements would clearly be desirable. A final refinementofthis image is showninFigure5.This enhancement was created by removing several shadesofgray below a certain threshold level. Ostensibly, this should show those areas that are most directly, or strongly illuminated. The water table at the site was encountered at a depthof118ftand is shown in Figure 6. Figure 6isalso a 64 level gray-scale modification created in the same manner as Figure 3. Enhancements to this image (Figure 7) were performed by taking a 32-level gray-scale image using Local Contrast Enhancement on the original 4096-color image and applying a Laplacian Edge Detection OperatortoFigure 6. Various features are clearly highlighted using the computer software as can be seen by comparing Figures 6 and7.Frames captured using the Video Toaster, once converted to the 16-bit standard, were manipulated using several off-the-shelf or public-domain image-pro cessing programs. Becauseofthe numberof colors, resolution, or sizeofthefiles(100,000+ bytes), only one package,PIXmate, was found suitable for useinthiseffort. PIXmateisan image-pro cessing software package published by Progressive Peripherals&Software, Inc.Thecapabilitiesofthe software are too extensive to detail here but it allows forsuchthings as(1)color and pallet mani pulation, (2) cutting and copying segmentsofan image, (3) color separation,and(4)image enhancement through suchbit-mapoperations as averaging, random ization, local-contrast enhancement, unsharpmasking, and median filters. Severalof these image-enhancement processeswere carried out on images capturedfromtapeofthe Henderson Road site. Figure 3 is a 64-level gray-scale imagemodificationofthe injection well at adepthof90ft.Itwas created by conFigure3:A 64-level gray-scale imageofthe injection well at the Henderson Road site, taken at a depthof90 feet. This image was converted from the original4096-color HAM mode.Notethe excessive brightness on the right sideofthe image and the large void on the left sideofthe image.173

PAGE 182

Field and Critchley Appalachian Karst Symposium, 1991ConclusionsLange, A.L. and Kilty, K.T., 1988,Detection and mappingofkarst conduits from the surfacebyacoustic andnatural potential methods:Contract Report Preparedfor the Superintendent,MammothCaveNationalPark,andtheEnvironmentalMonitoring Systems Laboratoryofthe U.S. EnvironmentalReferencesThe authors thank Dave PaigeandBart CasielloofNew Age Computers (College Park, Maryland) for providing access to and assistance in their video laboratory. Their assistance was invalua ble to this project and is gratefullyacknowledged . AcknowledgmentsVideo logging may be regardedasa very valuable borehole-geophysicaltechnique for displaying actual subsurface conditions; however, it may notbeenough,particularlyin contaminant hydrogeology. Capturing the videologinto computer picture files and manipula ting these files can greatly improveobservational interpretationsofsubsurficial conditions. As technology continuestoadvance, computer refinements willbecome ever more sophisticated. Whereas video enhancement hasbeenshown to be useful, limitations relatedto j the original video-tape quality are still a major impediment to the enhancement process. Improvements in video log ging, therefore, should begin with proper collectionofthe video logs, preferably with a professional geologist overseeing the logging operationtoensure thatimportant features are being given the nec essary care required for good visual dis play. Useofa Super VHS recordermayalso lead to better loggingoffeaturesandwould be more amenabletocomputerenhancement. ability to change the various shadesofcolor.Inone en hancement effort, the red componentofthe spectrumwasenhanced to give a more "realistic" look to the image rather than the predominant yellow hue givenoffby the artificial light used dur '. ing the video logging operation. Figure5:Gray-scale conversionofthe original HAM imageofthe 90-foot depth with selected background gray-scale levels removedtoallow for enhanced clarityofthose features most strongly illuminated. Figure 4: A 64-level gray-scale imageofthe 90-foot depth with Laplacian Edge Detection overlaid onto the original gray-scale image.Notethe enhanced viewofselected featuresinthe image. What cannot be shown in this paper are the color im ages -the original frames captured by the Video Toaster. These were more amenable to modification becauseofthe174

PAGE 183

Appalachian Karst Symposium, 1991 Field and CritchleyMilanovic,P.T.,1981,KarstHydro geology:WaterResources Publica tions, Littleton, Colorado, 434 p.McMahon,F., 1991, Newtek's Video Toaster: A new era in Amiga Video:AmazingAmigaComputing,v.6, p. 48-62. Molz, F.J.; Guven,0.;Melville, J.G.;andKeely, J.F., 1986,PerformanceandAnalysisofAquifer Tracer Tests withImplicationsforContaminant TransportModeling:EPA ProjectNumberEPA/600/2-86/062:Robert S.KerrEnvironmentalResearchLaboratoryofthe U.S. EnvironmentalProtectionAgency, Ada, Oklahoma, 88p.Molz,FJ.;Morin,B.H.; Hess, A.; Melville, J.G.; and Guven,0.,1989, The impeller meter for measuring aqui fer permeability variations: Evaluationandcomparisonwithothertests:Water Resources Research,v.25, p. 1677-1683. Nevius, J.G., 1987,A Hydrogeologic Analysisofthe Carbonate Aquifer and Associated Surficial Deposits in UpperMerionTownship.Pennsylvania:M.S. thesis (unpublished), GeologyDeparunent,UniversityofPennsyl vania, 85 p. Quinlan, J.F.andEwers, R.O., 1985,Groundwater flow in limestone ter ranes: Strategy rationale and procedure for reliable,efficientmonitoringofgroundwaterquality in karst areas:Proceedingsofthe Fifth National Symposium and ExpositiononAquifer Restoration and Ground Water Moni toring, Columbus, Ohio:National Water Well Association, Worthington, Ohio, p. 197-234. Sen, Z., 1987, Non-darcian flowinfrac tured rocks with linear flow pattern:JournalofHydrology,v.92, p. 43-57. Lange, A.L. and Quinlan, J .F., 1988, Mapping caves from the surfaceofkarstterranes by the natural potential method:Proceedingsofthe Second Conference on Environmental Prob lems in Karst Terranes and Their Solu tions (Nashville, Tenn.):NationalWaterWellAssociation,Dublin,Ohio, p. 369-390.Protection Agency,LasVegas, Nevada:TheGeophys ics Group,WheatRidge, Colorado, 40 p. Figure7:A 32-level gray-scale imageofthe1I8-footdepthofthe injectionwellinwhich Local Contrast Enhancementwas used to enhance the original 4096-colorHAMimage.Notehow clearly features are outlined with this typeofenhancement. Figure 6: A 64-level gray-scale imageofthe injection wellatthe HendersonRoadsite, takenata depthof118 feet. This image was also created from the original 4096-colorHAMimage.Notethe water table evident in the image.175

PAGE 184

Field and CritchleySen, Z., 1988, Dimensionless time-draw down plotsoflate aquifer test data:Ground Water,v.26, p. 615-618. Thrailkill,J.,1988,Drawdowninterval analysis: A methodofdetermining the parametersofshallow conduit flow carbonate aquifers from pumping tests:Water Re-176Appalachian Karst Symposium,1991sources Research,v.24,p.1423-1428.u.S.Environmental ProtectionAgency, 1988,RecordofDecisionforthe HendersonRoadNPL Site Injection Well Operable Unit. Hazardous Waste Management Division:U.S. EPA, RegionIII,Philadelphia, Penn.

PAGE 185

Appalachian Karst Symposium. 1991AnAttempt to ModelanAppalachian Karst Aquifer Using MODFLOWSaraA.Heller,Geology Department., CollegeofCharleston, Charleston, S.C. 29424ABSTRACTIn southeastern West Virginia, the Middle Mississippian Greenbrier Group, a limestone with some thin calcareous shales, is 600to900 feet thick and strikes about N 25 E where it crops out in Greenbrier County. This area was chosenfor aquifer modeling becauseofthe previous existenceofa detailed database, including dye tracing. MODFLOW, a finite-difference aquifer-modeling program available from theU.S.Geological Survey, has features that allow someofthe peculiaritiesofkarst aquifers to be incorporated. For purposesofmodeling, the Greenbrier Group was treated as an aquifer consisting of a single layer. The aquifer rests on a thick shale (the Lower Mississippian Maccrady Formation) whichisbroadly folded into a syncline, and sharply faulted upward in the southwest near the townofAsbury. The Maccrady Formation acts as a downward barriertoground-water flow, and a westward boundary to flow where it is faulted. Structural contours on the Maccrady were determined from geologic maps, and interpolatedtoan arrayofvalues filledtothe finite-difference grid. The finite-difference grid used for the model was oriented parallel to stratigraphic strike so thataquifer anisotropy could be simulated. The anisotropy factor was determined from directional measurementsof168 cave passage segments. Diffuse aquifer recharge (infiltrationofrainfall) was estimated from climatic records. Concentrated aquifer recharge (cave insurgences with capture zones on detrital rocks outside the grid area) were simulated by using recharge wells. Aquifer hydraulic conductivity was determined from well-pumping tests (diffuse flow) and dye-trace travel times (conduit flow). The model was run repeatedly to determine the effectsofvarying the input factors. Making the anisotropy or the hydraulic conductivity higher caused the hydraulic headtodrop and made the effectofthe cave insurgences less noticeable. All aquifer solutions showed a ground-water divide between Spring Creek and the Greenbrier River which was close to the one actually determined by tracer tests.HellerIntrOductionThe inherent complexityofkarst aquifers, such as concentrated recharge, discharge, and conduit flow, present great difficultiestonormal modeling methods. A well studiedWestVirginia limestone aquifer presents an opportunity to simulate water flow through a karst aquifer.Thepurposeofmodeling in this study wastoexperimentwithhow well the model approximates reality, and how sensitive the model istovarying the input factors. MODFLOW, a finite-difference aquifer-modeling program available from the U.S. Geological Survey (McDonald and Harbaugh, 1984), has some features which allow someof the peculiaritiesofkarst aquiferstobe incorporated. A commercial preprocessor was usedtogenerate the data files (translate structure contours, for example, to a tableofnumbers in the correct format). The resultsofthe model were gridded and contoured by a commercial graphicspackage.177The study area covers approximately 100 square milesofthe outcrop extentofthe Middle Mississippian Green brier Group in central Greenbrier County, West Virginia. The Greenbrier Group consistsof600 to 900 feetoflime stones, with a few thin shales. The Lower Mississippian Maccrady Formation, a red shale, underlays the Greenbrier Group. The limestone is partially capped by the LillydaIe Shaleofthe Upper Mississippian Mauch Chunk Group (Bluefield Formation). In the study area, the Greenbrier Group outcrop beltstrikes about N 25 E and stretches roughly between the Greenbrier River to the south, and about 15-20 miles northeast to Spring Creek (Figure1).The geologic structureofthe study areaisdominatedbybroad, asymmetric, northeasterly trending foldsoflow plunge, and severalenechelonreverse-faull/fold com plexes. Differential erosionofthe limestone has resulted in deep karst valleys trending along stratigraphic strike and receiving surface drainagefromsurrounding detrital rocks.

PAGE 186

HellerAppalachian Karst Symposium.1991Model AssumptionsBLUEFIELD FORMATION GREENBRIER GROUP MACCRADY ,",","FORMATIONo5 MI The geology, hydrology, and aquifer geochemistryof the Greenbrier Group has been previously studied byHeller(1980), Ogden (1976), Price andHeck(1939), Reger (1926), Chisholm and Frye (1975), and Jameson (1985). The major subsurface ground-water basins in Greenbrier County were delineated bydyetracing (Jones,1973).Finally, detailed cave surveys have been completed formany Greenbrier County caverns by Davies (1958), Stevens (1988), and theWestVirginia Association forCaveStudies. Any time a natural phenomenon is simulatedmathematically, simplifying assumptions must be made.Itis important to understand these assumptions as they affect the outcomeofthe simulation. In aquifer modeling,simplifications must be made about aquifer boundaries(cellidentification as barriers, constant heads,orwells), aquifer homogeneity, isotropy, thickness, hydraulic conductivity, and other input values. These simplifying assumptions are summarizedbyTable1,and discussed in detailbelow.Figure1:Geologic mapof the study area with locationsoftowns. The-fmite difference grid was aligned parallel tostratigraphic strike (Figure 2). The underlying Maccradyshaleacts as a strong downward barrier to flow, and, whereitisfolded or faulted upward, a westward barriertoground-water movement as well (flow is generally towards the southandwest). In fact, mostofthe flow through the Greenbrier Group occurs at or near the Greenbrier-Maccrady contact. The flow along the limestone-shale contact is sostrong178

PAGE 187

Appalachian Karst Symposium. 1991 HellerSimplifyingAssumptionsRealitySingle layer Several layers separated by thin leak v confinin!! units Thickness varies from Constant thickness approximatel y 600 feet at Spring Creek to 900 feetatthe Greenbrier River Single K Double K becauseofboth diffuse and conduit flow Anisotropy factor can Cave passages parallel to be determined from strike are largerthancrosscave passage frequency strike passages and orientations Anisotropy probably varies Homogeneous with thickness, facies changes, and degree ofkarstification Table1:Asummaryofthe simplifying assumptionsmadefor modeling purposes.thatthere areover60milesofmapped caverns (known as "contact caves")atthis horizon. Because the Maccrady shaleissuch a significant hydrologic barrier, structure contoursofthe Greenbrier-Maccrady contact were constructedfromgeologic maps (Figure 3). These structural contourswerefitted to the finite-difference grid dimensions, and the bottom elevations for the Greenbrier Group were interpo lated for each cell. Although other thin shales exist in the GreenbrierGroupthat will perchorconfine water, thesethinshalesarenotnearly as hydrologically significant. TIlus for modeling purposes, the GreenbrierGroupwas treated as a single layerofconstant thickness. In the study area, theoverlyingshaleofthe Bluefield formation is regionallyextensiveenoughtoconfinetheGreenbrierGrouponlyin the synclinal area belowMuddyCreekMountain (betweenFortSpring and Asbury, Figure 1). MODFLOW canbeprogrammed to determine whether now is confinedorunconfined basedonthe computed hydraulic head and the elevationofthe topofthe aquifer,andthen modify the flow equations accordingly. SpringCreek,MuddyCreek, Mill Creek,andthe Greenbrier River (Figure 4) act as the major ground-water discharge zones for the area, and are treated as constant-head cellsinthe grid. This means that the hydraulic head doesnotchange for these surface streams (in reality, it probably..... :':::::-.::.';';'m.::xW;;1'-:';'u.o.'.mmfm... .. mEI-:mf@1.'m Figure2:Finite-difference gridshowingcell identifi cation. Culverson and Milligan Creeks appear as constantheadcellsinthis example.CAVEINSURGENCECONSTANTHEAD rx:;:I NOFLOWBOUNDARY179 x ..:' ..-::: ..: .......

PAGE 188

Heller180Appalachian Karst Symposium. 1991Figure3:(At left) Structural contours (feet abovemeansea level) for the contact between the Maccrady Shaleandthe Greenbrier Group. varies by 20 feet or less). Culverson and Milligan Creeks were less easy to characterize because these partiallysegmented streams may be perched or evendryduringsomepartsofthe year. For experimental purposes, therefore, the model was set up two ways: first treating these streams as constant-head cells, and later removing themassuch, but instead treating their downstream termini ascaveinsurgences (recharge wells, to the model). Diffuse-aquifer recharge was estimated from climatic records. Concentrated-aquifer recharge occurs assmallstreams flow along the surfaceofthe detrital rocks andarecaptured by caves (both mapped and those only suspected) in the Greenbrier Group. These cave insurgencesweretreated as recharge wells in the grid. Davis Spring,themajor conduit spring that drains the area southofMaxwelton, wastreatedas a constant-head cell along theGreenbrier River, and notasa pumping well. Although theareadoes contain some widely spaced domestic water wells,thedischarge from theseistrivial compared to the capacityandhigh conductivityofthe limestone aquifer. So thesewellswere ignored for modeling purposes. Short-duration aquifer pumping tests (Heller,1980;Ogden, 1976) and dye-tracing (Jones, 1973) have deter mined the hydraulic conductivityofthe Greenbrier Group. The resultsofeight pumping tests yielded an averageconductivityof5.1x 104 feet per minute (7.3 x 10-1feetperday). Although it is unknown whether anyofthewellstested were receiving conduit flow, this rateofhydraulic conductivity is rather typical for diffuse-type flowinfractured aquifers (Heath, 1983). Dye tracing, on the other hand, which directly measures travel times via conduit flow, yielded an average hydraulic conductivity fortheDavis Spring basinof1.4 feet per minuteatlow flow(orabout 2000 feet per day). Although these two estimatesofhydraulic conductivity for the Greenbrier Grouparequite different, it is not unusual in karst aquifers tohaveboth diffuse and conduit modesofflowatthe sametime(Shuster and White, 1971). For modeling purposes,however, a compromise valueof1x10-2feet per minutegavethe most realistic solutions. Hydraulic conductivityinthe Greenbrier Groupisalsoprobably highlyanisotropic,with flow paralleltostratigraphic strike much greater than that across strike. MODFLOW allows the user to insert an anisotropyfactor(TRPY) to account for this. The valueofthis factorwasestimated by compiling a cumulative length-orientation diagram for 168 straight cave passages (Figure 5).Thesewere then subdivided into strike-aligned versus cross-strike passages, resultingina ratio 3.67 to1.Because thefinitedifference grid was oriented with the y axis paralleltostrike, the y-to-x conductivity ratio (TRPY) was assigned this value. In reality, however, strike-alignedcave

PAGE 189

Appalachian Karst Symposium. 1991HellerModel Resultspassages are generally largerinvolume than cross-strike passages, so this TRPY shouldberegardedasa minimum value. Figure5:Cumulative length-orientation diagram for168straight cave-passage segments .NFigure 4: (At left) Mapofstudy area showing surface drainage, locationsofknown contact caves (arrows), and subsurface drainage dividesofthe Davis Spring basin (dashed lines). The model was run repeatedly to detennine the effectsofvarying the input factors. Each result was compared with: (1) mapped water-table elevations for the area, based on 35 wells (Figure 6), and (2) drainage divides for the Davis Spring Basin, based on dye-tracing results (Figure 4). Model results that were not reasonably similartoreality (given the sparsity and noisinessofthe welldata)were discarded. Model solutions that were realistic often showed "dry" cells around the edges, near the no-flow boundaries. This occurrence could be expected, because the aquiferisthin and topographically high in these areas. A typical model solutionisshown by Figure7.This result treats Milligan and Culverson Creeks as constant head cells, uses a hydraulic conductivity of 1x10-2 fl/min, and a TRPYof3.67. The ground-water divide between Spring Creek and the Greenbrier River liesatabout 2040 feetinthis example, and coincides (roughly) with the one actually detennined by dye tracing (Jones, 1973). About aIIIII ," ,............. '--\\\\ .....1\+ 5 MI .....,, \ \,,I,DAVIS SPRING /----" \ ," "\,,I\\ ,....."-......S'IJ,. Il}g Creek, "\,I\ \ ) '-"'" .. l ,...._J""( \, \ .... I ......_'" 181

PAGE 190

Heller182Appalachian Karst Symposium. 1991Figure6:(At left) Contoursonthe water table (feet) above mean sea level) plotted from measurementsof35wells (shown as small dots). Figure 7: (Next page, at left) A typical model solutionfora run which included MilliganandCulverson Creeksasconstant-head cells, used a hydraulic conductivityof10-2 ,a TRPYof3.67, and an aquifer thicknessof1000 feet. Figure8:(Next page,atright) A model solution foranaquiferthicknessof200feet.Thesolution usedahydraulic conductivityof102 ,a TRPYof10, and treated Milligan and Culverson Creeks as recharge wells rather than constant-head cells. dozen cells went dry during this run. A model runwiththe same input values, but which treated MilliganandCulverson Creeksasrecharge wells rather than constant head cells was similar, although water levels were slightly higher. Several model runs experimented with the shapeoftheaquifer. If the Greenbrier Group was treated as a single 1500-foot layer with a horizontal geometry, the solution appeared surprisingly similar to thatofFigure 7, although the Spring Creek Greenbrier River divide shifted slightly to the southeast in these cases. This seems to indicate that, in spiteofthe complex natureofthe foldingoftheaquifer and its impermeable base, the hydraulic constraints (elevationofconstant-head cells, locationsofflow barriers) are more importantincontroUing the water flow.Ifthethicknessofthe aquifer is reduced while the complexly folded natural geometry is maintained, almost alloftheaquifer flow becomes confined at a thicknessofabout200feet, and water levels rise too hightoberealistic. A good solution was obtained in this case by increasing theTRPYto 10 (Figure8).This solution may be the most realistic for approximating the potentiometric surfaceofthe lowest memberofthe Greenbrier Group, the Hillsdale Limestone, which is confined by a thin shale. In general, decreasing the anisotropy factororthehydraulic conductivity had the direct effectofelevatingthewater table. By allowing the aquifer to drain less freely, this also resulted in maximizing the effectsofthe conduit recharge to the aquifer. Normally the low-flow cave-insur gence recharge had minimal effects on the model solutions unless the TRPY was less than 5orthe hydrualic conduc tivity less than10-2ft/min. All model solutions that contained Spring Creekandthe Greenbrier Rivers as constant-head cells resultedinaground-water divide that coincided roughly with theonedetermined by dye tracing, and that varied little in position no matter how other factors were manipulated. Thissuggests that the locationofthis divide is a direct resultoftheelevationofthese two major ground-water discharge zones. This hypothesis is supported by a few modelruns

PAGE 191

Appalachian Karst Symposium. 1991183Heller

PAGE 192

Heller Figure 9: (At right) A model solution that used no constant-head cells at all, and that treated all springs or cave insurgences as rechargeorpumping wells. The hydraulic conductivity used was 102 thelRPYwas 5, and the aquifer thickness was 1000 feel. in which all constant-head cells were removed, and all known conduit springs were treated as pumping wells (Figure 9). The interesting result is a very flat water table (note that the contour interval is only twofeeL)with gener alized flow from north to south. This solution is probably highly unrealistic, however, as there are likely tobemany as yet unknown diffuse springs that discharge into the two major rivers, and that could not be simulated accurately.Conclusions1.The most important controlling factors in the solution for hydraulic head seem tobethe hydraulic conductivity and anisotropy factor, followed by aquifer thickness as it affects confined versus unconfined flow.2.The locationofthe ground-water divide between Spring Creek and the Greenbrier River seems tobea direct resultofonly the elevationofthese discharge zones treated as constant-head cells. 3. The complexity in shapeofthe aquifer boltom, exceptasit affects the grid boundaries, seems to have little effect on the model solution.AcknowledgmentsThe author extends gratitude to Alan JohnsonofGen eral Engineering Laboratories for his patient instruction and trouble-shooting services with both hardware and soft ware. Appreciation is also extended to the CollegeofCharleston Geology Department which provided the computer time, and to Mitchell Colgan for reviewing the manuscript.ReferencesChisholm, J.L. and Frye, P.M., 1975, Recordsofwells, springs, chemical analysesofwater, biological analysesofwater, and standard streamflow data summaries from the Upper New River Basin in West Virginia: U.S. Geological Survey Basic Data ReportNo.4,8Op.Davies, W.E., 1958, CavernsofWest Virginia:WestVirginia Geological and Economic Survey Volume 19, 350p.(Reprinted in 1965 with a 72-page supplement as volume 19A.) Heath, R.C., 1983, Basic ground-water hydrology: U.S. Geological Survey Water-Supply Paper2220,84p.184 Appalachian Karst Symposium.1991

PAGE 193

Appalachian Karst Symposium, 1991Heller,S.A.,1980,AHydrogeologicStudyofthe Greenbrier Limestone KarstofCentral Greenbrier County. West Virginia:Ph.D. dissertation (unpub lished), West Virginia University, 167 p. Jameson, R.A., 1985,Structural Segments and the Anal ysisofFlow Paths in the North CanyonofSnedegar Cave. Friars Hole Cave System. West Virginia:MS thesis (unpublished), West Virginia University, 421 p. Jones, W.K., 1973,Hydrologyoflimestone karst in GreenbrierCounty,WestVirginia:West Virginia GeologicandEconomic Survey Bulletin36,49p.McDonald, M.G. and Harbaugh, A.W., 1984, A modular three-dimensional finite-difference ground-water flow model:U.S. Geological Survey TechniquesofWater Resources Investigations,ChapterAI,528 p. Ogden, A.E., 1976,The Hydrogeologyofthe Central185Heller Monroe County Karst. West Virginia:Ph.D. disserta tion (unpublished), West Virginia University, 263 p. Price, P.H.andHeck, E.T., 1939, Greenbrier County:West Virginia Geological Survey County Report, 846p.Reger,D.B.,1926, Mercer,Monroe,andSummersCounties:West Virginia Geological Survey County Report,963p.Shuster,E.T.andWhite,W.B.,1971,Seasonalfluctuations in the chemistryoflimestone springs: A possible means for characterizing carbonate aquifers:JournalofHydrology,v.14,p.93-128. Stevens,PJ.,1988, Cavesofthe Organ Cave Plateau, GreenbrierCounty,WestVirginia:West Virginia Speleological Survey Bulletin9, 200 p.

