USF Libraries
USF Digital Collections

Self-potential response to rainfall changes over plugged and unplugged sinkholes in a covered-karst terrain

MISSING IMAGE

Material Information

Title:
Self-potential response to rainfall changes over plugged and unplugged sinkholes in a covered-karst terrain
Physical Description:
Book
Language:
English
Creator:
Bumpus, Peter
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Collapse conduits
Hydrology
SP
Groundwater
Self-potential
Sinkholes
Dissertations, Academic -- Geology -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: For the protection of wetland and water resources it would be beneficial to understand when collapse conduits function as recharge points to the underlying aquifer. Inexpensive, noninvasive methods to detect recharge are desirable. Previous studies show negative self-potential (SP) anomalies over sinkholes that correspond to the expected electrokinetic effects of groundwater flowing downward through a conduit. SP surveys are less labor-intensive than high-resolution 3D GPR and resistivity, and continuous long-term monitoring is possible. However, before SP surveys can be reliable indicators of flow, SP contributions from ET, conductivity changes, redox reactions, thermoelectric effects, cultural noise, and lateral flow must be understood. A year of continuous SP monitoring was combined with high-resolution 3-D GPR surveys, and intermittent water table monitoring over two small covered-karst sinkholes in Tampa, Florida. Positive and negative SP anomalies episodically manifested over conduits, suggesting that conduit flow is dynamic, not static. Three distinct SP flow regimes in the conduits are postulated: fast flow open to the aquifer, slow flow open to the confining layer through the collapse conduit walls, and a conduit, plugged high enough to behave like the rest of the confining layer. SP responses after rain events also appear to measure the effects of two moving Gaussian wetting front curves, one striking the monitoring electrode, one the reference. viii The wetting front volumes are differently dispersed by traveling different distances. By comparing curve shapes for all possible pairs of electrodes, it may be possible to establish surficial infiltration and flow patterns. Temporal SP response clearly shows SP is also affected by soil conductivity, rainfall history, and cultural noise. Ultimately, SP changes too frequently to rely on measurements many hours or days apart. Over the course of the year, the electrodes became less responsive to rainfall and more erratic. Extremely wet and dry conditions seemed to affect responses. The porous faces of the electrodes or the bentonite clay gel used to enhance contact may decline. It appears a better design for electrodes and electrode contact needs to be developed. To test the intermittent behavior hypothesis, more conduits need to be studied, and moisture and SP must be studied concurrently. Several reference electrodes placed in various topographic, vegetative, geologic, and climatic settings could help distinguish groundwater flow from other SP sources. SP is a valuable research tool; however external complexities such as cultural noise, sinkhole lithology, and the state of the unsaturated zone make SP data difficult to interpret without ancillary information.
Thesis:
Thesis (MS)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Peter Bumpus.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

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:
usfldc doi - E14-SFE0004581
usfldc handle - e14.4581
System ID:
SFS0027896:00001


This item is only available as the following downloads:


Full Text

PAGE 1

i Self Potential Response to Rainfall Changes Over Plugged and Unplugged Sinkholes in a Covered Karst Terrain by Peter B. Bumpus A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Geology College of Arts & Sciences University of South Florida Major Professor: Sarah Kruse, Ph.D. Member: Mark T. Stewart, Ph.D. Member: Mark Rains, Ph.D. Date of Approval: July 8, 2010 Keywords: collapse conduits hydrology, SP, groundwater, se lf potential, sinkholes Copyright 2010, Peter B. Bumpus

PAGE 2

i Acknowledgments Thanks is given for the opportunity to have pursued this degree, to all the gracious and accomplished professors who taught me, and especially to my mentor, Sarah Kruse.

PAGE 3

i Table of Contents List of Tables iii List of Figures i v Abstract vii Chapter 1: Introduction 1 Self Potential 4 Chapter 2: Study Site 9 Hydrostratigraphy 10 Cultural Factors 12 Chapter 3: Methods 13 Ground Penetratin g Radar (GPR) 13 Self Potential 14 Magnetics 16 Terrestrial LiDAR Scanning 16 EM 17 Resistivity 17 Soil Moisture 17 Chapter 4: Results 18 Rainfall Before Data Collection 18 Voltage Response in Electrodes 18 Cultural Noise 18 Natural Diu rnal Variability 19 Buried Metal 19 Response to Rain Events 20 Dry Season: January to May 20 Wet Season 22 Return to Dry Season 25 Conduit Potentials Through Time 25 Chapter 5: Discussion 28 Rain and Flow 28 Initial Spike 29

PAGE 4

ii Non Linear De cay 30 Rise in Potential 31 Soil Conductivity 31 Anomalous Electrode Behavior 32 Heavy Rain 32 Drought 32 Electrode Design 34 Dense Roots 36 Diurnal Changes 36 Vegetation 37 Temperature 38 Evapotranspiration (ET) 39 Flow in Collapse Cond uits 40 Lateral Flow 41 Intermittent Flow 42 Long Term Electrode Behavior 45 Chapter 6: Conclusion s 47 Figures 50 References 112 Appendix A: Tables 115 About the Author End Page

PAGE 5

iii List of Tables Table I: Soil Layers from Auger Buc kets on SP Site 116 Table II: Soil Layers from Rotary Borings on SP Site 117 Table III: Stratigraphy of Floridan Aquifer 118 Table IV: Monthly Average Rainfall for Tampa, Fl 119 Table V: Saturation versus Resistance 120 Table VI: Rain Events Sept 2008 Feb 2010 121 Table VII: Water Tables 125

PAGE 6

iv List of Figures Figure 1: Electrical Triple Layer 51 Figure 2: Study Site 52 Figure 3: Grid Layouts 53 Figure 4: Schematic Cross Section & Depth to Confining Clay 54 Figure 5: North Conduit E W El ectrode Line 55 Figure 6: North Conduit N S Electrode Line 56 Figure 7: South Conduit N S Electrode Line 57 Figure 8: South Conduit E W Electrode Line 58 Figure 9: 250 MHz GPR Vertical Section through North & South Conduits 59 Figure 10: Electrodes Re lative to Roots 60 Figure 11: Petiau Electrodes and Burial 61 Figure 12: Magnetic Anomalies 62 Figure13: LiDAR 63 Figure 14: EM 31 Survey Horizontal Dipole 64 Figure 15: EM 31 Survey Vertical Dipole 65 Figure 16: Soil Resistance 66 Figure 17: Typ ical Electrode Responses 67 Figure 18: SP Maximums and Minimums versus Time of Day 68 Figure 19: Monthly Potential Profiles with Cultural Noise Removed 69

PAGE 7

v Figure 20: Monthly Potential Profiles with Cultural Noise Removed 70 Figure 21: Monthly Potential Profiles with Cultural Noise Removed 71 