PAGE 194

Plate Appalachian Karst Symposium.1991 BuriedTreasures PlateI:Entrance to museum Exhibit,Buried Treasures: Cavesofthe Virginias.This traveling exhibit, designed by the Virginia MuseumofNatural History, emphasizes the origin of, the biology of, and man's interaction with caves, including history, archeology, exploration, and environmental issues.Theexhibit was funded, in part, by theCaveConservancyofthe VirginiasandtheRichmondArea Speleological Society.Thepremiere showingofthe exhibit occurredatRadford University during the Appalachian Karst Symposium.Photograph by KarenM.Kastning.PlateJ:Environmental display within theCavesofthe Virginiasexhibit shown above. Trash collected from several sinkholes within the Virginias has been arranged with a water spigot10emphasize that pollutants introduced in sinkholes can easily enter drinking water with littleorno filtration.Thesign asks the question, "Would you want to drink water that had flowed through this trash?"Photograph by KarenM.Kastning.186

PAGE 195

Appalachian Karst Symposium, 1991 WernerHydrochemical Characteristicsofthe Greenbrier Limestone KarstofEast-Central West VirginiaEberhard WernerP.O. Box 795 Morgantown, WV 26507-0795ABSTRACTLimestonesofthe middle Mississippian Greenbrier Group comprise only about10%ofthe surfaceinsouthern Randolph and northern Pocahontas counties, but are very important hydrologically. The areaisin the folded-plateau region,justwestofthe Valley and Ridge Province, with 300 mofstructural relief and 700 moftopographic relief. Because clastic rocks overlying the limestones form mostofthe surface, there is relatively little karst-landform development on the surface, although caves are common, and streams usually sink during the drier partsofthe year as they encounter the limestone. Muchofthe karst develop ment occurs as underdrainsofsurface valleys, although in a numberofcases interbasinal stream piracy does occur. Water analyses,ofabout 250 samples from 50 locations, show considerable variability, even for sam pling stations very close to each other. Statistical analysis places the samples into groups characteristicofsurface streams, high-flow conduit springs, low-flow conduit springs, and diffuse-flow springs. Factor analysis showsthat the chemical components group into carbonate-dissolution products, carbon-dioxide controlled parameters, and pollution constituents. Controlling hydrologic factors appear to be the typeofrechargetothe karst system that controls the carbon-dioxide related parameters, and the lengthoftime and degreeofcontactofwater with the rock within the system, controlling the dissolution products. Pollution constituents are not related to geological factors, but appear to be strictly anthropogenic. Discharge measurements were seldom made during water sampling, but determinationsofchemical denudation are possible in a few cases. Results range from a lowofImm/katoa highof20mm/kaoflimestone removed by solution. Little correlation can be clearly demonstrated between calculated solutional denudation and any other parameter, although there appears to be an inverse relation between average gradi ent and calculated denudation rate.IntroductionThe karst on the eastern edgeofthe Appalachian PlateausProvince is developed on the Middle Mississippian Greenbrier Group. Unlike most karst areas, the northernhalfof the Greenbrier karst has relatively little surface exposureoflimestone (Figure 1); however, the limestones underlie mostofthe area at depths shallow enoughtobe intherangeofgroundwater circulation. Therefore, the area is dominated by karstic hydrology and mostofthe water inthearea has characteristicsofkarst water. This paper summarizes hydrochemical aspectsofthe Greenbrier karstofnorthern Pocahontas and southern Randolphcounties in eastern West Virginia. The physical hydrologyofthis same region was summarized in an earlier paperbyMedville and Werner (1977). Mostofthe 187 data used to develop the discussionofthis paperarederived from investigationsofthe Edray Fish Hatchery springsincentral Pocahontas County, and the Elk River drainage basin in northwestern Pocahontas County and south western Randolph County. Detailed reportsofthe hydro chemistryofthese twoareasare in preparation.StudyAreaThe study area is located at the eastern edgeofthe Appalachian Plateau (Allegheny Plateau) Province. Oneofthe main streams draining the area, the Greenbrier River, is generally considered to be the boundary between the plateaus and the Valley and Ridge ProvincetotheeastGeologic structures underlying the area are broad, gentle folds that keep the dipofbeds quite low, no more than a

PAGE 196

Werner Appalachian Karst Symposium,1991 GreenbrierGroupoufcropSamplelocationslloIo /::,. o5streamslargeconduitspringssmallconduitspringsdiffusesprings10kilometers20 4 Figure1:Location mapof lhe study area. Not all samplingsites are shown in congested areas. Water types indicated for sample locations are based on field observations ralher lhan lhe chemical characterof lhe water. Outcrop extentofthe Greenbrier Group taken from Cardwell, Erwin, and Woodward(1%8).188

PAGE 197

t-----+OMaccrady and Pocono GroupsAppalachian Karst Symposium. 1991fewdegrees in mostofthe area. Faultingisprevalent, but displacement on any single fault is generally measured in millimeters or centimeters (Eddy and Williamson, 1968; Werner, 1972), therefore noneofthese faults would appearonthe typical geologic map. The karst-forming rock in the areaisthe Middle Mis sissippian Greenbrier Group (Figure 2) which consists primarilyoflimestones with a few marker bedsofshale or shaley limestone. Thickness varies from about 100 m at the northern endofthe study area to 150 matthe southernend.The rocks underlying the Greenbrier are shales and sandstones, and can be disregardedinany discussionof mst development for this area. Overlying the Greenbrieraresequencesofsedimentary rocks that are mostly nonmarineclastic rocks, but repeated incursionsofthe sea, es pecially during the depositionofthe frrst hundred metersofrockafter Greenbrier time, have formed a numberofthin, impure limestone beds. Muchofthe sandstone and shaleinthis sequence also contains calcite as cementorveins.Eventhe purely non-marine sequence, a considerable thicknessabove the last marine limestone, contains a numberoffresh-water limestones. Thus, there is no sharp boundarybetween the carbonate and non-carbonate rocksin this area. Precipitation averages 1200 mm/yrto1500 mm/yr. Runoff determined from U.S. Geological Survey gauging slJuions in the area ranges from about 600 mm/yr to 900 mm/yr. The Greenbrier River drains most of the eastern partof the study area, and the Elk River drains the western part. Interbasinal subterranean stream piracy through solution conduits iscommon,but, for the most part this is confined within the major river drainage basin (as far asisknown).Theonly significant piracy between major surface drainage basins occursina tributarytothe TygartValleyRiver, northofthe Elk River basin, into a solution conduit which eventually discharges into the Elk Riverbasin(Medville and Werner, 1977).LITHOLOGY ''-r.... *' I Gl I fYl(;) 11.0I 0 II (')C.*(' I Clt:1-,---,-T ")"'\ ..' 'J.:'.Yr;> J J.I T ( -----:::--;:::;-=--==---1--_UNITMauch Chunk Group0.... :::::J o0::: <..9 0:::W0::: m zwW0::: <..9 Wernermeters150100-50The Greenbrier limestone terrane has undergone karsti fication throughout, but in the northern partofits extent, including the study area, there are few sinkholes. Thisisbecause limestone comprises less than 20%ofthe surface outcrop. Southofthe study area,inGreenbrier and Monroecounties, surface exposuresoflimestone are propor tionally greater and sinkhole developmentisintense.Methods.Approximately 260 water samples were collected fromnearly80 locations (springs, and surface and cave streams) during a two-year period from October 1973 to October1975.Chemical analyses for constituents importantincarbonate dissolution were performed on 230ofthese samples,and several derived values were computed. These datawereusedtodevelop the discussioninthis paper. Table 1 189 Figure2:Stratigraphic column for the study area. Lith ologies and thickness are representativeofthe centerofthe study area. The section thinstothe north, and thickenstothe south, primarily through corresponding changesinthe lower unitsofthe Greenbrier Group. Hydrologically im portant aquitards and aquicludes are indicatedbyasterisks. gives the descriptive statistics associated with these data. Most locations were sampled onetothree times, but sever al locations were sampled more often. In particular, the springs associated with the Edray State Fish Hatchery were sampled about 20 times each, and the streams and springsinthe upper Elk River drainage basin were sampled five or more times. Temperature and specific conductance were measuredinthe field. Although pH was usually measured with a

PAGE 198

WernerVariable median range 1st decile 9th decile pH 7.76 6.87/8.35 7.37 8.08 Ca++ 34.0 6.9/92.9 15.2 63.1 Mg++ 2.62 0.49/12.44 1.46 4.52HC03-102.5 16.0/298.5 44.3 185.1 Cl5.6 0.5/67.6 1.6 18.4N03-3.3 0.2/13.7 0.4 6.1 SIc -0.19 -2.11/1.15 -1.22 0.53 SId -0.66 -2.48/0.67 -1.64 0.07 pC02 -2.80 -3.63/-1.94 -3.16 -2.48 Ca++/Mg++ 7.2 2.5/52.3 5.0 10.1 Table1.Descriptive statisticsofwater-sample analyses. pH meter in the. field (at the vehicle), generally afewmInutes after collectIon, some collected samples were not measured until returned to the laboratory three or four hours after collection. The portable instrument was very sampling locations, probably becauseofhigh humidity, so attemptsatin situpH measurements were abandoned. A later setofmeasurements with a more stable meteratsomeofthe springs showed that pH varied by 0.1 units over periodsofa few minutes, and that up to 24 hours later on a sample collected ma plastiC bottle andkept cool agreed with those madeinthe field to within less than 0.1 units. Analysis for bicarbonate was based on titration for alkalinity with hydrochloric acid to the inflection point near pHof4.5. Calcium and magnesium determinations were by EDT A titration. The remaining analyses were not performed on all Chloride was determined for 225 samples by titratIOnWithmercuric nitrate using diphenylcarbazone indi cator, and nitrate was determined colorimetrically for 150 by cadmium reduction method. Due to prob lems m applymg the analytical techniques and the relative ly low levelsofthe remaining ions, noneofthese were incorporated into the data set for the statistical analysis. Sulfate analysis by the turbimetric method was performedonapproximately 20%ofthe samples, and never exceeded 10 mg/l, which was deemed the reliable resolution limit for the technique. On the10%ofsamples tested for sodi um and potassium by atomic absorption spectroscopy, no sample exceeded the detection limitofthe instrument for potassium (approximately 5 mg/l), and sodium was below 5 mg/l for all samples except a few with high chloride levels (but still below 10 mg/l). ,,:,alues calculated for the following from the data: saturation indices with respect to calcite and dolomite, initial equilibrium carbon-dioxide concentration, and calcium/magnesium (molar) ratios. Chemical-equilibium constants for these calculations were 190Appalachian Karst Symposium,1991obtained from a numberofsources (Hamed and Owen 1958; Plummer and Busenberg, 1982; Langmuir, 1971).'General Characterofthe Waters., The chemical characterofall the karst waters analyze{! Indicated that they were relatively nonpolluted. The principal dissolved components were those derivedfromlimestone solution, with virtually no influence from other sources. A significant although relatively minor exception was the additionofchloride from road-deicing saltsinthegroupofspringsatthe Edray Fish Hatchery (discussed previously in Werner, 1977). There has been little effect from or septic tanks as indicated by lowdissolved-nitrate levels, norisacid-mine drainage involvedasseen low dissolved sulfate levels. Although gypsum occursInsome areas in the Greenbrier limestones, none has been recorded in the immediate area, and low sulfate levels reinforce this. Thus, the data set used in this anal ysis would appear to represent as pure a setoflimestone karst waters as possible, without complicating factors.Spatial and Temporal Variations. Although, the water under discussion evolves largely In contactWiththe Greenbrier limestones which have significant chemical and physical variation, the variations seen in the chemical characteristicsofthe water do notcorrespond to limestone characteristics. Field observationsofthe springs under discussion, including traversing someofthe caves through which the water flows before emerging and dye .tracingtodelineate someofthe recharge-discharge connectIOns, has provided information for a general inter pretationofthe spring types foundinthe area. Various workers (Shuster and White, 1971; Harmon and others, 1972, 1973; Drake and Harmon, 1973)have cla,ssifie? karst waters on the basisoftheir positional prop ertIes Withregardtothe aquifer, and have attemptedtorelate this classification in one wayoranother to the chemi cal ofthe waters. In a very general way, allwaters m most karst terranes are mixturesoftwo end-member types, surface stream recharge and diffusing soil waters, that have passed through sequential stepsofchemical Surface waters are more or less in equilibriumWithatmospheric carbon dioxide, and soil waters areatequilibrium with soil carbon dioxide that may be about two ordersofmagnitude higher. Drake and Harmon(1973) ata classification comprising six water types -allo surfa.ce recharge, soil-zone recharge, conduit springs, diffuse spnngs, wells, and surface discharge that couldbedistinguished chemically on the basisofsaturationwithrespect to calcite and equilibrium carbon-dioxide partial pressure alone. Considering basic principles with regard to ground water, watersofa given terrane progress through several stages as they pass through the aquifer (Figure 3); thereisa source recharging the aquifer, water passes throughstor-

PAGE 199

AppalachianKarst Symposium,1991age in the aquifer to a discharge point, and then leaves the terrane. Applying this model to the real world can leadtosome difficulties.Forexample, field observations allow one to easily determine that water is obtained from a stream, but determining whetheritishydrologically aboveorbelow karst terrane requires additional information. Al though that determination is not usually difficult, an addi tional complication exists.ThediagramofFigure 3 seems to clearly distinguish the two typesofstreams, en tering and leaving,butit is possible that a stream entering one karst terrane hasjustleft another. In the caseofthe Greenbrier karst, lithologiesaresuchthat therocksequence is divided into several hydrologic units, eachofwhich would then present one cycleofthe model presentedinFigure 3. Furthermore, although not shown on the dia gram, some streams flow on the surface across a karst ter rane, being chemically modified by dissolving limestonefromtheir beds. Thus streams are not always easily classified,other than as streams. Somewhat lesser problems exist with spring waters. Shuster and White (1971) classified springs in the Nittany ValleyofPennsylvania on the basisofseasonal variationinchemistry as either conduitordiffuse springs. Since then, this classification has been applied in its originalorslightly modified form to other areas. Although it is usuFigure 3: Water routing through a karst aquifer.Thevertical column on the right is the fundamental sequenceofwater passing through any aquifer. The left partofthe diagram is more specific to the karst aquifer with the various forms each segmentofthe routing sequence maytake.Prominent (and most likely) paths are shown as solid lines and less likelyorlow mass-transfer pathsaredashed lines. Only natural karst paths are shown; some streams simply flow across, and wells may abstract waterfromthe aquifer; neither are shown. 191Wernerally not difficulttodetermine the physical classificationofa spring when adequate observational dataareavailable, this chemical classificationofthe spring water may be more difficult becauseofthe complex path this water may have taken through the various stations denoted on Figure 3, and the attendant complications in the chemical evolu tionofthe water. In the area under discussion in this study, springs are classified as being conduit (cave) springsoftwo types, "large" and "small", and diffuse springs. Differentiationbetweensmall-and large-conduit springs is not based entirely on discharge, although in gen eral, "large" springs have higher dischargesthando "small" ones.Themajor distinctionsaregradient steepness and characterofthe conduit. "Large cave springs" flow from relatively steep conduits with little poolingofwater. The cave passage behind such a spring tendstohold a wall-to wall stream flowing throughout. This results in faster flow-through rates and consequently less time for dissolv ingofrock and comingtochemical equilibrium. Conduit systems have less storage capacity and springs are more flashythanthe other types, and therefore have greater coef ficientsofvariation in all characteristics. Reviewing data presented by the workers cited above indicates that various propertiesofthese springs fall outside the ranges reportedbythem,soperhaps largecavesprings do not exist in their study areas. "Small cave springs" flow from conduits that are usu ally less steep and contain pools. This produces a longer flow-through time for systems that have the same sink-to rise distance, and consequently the waterhasmore time to dissolve rock and cometoequilibrium. These systems are also less flashy in discharge and chemical character. The principal sourceofwater for both typesofconduit springs ispointrecharge from sinking streams into solution conduits, but there is a contribution from seepage water infiltrating through soil and entering through fractures in the ceilingsofconduits. Becauseofthe differences in flow-through time and overall flow volume, small cave springs tend tobemore affected by the contributionofseepage waterandshow that as higher contentofdissolved solids and higher equilibrium carbon-dioxide partial pres sures. These springs seem to correspond to all conduit springsofthe other workers. Diffuse springs derive allornearly alloftheir water from surface water infiltrating through soil cover. Flow paths are tortuous and constricted and therefore discharge tends tobelimited more by hydraulic propertiesofthe fracturesorpores than by availabilityofwateratthe infIl tration source. Flow-through times tend to be very long, for the lengthofthe paths, and as a result, these waters reach saturation with respect to the rocks they contact. Temporal variationinthe springtypesis in agreement with the fmdingsofShuster and White (1971) and may be seen in Figure 4. Differences in seasonal fluctuations among springs is pronounced. Figure 4 shows coefficientsofvariation for pH and bicarbonate content that approxi-

PAGE 200

Wernermately double between diffuse springs and small-conduit springs, and again between smalland large-conduit springs, for the two year period. Although these example springs have more data available than do others in this study, comparable coefficientsofvariation were obtained for other cases in which multiple samplings were done. The sampling locations were classified based on field observations, prior to an analysisofthe chemical data. These classifications are the ones used in Figures 1 and 5.Itisnoteworthy that, in several places within the study area, all three typesofsprings appear within a few tens or hundredsofmetersofeach other, rising from the same stratigraphic horizon.Ifa classification based on chemical propertiesofthe waters is considered against a classifica tion based on field observations, similar groupings occur. This can be illustrated by plotting variables against each otherasin Figure 5. Paper plots allow for only two varia bles; perfect class groupings are expectedifonly two vari ables were required for classification, such as in the data setofDrake and Harmon (1973). As can be seen in Figure 5, clearly more than the two variables used are required for complete separationofgroups; however, despite consider able overlap, the general trendsofvariation among the wa ter classes are readily apparent. Stream samples are lowindissolved solids and carbon dioxide; large-conduit springs reflect higher carbon-dioxide levels within cave passages through which they pass (although dissolved solids do not appreciably increase probably becauseofvery fastf1owthrough times). Small-conduit springs, becauseoflonger residence times, increase in dissolved-solids content, but are in equilibrium with essentially the same cave atrnos-Appalachian Karst Symposium, 19910 400' 00 0 0300000 A SpC0 'ho.d" 0 0 0 0200do I:>. I:>.@ 0 1:>.1:>.CblJ;. 0 {fg,1:>.1:>.0600<9 0 0 I:>.'1? 10080 -3.0 -2.0pC02Figure5:Cross plotofspecific conductance (as represen ting total dissolved solids) against log equilibrium partial pressureofcarbon dioxide for all sampling locationsinthis study. Where more than one sample was taken at a location, the mean values were used in this figure. Classi ficationoflocations is based on field evidence. Open cir cles are stream samples, triangles are large-conduit springs, squares are small-conduit springs, and filled circles arediffuse springs. phere as the large-conduit springs. Diffuse springs have contributions from some higher carbon-dioxide levels,andlonger residence times, as well as more intimate contact with the rocks, allow for higher dissolved-solids concentra tions.StatisticalAnalysesFigure 4:Anexampleoftemporal variation for the three spring types. The springs are threeofthe springsatthe Edray Fish Hatchery. Bicarbonateisrelatedtothe meanofall values for the particular spring so as to show variation over time. Means for the springs are 60 mg/l for the large conduit spring (triangles) with coefficientofvariationof30%, 120 mg/l for the small conduit spring (squares) with coefficientofvariationof18%, and 195 mg/l for the diffuse spring (filled circles) with coefficientofvariationof7%. 0'"<.>:r: 1.0 1;1___ ill 0 00 6 o 6 0 6 0 6660666666 0 6 6 6 6 6 000 ,... el.6El:lfj 197311974I1975192 A varietyofstatistical tests were appliedtothe chemi cal data collected. In general, the resultsofthese testsconfirmed the hypotheses generated from field observations and general expectations based on the common understand ingofkarst hydrochemistry.CorrelationAnalysis.Pearson's Product Moment Correlation coefficients were generated for the chemical variables and two generated time variables (Table 2). One time variable (WD)isthedayofthe water year, that is, 1 October is set equalto1,1 November becomes 32, etc.; the other (DD) is number of days away from 1 April (middleofthe water year), thatis,both31March and 2 April are set to1,both 1 May and 2 March are equal to 30, etc. These variables were createdtotest the basisofcommonly held wisdom with regardtokarst-water chemistry. Discharge is expected to beatminimum at the beginning (and end)and maximum atthemiddleofthe water year. Most dissolved constituentsareexpected to holdaninverse relationship to discharge; ifso,they should correlate with DD. Conversely, dissolved mineral constituents have often been shown topeakwithan increase in discharge and then decline, though notinphase with the discharge. In this case, the WD variable

PAGE 201

Appalachian Karst Symposium,1991Table 2. Pearson's Product Moment Correlation Coefficientsofall water data.WernerpHT0.20*TNotes: Statistical significanceofeach coefficientSpC0.49** 0.24*SpCisdesignated as follows:Ca++0.44** 0.31 ** 0.94**Ca++**=alpha <0.001Mg++0.41** 0.24* 0.77** 0.75**Mg++*=O.OOIO.05N03-0.27* -0.110.39** 0.32** 0.23+ 0.35**0.41**N03-WD-0.20* -0.18+ -0.21* -0.30**-0.34** -0.21* -0.14-0.13-WDDD-0.17+ 0.57** 0.21*0.22* 0.23* 0.25** 0.07-0.18+ -0.03might be a better representationofa time variable expectedtorelate to the changesinconcentration. As expected for karst waters, the highest correlationwasbetween bicarbonate and specific conductance (0.96); correlation between calcium and bicarbonate (0.94) and cal cium and specific conductance (0.94) was nearly as high. Because the numberofobservations is fairly large, several other, weaker correlations are also statistically sig nificantatan alpha levelof0.0001. It is less certain that a cause-and-effect relationship is present, despite the appar ent high significance level, but relationships indicated byatleast someofthese correlation coefficients can be ration alized.Forexample, pH correlates positively with all the limestone dissolution variables (specific conductance and concentrationsofcalcium, magnesium, and bicarbonate) because as solutionoflimestone progresses, the dissolved constituents obviously increase and that reaction serves to increase the pH. Somewhat surprising is the general lackofcorrelation between the time variables and chemical variables. The best correlation is between temperature and the DD varia ble, at 0.57, indicating higher temperatures at the begin ning and endofthe wateryear thaninthe middle, which is hardly surprising. The relatively low correlation is most likely because mostofthe samples are spring samples thatLendto vary relatively little in temperature. Both calciumandmagnesium are negatively correlated with position inthewater year (WD), and slightly less positively correlated (at a lower significance level) with DD, perhaps indicating a relationship between these dissolved constituents and dis charge that is outofphase, or, as noted above, indicating flushingofthe more concentrated waters during the initial higher-discharge period and not a strong inverse relation ship with discharge.DiscriminantAnalysis.Discriminant analysisisnormally used in order to 193 group multivariate observations into classes.Itis necessaryto determine what the classes are to be on some basis, and so designate observations included in a training set. The procedure operates by assuming an equal probabilityofeach observation falling into eachofthe designated classes, and then developing a mathematical model (equa tion) that would calculate a high probability, ideally 1.0, that each observation actually falls into the group that has been predefined for each observation. In practice, for most real data sets, not all probabilities reach 1.0, nor is each observation in the training set actually assigned to the pre defined class. The procedure attempts the best classifica tion possible, given the data supplied. After the modelisdeveloped, it can thenbeapplied to other data not partofthe training set to classify those observations accordingtothe same criteria. For this project, observations on the water sources al lowed a classification into four basic water types streams, large cave springs, small cave springs, and diffuse springs. The samples taken at the Edray Fish Hatchery andatsever al surficial stream sites provided a sizable, easily classified subsetofdata. In the caseofthis data set,79observations were predefined as belongingtooneofthe four water types and used to train the discriminant program. Designatedimportant variables for this run were pH, temperature, con centrationsofcalcium, magnesium, bicarbonate, chloride, and nitrate ions, saturation indices with respect to calcite and dolomite, and equilibrium carbon-dioxide pressure. The discriminant equations were then applied to the train ing set and also to the mean valuesofthe same parameters for all sampling stations. The resultsofthis analysis were somewhat surprisinginhow well the procedure performed to classify both the observations includedinthe training set and the meansofobservations for the remaining data.Theobservations classed as streams were principally surface streams flowing across the limestone outcrop, but included those that had just entered the limestone outcrop as well as those that had traversed the entire outcrop. Also included were two

PAGE 202

Wernerobservations from cave streams taken at locations in the cave where the stream had not been underground for a long distance. For these "stream" samples, all classified as "stream" at a probabilityof1.0 except for one that came from a cave stream at a probabilityof0.97. "Large cave springs" are those with relatively high discharge and a fairlyhigh gradient (about 0.06). These are the main springsofthe Edray Fish Hatchery and their recharge is from stream sinks approximately one mile (air line distance) from the springs (Medville and Medville, 1976). Intuitive ly, one might expect some problem in differentiating wa ters from such springs from the stream waters. However, only one sample outof19 used in the training set as "large cave spring" (and, in fact, the only sample in the entire set) was misclassified (as a stream). About a third, however, rated a probability less than 1.0 (although always 0.8 or higher). "Small cave springs" have a lesser dis charge, and a lower gradient (about 0.03), and are knowntoissue from a cave passage (Salmon Cave) that has a consi derable numberofpools (Medville and Medville, 1976). All samples classified correctly; only one had a probabilityofless than 1.0 (at 0.98). The last group, "diffuse springs" was based on samples from a single tufa-deposit ing spring at the Edray site. Allofthese samples classi fied correctly with a probabilityof1.0. A pointofinterest in regard to the spring samples used in this set is that they are located within 1 kmofeach other, and all issue from the same stratigraphic horizon near the baseofthe Greenbrier Group. The chemistry is sufficiently distinct among these springs that they can be grouped into oneofthe three spring types simply on the basisofthe chemical analysisofa single sample. Unlike the setofsprings studied by Shuster and White (1971) that required a temporal factor for classification, the Edray springs apparently are sufficiently different from each othertonot require knowledgeofthat factor. Less accurate was the allempt to classify the remain ing locations. The meansofall samples takenateach location were classified against the discrimination model. Only 57%ofthese were classified the same as a classifica tion based on field observations. However, most disagree ments involve locations that are not very well known, and only 14%ofthe locations were clearly misclassified, prob ably resulting from too few or anomalous samples for the location. The experience here is quite different from thatofDrake and Harmon (1973). They found that only one vari able, saturation with respect to calcite, was sufficient to separate all their classes (with the exceptionofthe spring classes), and the additionoflog carbon-dioxide equilibrium pressure was sufficient to completely separate their six classes. In the present study, no single one or even two variables provided a very good separationofthe four de fined water classes. Oneofthe best two-variable determi nations is between specific conductance and log equili brium carbon-dioxide pressure (Figure 5). Although some 194Appalachian Karst Symposium, 1991senseofdiscrimination is apparent, thereisstill considera ble overlap. A significant difference exists between the terranesused for the two studies. Drake and Harmon (1973) obtained their water samples from areas in the Val ley and RidgeofPennsylvania, where the carbonates areallin one sequence, overlain by clastic rocks that contain al most no carbonate, even as cement. On the other hand, the Greenbrier GroupofWest Virginia is overlain by a mostly clastic sequence that contains a number of limestone beds as well as calcite-cemented sandstones. From a hydrochemical standpoint, then, water reachingtheGreenbrier limestones has probably already passed through a carbonate aquifer, however thin or impure it might be. Even though a fewofthe stream samples had quite low levelsofcalciumorbicarbonate, none were as lowasthemeanofDrake and Harmon's stream samples. Thus, becauseofmultiple cyclesofthe routing shown in Figure 3, waters from the Greenbrier terrane have had a more complex evolution than the central Pennsylvania waters.Factor Analysis.Factoranalysis was done using analytical and computed variables on samples derived from springs attheEdray Fish Hatchery; stream samples were excluded. Each data group large cave springs, small cave springs, and dif fuse springs was run separately.Ingeneral, four factors appeared (Table 3). The pollutionvariables chlorideandnitrate appeared as separate factors, which is not sur prising because the sources are so different. However,the"karst water" variables separated into two distinct groups. The resultsofrock solution calcium, magnesium, bicar bonate, and the saturation indices were associated on one Spring group Large conduit Small conduit DiffuseIcarbonatecarbonate carbon F solution solution dioxide 40% 31% 29% A 2 carbon carbon carbonate C dioxide dioxide solution 21% 31% 26% T 3 chloride chloride chloride014% 16%14%R4 nitrate nitrate nitrate13% 12%12%Total variation explained 88% 90% 81% Table 3. Resultsoffactor analysis on the springsoftheEdray Fish Hatchery area. Factor names are generalized from the variables actually composing them. Percentages are variation explained by the factor in the data set for that spring group.

PAGE 203

Appalachian Karst Symposium. 1991factor, and pH and equilibrium carbon-dioxide pressure associated on another factor. As noted above under correla tion analysis, pH did not correlate highly with the rock solution factors, indicating that there was independenceofthe variables.Theconclusion to be drawn from the statis tical analysis is that the final pHofthe spring water is more controlled by the partial pressureofcarbon dioxideofthe hydrologic system than by the amountofrock solu tion. This follows from the chemical pathway followedinthe evolutionofkarst water(seeWhite, 1988, p. 205); given an open system with water nearing saturation andtheunits in which the analytical values are measured, thepHchange is small relative to the amountofchange in the bicarbonate (and calciumormagnesium) ion concentration. Ogden (1976) performed factor analysis on his data set of water analyses from springs and wells in an areainMonroe County, West Virginia, approximately 70 km tothesouth and also underlain by carbonatesofthe Green brier Group. Although different in detail, his results were largely the same with the exceptionofone additional factorrepresentingthemagnesium-dolomitecomponent.Ca++/Mg++values for his water samples were lower although the reason for this is not clear. His area is closertothe Cambrian-Ordovician carbonates and there maybesome movementofwaterorweathered rock from that ter rane to the Greenbrier limestone terrane. The Greenbrier limestone itself has similar Ca++/Mg++ratios in both areas according to analyses given in McCue, Lucke, and Woodward (1939).DenudationRatesAlthough only a few discharge measurements were made at the timeofwater sampling, it is possible to deter mine instantaneouscarbonate flux rates (based on the dis solved calcium plus magnesium) for a numberoflocationsandtimes. For sixteenofthe springs, low-flow measure ments were made on one instance each. Several discharge measurements were madeofthe spring on the Elk River atthedownstream endofthe limestone outcrop, and it was possible to estimate flow for other sampling times from gauge recordsofthe U.S. Geological Survey kept at Web ster Springs. Discharge measurementsatthe EdrayHatch ery were difficult because water abstraction systems are installed at the spring orifices. However, it was possibletoobtain some direct discharge estimates and to relate these to records from a nearby U.S. Geological Survey gauge. Becauseofthe uncertainties inherent in estimatesofthis type, the results are offered here as rough estimates. Instantaneous flux rates for the low-flow measure ments on Back Allegheny Mountain (the northeastern part of the outcrop on FigureI)ranged from a lowof3.4 mm/ka (millimeters over the entire drainage basin per thousand years) to 15.4 mm/ka; in the Edray area (the southwestern partofFigureI),rates ranged from 1.4 mm/ka to 7.3 mm/ka; and for the Elk River basin (the western partofFigure I), the range was from 0.9 mm/ka 195Wernerto 3.0 mm/ka. One high-flow measurement for the down stream endofthe Elk River basin provided a rateof10.2 mm/ka. Estimated annual average carbonate flux rates couldbecalculated for two basins in the study area.TheU.S. Geological Survey has a recording gauge in the Elk River basin not far downstream from the karst area and one in Indian Draft below the karst terrane. Estimatesoftotal runoff are available for these two basins. When a weighted averageofwater analyses from this study and the average runoff are used to calculate carbonate-denudation rates, the results are 15.5 mm/ka for the Elk River basin, and 16.1 mm/ka for Indian Draft. These denudation rates are somewhat low compared to rates reported by other workers. Ogden (1982), working to the southofthis area, reports ratesof19.0 mm/ka to 22.6 mm/ka. The basin areas areofcomparable size, but there is a much more extensive outcrop and generally gentler relief in his area. The spring waters in his basins contain appreciably higher dissolvedcalcium and magnesium con centrations and are more saturated (Ogden, 1976), apparent ly reflecting longer residence timesinthe karst. Compared to other data summarized by White (1988, p. 218), all the rates derivedinthis study are very low. The average runoff reported from U.S. Geological Survey stations for this area is 600 mm/yrto900 mm/yr. Apply ing the regression equationofWhite results in denudation ratesof35.8 mm/ka to 50.5 mm/ka,ortwo to three times those derived from the chemical data. Analysisofthe con ditions prevailing in the study area relative to the "average" karst indicates that residence timeinthe limestoneismuch shorter here, mainly becauseofhigher gradients. How ever, this may not be valid because there seems tobea trend toward higher carbonate flux rates at low flow for those springs in higher gradient basins within the study area. That, combined with a lackofextensive outcrop accessible to water, may explain the low chemical-denuda tion rates for this terrane.Theonly direct measurementoferosionina karst channelinthe Greenbrier karst is that reported by Coward (1975) from acaveabout20kmto the south. He measured approximately 0.75 mmoferosion per yearinan active stream channel.Ifthat rate is comparedtothe overall denudation rate as calculated from the water chemistry, the implication is that approximately 2%ofthe surface is being actively eroded.Ifallofthis were occurringincave channels, then one would expect an average meter-wide cave passage approximately every kilometer (considering allowance for wettingofthe walls and floor).Ofcourse, it is unlikely that all dissolution isinconduits,orintraversable cave passage, so that estimatebecomestheupperlimitofactively-forming-cavefrequency, with the actual frequency considerably lower. It also does not consider formerly active, but now abandoned passages.

PAGE 204

WernerSummaryThe Greenbrier Group karst terraneofnorthern Poca hontas and southern Randolph countiesofWestVirginia has some unique characteristics in regard to its waters. In comparison to most other well studied areas, this karst area is characterizedbyhigh hydraulic gradients thatdonot allow the waters to dissolve as much limestone becauseofthe shorter average-residence time. Therefore, the average chemical-denudation ratesof15mm/kato 16 mm/ka are two to three times lower than average for karst terranesingeneral. There are three basic spring types tobefound in this area; these canbeclassified on the basisofphysical and chemical properties into large conduit, small conduit, and diffuse springs. The diffuse springs correspond well to dif fuse springs elsewhere, and the small conduit springs ap pear to correspondtoconduit springs studied in the karstofPennsylvania. The large conduit springs do not correspondtoanyofthe springs reported by the Pennsylvania work ers, and are probably a consequenceofhigh gradients in well developed conduits that are not common in the Valley and RidgeofPennsylvania. Multiple cycling through a basic karst-aquifer route for mostofthe waters in this terrane creates a more com plex evolutionofthe waters thanina terrane with more distinct carbonate-non-carbonate rock sequences such asinthe Valley and Ridgeorsomeofthe karst in the continen tal interior.ReferencesCardwell, D.H.; Erwin,RB.;andWoodward,H.P.,1968,Geologic MapofWest Virginia:WestVirginia Geological Survey, 2 sheets. Coward, J.H.C., 1975,PaleohydrologyandStreamflow SimulationofThree Karst Basins in Southeastern West Virginia, U.S.A.:Ph.D. dissertation (unpublished), McMaster University, 303 p. Drake, J.L.andHarmon,R.H., 1973, Hydrochemical environmentsofcarbonate terrains:Water Resources Research,v. 9,p.949-957. Eddy, G.E. and Williamson,D.B.,1968, The effectoffaulting inCassellCave,WestVirginia(abstract):National Speleological Society Bulletin,v.30,p.38. Harmon,RS.;Drake, J.1.; Hess, J.W.; Jacobson,RL.;Ford,D.C.;White,W.B.; Fish, J .E.;Coward,J.H.C.; Ewers, R.O.; and Quinlan, J.F., 1973, Geo chemistryofkarst waters in North America:Proceed ingsofthe 6th International CongressofSpeleology, Olomouc,v. 3,p.103-114. Harmon,RS.;Hess, J.W.; Jacobson,RW.;Shuster, 196Appalachian Karst Symposium,1991E.T.; Haygood, C.;andWhite, W.B., 1972, Chemis tryofcarbonate denudation in North America.Transac tionsofthe Cave Research GroupofGreat Britain,v.14,p.96-103. Hamed, H.S. and Owen, B.B., 1958,The Physical Chem istryofElectrolytic Solution:Reinhold, New York. Langmuir, D., 1971, The geochemistryofsome carbonate ground waters in central Pennsylvania:GeochimicaelCosmochimica Acta,v.35, p. 1023-1045. McCue, J.B.; Lucke, J.B.; and Woodward, H.P., 1939, LimestonesofWest Virginia:West Virginia Geological Survey, Volume12,560p.Medville, D.M. and Medville, H.E., 1976, Cavesandkarst hydrology in northern Pocahontas County:West Vir ginia Speleological Survey Bulletin6, 174p.Medville, D.M. and Werner, E., 1977, Karst hydrology and water chemistry in a mixed sedimentary terrain,inTolson, J.S. and Doyle, F.L. (editors),Karst Hydrogeol ogy Memoirsofthe 12th Congressofthe International AssociationofHydrogeologists:The UniversityofAla bama in Huntsville Press, Huntsville, Alabama, p.443457. Ogden, A.E., 1976,The Hydrogeologyofthe Central Monroe County Karst, West Virginia:Ph.D. disserta tion (unpublished), DepartmentofGeology and Geogra phy, West Virginia University, Morgantown, 262 p. Ogden, A.E., 1982, Karst denudation rates for selected spring basins in West Virginia:National Speleological Society Bulletin,v.44, p. 6-10. Plummer, L.N. and Busenberg, E., 1982, The solubilitiesofcalcite, aragonite, and vateriteinC02-H20solutions between 0 and 90 C, and an evaluationofthe aqueous model for the systemCaC03-C02-H20:GeochimicaetCosmochimica Acta,v. 46,p.1011-1040. Shuster, E.T. and White, W.B., 1971, Seasonal fluctua tions in the chemistryoflimestone springs: A possible means for characterizing carbonate aquifers:JournalofHydrology,v.14,p.93-128. Werner, E. 1972, Effectofsmall thrust faults on cave pas sage cross-section:National Speleological Society Bul letin,v.34,p.143-147. Werner, E., 1977, Chloride ion variations in some springsofthe Greenbrier limestone karstofWest Virginia,inDilamarter,RR.and Csallany, S.C. (editors),Hydro logic Problems in Karst Regions:Western Kentucky University, Bowling Green,p.357-363. White, GeomorphologyandHydrologyofKarst Terrains:Oxford University Press, New York, 464p.

PAGE 205

Appalachian Karst Symposium, 1991 Ogden, Hamilton, Eastburn, Brown. and PrideNitrate Levels in the Karst GroundwatersofTennesseeAlbertE.Ogden1 ,Kristie Hamilton1 ,and Edward P. Eastburn1 ,TeresaL.Brown2 ,and ThomasE.Pride,Jr.3lCenter for the Management, Utilization and ProtectionofWater Resources and the DepartmentofEarthSciences Tennessee Technological University Box 5033, Cookeville, 1N 38505 2First Tennessee Development District 207 N. Boone Street, Suite 800 Johnson City, 1N 37604 3Young, Morgan and Associates P. O. Box 208 Abingdon, VA 24210ABSTRACTNumerous wells and springs were sampledinthe Ordovician and Mississippian carbonatesofTennessee to determine nitrate levelsinorder to ascertain the impactofman's activities on groundwater quality. Ten municipal springsineast Tennessee were sampled monthly for one year, and the results showed that nitrate levels remained below 2.5 mg/l (nitrate-nitrogen). Springs recharged from agricultural lands had signifi cantly higher levels. Nitrate levels were also measured for sixteen domestic wells and twenty-four small springs in east Tennessee. The range and average for the wells were 0.02 to 2.64 mg/l and 1.05 mg/l, respectively. For the springs, the range and average were 0.02to5.60 mg/l and 1.95 mg/l, respectively.Fourlarge springsincentral Tennessee were also sampled monthly for one year. The average nitrate levels for the springs recharged primarily by storm-water runoff from the cityofCookeville ranged from 0.69to0.94 mg/l; whereas, a spring recharged by agricultural lands outsideofCookeville had an average nitrate levelof3.09 mg/l. Sixty-four wells and springs around Cookeville were also sampled. The range and average for the wellsinMississippian-age beds were <0.1 to31mg/l and 1.2 mg/l, respectively. The range and average for the wellsinOrdovician-age beds were <0.1 to8.1mg/l and 2.8 mg/l, respectively. Mississippian springs had a rangeof<0.1 to 2.9 mg/l with an averageof1.4 mg/l. Ordovician springs ranged from <0.1 to 3.3 mg/l with an averageof1.0 mg/l. Municipal water-quality data supplied to the Tennessee DivisionofWater Supply was tabulated from twenty-nine wells and springsinthe Cambrian-Ordovician carbonatesofeastern Tennessee for the 1981-85 period. Multiple measurements produced a rangeof0.01 to 3.27 mg/l and an averageof0.60 mg/l. Sixty six measurements from18municipally used wells and springsinthe Mississippian carbonatesofcentral Tennessee showed a rangeof0.01 to 3.67 mg/l and an averageof1.21 mg/l. Twenty-five nitrate measure ments from municipal water supplies in the Central Basin Ordovician carbonate aquifer yielded a rangeof0.01 to 2.81 mg/l and an averageof0.44 mg/l at six sites. The resultsofthis study show that nearly all samples were well below the10mg/l health limit for nitrate, but significantly higher levels occurredinrural areas where there is a greater acreageofagricultural activity. In general, nitrate levels were found to be higherinthe Mississippian carbonates than in the Cambrian-Ordovician carbonates. 197

PAGE 206

Ogden. Hamilton. Eastburn. Brown. and PrideIntroductionIn Tennessee, groundwater is the sourceofdrinking water for51percentofthe 4.76 million residents (Hutson, 1985). Approximately halfofthe state is underlainbyfractured and cavernous carbonate rock that provides little flltration ofcontaminants as they move towards the water table. This is particularly true for nitrate which can cause methemoglobinemia or "blue babies disease" in newborn infants. As a result, the U.S. Environmental Protection Agency (EPA) (1982) has adopted a drinking-water limitof10 mg/l nitrate-nitrogen. Natural background levelsofnitrate-nitrogen are generally less than 3 mg/l. Higher concentrations suggest the influenceofhuman-related sour ces such as crop fertilizers, septic tanks, land disposalofmunicipal and industrial waste, and animal wastes from livestock and poultry. Preliminary data from EPA's Sep tember I, 1989 press advisory showed that about 50 per cent of drinking-water wells tested had nitrate residue. Growing concern about the potential impactofman's activities on nitrate levels in groundwater has promulgated a numberofnational and regional surveys (U.S. EPA, 1978; Zurawski, 1978; Madison and Brunett, 1985; Young, 1986; Adams and others, 1986; Canter, 1987; and Soileau, 1988). Madison and Brunett (1985) found that the seven states within the Tennessee River drainage had lower than national levelsofnitrate with lessthan1to4 percentofthe wells exceeding the10mg/l drinking stan dard and less than17percent exceeding 3 mg/l. Soileau (1988) added EPA STORET data to Madison and Brunett's (1985) USGS W ATSTORE data and found a mean concen trationof1.3mg/lofnitrate-nitrogen and a rangeof<0.01to28.3 mg/l for 587 Tennessee Valley region wells sam pled from 1956 to 1986. Very fewofthese wells were locatedinthe stateofTennessee owingtoan extreme pau cityofdata. Thus, the purposeofthis paper is to present recent nitrate data that has been assimilated by the authors for the karst areasofcentral and eastern Tennesseetodemonstrate the wide variabilityofnitrate concentrations both areally and temporally.Sources of Data, Methodology,andHydrogeologyMostofthe data presentedinthis paper were collected and analyzed by the authors, utilizing the laboratory at the Center for the Management, Utilization and ProtectionofWater ResourcesofTennessee Technological University. Analyses were performed on preserved samples by a Tech nicon Autoanalyzer GTPC (Standard Method 429 [APHA, 1989]). In addition, recent unpublished data from local studies and municipal water-quality data suppliedtothe Tennessee DivisionofWater Supply (1981-85) was incor poratedinthesurvey. The water samples were obtained from wells and springsinthe flat-lying Ordovicianand Mississippian aged carbonatesofcentral Tennessee and the folded198Appalachian Karst Symposium. 1991Cambrian-Ordovician-aged carbonatesofeastern Tennessee (Figure1).Mostofthe research has been conducted inandaround Cookeville and Johnson City. Cookeville occurs on the Eastern Highland Rim Province which is underlainbyflat-lying Mississippian limestones (Figure 2). Streams originating on the Cumberland Plateautothe east flow over shales and sandstones until underlying limestone beds are intersected. Some streams sink when they reach the Bangor Limestone whereas others sink into the lower Monteagle, St. Louis,orWarsaw limestones. Subterra nean water moves through caves, pits, and solutionally enlarged fractures until emerging as spring flow. The lower Warsawissandy, contains shale beds, and acts asanaquiclude. As a result, numerous springs and caves are foundatthe St. Louis/Warsaw contact. The Fort Payne chert-rich limestone forms an areally extensive bench along the western edgeofthe Eastern Highland Rim Pro vince. Discharge from the Fort Payne aquifer occurs along the Highland Rim Escarpmentatthe Chattanooga Shale contact This water moves down the escarpment andthencommonly sinks into cavernous strata within the Leipers Catheys/Bigby-Cannon Ordovician formationsoftheCentral Basin Province. The geologyofthe Valley and Ridge Province of eastern Tennessee is significantly more complex. Surface water movesoffmountains composedofclastics and igneous rock and sinks into strongly folded and faulted Cambrian-Ordovician carbonate rocks comprising thevalley floors. The groundwater initially moves down gradient along bedding planes and fractures and then migrates along the strike (Ogden and others, 1989). Nearly allofthe karst occurs in the Knox Group because it is over 3,000 feet thick and thus has a large outcrop area. Groundwatersamples were also taken from springs in the Shady Dolomite and Copper Ridge Dolomite.ResultsTime-Series AnalysesofLargeSpringsAs partofa wellhead-protection project in eastern Tennessee, nine springs were sampled monthly between July 1989 and June 1990 (Ogden and others, 1990). John son City springs emerge from quartzite, and the recharge area is entirely within National Forest boundaries. The low levelsofnitrate are representativeofa hydrogeologic environment freeofman-made contaminants (Figure3).Hampton Springs emerges from thick river alluvium over lying dolomite. Nitrate levels are also low even though the entire townofHampton has individual home septic tanks. Blue, Big, Rockhouse caves, and Jonesboro springs emerge from limestone and exhibit conduit flow (basedondye tracing and geologic evidence). Gradual increasesinnitrate were observed during the winter months at these four springs. During winter months, plant uptakeofnitrates isata minimum and nitrates are availabletomigrate to the water table. In addition, nitrates canbeaddedtothegroundwater from septic tanks during winter monthswhen

PAGE 207

NashVille Dome Appalachian Karst Symposium. 1991 WulernV,lIey ot Tennusee River MhslulpplRlyerV.lley RimOgden, Hamilton, Eastburn, Brown. and Pride App,llchl,n roldbelt r::::':l 'tlll",.nd CItI.U'ou.""dMI"I"IOpl'...O'do",lcll" '..6 Delton"". 5111.1'1,"."mb,I... dolo""'1t 0,60\'lcll".' .. d C.,nb,I,1I 1i"""toOl' """I,I""nl, .. ,,,,d.lon, ""II.,lpD"".'.u: .... bd... '.,cIPn''', ..I.",.c" """h""""". '"d ""l.morptllcroc' C.",hl," ,hll,Figure1:Relief mapofTennessee showing physiographic provinces and major geologic structures. The region around Cookeville lies on flat-lying carbonate rocksofthe Eastern Highland Rim. The region around Johnson City lies on folded carbonate rocks in the Valley and Ridge Province (from Miller, 1974). the soil isatfield-capacity conditions. Overland flow to sinking streams through fields used by cattle is also great est during the wet winter and spring months.Leeand Hamilton springs emerge from dolostone, andthesuspected recharge areas do not exhibit the degreeofsurficial karst features seen in the limestone terranesofthe area. A significant componentofdiffuse flow is believedtoexist within these dolomitic aquifers. Other measured chemical parameters not presented in this paper showed lower coefficientsofvariation than at the limestone springs. Regardless, conduit versus diffuse flow does not appear to be a primary factor affecting nitrate levels. Land use in the recharge areasismore important. Hamilton, Lee, and Big springs have the highest levelsofnitrate; thisisattributedtothe recharge areas having a significant ly greater percentageofpasture lands than do other spring basins. Nitrate measurements werealsomade for these springs during a May 21, 1990 storm event (Figure 4). Figures 3 and 4 show that in general, nitrate levels vary little withinGEOLOGIC PROFILE OF THE WESTERN SLOPE OF THEVERTICALEXAGGERATION75Xo4000metersICUMBERLAND PLATEAU, PUTNAM COUNTY, TENNESSEEELEVATION2000ItCAPROCKSEQUENCE..: ....(SHALES, SANDSTONES....:::.:'.::CONGLOMERATES)-.':':.::.:Figure2:Geologic profileofthe western slopeofthe Cumberland Plateau, Putnam County, Tennessee. 199

PAGE 208

Ogden. Hamilton. Eastburn. Brown. and Pride Appalachian Karst Symposium. 1991Nitrate-N (mg/l) 2.5 each groundwater basin with little change taking place duetostorm events. As a result, any single measurement for nitrate will provide an estimateofthe yearly range within approximately .5 mg/l. A similar time-series analysisofnitrate levels was performed at four large springs in the Cookeville area (Fig ure 5). Allofthese springs exhibit conduit flow and show significant turbidity duringstorm events. Hidden Hollow, Pigeon Roost, and Big springs all receive muchoftheir re charge from storm-water runoff from paved areas in Cooke ville. In contrast, City Springs is recharged almost exclu sively by farmlands, and this is the likely cause for signifi cantly higher levelsofnitrate. Although City Springs has more nitrate, its overall water quality is much better than at the other springs. The cave water at City Springs sup ports a much greater diversityofspecies, and several blind cave fish have been observed (Pride and others, 1989).Domestic Wells and Small SpringsNitrate levels were also measured for sixteen domestic wells and twenty-four small springs in eastern Tennessee. The range for the wells was 0.02 to 2.64 mg/l, andtheaverage was 1.05 mg/l. For the springs, the range andaverage were 0.02 to 5.60 mg/l and 1.95 mg/l, respectively. Collar (1989) analyzed sixty-four wells and springsfornitrate levels in the Cookeville area.Hefound that Fort Payne (Mississippian) wells had a rangeof<0.1 to31mg/l with an averageof1.2 mg/l; whereas, the Ordovi cian-aged wells had a rangeof<0.1 to8.1mg/l withanaverageof2.8 mg/l. Springsinthe Fort Payne Limestone had a rangeof<0.1 to 2.9mg/1with an averageof1.4mg/l, and the springs in the Ordovician-aged carbonates had nitrate levels that ranged from <0.1to3.3 mg/l with an averageof1.0 mg/l. Collar(l989)found no statistically significant differencesinthese aver ages using the Mann-Whitney U Test. A surveyofnumerous wells and springs in the Cambrian-OrdovicianandMississippian carbonate aquifersofTen nessee has revealed thal nearly all waters are well below the safe drinking limitfornitrate-nitrogenof10mg/1.Table 1Summary and ConclusionsEvery three years the municipalities supplying public water must submit a full analysisoftheir water qualitytotheTennessee DivisionofWater Supply. This information was obtained and tabu lated for the 1981 1985 period and divi ded into groups based on the geographic distributionofprincipal aquifers in Ten nessee according to Bradley and Hollyday (1985). One hundred and four samples from 29 municipal wells and springsinthe Cambrian-Ordovician carbonate aqui ferofeastern Tennessee yielded a rangeinnitrate valuesof0.10 to 3.27 mg/landan averageof0.60 mg/l. Sixty-six sam ples from 18 municipal water suppliesinthe Mississippian carbonate aquifer of central Tennessee produced a rangeinnitrate valuesof0.10 to 3.67 mg/l withanaverageof1.21mg/1.Six municipally used wells and springs in the Ordovician carbonate aquiferofthe Central Basin Physiographic Province yielded a rangeof0.01 to 2.81 mg/l and an average of 0.44 mg/l nitrate-nitrogen from twenly five measurements.Public Drinking-WaterSupplies 2+---+-==--\-------------::::::=:--f-=:;;--------j Figure3:Monthly nitrate measurements at eight eastern Tennessee springs.10.50JulAugSepOctNov DecJanFebMarApr MayJunI89901BlueSpring BigSpring---RockHouseCave JohnsonCitySpring2.5 2 1.5 1 0.50JulAugSepOctNov DecJanFebMarApr MayJunI89901Jonesboro LeeSpring -..HamiltonSpring -+HamptonSprings1.5 f-------\------;.,L--------------->.,:---="''''-----l 200

PAGE 209

Appalachian Karst Symposium. 1991 Ogden. Hamilton. Eastburn. Brown, and Pride Jonesboro -+-LeeSpring -iiHamiltonSpring -bHamptonSpringsFigure4:Nitrate measurements at eight eastern Tennessee springs during a May1990storm event.-----------------Significantly higher levelsofnitrate are found in springs that largely grazed bycatLIe.Fertilizer application to croplands may be a factor, although thistypeofland use is minimal in the study areas. Septic summarizes the data presentedinthis report. In general, groundwaters in the Mississippian carbonates have higher levelsofnitrate than in the Cambrian-Ordovician carbon ates, although a great degreeofvariability occurs among formations within these two age groups. The time-series analysisofthe spring-water data has shown thatlitLIevariabilityofnitrate levels occurs throughout the year, but slight increases are observed during the winter when plantuseofnitrateisat a minimum and recharge is the greatest. A predominanceofconduit versus diffuse flow does not appear to cause a significant differenceinyearly average mtrate levelsorthe amountofchange resulting from smalltomoderate storm events. Adams, V.D.; Wilson, T.M.; Brown, R.T.; Gordon, J.A.; and Mills, H.H., III, 1986, Designofa statewide ground-water monitoring network: Tennessee Water Re-sources Research Center. UniversityofTennessee. Knoxville. Research Report No.13,199p.Acknowledgments ReferencesThis research was fundedinpart by the UniversityofTennessee Water Re sources Research Center (USGS) and the Center for the Management, Utilization and ProtectionofWater ResourcesTen nessee Technological University. Some matching funds were also provided by the First TennesseeDevelopment District. The authors thank the following students for assisting in this project: WalterCrawford,John Mason, and Scott Wheeler. tanks likely contribute some nitratetothe groundwater, but a clear relationship between septic-tank density and nitrate levels was not seen at the springs where the recharge areas had been delineated by dye tracing. Therefore, it is felt that more concern must be placed on agricul tural practicestoinsure groundwater pro tection.Jun20Jun20American Public Health Association, American Water Works Association, and Water Pollution Control Federa tion, 1989, Standard Methods for the ExaminationofWater and Waste Wat er. 14th edition: American Public Health Association Washington, DC. Bradley, M.W. and Hollyday, E.F., 1985, Tennessee ground-water resources, in National water summary 1984; Hydrologic events, selected water-quality trends, and ground water resources: U.S. Geological Survey Water-Supply Paper2275,p. 391-396. Canter, L.W., 198'7, Nitrates and pesticidesinground water:Ananalysisofa computer-based literature search,inGround Water Quality and Agricultural Practices: Lewis Publishers, Chelsea, Michigan,p.153-174. Collar, P.D., 1989, Radon in homes. soils. and caves. and water qualityofwells and springs in north central Ten nessee: M.S. thesis (unpublished), Tennessee Techno logical University, 342p.May23May23 -+-BigSpring -eJohnsonCitySpringMay21May21May20 ------A oApr202Nitrate-N (mgll) 2BlueSpring -iiRockHouseCave0.51.5oApr20May200.51.5201

PAGE 210

Ogden, Hamilton, Eastburn, Brown, and Pride Appalachian Karst Symposium, 1991Miller,R.A.,1974,TheGeologic His toryofTennessee:Tennessee DivisionofGeology Bulletin74, Nashville,63p.Ogden,A.E.;Curry,W.;andCum mings, J., 1989, Morphometric analy sisofsinkholesandcaves in Tennessee comparing theEasternHighlandRimandthe ValleyandRidgephysiogra phic provinces,inBeck, B.F. (editor),EngineeringandEnvironmental Im pactsofSinkholesandKarst: ProceedingsoftheThirdMultidisciplinary ConferenceonSinkholes and the Engi neeringandEnvironmental ImpactsofKarst, St. Petersburg Beach, Florida,2-4October1989:A.A. Balkema, Rotterdam and Brookfield, p. 135-142.Ogden,A.E.;Hamilton,K.G.; andBrown,T.L.,1990,Delineationofwellhead protection areas for munici pal-used springsofeastern Tennessee:Tennessee Water Resources Research Center, UniversityofTennessee, Knoxville, Report124,134p.Pride, T.E.; Ogden, A.E.;andHarvey, M.J., 1989, Biol ogyandwater qualityofcaves receiving urban runoff in Cookeville, Tennessee, USA:Proceedingsofthe 10th International CongressofSpeleology, Budapest, Hun gary,p.27-29.FebMar89 -BPigeonRoostJanDecNov -r HiddenHollowOctAugSep88 -+BigSpringJulJun-CitySpringNitrate-N(mgll)4.5,------------------------4f--------::;;-....,:::::"""--=::----------------I 3.5 f---------------1-----==---.-----1'<:::""----.j 3 r-----------\-----/---\c----I---I 2.5 f--------------\----/-----\ 2t-------------\1.5f------------k-lHutson,S.S,1985,Ground-waterusebypublic-supply systems in Tennesseein1985:U.S. Geological Survey Water Resources Investigations Report89-4092,1plate. Madison, R.J.andBrunett J.O., 1985, Overviewofthe oc currenceofnitrate in ground waterofthe United States,inNationalwatersummary1984; Hydrologic events, selected water-quality trends,andground water resources:U.S. Geological Survey Water SupplyPaper2275,p.93-105. Figure5:Monthly nitrate measurementsatfour central Tennessee springs near Cookeville. o MayIMississiooian Carbonates Ordovician Carbonates Data Source Wells Sorin s Wells Sorin s Ranl!e Averal!eRanl!"eAveral!e Ranl!eAveral!"eRanl!e Averal!e Wellhead Protection Project 0.02 2.64 1.05 0.45 -2.101.21 (Ol!den et al., 1990)SepticTankSurvey
PAGE 211

Appalachian Karst Symposium. 1991Soileau, J.M., 1988,Anoverviewofground-water nitrate and fertilizer nitrogen consumption in the Tennessee Valley region:Tennessee Valley Authority. Officeof Agriculture and Che,mical Development Bulletin 204:Muscle Shoals, Alabama, 29p.Tennessee DivisionofWater Supply, 1981-1985,Water Quality RecordsforMunicipal Water Supplies. Nash ville. Tennessee.u.S.Environmental Protection Agency, 1978,National Water Quality Inventory:1977ReporttoCongress: EPA 440/2/84-006,Washington, DC.U.S.Environmental Protection Agency, 1982, Maximum contaminant levels (Subpart BofPart 141, National Interim Primary Drinking Water Regulations):U.S. CodeofFederal Regulations. Title 40, Parts 100to149. Revised July1.1982,p. 315-318. 203Ogden. Hamilton, Eastburn. Brown. and Prideu.S.EPA Press Advisory, 1989,EPA Releases Interim ResultsofNationalSurveyofPesticides in Drinking Water Wells. Washington, DC.Wilson, T.M.; Reinhard, J.D.; Gordon, J.A.; Ogden, A.E., 1989, A studyofseptictanksystemsinSullivan County, Tennessee:Report to the Tennessee Valley Authority. Chattanooga, Tennessee,90p.Young, H.C., 1986, Distribution mapsofTennessee Val ley region:National Fertilizer Programs. Agricultural Programs. MajorLandUses. Livestock Inventions. Farm Products and Sales: Circular Z-211 TVA/OACD,Muscle Shoals, Alabama, 46p.Zurawski, A., 1978, Summaryappraisalsofthe nation's ground-water resources-Tennessee region, includingpartsofTennessee and adjacent states:U.S.Geologi-cal Survey Professional Paper 813-L,p.Ll-L35.

PAGE 212

Plate Appalachian Karst Symposium. 1991 PlateK:A sinkhole in rural PulaskiCounty, Virginia, filled with farm and household trash.Ithas been estimated that thousandsofthese sinkhole dumps exist within the groundwater recharge zonesofthe Appalachian karstofVirginia andWestVirginia. (See papers by Erchul, p. 147, Kastning and Kastning, p. 123, and Hubbard, p. 135, in this volume). Photograph by ErnstH.Kastning.204

PAGE 213

Appalachian Karst Symposium, 1991 Brown and EwersImpactsofBarnyard Wastes on Groundwater Nitrate-N Concentrations in a Maturely Karsted Carbonate AquiferofSouth-Central KentuckyCraigJ.Brown1and Ralph O. Ewers2IGeraghty&Miller, Inc. 180 Admiral Cochrane Drive Annapolis, MD 21401 2DepartrnentofGeology Eastern Kentucky University Richmond, KY 40475ABSTRACTIn maturely karsted areas such as the Sinking Valley groundwater basinofsouth-central Kentucky, where conduit developmentisextensive and soil cover is thin, human and animal wastes can pose a serious threat to the local groundwater system. The water chemistryofdomestic wells, other groundwater-access points, and surface streams throughout the basin were monitored between Juneof1987 and Marchof1988. In general, nitrate-N concentrations were less than 5 mg/l. Nitrate-N levels in one domestic well (Harris Well), however, exceeded the drinking-water standard. The suspected nitrogen source for the elevated nitrate N levels appears to be a barnyard within 200 feetofthe well; septic discharge from the house occurred down slope from the well and was also a potential factor. Nitrogen sources contributing to groundwater-nitrate occurrences throughout mostofthe basin consis ted primarilyofN-fertilizers, animal wastes and decomposed plant matter. Nitrates behaved as non-point source contaminants and were leached through the soil and epikarst and into the groundwater system during recharge events and/or under field-saturated conditions,bydiffuse autogenic recharge and diffuse flow. Following a rain event, nitrate-N levels at most groundwater-sampling locations would typically decrease initially, in response to dilution by concentrated autogenic recharge. Following this initial decrease, nitrate N levels would typically increase above base-flow concentrations as the diffuse autogenic recharge compo nent became a larger contributing factor toward groundwater flow. Sourcesofnitrate and other contami nants that were introduced at discrete locationsinthe limestone resulted in deleterious effects on local groundwater quality. Such was the caseofHarris Well, where a nearby barnyard is located over thinly soiled limestone. GroundwaterinHarris Well showed an inverse responseofnitrate-N concentrations, as well as specific conductance and hardness, to rainfall events. Following the event, these levels reboundedtoor near base-flow levels, indicating the existenceofa nitrogen source that was continually leached into the groundwater and well.IntroductionThe outcropofMississippian limestones along the Cumberland escarpmentineastern south-central Kentucky (Figure 1) is a unique and, thus far, relatively unstudied karst region. The areaischaracterizedbya thick sequence of soluble limestones, thin soil cover, moderate relief, andlowstructural control. The Sinking Valley groundwater basin within this regioniswholly agricullural and occupiesapproximately 52 square miles. The groundwater systemexhibits conduit flow and a focused pointofground205 water discharge, thus providing an excellent sellinginwhich to determine the behaviorofnitrates,toestablish the major sources contributing to groundwater nitrate, and to determine the effectsofgroundwater-recharge eventsonnitrates. Harris Well was among eleven sampling locations throughout the basin that were sampled at least biweekly between June 13, 1987 and February 16, 1988 (Figure 2). These locations included other domestic wells, springs, caves, and surface streams. In additiontoconcentrationsof

PAGE 214

Brown and Ewers Appalachian Karst Symposium, 1991 Similarly, Gerhart (1986) found that nitrate-N concen trations in a shaIIow, fractured dolomite aquifer are influenced by two different recharge mechanisms: direct recharge through fractures and sinkholes that affects nitrate concentrationsofgroundwater for twotothree days;andgradual recharge that occurs through small channelsandmicropores in the unsaturated zone and affects nitrate levels for several weeks or more. ent carbonate aquifers by proposing two end-membermodels, diffuse flow and conduit (or free) flow. Although greatly oversimplified, these end-member models canbecompared with discharge from actual carbonate aquifers. Gunn (1985) provided a conceptual model for conduit-flow dominated karst aquifers. The model examines the three main system components: inputs, stores, and transfer mechanisms. Depending upon the natureofthe karst aquifer,thestorage and movementofnitrate will behave in different ways. In dye-tracer studies in the Mendip Hills, England, Friederich and Smart (1981) found that dye spread laterally through the subcutaneous zone at flow rates on the order of 100 meters per day. However, dye was still presenttenmonths after its artificial injection into the epikarst, indi cating significant storage in this zone.SHOPVILLE MARETBURG 2040, I I III III I I I ,, II I /) : <-V;I I III / I ,, I I I , KENTUCKY 50 100 SCALE(11IInl Figure1:Locationofthe study area in three quadrangles in Pulaski and Rockcastle counties, Kentucky. nitrate-N and nitrite-N, monitored water-chemistry parame ters included calcium and magnesium hardness, alkalinity, specific conductance, pH, chlorides, and coliform bacteria. Calculated parameters PC02 and SIc were also determined at each sample location in an effort to characterize the groundwater recharge and flow components (Drake and Harmon, 1973). Environmentsofsampling locations in the study area displayed differences in certain water chemistry parameters, thus allowing the dominant recharge and flow mechanismtobe identified. A networkofrain gauges was distributed over the basin to monitor precipitation (Figure 2), while ground water discharge was monitored during muchofthe study at a karst window near the groundwater-basin discharge point. The periodofstudy was unusuaIly dry with respect to rainfall and low-flow conditions occurred between July and December.BackgroundGroundwater flowinmature karst has been the focusofseveral investigations in the attempt to characterize re charge, storage, and flow mechanismsofparticular aquifer systems. These properties can, in tum, be appliedtocon taminant (nitrate) behavior. White(1969)distinguished the flow behaviorofdifferAn important factor influencing groundwater nitrate-N levels is,ofcourse, the contributing nitrogen source.Thecomplexityofnitrate within the soil-groundwater system makes its behavior difficult to understand. There are sever al nitrogen sources potentially contributing to nitrate-N levels in the Sinking Valley groundwater basin, including animal and human wastes, fertilizers, and decayed plant matter. The increasing confmementofanimals in largenumbers for meat, milk, and egg production has caused large quantitiesofanimal waste to accumulate in small areas (Wadleigh, 1968), creating the potential to contaminate underlying groundwater. Aldwell and others(1983)reported that waste from farmyards and other areasoflive stock concentration is the most common and widespread sourceofgroundwater pollution in Ireland. Brown(1990)found that diffuse autogenic recharge had the greatest over all effect upon nitrate-N levels in a conduit-dominated karst aquifer in south-central Kentucky. Localized groundwater conditions were greatly affected by concentrated nitrogen sources (barnyards), however, and exhibited excessivelevelsofnitrate-N and coliform bacteria. Studies by the U.S. Environmental Protection Agency show that 29%ofthe U.S. population disposesofdomes tic waste through individual on-site disposal units,andinvestigators estimate that manyofthese septic tank-soil absorption systems are not operating satisfactorily (Scalf and others, 1977).Inaddition to nitrates, other products of human waste associated with septic drainage include bac teria, suspended solids, biological oxygen demand (BOD),206

PAGE 215

Appalachian Karst Symposium, 1991 Brown and EwersHydrogeologicSettingUsing terminology proposed byWhite(1969),theSinkingValley groundwater system can be classified as a conduit-dominated carbonate aquifer This categorization is supported by the presenceofnumerous sinkholes,dryval leys, sinking streams, extensive solu tional development, and characteristically high groundwater flow rates throughouttheSinking Valley region. A thick sequenceofMississippian limestones within the Newman Forma tion makes up mostofthe Sinking Val ley aquifer system. The Ste. Genevieve and upper member limestones have un dergone significant dissolution and pro vide the pathway for most groundwater flow. The underlying St. Louis Lime stone contains abundant chert in the upper horizon that inhibits dissolution. Overlying these limestone units are mud stones, sandstones, and limestonesofthe Pennington Formation, forming the up per partofknobs and ridges, and conglo merates and sandstonesoftheLeeForma tion (pennsylvanian) that cap the hills. following harvest. Soil conditions in autumn promote rapid nitrificationoforganic nitrogenand, following harvest, can leadtoconsiderable leachingofni trate ions (Addiscott, 1986). Previous dye-tracing studies have demonstrated that groundwater flowinthe basin follows a dendritic pattern, con verging toward a central conduit that dis charges through a karst spring into Buck Creek, a surface stream to the west (Romanik, 1986). The conduit network transports groundwater down dip in a southerly direction before making a shift in the southern partofthe basin to follow strike toward the southwest. Short Creek, a karst window, was a major monitoring point immediately upstream from the ground water-basin discharge spring. A surveyoflanduse, agricultural practices, and septic disposal was performedtodetermine the major sourcesofnitrates in the Sinking Valley. The survey data indicate that although the basin was not farmed intensively, nearly 20%ofthe area was used to grow crops including com, tobacco, hay, and alfalfa, whereas over 35% was dedicatedtopasture. Livestock are raised in pastures and barnyards throughout the basin; such locations represent a concen trated areainwhich nitrogenisintroduced into the ground.EXPLANATIONSampleSite RainGauge 80Anadditional, but somewhat less influential sourceofnitratesingroundwater results from the leachingofnitrates viruses, ammonia,chlorides, phosphates, and sodium, aswellas waterborne diseases such as infectious hepatitis. Figure2:Locationofsampling sites and rain gaugesinthe Sinking Valley Groundwater Basin. Applicationofnitrogen fertilizer to agricultural areashasincreased tremendously since the endofWorld War II,andfertilizers are generally considered tobethe primary Sourceofnitrate concentration (Madisonand Brunett,1985).Halberg (1986) ascertained that a storeofnitrogenfromfertilizer exists in the karst soil mantleinIowa. Leachingofnitrate-N into the groundwater system wasfoundtobedirectly related to recharge eventsinthe spring. 207

PAGE 216

Brown and Ewers Appalachian Karst Symposium.1991location where nitrate-N concentrations exceeded thedrinking-water standardof10mg/l, and remained above thelimit for mostofthe study (Figure 5). Nitrate-N concentra tionsinthis well behaved differently than other samplelocations in that nitrate-N, along with hardness and specific conductance (Figure 6), demonstrated a negative correlation with rainfall events and obtained highest concentrationsbetween rain events. Nitrate-N in groundwater at all other locations increased only in response to a recharge-produc ing rain event. A comparisonofgroundwater parametersatHarris Well and Short Creek during low and highflowis provided in Table1.Parameters in both sample loca tions decreased following the increased groundwaterflowcaused by recharge events between October 4, 1987andJanuary 20, 1988; decreased nitrate-N levels inShort Creek over this period reflect the flushing and dilution of2 '"<> 03 a.<> 4 0 E 11 5 'c:::Jc: 6 <> '3r:r 7lU z8 9200240 260 320 300 400Julian Days2430.5 3.5 <>Z 1.5z 2.5 .s. 0 0.9 0.8 2 '" E Ql 00.70 "c '" 03 Co E0.6 a.E(i).2-0 4EC0.5 Eco-0'" C :J'ccoO 5 8E. 0.4 :::Jc:()Qlco 6 "-'3 0> r:r E7lU0.2 z0.18 09160200240260320 300 400Julian DaysHarris Well Nitrate-N Levels in the Groundwater SystemNitrate-N levelsingroundwater at0 -f----,....--.,---r--r--r--r----r---.--.,--.,..='--r----tmost sampling locations, including basin160discharge monitored at Short Creek, exhibited a strong relationship with Figure3:Nitrate-N concentrations and precipitation vs. time at Short Creek. recharge-producing rain events. Early oninthe courseofa rain event, groundwater chemistry was largely representativeofconduit flow,orquickflow, supplied dominantly by concentrated autogenic re charge. Low levelsofnitrate-N, calcium hardness, and specific conductance were detectedingroundwater at this time (Fig ure 4). As diffuse recharge (infiltration) and diffuse flow began toplayalarger role in groundwater flow, nitrates storedinthe soil and epikarst were leached into the groundwater system, resultinginan increase in nitrate-N levels. This in crease usually peaked after the rain event and remained above base-flow levels for several days. Septic-disposal practices in the study area were found to rangefTomseptic tank leach field systems to direct dis chargeofraw sewage onto the ground surface. Nitrate-N levelsinthe Sinking Valley groundwater system demonstrated a seasonal relationship. Groundwater nitrate-N levels in Short Creek (Figure 3), representativeofconduit flow, were found to increase in the spring months, apparently resulting from the applicationoffer tilizers. During summer and early fall months and base flow conditions, nitrate-N concentrations were relatively stable, as low rainfall and the effectsofevapotranspiration prevented the leachingofnitrates through the soil and epi karst and into the groundwater system. At the onsetofsignificant groundwater recharge and the cessationofevapo transpiration,groundwaternitrate-N levels throughout thebasin increased sharply as nitrates in the soil and epikarst were carried by recharge (infiltration) into the groundwater system. Following the major periodofgroundwater rechargeinthe fall and early winter, nitrate-N concentrations in groundwater declined, probably due to the depletionofnitratesinthe soil and epikarst, coupled with dilution by increased groundwater flow. Thus, seasonal trends were observedingroundwater nitrate levels. Harris Well was the only sampling Figure4:Calcium hardness, specific conductance, and precipitation vs. timeatShort Creek. 208

PAGE 217

Appalachian Karst Symposium, 1991 Brown and Ewersnitrates stored in the soil and epikarst.Inaddition to ni trate-N, other parameters in Harris Well were anomolously high, including specific conductance (737 chlorides (52.5 mg/l), and fecal coliform (400 colonies/lOOmI).Basedonthestrikinglydifferentrelationshipofnitrate-N with rainfall from therestofthe groundwater system, the significantly higher levels, and the associated parameters indicativeofanimalorhuman waste, it is appa rent that the groundwater nitrate originated from a major source other than nitrogen fertilizers, as was attributed to other sample locations. GroundwaterinHarris Well appeared to reflect con-taminationbylocalizedanimalandpossiblyhuman wastes. The probable nitrogen source was a cattle barnyard approximately 180 feetsoutheastand uphill from the uncased well. Another, less likely source exists about 50 feet westofthe well, where sewage was discharged from thebackofthe house onto the ground. Contaminants from this source are downgradient from the well and would have a more tortuous path to travel. From the nitrate-N and precipitation vs. time relation ship it is evident that nitrate was introduced to and stored within the soil and epikarst and continuously leached into the groundwater and Harris Well through diffuse and con duit (mixed) flow. Following a rain event and the resul tant flushing and dilution effects causedbyconcentrated recharge and conduit flow, nitrate-N con centrations in the groundwater decreased. Figure5:Nitrate-N concentrations and precipitation vs. time at Harris Well. Figure 6: Calcium hardness, specific conductance, and precipitation vs. time at Harris Well.ConclusionsThinly mantled, conduit-dominatedkarstaquifers in eastern south-central Kentucky such as the Sinking Valley ba sin are extremely vulnerable to contami nants from the surface. Although it was determined that nitrogen fertilizers werenotapplied in quantities sufficient to generateexcessivenitrate-N levels in groundwater, animal husbandry and inef fective septic systems can resultinhigh levelsofnitrates, as well as bacteria and other associated contaminants, in localizedgroundwater environments. Inmostsampling locationsofthe Sinking Valley groundwater system, ele vated groundwater nitrate-N levels (above 1 mg/l) appeared to result mostly from the applicationoffertilizers in the springandsubsequentleachingby rainfall events. Increased nitrate-N concentra tions exhibited strong correlations with recharge events following the growing season. In Harris Well, however, a nitrogen sourceofa different nature was apparent, as indicated by much higher concentra tions, different behavior, and associated occurrence with elevated levelsofchlor ides and coliform bacteria. High nitrate N levels occurred in Harris Well between rain events,butexhibited sharp reduc tions immediately following rechargeevents,apparentlyassociatedwithflushing and dilutionofgroundwater affected by barnyard wastes and/or human wastes.982 U>.,.t::U 3 ,..t::Q. 4 E (; 5 :g=>C 6 > ':; 7 0; za400360 320280 240200Julian Days2120 19 18 17 16 15 14 13 "Ol 12 .s "z10 ., 9 8 Z 7654321 160 0.9E0.8 2 U>u.,';;-.t::u 0 :..c 3E0.7 .t::Q. Ec0.6 4E ..'0'" C :l.m 0 .c 0.5 5 C8t:.=>(.)Cm.,(.) 0.4 6 'iii'> Ol:; E CTUJ 0.3 7 0; z0.2 8 0.1 9 160200240 280320360400Julian Days209

PAGE 218

Brown and Ewers Appalachian Karst Symposium,1991Table1:Groundwater chemistry parametersatHarris Well and Short Creek during lowand high-flow conditions. Nitrate-N Total Hardness Chloride Specific Fecal ColifonnDateConductance (mg/l) (mg/l)(mg/l) (unhoslcm) (per 100ml)HARRIS WELL: October 4, 1987 16.26 324 400* Low flow 52.5 737 High flow January20,19888.33131-584-SHORT CREEK: October 4, 1987 1.67 178 4.25 387 20* Low flow High flow January 20, 1988 0.5531-140-*Measured December 2,1987.ReferencesAddiscott, T., 1986, Farmers, fertilizers and the nitrate flood:New Scientist,v. 8, p. 50-54. Aldwell, C.R.; Burdon,OJ.;and Sherwood, M., 1983,Impactofagricultureongroundwaterin Ireland:Environmental Geology,v. 5, p. 39-48. Brown, C.J., 1990,The EffectsofGround-Water Recharge on Nitrate-NConcentrations in a Conduit-Dominated Carbonate Aquifer, the Sinking Valley Ground-Water Basin, South-Central Kentucky:M.S. thesis (unpub lished), Eastern Kentucky University, Richmond, 145p.Drake, J.J. and Harmon, R.S., 1973, Hydrochemical environmentsofcarbonate terrains:Water Resources Research,v. 9, p. 949-957. Friederich, H. and Smart,PL,1981, Dye tracer studiesofthe unsaturated-zone rechargeofthe carboniferous lime stone aquiferofthe Mendip Hills, EnglandinBeck, B.F. (editor),Proceedingsofthe Eighth International Con gressofSpeleology, Bowling Green, Kentucky, July18-24, 1981: National Speleological Society, Hunts ville, Alabama, v. 1, p. 283-286. Gerhart, J.M., 1986, Ground-water recharge and its effects on nitrate concentration beneath a manured field site in Pennsylvania:Ground Water,v.24, no. 4, p. 483-489. 210 Gunn, J., 1985, A conceptual model for conduitflowdominated karst,preprint fromInternational Symposium on Karst Water Resources, Ankara(inpress), p. 1-10. Halberg, G.R., 1986, Overviewofagricultural chemicals in groundwater:Proceedingsofthe AgriculturalImpactson Ground Water Conference. Omaha, Nebraska,August1986: National Water Well Association, p. 1-63. Madison, R.J. and Brunett, J.O., 1985, Overview oftheoccurrenceofnitrate in ground waterofthe United States,inNational water summary 1984; Hydrologic events, selected water-quality trends, and ground-water resources:U.S. Geological Survey Water SupplyPaper2275, p. 93-105. Romanik, P.B., 1986,Delineationofa Karst Groundwater Basin in Sinking Valley, Pulaski County, Kentucky:M.S. thesis (unpublished), Eastern Kentucky Univer sity, Richmond, 93 p. Scalf, M.R.; Dunlap,WJ.;and Kreissl, J.F.,1977,Environmental EffectsofSeptic Tank Systems:EPA60013-77-096,31 p. Wadleigh, C.H., 1968, Wastes in relation to agriculture and forestry:U.S. DepartmentofAgricultureMiscellaneous Publication 1065,23 p. White, W.B., 1969, Conceptual models for carbonateaquifers:Ground Water,v. 7,p.15-21.

PAGE 219

Appalachian Karst Symposium. 1991 Ogden. Redman. and BrownPreliminary Assessmentofthe ImpactofClass V Injection Wells on Karst GroundwatersAlbertE.Ogden1 ,RonaldK.Redman2 ,and TeresaL.Brown3 1 Center for the Management, Utilization and Protection ofWater Resources Tennessee Technological University Cookeville, 'IN 38505 2DepartmentofBiology Tennessee Technological University Cookeville, 'IN 38505 3First Tennessee Development District 207 N. Boone Street, Suite 800 Johnson City, 'IN 37604ABSTRACTTheU. S. Environmental Protection Agency has contracted the authors to investigate the impactofClass V injection wells on groundwater in karst terranes to determine the need for regulationsoverthese shallow injection methods. Any sinkhole that has been modifiedtobetter accept drainage, including storrn water runoff, is considered a Class V injection well. In addition, service-station-bay drains that lead to septic tanks, pits,ordrywells are a focusofthe study.Thestudy is being conducted in Cookeville and in Johnson City, Tennessee to compare contaminant transport in both flat-lying and folded rock. To date,over75 service stations have been surveyed astobay drain-disposal methods. Sampling sites have been chosen to document potential impacts. Groundwater flow paths and velocities have been determined using dye-tracing methods. Majorcavesystems underlie both Cookeville and Johnson City. Both cities have modified sinkholestobetter accept storm-water runoff. Initial samplingofsome springs has shown a significant degradationofwater quality and an impact on the diversity and densityofbenthic macro-invertebrates. An additional objective ofthe study istoreview the Class V injection-well regulationsofeastern states containing karst and to formulate a model regulation to protect karst groundwater. This review has shown that most states do not provide for specific protectionofvulnerable karst waters. In fact, several states con taining large karst regions have no Class V injection program at all.IntroductionPart Cofthe Safe Drinking WaterAct(PublicLaw93-523) authorizes the U.S. Environmental Protection Agency(EPA) to establish regulationstoassure that potable groundwater is not endangeredbyunderground injectionofwaste.Guidelines for underground injection and the classi ficationofinjection wellscomeunder Part 146.04ofthe FederalUndergroundInjection Control(UIe)program(U.S.EPA, 1981). This program created five classesofunderground injection wells. Classes I through IV includeSuchcategories as radioactive and hazardous waste injection anddisposalofbrines from the oil and gas industry. 211 Class V wells are generally definedasthose which inject only non-hazardous fluids intoorabove strata that contain an underground sourceofdrinking water (USDW). USDWs not only include aquifers that are currently serving as drinking-water sources but also aquifers which areofacceptable quality for possible future use. Class V wells include any typeofinjection well not covered in the UIC definitionofClasses I, II, IIIorIV.EPAhas classified Class V injection wells into six groups based in part on the expected qualityofthe injected fluid. The foHowing is a listingofthese groups as found in regulations for the StateofFlorida which is particularly sensitive to karst aquifers:

PAGE 220

Ogden. Redman, and Brown Group1wells associated with thermal energy ex change processes, that include air-conditioning return flow wells and cooling-water return-flow wells. Cooling-water return-flow wells maybepartofa closed-loop system, with no hazardous additives.Group2recharge wells, saltwater-intrusion barrier wells, connector wells, and subsidence-control wells (associatedwith aquifer overpumping).Group3-wells that are partofdomestic waste-treat ment systems, swimming-pool drainage wells,injec tion wells used in experimental technologies, wells used to inject spent brine into the same formation from which it was withdrawn after extractionofhalo gens or their salts.Group4 non-hazardous industrial and commercialdisposal wells, that include laundry waste, dry wells, sand-backfill wells, and nuclear disposal wells used to inject radioactive wastes (provided the concentrationsofthe waste do not exceed drinking water standards contained in Chapter 17-22, FAC) and injection wells used forin siturecoveryofphosphate, urani ferous sandstone, clay, sand, and other minerals ex tracted by the borehole-slurry mining method (lig nite,tarsands, oil shale, coal).Group5lake-level drainage and storm-water drainage wells.Group6geothermal wells and "other" wells. The U.S. EPA has funded twenty-four projects nation wide underitsShallow Injection Well Initiatives Programtohelp evaluate the impactofClass V injection wells on groundwater and to establish best management practices. The authorsofthis paper were chosentohelp evaluate Class V injection-well practices in the karstofTennessee. Any sinkhole that has been modified to accept waste, in cluding storm-water runoff, is considered a Class V injec tion well.ObjectivesThe primary goalofthe proposed study is to deter mineifClass V injection wells have created a groundwater pollution problem. Evaluating the effectsofservice station-bay drains (Group 6) that lead to septic tanks, pits, or dry wells are a primary targetoftheinvestigation. Other high-priority Class V injection wells that are being investigated include: I) Agricultural and municipal drainage into improved and unimproved sinkholes (Group 1).2)Industrial drainage into sinkholes (Group 4).3)Domestic wastewater drainage into sinkholes (Group 5). 212Appalachian Karst Symposium. 1991Additional objectivesofthe researchare:I)Review case historiesofcontamination from Class V injection wells in karst through a literature review.2)Evaluate the effectiveness and applicabilityofUIC Class V injection-well regulations in states having karstic limestone terranes and provide suggestionstomodify regu lations for better protectionofhuman health and the envi ronment.3)Sample selected runoff waters entering sinkholes and analyze chemical constituents in nearby wells and springs to determine contaminant levels. Analyses for benzene-toluene-ethylene-xylene(BTEX),methyl tertiary butyl ether (MTBE), total petroleum hydrocarbon (TPHC), ethylene glycol., zinc, and lead will help evaluate the im pact from service-station-bay drains and urban storm-water runoff. 4) Perform a benthic macro-invertebrate study at sites being sampled chemically to document effectsofcontami nants on biota in streams. 5) Demonstrate the applicabilityofdye tracingtoshow the connection between selected sourcesofpollution and contaminated wells and springs.HydrogeologyoftheStudySitesTwo geologically different karst terranes in Tennessee were chosen so that the results would have applicability throughout muchofthe karst in the eastern United States(seeFigure1ofOgden and others, this volume, p.199).Around Cookeville, the Mississippian-aged carbonates of the Eastern Highland Rim Province are flat-lying, allow ing groundwater to move along a wide rangeoforienta tions corresponding to joint and photo-lineament trends (Ogden and others,1989).In the Valley and Ridge Pro vince around Johnson City, the Ordovician-aged carbonates are complexly folded and faulted, and groundwater moves predominantly along stratigraphic strike within solutional ly enlarged bedding planes. Mostofthe groundwaterflowpathsinand around Cookeville have been documentedbyFaulkerson and others(1981)and Hannah and others,(1989).These traces have delineated the boundaries of three spring-water basins that receive storm runofffromClass V injection wells. No groundwater tracing had been conducted in Johnson City until initiationofthis project. Field investigationstodate show that muchofthe storm drainageinnorthern Johnson City flows through a cave system and emerges at a spring along Knob Creek.MethodsSamples were collected where water enters Class V injection wells and at springs influenced by these waters. As a control, samples were also gathered from a sinking stream and a spring with a predominantly forested recharge

PAGE 221

Appalachian Karst Symposium, 1991area. Sampleswereanalyzed for constituents expected from service stations as well as from agricultural practices. Dissolved oxygen(00),pH, conductivity, and temperature were measured in the field with a Hydrolab field monitor. Laboratory analyses used as indicatorsofcontamination fromagriculturalactivities and septic tanks included nitrate, chloride, fecal colifonn, and fecal streptococcus bacteria. Zinc, chromium, lead, BTEX, MTBE (a gasoline additive), TPHC, and ethylene glycol levels were measured as indicatorsofwaste products and leaks from service sta tions and parking-lot runoff.Toenhance the interpretationofthe impactofClass V injection wellsonspringwaters,samplesofbenthic macro-invertebrates were collectedatriffles and poolsatboth the injection points and the springs. Samples were collected with a modified kick net and a Surber sampler (0.09 m 2 ). All organisms were preserved in a 10% for malin solution, enumerated,andidentified to the lowest taxonomic level possible. Groundwater tracing has been conducted using fluores cein and rhodamine dyes and activated charcoal detectors. Optical brighteners and cotton detectors have also been used for tracing. A control packet was placedata spring knownnotto be hydrologically connected to the tracer in put site during each test.Results to Date Service-Station-BayDrainSurveyApproximately75sites have been visited in the John son City and Cookeville areas to detennine service-station bay drain disposal practices. This survey yielded the fol lowing results:1)Mostservice stationsclaimto have their bay drain connectedtothe city sewer.2)Others have no idea where the drain leads to.3)Only a few say that the drain leads to a septic systemordrain well. These results have made it very difficult to document if degradationingroundwater quality is occurring. Person nel [rom regulatory agencies with the authority to perfonn detailed on-site inspections willbeneeded todetennineactual disposal practices. It is anticipated that this survey willbeexpanded to the Jefferson City Morristown areas where significant karst occurs within the city limits.GroundwaterTracingFourgroundwater traces from Class V injection wellsinJohnsonCityhave been conducted to date.Theseresults have documented the groundwater flow paths to a spring along Knob Creck that drains muchofthe northern 213Ogden, Redman, and Brownpartoftown. Six traces were conducted in the Cookeville area. Groundwater travel times are very rapid and areinthe orderofseveral thousand feet per day. Groundwater tracing efforts will continue within these drainage basins.Water-Quality SamplingThehigh costsofanalyzing for organic compounds re quires detail scrutinyofmonitoring locations before sam pling begins.Atthispointin the research, fifteen sites have been sampled around Cookeville and Johnson City. Field parameters such as pH, conductivity, and dissolved oxygen indicate degraded wateratsprings drained by urban runoff.Thesamplesarepresentlyatthe lab awaiting analysis for metals and organics. More sites will be sam pled during the present wet season and the data comparedtosubsequent sampling during the dry season.BenthicMacro-invertebrateSurveyMany springs and sinking streams willbesampled for the density and diversityofbenthic macro-invertebrates be fore the research is concluded. These sites include springs that drain from Class V injection wells within the city limits, rural springs that drain from forested country-side withsomeagricultural activity, and waters entering sink holes that have been modified to accept drainage. In springs polluted by Class V injection wells in urban settings, the numberofbenthic macro-invertebrate taxonomic groups was found tobealmosthalfthatofsprings not affected by injection wells. In the springs polluted by injection wells, only 4 to 5 taxonomic groupsofbenthic macro-invertebrates were found, whereas,inspringswhereinjection wells arenotin the drainage basins, up to 9 taxonomic groups were found (Table 1). Urban Drained Rural Drained Taxonomic Group Springs Sorings Chironomidae 6742Eohemerootera 0223Plecootera 098Coleootera435Crustacea 145Gastroooda 0 6 IsoDoda 3107 Nematoda402Trichootera 089Total115647 Table1.Benthic macro-invertebrate taxonomic groups and numbers foundinurban springs receiving pollution from Class V injection wells versus rural springs not receiving Class V recharge from injection wells.

PAGE 222

Ogden, Redman, and BrownState SurveyofClass V Injection-Well RegulationsA regulatory survey and literature review was con ducted to gather information on the scopeofgroundwater contamination problems associated with Class V injection wells and to compare the regulatory frameworks existing within state programs. This will assist in developing re commendations for improved permitting, monitoring, and control over a varietyofshallow injection wells. Karst limestone terranes provide a "worst-case" setting for this study because the underlying aquifer systems are extremely vulnerable to pollution by surface runoff and waste-dis posal practices. A telephone surveyofstates with significant occur rencesofkarst limestone bedrock was undertaken in order to: communicate with regulatory staff responsible for underground injection control(DIC)orgroundwater protection programs, obtain copiesofthe states'mcregulations pertain ingtoClass V wells, document case historiesofcontamination from Class V wells, and gain ideas and information on alternatives for pre venting groundwater contamination by Class V in jection wells. Regulatory agencies in sixteen states have been inter viewed, including the Alabama DepartmentofEnviron mental Management, Arkansas DepartmentofPollution Control and Ecology, Florida DepartmentofEnvironmen tal Regulation (DER), Georgia DepartmentofNatural Re sources, Georgia Geological Survey, Illinois DepartmentofHuman and Natural Resources (DHNR), Indiana Depart mentofEnvironmental Management, Kentucky DivisionofGroundwater Protection, Minnesota Pollution Control Agency (MPCA), Missouri DepartmentofNatural Resour ces, New York DepartmentofEnvironmental Conserva tion (DEC), Ohio Environmental Protection Agency, Pennsylvania DepartmentofEnvironmental Resources, TennesseeDepartmentofHealth andEnvironment(TDHE), Texas Water Commission, Virginia Water Con trol Board, and the West Virginia DivisionofNatural Re sources. Tenofthe sixteen states interviewed reported having no rules dedicated to Class V wells. Mostofthese do, however, give special approval to groundwater return-flow wells for heat-pump and air-conditioning systems. Several commented that misused or abandoned Class V wells are regularly discovered during investigationsofunrelated complaints and violations. An Indiana regulator stated that so many different branches come in contact with Class V 214Appalachian Karst Symposium. 1991wells, that the reports are scattered and formal rules had never been organized. He advocated the implementationofa permitting system with permit fees fmancinginspections and enforcement. Minnesota prohibits any useofwells for injection purposes (except return-flow wells) and requires landowners to seal them as they are discovered. Investigators have found a numberofsmall-quantity generators illegally dis posingofhazardous wastes in septic tanks and shallow wells. The MPCA also cited several instancesofClass V wells being used for sewage disposal (cesspools). Investi gatorsofunderground storage tanks often find ClassVwells being used inappropriately when conducting site investigationsatservice stations and auto-maintenance shops. The MPCA attempts to inform the public about "disposal wells" and contamination ofgroundwaterbydis tributing newsletterstothe agricultural and business com munities. Illinoisisanother state that uses a public-information campaigntocontrol Class V injection wells. The DHNR maintains "good relationships" with county Farm Bureau associations and soil-conservation districts. These groups encourage membersofthe agricultural community to di vert feedlot, fertilizer, and other polluted runoff awayfromwells and sinkholes. DHNR maintains a registryofdry wells used for flood control that is updated regularly with a questionnaire sent out to developers, municipalities, con tractors, and others. The useofretention basins to pretreat stormwaters and allow gradual percolation/evaporation is common in Illinois. The Missouri DivisionofGeology and Land Survey issues a joint permit with the DivisionofEnvironmental Quality for large commercial or institutional heat-pump withdrawal/injection wells. The construction and opera tion permitisdesigned to control the temperature differen tialata radiusof200 feet from the well. Observation wells are installed at this distance for the purposeofrecord ing monthly temperatures and total dissolved solids. The permit also requires random inspections and an annual report by the permittee. No other Class V wells are approved in Missouri except those constructed for remedial purposes. Ohio and Arkansas have conducted inventoriesofClass V wells in their states f9r the purposeofdeveloping appropriate regulations. Ohio will have regulations avail ablein1991that will contain special provisions for karst terranes. The statesofAlabama and Tennessee have a registra tion requirement for Class V injection wells. The regula tors acknowledge that, with limited tracking and enforce ment capability, they are notifiedofonly a small fractionofthe disposal wells in use. Regulators determine if there is a possibilityofadverse impact on an underground sourceofdrinking water based on the typeoffluid injection and

PAGE 223

Appalachian Karst Symposium, 1991other general infonnation submitted. Class V wellsareexempt from sitingandconstruction requirements, al thoughTDHEreserves the righttodisapprove any well that is considered tobeofsubstandard construction. Ala bama requires new Class V wells tobedrilled by a licensed well driller and issues a ten-year operating penniL TDHE requires owners/operators to apply for a Plugging and AbandonmentPennitthat justifies well abandonment and outlines a closure plan using a cement plug. Georgia, Florida, andWestVirginia have the most comprehensive regulations covering Class V wells that have been reviewed to date. InWestVirginia, owners/ operatorsofClass V wells must notify theChiefofthe Water Resources Divisionofthe following infonnation: construction featuresofthe well, nature and volumeofinjected fluids, alternative meansofdisposal available, en vironmental and economic impactsofwell disposal and its alternatives, facility name, location, ownership, legal con tact, nature and typeofinjection well, and operating sta tus. All Class V wells are authorized by rule for a periodoffive years. When theChiefdiscerns a potential viola tionofdrinking-water standards, an individualpennitcanbeissued. The pennits may include conditions for opera ting, mechanical integrity, monitoring, reporting, and plugging and abandonment. Before Class V wells are approved in West Virginia, regulations specify that hydraulic connections with under ground sourcesofdrinking water should be considered in addition to the potentially affected population. The areaofconcern is called the "zoneofendangering influence", which is defined as the ... horizontal distance from the in jection well in which the pressures in the injection zone maycausethe migration ... "ofcontaminants into an USDW. The regulations suggest two possible methods for detennining the zoneofendangering influence. The first takes an analytical approach using the Theis equation, based on the following assumptions. the injection zone is homogenous and isotropic, the well penetrates the entire thicknessofthe injec tion zone, the well diameter is insignificant comparedtothe radiusofinfluence when injection time is longer than a few minutes, and emplacementoffluid into the injection zone creates an instantaneous increaseinpressure. Because such idealized aquifer characteristics occur very rarely in the real world, the regulations also allow a more qualitative approach for determining the zoneofen dangering influence. A fixed radiusofnot less than 1/4 mile can be used for a wellora close clusterofwells basedonthe following considerations: 215Ogden, Redman. and Brownchemistryofthe injection and fonnaLion fluids, geology and hydrogeology, population, groundwater use and dependence, and historical practices in the area. Regulations in West Virginia specifically mention two typesofClass V wells often found in carbonate for mations: wells for waste disposal into solution cavities, and sinkholes used for the disposalofsewage or any other waste. Alabama, Florida, Georgia and West Virginiarequirethat Class V wells be constructed by water-well contractors licensed within the respective states. None offer specific construction-design standards owingtothe varietyofwells and their uses. Florida, however, reserves the right to apply Class I design standards if the situation warrants. The DER also requires a well-completion report and may ask for samplesofformations penetrated. Georgia requirescasingfivefeetintotheinjectionzoneandgrouting/sealingofthe annular space. The DER can impose operating, pretreatment, moni toring, and reporting requirements on cooling-water return flow wells with additives, experimental/remedial wells, spent-brine return-flow wells, non-hazardous commercial/ industrial disposal wells, geothermal wells, and "other" wells. The latter three must be plugged with cement when abandoned to prevent the movementoffluids between USDWs. One concern aroused during the interviews was that the recently issued NPEDS municipal stonn-water regula tions will increase the useofClass V injection wells for controlling urban runoff. With many states having such loose regulatory control over Class V wells, there are im plications for impacting groundwater quality with re-routed storm-water drainage. The New York DEC has begun to address this pro blem with its State Pollutant Discharge Elimination Sys tem.Ifa facility has a discharge, the DivisionofWater Resources must see that the discharge meets minimum standards and that groundwater under the site meets desig nated standards. Detention-basin discharges draining large parking areas are required to have no visible oil and grease,orelse an oiVwater separator will be required. The regula tor stated that non-discharging collection basins that allow storm water to percolate/ evaporate are not monitored ex cept in heavily industrializedorpotentially contaminatedareas.During the interview, the regulators seem interestedindeveloping practical alternatives for managing Class V

PAGE 224

Ogden. Redman. and Browninjection wells, and several requested copiesofourfinal report. Most noted the wide rangeoflocations, uses, and typesofClass V wells as being the main barrier to a uni form regulatory program.Summary and ConclusionsOurpreliminary results have shown that many more Class V injection wellsoccurin thestudyareasthan anticipated. Initial sampling shows that degraded surficial recharge to these wells adversely impacts groundwater. Regulatory authority willbeneededtoassess the impactofindividualservice-station-baydrainsongroundwaterquality.Inaddition to the regulations from the states,wehavecompiledanumberofcasehistoriesandlocal/model ordinances from wellhead-protection programs, watershed districts, and areas with sole-source aquifer designation. These willbeevaluatedanddeveloped into recommended guidelines and "best-management practices" for preventing contamination by Class V injectionwells.SomecombinationoftheWestVirginia, Florida, andGeorgiaUICregulationswouldprovidearelativelyflexible framework for regulating Class V injection wells. It is anticipated that the useofClass V injection wells forstorm-waterdrainageandothertypesofdisposal will increase in rapidly developing urban areas andevenin some smaller communities.Theimportanceofevaluating andmonitoringtheimpactofthese wellsonUSDWsshouldbeemphasized.Wesuggestthatdye-tracingtechniqueswouldbeapplicableinkarstterranestodelineate the potentialzoneofimpact.Otherfactors to consider would include the densityofdisposal wells in the area, the timeoftravel between the injection pointandthe nearest dischargeorwithdrawal point,otherpotentially impacting landuses, possible chemical interactions among injected fluids, and anticipated alterationstothe subsurface flow systems caused by additional volumesoffluids. 216Appalachian Karst Symposium. 1991AcknowledgmentsThisproject isbeingfundedbythe U.S. Environ mental Protection Agency with matching funds provided by the Center for the Management, Utilization and Protec tionofWaterResources Tennessee Technological Uni versity and the First Tennessee Development District.ReferencesFaulkerson,J.;Burden, D.; Burden, K.; Edwards, C.;Kinley,T.;Lee,T.;Sparks,V.;Starnes,D.D.; Walls, E.;andWebster,S.,1981,Karst Hydrology, Morphology.andWater Quality in the VicinityofCookeville. Tennessee:Report to the CityofCooke ville, Tennessee Technological University, Cookeville, Tennessee,67p. Hannah, E.D.; Pride, T.E.;Ogden,A.E.;andPaylor, R., 1989, Assessing groundwater flow paths from pol lution sources in the karstofPutnam County, Tennes see,inBeck, B.F. (editor),Engineering and Environ mental ImpactsofSinkholes and Karst: Proceedingsofthe ThirdMultidisciplinary Conference on Sinkholes and the Engineering and Environmental ImpactsofKarst.St PetersQurg Beach. Florida.2-4October1989: A.A. Balkema, Rotterdam and Brookfield, p. 183-188. Ogden, A.E.; Curry, W.A.;andCummings, J.L., 1989, Morphometric analysisofsinkholes and caves in Ten nesseecomparingtheEasternHighlandRimandthe ValleyandRidgephysiographicprovinces,inBeck, B.F. (editor),Engineering and Environmental ImpactsofSinkholes and Karst: Proceedingsofthe Third Multi disciplinary Conference on Sinkholes and the Engineer ing and Environmental ImpactsofKarst. St Petersburg Beach. Florida.2-4October1989:A.A. Ba1kema, Rotterdam and Brookfield, p. 135-142. U.S. Environmental ProtectionAgency,1981,40CFRParts122and146UndergroundInjectionControl Program criteria and standards:Federal Register,v.46, no.190,p.48243-48255.

PAGE 225

Appalachian Karst Symposium, 1991 JonesThe Carbonate Aquiferofthe Northern Shenandoah ValleyofVirginia and West VirginiaWilliamK.JonesEnvironmental Data P.O. Box 490 Charles Town, WV 25414ABSTRACTA Cambrian and Ordovician carbonate sequence over 12,000 feet (3,700 m) thickisexposed westofthe Blue Ridgeinthe northern Shenandoah ValleyofVirginia and West Virginia. The rocks are highly folded and faulted and the nearly vertical attitudeofthe rocks tendstoforce groundwatertofollow more permeable bedding planes paralleltothe stratigraphic strike. Dye-tracer tests have generally shown a half-radial flow pattern with travel timesofthreetofive months for distancesofone to three miles (1.6 to 4.8km).Karsti ficationofthe land surface and the aquifer is subdued duetothe high-density fracture system, that tends to disperse rather than concentrate groundwater flow, and the generally low hydraulic gradients (dh/dl:= 0.01). Groundwater circulation may dip well below local base level and solution cavities have been observed at depthsof150 feet (45m)below major surface streams. Permeabilityisonly through secondary fractures al though the groundwater appearstobe a continuum; so the aquiferistermed "fracture-diffuse". Both tracer test and pumping-test results show that permeability is almost one orderofmagnitude higher paralleltothe strike than normal to the strike. However, potentiometric contours tend to parallel the strike, so the hy draulic gradient is normaltothe strike. The almost vertically dipping rock and the ready communicationofsurface sinkholes with groundwater suggest that the aquifer is generally unconfmed. However, most wells are drilledtodepths greater than 100 feet (30m)in spiteofrelatively shallow, (average about 30 feet or10m) static water levels in the region. Storativity values calculated from several multi well pumping tests range from 0.01 to 0.001 and suggest confined flow conditionsinthe deeper fractures despite the absenceofan obvious confining layer.IntrOductionThe northern partofthe Shenandoah Valleyisa broad carbonate lowland bounded by the precambrian Blue Ridge Mountains to the east and clastic ridgestothe west. The study area lies within the Valley and Ridge physiographic province and includes partsofClarke and Frederick coun tiesinVirginia and Berkeley and Jefferson countiesinWest Virginia (Figure1).The carbonate sequenceissplit into two sections by the Martinsburg Shale which crops out along the Frederick-Clarke and Berkeley-Jefferson county borders. The western carbonate belt is narrower, higher, and generally exhibits more relief than the eastern belt. This paper concentrates on the hydrogeological set tingofthe eastern carbonate belt which is exposed over about 75%ofClarke and Jefferson counties. An overviewofthe karst settingofthe northern Shenandoah Valley was presentedinmapsbyHubbard (1983, 1990). Regional groundwater studies have been conducted by Bieber (1961), Hobba and others (1972),217EASTERNCARBONATEBELTSFigure1:Location mapofstudy area showing major carbonate outcrop belts.

PAGE 226

Jones Appalachian Karst Symposium. 1991Hobba (1976,1981), Kozar and others (1991), Trainer and Watkins (1975) and Wright (1991). Detailed site studies were presented by Jones and Deike (1981), Jones (1987), and Jones and Jones (1988). Papers by Hack (1965) and White and White (1974) discuss the geomorphologyofthe region. 1000 feet long), Frederick County has17caves(1over 1000 feet long), and Clarke County has 5 caves ( none over 100 feet long). The more significant karst features tendtodevelop near streams (Hack, 1965). This may be especially pronounced near entrenched streams where the hydrologic gradient is steepened and groundwater flow rates are accelerated. Figure2:Generalized geologic column for the northern Shenandoah Valley.Geologic SettingPRECAMBRIANSHADYFM(orTomstownDolomite) VJEVERTONFM ANTIETAMFMCATOCTINFMHARPERSFM STONEHENG FM(or Beekmantown Group)EL8ROOK FM ORDOVICIANROCKDALERUNFM(or Stones RiverFM)CAMBRIANROMEFM(or Waynesboro Fm) ORANDA'&' EDINBURGFM(or Chambersburg Ls)LINCOLNSHIRE& NEW MARKETLSPINESBURGSTATIONDOLOMITEMARTINSBURGFMCONOCOCHEAGUEFMConduit development is minimal in the eastern carbonate belt. Only partofthe dyerecovery from one water-tracing test, outoffourteen tests conducted to date, exhibited the travel times and recovery pattern associated with conduit flow. The results from three tracer testsinthe western beltinBerkeley County were more charac-12150ftThe developmentofkarst features is generally subdued in this region. The most obvious karst features are exposed ridgesofcarbonate rocks trending paralleltothe strike and numerous large springs. These springs have mean flows from 1to4 cfs (0.028 to 0.113 ems) and show little response to individual storm events. Sinkholes tend to be small and shallow, and most caves in the eastern belt are small, isolated features. Berkeley Countyhas 50 caves(3over 1000 feet long), Jefferson County has 42 caves(3over The carbonate sequenceisa seriesofmassive limestones, thin, shaley lime stones, dolostones, and occasional bandsofclastic, shaley sandstones (Figure 2). The highest water yields are from the Stonehenge (Beekmantown) and Conoco cheague formations. The Beekmantown Group also contains the highest densityofmapped sinkholes, with 5 sinkholes/ square mile in Jefferson County (Kozar and others, 1991). The study area lies in the Massanutten Mountain syn clinorium and is underlain by Cambrian and Ordovician carbonates. The carbonate sequence is about 12,500 feet (3,800 m) thick and is highly folded and faulted. The east limbsofsynclines and the west limbsofanticlines are oversteepened and locally overturned. Many low-angle faults dip to the east. The strike is N 20 E,butsomeofthe fold axes trend in a more northerly direc tion than the strikeofthe formational outcrop belts. The roleoffaultsincon trolling groundwater movement is not clear. Hobba (1981) felt that faults act as groundwater corridors and that 67%ofthe large springs in Jefferson County were located onornear faults. However, detailed site investigations (Jones and Deike, 1981) suggest that mostofthe fault zones are well cemented with calcite and probably act as a partial barrier to groundwater. This could force ground water to the surface at some points and may result in increased solutional activ ity on the upgradient sideofthe fault. 218

PAGE 227

Appalachian Karst Symposium. 1991teristicofconduit-flow tests, but a diffuse-flow compo nent was still present. The general lackofconduit-flow development is attributed to a combination of:(I)High density fracture system which tends to disperse ratherthanconcentrate groundwater flow; (2) Generally low relief and iow hydraulic gradients(dh/dl '" 0.01); (3) Lackofsur rounding non-carbonate highlands to concentrate aggres sive water at certain points on the carbonates. The nearly vertical attitudeofthe rocks tends to force groundwater to flow in troughsofmore permeable units or fractures parallel to the strike. The potentiometric con tours also tend to parallel the strike, so groundwater movement predicted orthogonal to the potentiometric con tours must move across lithologic barriersoflower per meability (Jones and Jones, 1988). The aquifer has entirelysecondary permeability and is strongly anisotropic. Groundwater does appeartobe a continuum so the aquiferistermed "fracture-diffuse".HydrologyTotal annual precipitation for the study area is about38inches (965 mm) and potential evapotranspirationis30 inches (762 mm). Discharge records at Leetown (Jones and Deike, 1981) show13inches (330 mm)ofrunoff so the actual evapotranspiration is probably about 25 inches(635mm) and groundwater recharge is about 9 inches (230 mm) annually.Jonesstorm events. Dye recovery lagged precipitation by four to five days during a test to Prospect Hill Spring in Clarke County (Figures4 and5).The assignmentofdistinct catchment areas to indivi dual springsinthe area is probably not practical becauseofthe diffuse natureofgroundwater movement. The springsallappear to draw water from a common aquifer, so groundwater divides between springs are not sharp. Dye was also foundinpumped wells during severalofthe tracer tests. The sizeofthe recharge area may be estima ted from discharge records and a water-budget calculation (Jones and Deike, 1981; Wright, 1991), but pollutants may arrive at a spring from distances well outsidethecal culated recharge area. Groundwater flow appears to approach darcian condi tionsifa relatively large volumeofthe aquifer is consi dered. Velocities are low enough that the flow should be generally laminar, and conduit flow is rare and integrated into the regional potentiometric gradient. Trainer and Walkins (1975) estimated the average transmissivity(T)ofthe upper Potomac River Basin carbonatesat500 square feet/day (sq ft/d) (46sqmid)and average storativity at 0.03. Multi-well pumping tests from the Rockdale Run Formation and the Stonehenge LimestoneinMaryland gave T valuesof17,000 sq ft/d (1580 sqmid)and 26,700 sq ft/d (2480 sq mid) and storativitiesof0.02 and 0.01 respectively. Kozar and others (1991) estimated Fourteen groundwater-tracer tests have been conducted in the eastern car bonate beltofJefferson and Clarke coun ties (Jones and Deike, 1981; Jones, 1987; Kozar and others, 1991). Sinking streams are rare in this area, so the dye was injected into sinkholes and flushed into the aquifer using 1000Lo2000 gal lons (3.8 to 7.6 cubic meters)ofwaterfromtank trucks. Dye recovery was at multiple springs one to three miles (1.6to5 km) from the injection point. Traveltimes were threetofive months with the exceptionofone test site situated along a fault where partofthe dye was recovered in less than two weeks (Kozar and others, 1991). Movementofthe tracerisoften par allelto the stratigraphic strike'and some times almost parallel to the potentio metric contours. The flow pattern is generally half-radial and travel times are not very dependent on straight-line dis tances between the injection and recovery points (Figure 3). The flow rates do not appear to be linear with respect to time. Dye recovery from most quantitative testsinthe area appears to correlate withTRACERTEST FROMGILPINSINK,CLARKECO.,VIRGINIA3400FTPROSPECTHILLSPA. ----1-S-0-0-A-Y-S--.n Q=1.2 CFSPOTENTIOMETRIC GRAOIENTCARTERHALLSPRQ=2.2 CFSFigure3:Patternofdye recovery from the injectionof1Iboffluorescein sodium dyeinGilpin sinkhole near Boyce, Clarke County, Virginia. The slopeofthe potentiometric surfaceisshown by the stippled arrow. 219

PAGE 228

Jones Appalachian Karst Symposium. 1991164154144 ...-.. BLUE0 T""" >< (f) W I ()Z --.... 0...()W IT: 0...1049484 0'9o...,----------------------------------, zPROSPECTHILLSPRING,CLARKECO.,VIRGINIA ::J 80 -+-----------------------------R------j---1<{ 70-+----------------------------1-1----1o --.... 60-+----------------------+-+-----1W W 20 w t5 10-+-------------1J-------------1I------------1::JLi 74114124134TIME (DAYS)Figure 4: Dye recovery and precipitation at Prospect Hill Spring, Clarke County. The blue dye, Tinopal CBS-X, and the green dye, fluorescein sodium, were injected on the same date, but the blue dye was injectedina sinkhole 2.25 miles westofthe spring and the green dye in a sinkhole 0.6 miles westofthe spring. Note the long travel times and the earlier arrivalofthe blue tracer. Jefferson County transmissivity at 4000 sq ftld (372 sqmid)parallel to the strike and at 500 sq ftld (46 sqmid)perpendicular ta the strike, using the gradientofthe water table and stream flow recession. This indicates an isotropyratioofabout8:1 in the strike direction. Several multi-well pumping tests have been conduc tedinthe area, but manyofthe assumptions behind aqui fer tests are not metbyfield conditions. Transmissivity values from these tests seem to range between rather narrow limits, but they are not constant at all times and places (Trainer and Watkins, 1975). The resultsofaquifer testsinthis area are not exactly reproducible and different typesofanalysis will give somewhat different values from the same test. The validityofthe useofthe analysis and the degree to which the results are representativeofthe aquifer mustbequestioned for all pumping testsinthis re gion, but a certain broad consistency in the results builds some confidence in the applicabilityofthe technique. A pumping test conductedinthe Rockdale Run For mation at Leetown, Jefferson County,isshowninFigureLAGCORRSE-7-.126.236-6.091.229-5.670.224-4.568.218-3-.117.213-2-.096.209-1-.160.2040-.211.2001-.217.2042-.124.2093-.109.2134-.138.2185-.030.2246-.086.2297-.034.236( ( ( ( ((( ( ( (((( ( (o) )) ) ) ) ))) ) )LAG TIMEINDAYSFigure5:Plotofthe cross-correlationofblue dye recovery and stonn events at Prospect Hill Spring, Clarke County. Dye recovery peaks lagged stannpeaksby 4 to 5 days during this test. 220

PAGE 229

Appalachian Karst Symposium.1991 900o2100CARBONATES1800Figure 6: Plan viewofarrangementofobservation wells and pumpedwell"A" for pumping test conductedatLeetown, Jefferson County, West Vir ginIa. Note the presenceofan impermeable shale boundary westof the testareaJonesthat the test is dealing with a confined aquifer. Storativitiesofconfmed aquifers are generally less than 0.01, so the val ues for this region are right on the boun dary between confined and unconfined. Also, highly fractured aquifers may have low storativities under water-table condi tions. There is no obvious confining bed, although a more open solutionally enlarged zone may exist at some depth. Wells drilled in the area generally go much deeper than the potentiometric sur face to obtain an adequate water supply. The aquifer is in ready communication with the surface at sinkholes. In short, the aquifer appears to have both confmed and unconfmed characteristics. One pos sible interpretationofsomeof the irregu larities in the pump-test results is that the initial drawdown and releaseofwater from the aquifer is under confined condi tions, but water-table conditions develop after some periodoftime and specific yield begins to control the flowofwater to the well. This may also explain why many wells in the area produce turbidwaterafterpumpingfor 15or20 minutes.6.Well "A" is 120 feet (37m)deep and was pumpedat100 gpm (0.38m 3/min)for 48 hours. The boundaryofthe carbonates and the Martinsburg Shaleisabout 1200 feet (366 m) westof the test area. Transmissivities calcu lated using the Theis nonequilibrium formula were 7280 sq ftld (676 sqmid)atwell "C" and 15,400 sq ftld (1430 sqmid)atwell "D". Storativity was 0.001at"C"and 0.002at"D". Another calculation method, the Jacob time-drawdown method,ispresentedinFigure 7 for well "D". This analysis gives a transmissivityof10,600 sq ftld (985 sqmid)and a storativityof0.03. One major sourceofnoise may be the assumption00WELLIIDII0.1 __.2..-...t=' 0o = 100GPMW W0.2r = 622FEET l:!:s = 0.33FEET10 = .045DAYSZ0.3 ---0 Cl T= !.9,60_0 saFT/DAY 0.4 '"
PAGE 230

JonesReferencesBeiber, P.R., 1961, Groundwater featuresofBerkeley and Jefferson Counties, West Virginia: West Virginia Geological Survey Bulletin21,79p.Hack, J.T., 1965, Geomorphologyofthe Shenandoah Valley, Virginia and West Virginia, and originofthe residual ore deposits: U.S. Geological Survey Profes sional Paper 484, 84 p. Hobba, W.A., Jr., 1976, Groundwater hydrologyofBerke ley County, West Virginia: West Virginia Geological Survey Environmental Geologic Bulletin13,21p. Hobba, W.A., Jr., 1981, Groundwater hydrologyofJeffer son County, West Virginia: West Virginia Geological Survey Environmental Geologic Bulletin16,21p.Hobba, W.A., Jr.; Friel, E.A.; and Chisholm, J.L., 1972, Water resourcesofthe Potomac River Basin, West Virginia: West Virginia Geological Survey. River Basin Bulletin 3, 110 p. Hubbard, D.A., Jr., 1983, Selected karst featuresofthe northern Valley and Ridge province, Virginia: Virginia DivisionofMineral Resources Publication 44, one sheet. Hubbard, D.A., Jr., 1990, Geologic and hydrologic com ponents maps for Clarke County, Virginia: Virginia Division Mineral Resources Publication 102,2sheets. Jones, W.K., 1987, Overviewofthe groundwater resour cesofClarke County, Virginia, with emphasis on the 222 Appalachian Karst Symposium. 1991 carbonate aquifers westofthe Shenandoah River: Clarke County Ground Water Protection Plan, p.7.17.22. Jones, W.K. and Deike, G.H., III, 1981, A Hydrologic Studyofthe Watershedofthe National Fisheries Center at Leetown. West Virginia: Environmental Data, Frankford, West Virginia, 74p.Jones, W.K. and Jones, L.E., 1988, The studyofground water movementinhighly fractured aquifers: Proceed ingsof21st IAH Congress, Guilin. China,p.620. Kozar, M.D.; Hobba, W.A., Jr.; and Macy, J.A., 1991, Geohydrology and water qualityofJefferson County, West Virginia, with emphasis on the carbonate area: U.S. Geological Survey Water Resources Investigations Report (open me). Trainer, F.W. and Watkins,EA.,Jr., 1975, Geohydro logic reconnaissanceofthe upper Potomac River Basin: U.S. Geological Survey Water-Supply Paper 2035,68p.White, W.B. and White, E.L., 1974, Base-level controlofunderground drainage in the Potomac River Basin, in Rauch, H.W. and Werner, E. (editors), Proceedingsofthe 4th Conference on Karst Geology and Hydrology: West Virginia Geological Survey, p. 41-53. Wright, W.G., 1991, Groundwater hydrology and quality in the Valley and Ridge physiographic provincesofClarke County, Virginia: U.S. Geological Survey Water Resources Investigations Report 90-4134, (open file).

PAGE 231

Appalachian Karst Symposium, 1991 Zewe and RauchInfluenceofHydrogeologic Setting and LineamentsonWater-Well Yield in the Great Valley Karst TerraneofEastern West VirginiaBrad T. Zewe and Henry W. RauchDeparunentofGeology and Geography West Virginia University 425 White Hall Morgantown, WV 26506ABSTRACTA studyof419 water wells was done in the Cambrian-Ordovician carbonate rock terraneofBerkeley and Jefferson counties in West Virginia, to determine how hydrogeologic factors influence well yield. The study area covers about 280 square miles and is dominatedbythe Massanutten Synclinorium that contains many tight folds and faults. Drillers' well yieldsingallons per minute were utilized from various sources. Topography shows a very strong correlation with well yield,asexpected. Yields from valley wells are the greatest (25 gpm, median yield), compared to slope and upland wells (15 gpm, median) and hilltop wells (8 gpm, median). The carbonate terrane was subdivided into poorly cavernous(PC)and moderately cavernous (MC) rock types. Two cave indices were derived from lengths of caves over 50 ft long with natural entrances: the cave density (CD) index(toLaicave length / carbonate rock area), and the average cave length (ACL) index. PC rocks were judged to have an ACL indexofless than 250 ft and a CD indexofless than 75 ft/mi 2 MC rocks have higher index values. The median yieldofPC rocks (15 gpm) is significantly higher than that for MC rocks (12 gpm). MC rocks show no significant correlation between lineaments or faults and well yield. PC rocks exhibit strong well yield with short lineaments, averaging 33.5 gpm within 0.05 kmofsuch lineaments, or 2.8 times the yieldofmore distant wells. These strong trends are evident for all topographic settings except hilltops. PC-rock wells, each located within 0.10 kmoftwo short lineaments, average about 125 gpm, or about eight times greater than non-lineament wells. Wells in PC units were also found to have significant ly higher yields within 0.20 kmofa thrust fault. In summary, short lineaments are well suited for ground water explorationinPC-carbonate aquifers.IntrOductionThe main purposeofthis researchistodetermine how hydrogeologic setting, and especially surficial lineaments, relate to water-well yield, with the intentionofimproving groundwater explorationinthe study area. The study area is located in the Valley and Ridge Province within the eastern panhandleofWest Virginia, including allofBerkeley and Jefferson counties, as showninFigure I. The carbonate units range in age from the Cambrian Tomstown dolomite to the Devonian Helderberg limestones. The Cambrian-Ordovician carbonates, com223 posedoflimestone and dolostone units, represent approxi mately 95%ofthe total carbonate area and are situated within the Great Valley region, bounded by North Moun tain on the west and the Blue Ridge on theeastThe Great Valley is structurally dominated by the Massanutten Syn clinorium, where bedrock is intensely faulted with many minor folds that trend approximately N 20E.The hydrology and hydrogeology within the study area have been described by Bieber (1961) and Hobba (1976, 1981). The carbonate units are dense, causing groundwatertoflow primarily through secondary openings such as frac tures, joints, bedding partings, and solutionally widened

PAGE 232

Zewe and Rauch KEY: DCa'bonate mCla 11c Metamorphic ManlMbufQ Figure1:Location mapofthe Berkeley and Jefferson counties study area within West Virginia. openings. These secondary features not only determine flow direction but also influence whether a well will pro duce adequate quantitiesofwater.Previous InvestigationsThe relationship between well yield and lineament proximityincarbonate rocks was originally described by Lattman and Parizek (1964) within the Valley and Ridge rocksofcentral Pennsylvania. They reported increased water-well yields near fracture traces (short lineaments) associated with enhanced solutional and fracture porosity and permeability along fracture zones. Siddiqui and Parizek (1971) elaborated on this lineament study, investigating hydrogeologic factors that influence well yields. They concluded that the controlling factors on well productivity are fracture traces, rock type, dipofthe bedrock, topography, and structural setting, in orderofdecreasing importance. LaRiccia and Rauch (1977) studiedthe effectofphotolineaments on well yield in the carbonate rockofFrederick Valley, Maryland. They found significantly higher yields where wells are located within 200 ft and especially within 100 ftofa photo lineament (short or long straight lineaments). Rauch and Plitnik (1984) studied the effectoflinea ments on well yield in the Cambrian-Ordovician rocksofHagerstown Valley, Maryland, located adjacenttothe present studyarea. Most carbonate units there are poorly cavernous (diffuse flowincharacter), and exhibit high yields to wells near lineaments, but one highly cavernous unit showsnorelationship with well yield. Ogden (1976), 224Appalachian Karst Symposium. 1991Heller (1980), and LaRiccia and Rauch (1977) have also found that highly cavernous (conduit) aquifer systems are areas where lineaments are not useful as exploration tools for groundwater.MethodsLineaments were defmed as being linear (straight line ament) or uniformly curved (curvilineament) features iden tified as stream segments,dryvalleys, aligned sinkholes, aligned meander bends in channels, or tonal streaks in soil. Both short lineaments (0.5 2.0kmlong) and long linea ments (>2.0kmlong) were mapped. Lineament mapping proceeded in two phases; first lineaments were mapped based solely on topographic contours and streams shown on the 7.5-minute topographic quadrangles, and then linea ments were mapped from black-and-white aerial photos with a scaleof1:20,000. The LANDSAT lineaments used in this investigation were mapped by Hobba (1976, 1981), and have lengths ranging from 9 to19miles. Figure 2, partofa 7.5-minute topographic quadrangle that was used as a base map, shows examplesofmost typesofmapped lineaments. Well-yield data were obtained for 419 wells situated within the Cambrian-Ordovician units. About 20 percentofthese data were gathered from the U.S. Geological Sur vey reportsofBieber (1961) and Hobba (1976,1981), and about 80 percentofthe well-yield data were derived from drillers' permit reports for 1984-1989 that are available at the county health-department offices. All well locations designatedonthe drillers' permits were first field checked for accuracy and then plotted on the topographic base maps. All well plotting was done only after all linea ments were identified, to eliminate biasesinlineament mapping that could result from knowing locationsofhigh yielding wells beforehand. The lateral distance from each well to the nearest lineament was measured directly on the base map with an accuracyofO.OIkm. Graphical plots were constructedofwell yield versus parametersofthe hydrogeologic setting at the wells andofmeasured lineaments. Apparent visual trendsinthe data were statistically tested for strength using the nonparametric Mann-Whitney U test as described by Siegel (1956).Hydrogeologic InfluencesonWell Yield Topographic InfluencesA topographic positionofvalley, slope, upland,orhilltop was designated for each well. Valley wells have a median yieldof25 gpm, compared to15gpm for slope and upland wells and 8 gpm for hilltop wells. Slope and upland wells were grouped together because their yields are not significantly different Valley wells have significantly higher yields than do hilltop and slope/upland wells at the

PAGE 233

Appalachian Karst Symposium. 1991 Zewe and RauchSCLShortCurvilineamentLSL LongStraightLineament0.5Scale(lem)oKEY:SSL-ShortStraightLineamentLCL-LongCurvilineament(Notshown) \...\ IL/. '---..jLj SCL-.. J o. ISo II '1\ II"11 -"111/--" \I/' .\'f' \\ \1-, II ...., .1/'II "" Figure 2: Partofthe Shepherdstown 7.5-minute Quadrangle, showing typical typesofmapped lineaments.0.()()()()5and 0.0005 alpha probability levels, respectively, based on Mann-Whitney U statistical tests. Slope/upland wells have significantly higher yields than hilltop wells at the 0.05 alpha probability level. It has been thoroughly documented from past studies that valley positions are favorable locations for higher than-average yielding wells. There are severalreasons for this. Valleys are areasofdischarge where the water table is usually closer to the surface when comparedtoother topographic positions. Also, valleys preferentially form along areasofstructural weakness such as fracture zones and thus promote differential erosion. These fractures would then be responsible for increased porosity and permeability associated with valleys, especially where cavities formedbydissolution areprevalent.Well-Depth InfluencesWell-depths, available for 401 analyzed wells locatedincarbonate terranes, were plotted against well yield. A notable drop-off in yield is apparent for wells deeper than 200ft.Shallow wellsft deep) have a median yieldof20 gpm, compared to just 8 gpm for wells with depthsof>200ft.This trend is strongly significant at the 0.00005 alpha probability level. A secondary zoneofmoderate well production was observed between the 200 footand 300-foot-depth levels, where wells have a median yieldof15gpm. Higher well productivity at shallow depth probably represents an intensely weathered zone near the surface where well yields are high compared to the deeper, un weathered zone. Rock fractures tend to be more numerous and open, and solution cavities are more common within 200 ftofthe surface. Valleys were also found to have a greater proportionofwells less than 200 ft deep thandoother topographic settings.StratigraphicInfluencesWells in carbonate units have an overall median yieldof15gpm with wellsinindividual carbonate beds having median yields ranging from 5 to 35 gpm. Table 1 shows a summaryofthe yields obtained from the Cambrian Ordovician units. Lithologic descriptions and unit thick nesses were obtained fromDeanand others (1987).KarstCharacterInfluencesThe carbonate terrane was subdivided into poorly cavernous(PC)rock units and moderately cavernous (MC) rock units,basedon inference from known cavesasrepor tedbyGulden and Johnson (1984). The lengthsofthose caves with natural entrances and with over 50fLofpas sages were utilized to derive two cave indices,thecave den sity (CD) index and the average cave length (ACL) index. Our analysis was restricted to these cavesbecause they probably best represent the larger populationofall caves in the study area, including caves without entrances. The 225

PAGE 234

Zewe and Rauch Geol. Carbonate Time Rock Scale Type Range In Thickness (feet) Lithologic Description Appalachian Karst Symposium. 1991 Average Cave Cave Number Median Length DensityofTest Well Index+Index+ Wells Yield(feet)(feet/mile 2 ) (gpm) Ornc 540-860 Hard thinly bedded limestone with 314 430.34260some pure limestone beds. R D Obps 0-500 Gray dolomite with calcite veins* *3350and chert nodules common. V I0002400-2750 Limestone and dolomite with dolo168 42.3 62 20 C mite predominant in the upperpartI A Obs 800 Limestone with argillaceous lamina-**3012N tions. Silty laminations are common in the lower member. Cc 2200 Limestone with interbedded dolomite. 56 4.3 16712Sandy dolomite at base. C A Ce 2000 Argillaceous, dolomitic limestone 856 84.6 8715M with shaly interbedsinthe lower B one-third. R I Cwy 1000 Dolomite and sandy limestone with 842 240.21112A sandstone and shale in upper part. NCt1000 Massive dolomite and sandy dolomite 786 164.6135 with some interbedsofpure limestone. Table1:Descriptions and well yields for carbonate rock units in the study area. Omc Middle Ordovician Carbonates Cc Conococheague Formation Obps Pinesburg Station Dolomite Ce Elbrook Formation Obrr Rockdale Run Formation Cwy Waynesboro Formation Obs Stonehenge Limestone Ct Tomstown Dolomite +Based on only natural caves >50 feet long.*-No naturalcaves>50 feet in length are within these units. PC rocks were judged to have an ACL indexofless than 250 ft and aCDindexofless than 75 ft/mi 2 TheMCrocks were determined to have an ACL greater than 250ftand a CD index greater than 75 ft/mi 2 Noneofthe car bonate rock units were judged to be highly cavernous, as for example the Greenbrier Limestone in West Virginia. Table 1 shows the CD and ACL index for eachCambrian Ordovician carbonateunitFigure 3 shows the spatial dis tributionofPC and MC units. The MC units are located adjacent to areasofclastic rock at higher elevation where an increase in concentrated recharge owing to allogenic runoff may have led to increasing solutional developmentofthese rocks, even though some arepredominantly dolo mitic. The overall median yieldofPC carbonate rocks is15gpm and the median yield for MC rocks is 12 gpm. This slight difference in well yieldisstatistically significantatthe 0.10 but not at the 0.05 alpha probability level, indi cating a moderate trend whereby the least cavernous rocks 226 have higher well yields on average. It is believed that the PC rocks represent diffuse-flow aquifers with more uni formly spaced fractures and small cavities that are more conducivetohigh yields on average.Structural InfluencesWells positioned on synclines in carbonate terrane show a slightly higher median yield (15 gpm) than wells positioned on anticlines (10 gpm) (based on the 136 wells located within 0.50 kmofa structural fold axis). However, statistical tests show no significant difference in well yield between synclines and anticlines, at the 0.10 alpha probability level. Carbonate units as a whole have a significant correla tion between well yield and thrust faults. These major faults are usually oriented parallel to stratigraphic strike. Carbonate wells within 0.25 kmofa thrust fault were foundtohave significantly higher yields than more distant

PAGE 235

Appalachian Karst Symposium. 1991KEY: tJ CLASTIC METAMORPHIC 19 KARST \" IDpOORLYmil..CAVERNOUSKARSTFigure3:Spatial distributionofpoorly cavernous and moderately cavernous carbonate-aquifer units within the study area. wellsat0.05 alpha. This trend was re-analyzed, breaking the carbonate units intoPCand MC units. MC unitsshownographicalwell-yieldtrend with thrust-fault proximity,butPCunits were found to have significantly higher yields (at 0.05 alpha) when located within 0.20 Ianofa thrust fault.ForPCunits, wells near such faults have a median well yieldof25gpm compared to 10 gpm for wells >0.20 km from such faults.Thrustfaultsareprobably accompanied by permeable fracture zones that enhance well yield. Transverse faults and normal faults, usually oriented nearlyperpendiculartostratigraphicstrike, are also apparently associated with higher well yields. However, this trend is not statistically significantat0.10 alpha.Lineament Associations with Well Yield LANDSAT LineamentsA graphicalplotofwell yield versusdistance to the nearestLANDSATlineament showed that wells within 0.10 kmofaLANDSATlineament have significantly higher yields at 0.10 alpha than wellsatgreater distances (0.10 km to 0.50 km). When subdividing the well data into PC andMCunits, PC units maintain and strengthen the well-yield correlation within 0.10 km that is statis tically significant at 0.05 alpha, but no LANDSAT linea227Zewe and Rauchment correlation was observed for yieldsofwellsinMC units.Long LineamentsA graphical plot for long, straight lineaments (LSL's) showed a well-yieldtrendwith lower yields associated with wells located <0.20 Ian from a LSL. This trend is only slightly significantat0.10 alpha.Thestatistical results indicate that long, straight lineamentsdonotstrongly relate to well yield.Nowell-yield trend was foundtobe associated with proximal distance to long curvilineaments (LCL). Because onlyoneoutof23 wells located within 0.20 Ianofa long lineament (LSLorLCL)issituated in a MC unit, these long-lineament results are indicativeofPC units. In an attempt to search for overlooked long-lineament trends, topography was held constant using the same data,butnosignificantwell-yieldcorrelationswith long lineaments were apparentat0.10 alpha. These data were again re-analyzed keeping well depth constant. Wells <200 ft in depth showed no well-yield relationships with long lineament proximal distance. Deep wells (>200 ft in depth) were found to have significantly lower yields when located within 0.20 Ianofa long lineament, at 0.05alphaAlso, no significant well-yield associations were found with either long-lineament lengthororientation for wells within 0.20 Ianofa long lineament. In general, therefore, longlineamentsshouldbeavoidedingroundwaterexploration.Short Straight LineamentsShort straight lineaments (SSL's) were next examined for well-yield associations.Themedian yieldofwells within 0.05 kmofa SSL (30 gpm) isovertwice thatofwells located from 0.05 km to 0.50 Ian from a SSL, and this trend is significantat0.05 alpha. Also, wells located within 0.30 Ianofa SSL have significantly higher yields (15gpm,median)at0.01 alpha than wells located at greater distances (0.30 Ian to 0.50 Ian). These two resultsindicatethatshortstraightlineamentsarestrongly associated with high well yields and that average well yield becomes progressively higher as such lineaments are approached within 0.30 km. Topography was then held constant using the same da ta. Wells positioned as slope/upland and hilltop show sig nificantly higher yields at 0.05 alpha when located within 0.20 km and 0.30 kmofa SSL, respectively. However, valley wells show no significant correlation between yield and proximal distance to SSL'sat0.05 alpha. Well depth was then held constant. Wells <200 ft deep have slightly significant higher yields at 0.10 alpha when located within 0.10 kmofa SSL, and wells >200ftdeep show significant increasesinwell yield at 0.01 alphaiflocated within 0.30 kmofa SSL lineament.

PAGE 236

Zewe and RauchOtherSSLparameter associations with well yield were then analyzed for wells within 0.30kInofa SSL. Statistical tests indicated no strong relationships between well yield and SSL lengthorSSL orientation.Appalachian Karst Symposium. 1991yieldsof150 gpm and 100 gpm. Both wells are located within a poorly cavernous(PC)unit, indicating that linea ment density is an important factor determining high yield in such units. Short curvilineaments were investigated for well-yield associations. A median well yieldof20 gpm was found for wells within 0.20 kmofa short curvilineament (SCL) compared to a median yieldof13gpm for wells located at greater distances (0.20kInto 0.50kInaway). This trend is statistically significantatthe 0.005 alpha level. These data were then re-analyzed holdingtopography constant. Valley wells and slope/upland wells have an apparentwell-yieldcorrelationwithSCLproximal distance, indicating increased yields within 0.20 kmofa SCL lineament; these well-yield trends are statistically significantat0.10 alpha and 0.05 alpha, respectively. Hilltop wells show no well-yield relationships with SCL proximal distanceat0.10 alpha. Well depth was then held constant, subdividing the data into two groups. Deep wells (>200 ft deep) and shallow wellsft deep) were both found to have high yieldsiflocated within 0.20 kmofa SCL,at0.05 and 0.01 alpha, respectively. The deep-well trends found with SSL's and SCL's support the idea that short lineament fractures in the carbonate units penetrate deeper than 200 ft and influence well yields.Short CurvilineamentsLineament parameters were next in vestigated for wells within 0.20 kmofa SCL. A SCL orientationof300 to 330 degrees (across-strike) was found to be associated with significantly higher well yieldsat0.05 alpha. SCL's with a radiusofcurvatureof0.5 km to 1.0 km were also foundtobe associated with increased well yields compared to nearly straight SCL's, but this trend is only significantat0.10 alpha.SCLlength was not found to be associated with well yield.Short LineamentsWell-proximitydataregardingshortlineaments (SCL's and SSL's) werenextgrouped together and examined. A graphical plotofwell yield versus proximal distance to the nearest short lineament for moderately cavernous (MC) units showed no apparent well-yield trend. A similar graphical plot was made forPCunit data, (Figure4).This figure shows that wells in PC rocks exhibit strong short-lineament trends with higher yields in wells near such lineaments. A 33.S-gpm-median yieldisevident within 0.05kInofa short lineament and a 17.5 gpm-median yield occurs within0.20kInofa short lineament; these both indicate significant correlationsat0.05 and 0.001 alpha, respectively. Data for well yield versus short lineament proximal distance were then analyzed for eachofthe carbonate units listed in Table1.AllofthePCunits that could be statistically tested were foundtohave significant well-yield correlations with short lineamentsat0.01 to 0.10 alpha, whereas threeofthe four MC units showed no well-yield relationship with short lineament proximityat0.10 alpha. Therefore, moderately cavernous units, that should exhibit somewhat less diffuse-flow character than poorly cavernous units, are not toberecommended for using short-lineament locations in groundwater exploration. When holding topo graphy constant, valley and slope/upland wells have signi ficantly higher yields within 0.20 kmofa short linea-mentat0.05 alpha. However, hilltop wells show no well-yield :2: 6050 I2..4C;::f---30 0.300.400.50 r----r---' .I ----,-r-..--ffi Because short, straight lineaments i 20and short curvilineaments have many similar associations with well yield, their data sets were combined for comparisons10with relative degreeofcavern development...'--,-" ---'---' .I .I. I..iI. I.IThe influenceoflineament intersec tions could not be analyzed because too few exist; however, the only two car bonate wells located in close proximity.10km)oftwo short lineaments have PROXIMITY DISTANCE BETWEEN WELL ANDNEAREST SHORT llHEUIEKT'11m) Figure 4: Plotofwell yield (gpm) versus proximal distance (km) to the nearest short lineament for wells in poorly cavernous carbonate aquifers. Solid horizon tal lines represent the median yield for each 0.05kInclassofproximal distancetoa short lineament. 228

PAGE 237

Appalachian Karst Symposium. 1991correlation with short lineament proximal distance. Lineament orientation was then analyzed with respect to well yield for wells within0.20kinofaSSLorSCL. MC units show no well-yield association with short-linea ment orientation.PCunits show no well-yield relation ship withSSLorientation, but have a statistically significantwell-yield correlation withSCLorientationatthe 0.05 alpha level. SCL's withanorientationof300 to360degrees (across-strike) are associated with higher yields than SCL's at other orientations.Lineament ExpressionDry-valley lineaments werecomparedwith stream lineaments for their influenceonwell yield inMCand PC units. Only wells within0.10kmofa short lineament were considered in the analysis. MC units have signifi cantly higher well yields(20gpm, median)nearlinea ments representing dry valleys than near lineaments repre senting streams,at0.05 alpha. However,PCunits have higher well yields associated with stream lineaments than with dry-valley lineaments, but not significantly higher at 0.05 alpha. Yieldsofwells fromPCunits and located near.20km of) stream lineaments are not significantly different from the yieldsofwells near streams without lineaments.PCunitwells located within 0.10kmofa stream have a median yield nearly three times the overall median yield forPCunits. Thus, streams are important locations for high yielding wells inPCunits, whether or not the stream is a mapped lineament. A more detailed investigationofthese wells within 0.10 kmofshortlineaments was also made. Wells in close proximity to perennial-stream lineaments appear to be higher producing than wellsnearintermittent-stream lineaments(notdryvalleys), for bothMCandPCrock units.Thethree wellsnearperennial-stream lineaments have yieldsof30,100, and 150 gpm.Incontrast, two wells inPCunits and locatedatthe headofintermittent streams have yieldsofjust5 and 6 gpm, possibly indi cating that these are poor producing areas.Primary Conclusions for the Study AreaI.Topography is an extremely important factor impact ing well yield. Valley wells have the highest median yield (25 gpm), followed by slope/upland wells (15 gpm) and hilltop wells (8 gpm). 2.Theoptimum depth in regard to well yield is less than 200 ft(20gpm, median), with moderate yields for depths between200and 300 ft (15 gpm, median) and low yields for depths >300 ft (8 gpm, median). The optimum zoneofenhanced fracturing and secondary permeability therefore is within 200 ftofthe surface. 3. Poorly cavernous (PC) rocks have significantly higher yields than moderately cavernous (MC) rocks. Wells 229Zewe and RauchwithinPCrocksprobablydrawwaterfrom morenumerousandbetterintegrated fracturesandbed partings thandowells in MC rock.4.Wells inPCrocks have significantly increased yields within 0.20 kinofa thrust fault (25gpm,median), within 0.10 kinofaLANDSATlineament (20 gpm, median), and within 0.20 kinandespecially within 0.05 kinofa short lineament. The very best PC rockwellsarelocatedwithin0.10kmoftwoshortlineaments, whereeighttimes higher yield (125 gpm) may be expected. ApparentlyPCrocks are excellentdiffuse-flowcarbonateaquifersthathavewell integrated networksoffracturesandsmall solution cavities along lineament/fracture zones. 5. No positive correlations between well yield and linea mentswerefound tobeassociated with MC rocks. Cavernous rocks may possiblybeassociated with nar row fracture zones, and widely spaced joints and bed ding partings as noted by Rauch and Plitnik (1984). Also, such rocksmaypossiblybenon-conducive to high well yield becauseoflowered water tables near conduits underlying the lineaments, where any fracture zones couldbewell drained, especially during timesoflow flow.6.Wells located within 0.10 kinofa stream inPCunits have a median yield nearly three times the overall median yield forPCunits, indicating that streams make good locations for wells in such rocks.ReferencesBieber, P.P., 1961, Ground-water featuresofBerkeley andJeffersoncounties,WestVirginia:West Virginia Geological and Economic Survey Bulletin21, 79p.Dean,S.L.; Kulander, B.R.;andLessing,P.,1987, Geologyofthe Hedgesville, Keedysville, Martinsburg, Shepherdstown, and Williamsport quadrangles, Berkeley and Jefferson counties,WestVirginia:West Virginia Geological and Economic Survey,WVMap-WV31.Gulden, R. and Johnson, M., 1984, Cavesofthe eastern panhandleofWestVirginia:West Virginia Speleo logical Survey Bulletin8,135p.Heller, S., 1980,A Hydrogeologic Studyofthe Greenbrier Limestone KarstofCentral Greenbrier County, West Virginia:Ph.D. dissertation (unpublished), DepartmentofGeology and Geography,WestVirginia University, 167p.Hobba,W.A.,Jr.,1976,GroundwaterhydrologyofBerkeleyCounty,WestVirginia:West Virginia Geological Survey Environmental Geologic Bulletin 13,21p.

PAGE 238

Zewe and RauchHobba, W.A., Jr., 1981, Ground water hydrologyofJef fersonCounty,WestVirginia:WestVirginiaGeological Survey Environmental Geologic Bulletin 16,21p. LaRiccia, M.P.andRauch, H.W., 1977,Waterwell productivity related to photo-lineaments in carbonatesofFrederick Valley, Maryland,inDilamarter, R.R. and Csallany, S.C. (editors),Hydrologic Problems in Karst Regions,Western Kentucky University, Bowling Green, Kentucky, p. 228-234.Lauman,L.H.andParizek, R.R., 1964, Relations be tween fracture traces and the occurrenceofground water in carbonate rocks:JournalofHydrology,v.2, p. 7391.Ogden, A., 1976,The Hydrogeologyofthe Central Mon roe County Karst, West Virginia:Ph.D. dissertation 230Appalachian Karst Symposium, 1991(unpublished), DepartmentofGeology and Geography, West Virginia University, 263 p. Rauch, H.W. and Plitnik, M., 1984, Useoflineaments in exploration for ground water in karst terrainofthe Hagerstown Valley, Maryland,inGeologic and Geotech nical Problems in Karstic Limestoneofthe NortheasternU.S.:AssociationofEngineering Geologists and Amer ican SocietyofCivil Engineers,18p. Siddiqui, S.H. and Parizek, R.R., 1971, Hydrogeologic factors influencing well-yields in folded and faulted car bonate rocks in central Pennsylvania:Water Resources Research,v.7,p.1295-1312. Siegel, S., 1956,Nonparametric Statisticsforthe Beha vioral Sciences:McGraw-Hill Book Co., New York, NY, p. 116-127.

PAGE 239

Appalachian Karst Symposium. 1991On Calculating the RiskofSinkhole CollapseBarryF.BeckFlorida Sinkhole Research Institute UniversityofCentral Florida Orlando, Florida 32816ABSTRACTIt is often important to be able to calculate the statistical riskofsinkhole collapse for a specific sitefor insurance purposes, for pre-purchase evaluation, and for engineering the developmentofthe site. Most damaging sinkholes that occur today are cover-collapseorcover-subsidence sinkholes: the resultsofthe downward movementofunconsolidated mantling sediment into dissolved voids in the karstic host-rock.Theprocessofsinkhole development is continual and often repeated. Individual eventsofcover collapseorsubsidence are stages in the fonnationofthe larger landfonn tenned a doline (or simply, a large, ancient sinkhole). It is possible to have a doline-dotted landscape where sinkholes are not collapsing today,orto have sinkholes collapsing (or subsiding) in an area where large surficial dolines are not present. Therefore, the calculationofthe riskofsinkhole collapse must be based on statisticsofrecent eventsofcollapse (or subsidence), not on the numberofdolines that are detectable on topographic mapsoraerial photographs. Obviously this requires the compilationofan exhaustive data base on recent collapse--from local residents, governmental agencies, newspaper reports, insurance reports, consulting engineers' records, etc. Such statistics must be in the fonnofthe numberofcollapses over an intervaloftime over some areaofland:e.g.23 collapses/30 years/25 square miles. In order to use a significant numberof data points in the calculation, the risk factor will usually need to be calculated over a large area--16or25 square miles, for example. Because the numberofsinkholes that develop will vary based on the lithologyofthe soluble rock, the statistical area mustallbe underlain by the same host-rocks. Moreover, the numberofsinkholes that develop will also vary with the surficial geomorphology; therefore, the areaofanalysis must also lie within a single geomorphic setting. When the areaofstatistical recordisselected appropriately, the average numberofcollapses per square mile per year can be calculated. A risk factor for the proposed site may thenbecalculated by multiplying by the areaofthe site and by the numberofyearsofanticipated lifeofthe property.A1,500-square-foot ranch-style home covers0.()()()()5square miles, so the riskofsinkhole collapse affecting an individual home is usually very small. However, a 170 acre industrial park in the same area has a5,OOO-timesgreater risk.Ifthe final risk factor appears to have significant impact on the proposed useofthe site, andifcost-effective engineering modifications are applicable to avoid the damage, then it is pertinent to conduct a geologic/geophysical/geotechnical investigationinorder to delineate the specific areas that are most hazardous. The simplest and most effective methodofmitigating the hazard is to redesigntheproject to avoid the hazardousarea.BeckIntrOductionGeological/geotechnical investigations to evaluate sinkhole hazards are often requested as partofthe planning phaseofa site evaluation. Such investigations may also be required as partofa remedial action,inwhich case the investigation is considerably more focused because damage has already occurred at a specific location. However,inthe planning phase it is necessary to examine large tractsoflandina cost-effective manner. 231Tobe cost effective, it is important that the evalua tion take place in a hierarchical manner.First. it must be determinedifsinkholes are asignificant hazard in this area,for the specific intended useofthe property. Second, cost effective measures that could be taken to avoid the hazard must be identified.Ifthe sinkhole hazard is significant, and if there are appropriate measures by which the hazard may be minimizedoravoided, then, and only then, is it warranted to conduct a detailed, site-specific investigation.Thus, the first step in evaluating the riskofsinkhole

PAGE 240

Beck damage at a site plannedfordevelopment, is a statistical analysisofthe riskofsinkhole collapse.Almost All SinkholesthatCollapseandCause Damage Today Are Subsidence Sinkholes,EitherCover CollapseorCover SubsidenceGeologically, a sinkholeisa closed depression in the land surface, a basin; generallyofmoderate dimensions, measured in ten's and even hundred'soffeet in diameter. It functions as a drainage basin, funneling water into the sub surface.Itowes its origin to the fact that the underlying bedrock dissolves in groundwater more rapidly than most rocks. Most sinkholes are formed on limestoneordolo stone, although a few less common rocks are also soluble. Although there are generally fourorfive different identified typesofsinkholes (Beck, 1988; White, 1988; Culshaw and Waltham, 1987), these are the resultofonly two different processes: the transportofsurficial material downward along solutionally-enlarged channels, or collapseofthe rock roofs over large bedrock cavities (Figure1).Williams (1985) explains how deep solution pipes develop as master drainage channels connecting the shal low epikarstic zone to the deeper true karst groundwater.Ina terrane where limestoneisexposed at the ground sur face, water flows over the limestonesurface toward these dissolved "pipes" and then downward. Because water con verges on these pipes, the limestone around them is more rapidly removed by solutional 'corrossion and erosion, resulting in a bowl-shaped depression.This is the classicsolution sinkhole(Figure IA).Itdevelops imperceptibly and is not generally an engineering hazard except for the fact that additional voids may underlie the solution pipe. Insoluble residue from the limestone may be left be hind on the surface as a thin soil that tends to accumulate in the bottomofthese sinkholes.Ifthe residual soil becomes thick enough, the terrane is termed a subsoilkarstHowever,ifthe limestone is mantled by sedimentsofan outside origin (for example, marine sands or glacial drift), the terrane is termed a mantledkarstWith respecttoprocesses that form sinkholes, the two situations are the same. The overlying, unconsolidated sediment may sim ply be termed "cover" and the resulting sinkholes are cover-collapse or cover-subsidence sinkholes(seeFigure 1 for examples). Conditions within the limestone are still the same as those described above, but the basin and the solution pipe that drains it are covered and in filled with loose sediment Precipitation will now infiltrate through the sediment and seep downtothe limestone surface. There it will mi grate along the bedrock surface to the solution pipe that drains the area, and through that, deeper into the limestone aquifer. The loose sediment directly above the solution pipe may gradually erode and be transported down the pipe, 232Appalachian Karst Symposium, 1991with the aidofinfiltrating water, leaving a soil cavity over the pipe.Ifthe sediment is somewhat cohesive, this void may grow larger and larger over time. More strata within the overburden sediment may cause the cavIty to grow laterally, with a flat roof. Eventually the upward growthofthe void may leave only a thin roofofsoil that is not strong enough to support its own weight, and the ground surface collapses. At that moment, a large, gaping sinkhole suddenly appears:a cover-collapse sinkhole(Figure IF). The limestone and the pipe may be visible in the bottomofthe hole. On the other hand,ifthe mantling sedimentisrela tively non-cohesive, the soil cavity may erode upward rapidly, without growing very large. In fact, as the roofofthe cavity "crumbles" and is deposited on the floorofthe cavity, the cavity simply migrates upward without grow ing larger, like a bubble rising through a liquid. When this reaches the surface, a small hole suddenly appears; this is also a cover-collapse sinkhole. In this case all that is visible in the holeisloose overburden sediment. The engineering implications and hazards to man's structures resulting from cover-collapse sinkholes are obvious and may be catastrophic.Itis important to understand that the positionofthe solution pipe and the processes Qperating in the subsurface have not changed and that this erosion process will continue in the future, with further potential collapses. However, the time frame is variable, sporadic, and unpredictable.Ifthe sediment overlying the limestone is very loose, the processofdownward erosion and surface subsidence may take place gradually and continuously.Assand is transported down the pipe grain by grain, sand from above settles down to take its place. A roofed cavity is never formed, but the sediment is slowly eroded from beneath the ground surface causing a slow, gradual settling. The rateofsubsidence may be only inches per year, but it is still a sinkhole:a cover-subsidence sinkhole(Figure IE). Such sinkholes can cause crackingofrigid structures built over the subsiding areaand have caused buildings, including homes,tobecondemned and declared a total loss. Such gradual sinkholedevelopment may also occur where a clay stratum overlies limestone and is below the water table. As support is eroded from beneath the clay by erosion down the solution pipe, the clay becomes looser and may absorb more water. This loose, very plastic, water-saturated clay may gradually "sag" under the weightofthe overlying sediment or structure. Because the basic causeofthis subsidence is erosionofmaterial downward into the dissolved voids in the limestone, this is a sink hole. Because the cover sediment gradually subsides, this tooisa cover-subsidence sinkhole. The sudden failureofa cover-collapse sinkhole creates a bowl-shaped, funnel-shaped, or cylindrical depressioninthe ground surface. However, after the sudden development

PAGE 241

Appalachian Karst Symposium.1991ofthis basin, natural processes will begin to fill it. The sides will erode inward creating a broader, shallower depres sion. The depressed area maybewet and swampy, possi bly even forming a small lake, where organic growth will infill the basinovertime. These processes, and others, may infill the sinkhole completely, eventually leaving no signofits previous existence. Such a feature is now termed aburied sinkhole(Figure 1D).1 A: SOLUTION SINKHOLE 1 C: SUBSIDENCE SINKHOLEBeckBuried sinkholes have serious engineering implica tions. Obviously, the original sinkhole is subject to reac tivationifthe sediment fIlling the solution pipe is eroded further down into the limestone. That is, it may collapse again. The continued, repeated collapseofa cover-collapse sinkhole will sporadically, but inexorably, erode surface1 B: CAVE COLLAPSE SINKHOLE 1 D: BURIED SINKHOLEnon-cohesive..........cover......sedIment .............'".................................".'..-"..,. ......,". 0' ........'1E (above):Lefttoright,stagesinslowdevelopmentofcoversubsidencesinkhole. 1F (below):Lefttoright,stagesindevelopmentofcovercollapse sinkhole.-:.coh;slve ---: cover ---------sedlment-----Figure1:Typesofsinkholes. E and F are subtypesofC. The processesoforigin are discussedinthe text. (From Culshaw and Waltham, 1987.) 233

PAGE 242

Beck Appalachian Karst Symposium. 1991Figure2:Numberofnew sinkholes per year in Orange andSeminole counties, Florida, for the years 1961-1986. (damaging cover-subsidence sinkholes should also be in cludedinthe data base). It is also critical that the data basenotcontain extraneous entries so old that the data collec tion cannot be complete for that time. For risk-analysis purposes, it is only valid to use data for that periodoftimeoverwhich the data collection appears tobeuniformly effective. However, when plot ting the geographic or geologic distributionofsinkholes, the older data is still useful, although it must be recog nized that it is probably biased toward the larger sinkholes because these are the most noteworthy and memorable.4131111'9Periodofmostcompleterecords366545 33 8n=1294Frequency based on general public records :l 20 1---------------17.7=--c en 15 Z 'C41-g 10 Q.41II:'0.. 5 41.Q E :::3 ZFigure 2 shows FSRI data on sinkhole collapses in the Orlando, Florida, area (Wilson and others, 1987). From 1981 to 1986, there was an averageof10.8 sinkhole collapsesperyear. However, from 1974 through 1986, the average is only 5.8 sinkholesperyear. The data prior to 1981 is apparently much less complete, even though an effort was madetodiscover all old reports from newspaper files, insurance records, engineers' reports, govemment agencies' reports, etc. Ketelle and others (1988) and Beck and others (1991) found similar results. Even when an exhaustive effortismade to discover past data on sinkhole collapses, the memory and recordsofthe general public only appear to be reliable for threetofive years. Obvious ly the best methodofcompiling a valid data base is a con tinuing data-collection effort over many years. It is important to note thatmaporair-photo data on the locationsoflarge. ancient sinkholes is not appropriateforthis data base.First, there is no yearofcollapse asso ciated with these features, so they cannot be usedina tem poral analysis. Further, as pointed out in the discussion above, these large basins are probably the resultofrepeated collapses over many hundreds and thousandsofyears. All fourofthe aforementioned typesofsinkholessolution, cover collapse, cover subsidence, and buried--are the resultofdownward movementofmaterial through the limestone along solutionally enlarged channels. However, dissolutionofthe limestone may also produce cavesatdepth. These processes may result in a situation where a large cave is close to the limestone surface with only a thin rock roof over the void.Ifthis rock roof suddenly collapses into the cave, abedrock-collapse or cave-collapse sinkholeis formed (Figure IB). However, White (1988,p.359) states"because the rateofsurface down wasting is extremely slow on a human time scale, [bedrock] collapse dolines are rarely observed in the actofcollapsing." (Doline is the technical synonym for sinkhole.) Waltham (1989, p. 17) says that bedrock-collapse sinkholes are "rare, except on a geologic time scale." The Florida Sink hole Research Institute (FSRI) has reports on more than 1,700 sinkholes that have developed in Florida in approxi mately the last 25 years, and not oneofthese was definite ly a bedrock-collapse sinkhole.Allofthem were cover collapseorcover-subsidence sinkholes.A statistical risk assessment for sinkhole damage, either by collapse or by subsidence, must be conducted specifically for the site under evaluation. It is critical that the data base used for this evaluation be appropriate for the intended purpose. The data base should contain pertinent information on all, or as many as possible of, the sink holes that havedeveloped recently. In order to establish an appropriate time frame,itis necessary that the data base include the dateofcollapse, oratleast the yearofcollapse Cover-collapseandcover-subsidence sinkholes are together termed "subsidence sinkholes," because the basic modeoforigin is subsidenceofthe overburden, whether slow or rapid. Engineers have termed this upward erosion process"ravelling" and refer to these sinkholes as ravelling sinkholes.Almost allofthe damagetoman's structuresbysinkholes is from the subsidence variety.To Calculate the RiskofSinkhole CollapseItIsFirstNecessary to Have a Valid Data Base on Modern Collapsessediments. The sinkhole cannot get significantly deeper, because the slowly dissolving limestone underlies the bot tom, but it will gradually grow wider as material from the side slopesiseroded into the basin and then down the pipe into the subsurface. Thus, the very large sinkhole basins that we see today on a mantled karst landscape are probably not the resultofa single collapse, but ratherofa slow, sporadic erosionofthe ground surface over thousandsofyears. As these sinkholes grow they may eventually merge due to cover erosion, forming even larger, irregular, compound basins.Ofcourse, the long-term landforms are not pertinent toourdiscussion herein. However,it is criticaltounderstand that these large sinkhole basins develop through repeated collapse, on the same feature.234

PAGE 243

Appalachian Karst Symposium, 1991Moreover. several recent investigations have shown that theselarge, ancient sinkholes cannot even be used as indicatorsofareas where sinkholes are now likelytoform."In areasofcovered karst. ... the distributionsofancient sinkholesdonotpredictmodemsinks" (Upchurch and Littlefield. 1987.p.363).Whetherthis is the resultofchanging conditions--suchasrainfall.sealevel.orgroundwaterlevel--orsimplybecausethecollapsesaresoinfrequent that they arenotnoted during the normal. brief (circa twenty-five years) periodofrecord. is a moot point.In mantled karst ....newsinkholes are appearing most commonly in the areas thatdonot correspond with areasofhigh old sinkhole density"(Bahtijarevic. 1989, p.80).Forthe purposeoflong-term planning, such as for the design lifeofa is important that the sinkhole data base encompass a sufficient time span to include climatic extremesofrainfall and drought.These extremes mayactas a "threshold" to totally modify the sinkhole incidence in anarea.Forexample, in the vicinityofChiefland on the cen tral western coastofFlorida, the data base compiled by the FSRI contained records on less than ten sinkholes from the late 1960's through 1990. This is probably somewhat low due to incomplete reporting. In late February. 1991 the Chiefland area received more than 10 inchesofrain in one weekend.Atthat time. more than a hundred sinkholes collapsed in the immediate Chiefland area. Obviously, the incidenceofsinkholes calculated for the Chiefland area asofDecember. 1990, would have been far different than it wouldbeifcalculated now.Inwestern-coastal Florida, annual rainfall was unusu ally high during the late 1950's--significantly greater than it has been during the period for which the FSRI has been collecting data on sinkhole collapses. Furthermore,justprior to this time. in the mid-1950's, a three-year drought occurred, ending with the driestyearsince 1900.Therecords on sinkhole collapse compiled by the FSRI are notvalidforeitheroftheseclimaticextremes,yetthe possibilityofsuch conditions occurring again is obviously significant. Thus. even though the data that have been compiled in the Institute's data base are the most complete available and as thorough as professionally possible, they may notbetotally indicativeoflong-term sinkhole riskinthe area.Forthis reason, andotherreasons discussed previously,statistics on sinkhole collapses must generallybeaccepted only as a minimum.RiskofSinkholeOccurrenceIs a FunctionoftheNumberofSinkholesthatDevelopPerUnit AreaPerUnit TimeA valid data base on recent sinkhole occurrence, such as discussed above, may be used to calculate the minimum occurrence rateofsinkholes. However. the risk factor 235Beckmust be calculated in a valid fashion in order for this tobeapplicabletoa specific site being evaluated. Ketelle and others (1988), studying sinkhole collapses in eastern Tennessee, found a maximum occurrence rateof1.69 sinkholes/mi 2/yearin Hamblin County. However,themorecommonlyencountered "high" incidence raterangedbetween 0.1 and 0.7 sinkholes/mi2/year. In the Orlando, Florida, area, Wilsonandothers (1987) found the highest sinkhole rate tobeapproximately 1 sinkhole/mi2 /25 yearsor0.04 sinkholes!mi2/year. However,asnoted previously, the older partofthe data base was inapplicable and the Orlando rate maybeapproximately 0.1 sinkholes! mi2/year,basedon the more up-to-date partofthe data. In ordertohave a statistically valid estimateofsink hole-occurrence rate. a significantnumberofsinkholes mustbecounted. Assumingatleast ten. and preferably more, sinkholes and an occurrence rateof0.1sinkholes/ mi2/year, it is suggested thatthe productofthe area over which the sinkholes are counted (in square miles) times the numberofyearsofrecord, should be at leastJOO,as a general ruleofthumb. Obviously, most sites for developmentare only fractionsofa square mile. Therefore, in order to collect statistically valid data, it is necessary to select a larger study areaofseveral square miles surround ing the site. that willbepresumed toberepresentativeofthe site. However, a larger sampling area cannotbeselected at random, for example by drawing a radial distance around the site.The statistical sampling area mustbegeological ly consistent, i.e.the bedrock geology underlying the site mustbesimilar for the entire sampling area. Ketelle and others (1988) show that there are distinctly different ratesofsinkhole occurrence in karst developed on different lime stone formations. In eastern Tennessee, the Conasauga Group has a high rateof0.29, the Knox Group 0.13, and theChickamaugaGroup0.04. sinkholes/mi 2 /year. It seems logical that geological consistency mustbestringentto be valid.Forexample, a given formation may contain a carbonate member and a shaley member. Ob viously, the incidenceofsinkhole occurrence will be differ ent for each. Furthermore,the statistical sampling area must be geomorphically and hydrologically consistent.Wilson and others (1987) found that the occurrenceofsinkholes in the Orlando, Florida area correlated with geomorphic ridges, that were also high recharge sites.Theentire region is underlainbysimilar bedrock geology. Therefore,inthe Orlando region, a statistical sampling area shouldbeeither all locatedonthe ridges,orentirely on the inter-ridge lowlands. A sampling area which spans both provinces would average "apples and oranges" and would arriveata meaninglessresultOther geomorphic and hydrogeologic constraints may modify the occurrence rateofsinkholes and these mustbeevaluated carefully before selecting a sampling area applicable to the site under evaluation.

PAGE 244

BeckSelectionofanAcceptable LevelofRiskMustBeTailoredtotheIntendedUseofthePropertyandtotheCostofA voidingPotentialDamageNote that the impactofa sinkhole on a large shop ping centeroran industrial park maybeminimal. As a first consideration, the oddsarethat it will only impact the parking area, where the chances for a serious accident are minimal. Secondly, assuming no unusual human tragedy, the cost incurred in repairs and lost revenue will probably be minor compared to the overallcostofthe project. However,ifthe site is tobeused for an airport,ora nurs ing home, the impact could be far different. A sinkholeinan airport runway could cause a major tragedy, and the extensive areaofmajor airport runways raises the oddsofoccurrence.Ona nursing home property, a sinkhole would probably not cause extensive damageper se,but the lossofconfidence from the residents and their families could cause major lossofrevenue.Ifthe statistics on sinkhole collapse indicate that the site being considered has an unacceptable risk,itis impor tant to evaluate whether preventative measuresarepossible and what the cost will be, before investing in a site speci fic survey to pinpoint areasofpotential sinkhole activity.One must balance the costofsite investigation plus the costofengineering modifications to avoid sinkhole dam age, against the potential cost and impactofthe sinkhole.Ifthe odds are I: 100 that a sinkhole will collapse on site andifthe potential damage is $50,000, then it is only justified to spend $500 to avoid the damage. The proba bilityofsinkhole damage to a 2,OOO-square-foot suburban home in northern Pinellas County, Florida, is 1:5,555 (Beck and others, 1991). Therefore, it is only warranted toinvest$18.00toavoidcompletedestructionofthe $100,000 home, assuming a 50-year planned life. In such cases, insurance is the only practical solution; engineering modifications are simply not warranted. However, as the sizeofthe site being developed increases and the potential costofdamage increases, it may become cost effective to attempt to avoid sinkhole damagea priori.This is particularly true in the planning phaseofthe developmentoflarge sites, which may have significant oddsofa sinkhole collapsing somewhere on the site.The most cost-effective methodofavoiding a specific sinkhole hazard is simplytoresite the structure that is at risk.This assumes a geologic investigation has identified a specifi cally hazardous area. Plans should place a building away from a hazardous area, that may instead be used for a park ing lotordecorative landscaping.Ifit is not possible to redesign plans to avoid a hazardous site, there are obviouslymany different ways in which engineers could render a structure "sinkhole proof'--everything from deep pilings to a rigid foundation that would bridge over any sinkhole void. However, thecostofthe modifications mustbebalanced against the riskofsinkhole occurrence and poten236Appalachian Karst Symposium. 1991tial costofdamages.ReferencesBahtijarevic, Aida, 1989, Sinkhole densityofthe Forest City Quadrangle,inBeck, B.F. (editor),Engineering and Environmental ImpactsofSinkholes and Karst: Pro ceedingsofthe Third Multidisciplinary Conference on Sinkholes and the Engineering and Environmental Im pactsofKarst. St Petersburg Beach. Florida,2-4Octo ber1989: A.A. Balkema, Rotterdam and Brookfield, p. 75-82. Beck, B.F., 1988, Environmental and engineering effectsofsinkholesnTheprocessesbehindthe problems:Environmental Geology and Water Science,v. 12, p. 71-78. Beck, B.F.; Estes, E.; and Sayed, S., 1991, The sinkhole hazard in Pinellas County: A geologic summary for planning purposes:Florida Sinkhole Research Institute. UniversityofCentral Florida, Report 90-91-1,55p.Culshaw, M.G. and Waltham, A.C., 1987, Natural and artificialcavitiesasgroundengineeringhazards:Quarterly JournalofEngineering Geology,v. 20, p. 139-150. Ketelle, R.H.; Newton, J.G.; and Tanner, J.M., 1988, Karst subsidence in East Tennessee,inProceedings 2nd Conference on Environmental Problems inKarst Terranes and Their Solutions:National Water Well Association, Dublin, Ohio, p. 51-65. Upchurch, S.B. and Littlefield, J.R. Jr., 1987, Evaluationofdata for sinkhole-development risk models,inBeck, B.F. and Wilson,WL(editors),Karst Hydrogeology: Engineering and Environmental Applications: Proceed ingsofthe Second Multidisciplinary Conference on Sinkholesandthe Environmental ImpactsofKarst. Orlando. Florida,9-11February1987: A.A. Balkema, Rotterdam and Boston, p. 359-364. Waltham, A.C., 1989,GroundSubsidence:Blackie, London, 202 p. White, W.B., 1988,GeomorphologyandHydrologyofKarst Terrains:Oxford University Press, Oxford, 464 p. Williams, P.W., 1985, Subcutaneous hydrology and the developmentofdoline and cockpit karst:Zeitschrift fUr Geomorphologie.N.F.,v. 29, p. 463-482. Wilson, W.L.; McDonald, K.M.; Barfus, B.L.; and Beck, B.F., 1987, Hydrogeologic factors associated withrecentsinkholedevelopmentin theOrlandoarea, Florida:Florida Sinkhole Research Institute, UniversityofCentral Florida. Report87-88-4, 109 p.

PAGE 245

Appalachian Karst Symposium. 1991 Mata and HaynesA Suggested Strategy for Characterizing the Hydrogeologic RegimeofKarst Terranes in the Valley and Ridge ProvinceMata, Larryl and Haynes, John T.21 The Bionetics Corporation P.O. Box 1575, EPA-EPIC Warrenton, VA 22186 2NUS Corporation P.O. Box 6032 910 Clopper Rd. Gaithersburg, MD 20877-0962ABSTRACTThe Valley and Ridge Province is relatively closetothe East Coast Megalopolis and many areasofit are coming under increased pressure from land developers. Growth will increase the likelihoodofmishaps involving hazardous wastes and hazardous materials, and will also increase the need for municipal solid waste landfIlls, sewage-treatment systems, and other solid-wastehandling systems. The abilitytocharac terize the hydrogeologyofa site planned for development is important, and for some typesoffacilities, such as most generators or handlersofhazardous wastes, hydrogeologic studies are required under RCRA regulations. An understandingofgroundwater flow paths and flow rates is critical in the early detectionofreleased contaminants. This paper discusses a strategy for characterizing the hydrogeologic regimeofkarst terranes. The authorscaLIfor a comprehensive strategy that incorporates the best available combinationofinvestigative tools.IntrOductionKarst terranes in the Valley and Ridge Province aremorevulnerable to contamination than manyothergroundwater systems because weathering has produced so lutionally enlarged conduitsincarbonate rock along which movementofcontaminants occurs very rapidly and effici ently. Many karst terranes in the region that are under the most pressure from developers primarilyinthe Great Valley, developed on the Cambrian-Ordovician carbonate sequence are characterized primarily by a diffuse-flow karstic aquifer. In such terranes, traditional strategies for groundwater monitoring, based on flow through granular media, are inappropriate. At the same time, however, a monitoring strategy based solely on tracer studies, as advocated by some geologists, may also be inappropriate because tracing materials usually become dispersed. A tra cer such as dye may take weeks or months to be recovered, and then it may turnupat random, at all, some, or noneofthe monitoring stations. Thus, studies using only this method for establishing flow routes are often inconclusive at best. 237 Owing to the diversityofflow regimesinthe region, itisour opinion that a comprehensive groundwater moni toring strategy should be used to determine the degreetowhich a given carbonate aquifer exhibits the characteristicsofa karst aquifer. Such a strategy would involve some combinationofremote-sensing techniques (both photo geologic and geophysical), groundwater tracing studies combined with cave mapping, and geochemical analysesofkarst springs.DiscussionDuring the initial phaseofa characterization process, for any carbonate aquifer, photogeologic techniques and field surveying should be usedtoproduce a detailed map of sourcesofrecharge and pointsofdischarge. At this phase it should be obviousifthese features are readily visible(e.g.sinkholes/swallets, sinking streams, or springs) or if such features are muted or completely maskedbyalluvium or colluvium. Once the map has been prepared, it can be used for planningofgeophysical investigations, geochem ical analysesofspring waters, tracer studies or a combina tionofthese. The most important pointisthat,inmany

PAGE 246

Mata and Haynescases, nosingleinvestigative technique will be suitable for characterizing the full rangeofpossible conditions(i.e.the spectrum from diffuse flow to conduit flow) that may be present within a given carbonate aquifer.Forexample, in carbonate aquifers that are overlain by thick (10 morgreater) sequencesofalluvialorcolluvi al materials, the water table is commonly above the upper surfaceofthe bedrock. Such aquifers occur in many areas along the western slopeofthe Blue Ridge Mountains in Virginia, but particularly from Buena Vista north to Front Royal, where colluviumandalluvium derived from the mass wastingofthe Blue Ridge massif cover large areas along the eastern edgeofthe Shenandoah Valley that are underlain by carbonate rocks (cf. Gathright and others, 1978a,b). Becauseitis commonly the hydrogeologyofthe overburden rather thanofthe carbonate bedrock that de termines how fast groundwater travels through the subsur face in such aquifers, the flow will normallybediffuse. Thus, even though a carbonate aquifer is presentatdepth, the carbonate strata are almost always completely within the phreatic zone and the carbonate aquifer has littleorno influence on the hydrogeologic behaviorofthe entire aqui fer. In fact, because the overburden is a granular medium, the aquifer may be more readily characterized using tradi tional conceptsofDarcian flow. It is very important to realize that tracer studies(e.g.Quinlan and Ewers, 1985) may beofonly limited value. A monitoring strategy based solely on tracer studies is applicable only in areas with carbonate aquifers that have little overburden and that are characterized primarily by conduit flow in the vadose zone. In such aquiferstracer recovery usually occurs in a few days to,atmost, three to four weeks. In areas where carbonate aquifers are overlain by relatively great thick nessesofoverburden, however, tracer studies alone will commonly produce results that cannotbeinterpreted with out much ambiguity owing to extended recovery times for tracers (several months to years) and high ratesofdisper sion and diffusion that occur in such aquifers. In these combination overburden/carbonate-rock aqui fers,the useofspring-water geochemistry to characterize the aquifer (Shuster and White, 1971) will probably also beofless value as an investigative tool. This is because the residence timeofthe groundwater is relativelygreat, allowing for greater geochemical interaction, notjustbe tween the groundwater and the carbonate rock, but between the groundwater and sediments in the overburden. As a result, even though large springs can and do occur locally, they are most likely discharging water that has traveled primarily through the overburden and has had littleorno contact with the carbonate bedrock. Therefore, in the contextofthe investigative methodsofShuster and White (1971), it may notbepossible to interpret the results ob tained from analysisofspring waters without ambiguity. Certain geophysical methods(e.g.ground-penetrating radar, seismic refractionorreflection) can be helpful in locating subsurface voids and in determining the thickness 238Appalachian Karst Symposium, 1991ofthe overburden in such areas without resorting to drill ing, the most costlyofall investigative techniques even though it provides a direct "look"atthe subsurface. Re sultsofgeophysical surveys mustbeinterpreted carefully, because the results will quite commonlybeambiguous (Greenfield, 1979). As mentioned above, photogeologic techniques such as aerial-photo interpretation canbeeffective tools for identifying geomorphic features that may aid in investiga tionofthe hydrogeologic regimeofa carbonate aquifer. Fracture-trace analysis, a typeofaerial photographic anal ysis, can be very helpful in identifying potential pathwaysofgroundwater migrationincarbonate aquifers (Mata and others, 1986).Notethat with all remote-sensing tech niques it is imperative that the findings be field checked.ConclusionsWith the useofa comprehensive strategy that takes advantageofall available investigative methods and tools, it is possible to gain a better understandingofthe hydro geologic characterofa given carbonate aquifer, no matter whether that aquifer is a "classic" karst aquifer( that is overlain by minimal overburden and is characterized principally byconduitflow in the vadose zone),orwhether the aquifer is buried beneath relatively great thick nessesofoverburden and is a diffuse-flow aquifer that is primarily in the phreatic zone. Making this determination is critical in the choiceofan appropriate groundwater monitoring strategyatany regulated waste-handling facility located above a carbonate aquifer in the Valley and Ridge Province, but especiallyinthose areas where devel opment is occurring most rapidly.ReferencesGathright, T.M., II; Henika, W.S.; and Sullivan, J.L., III, 1978a, Geologyofthe Grottoes quadrangle, Vir ginia:Virginia DivisionofMineral Resources Publica tion 10,text and 1:24,000 scale map. Gathright,T.M., II; Henika, W.S.; and Sullivan, J.L., III, 1978b, Geologyofthe Crimora quadrangle, Vir ginia:Virginia DivisionofMineral Resources Publica tion13,text and 1:24,000 scale map. Greenfield, R.J., 1979, Reviewofgeophysical approaches to the detectionofkarst:Bulletinofthe AssociationofEngineering Geologists,v.16, p. 393-408. Mata, L.; Fauss, L.M.; and May, J.P., 1986,Useoffracture trace analysisinkarst terranes for environmental remedial action planning -A case study,in Proceedingsofthe Environmental Problems in Karst Terranes and Their Solutions Conference. Bowling Green, Kentucky:National Water Well Association, Dublin, Ohio, p. 263-277.

PAGE 247

Appalachian Karst Symposium, 1991Quinlan, J.F. and Ewers, R.O., 1985, Ground water flow in limestone terranes: Strategy, rationale, and procedure for reliable, efficient monitoringofground water quality in karst areas,in Proceedingsofthe 5th National Symposium and Exposition on Aquifer Restoration and Ground Water Monitoring, Columbus, Ohio:National 239Mata and HaynesWater Well Association, Worthington, Ohio, p. 197 234. Shuster, E.T. and White, W.B., 1971, Seasonal fluctua tionsinthe chemistryoflimestone springs: A possible means for characterizing carbonate aquifers:JournalofHydrology,v.14, p. 93-128.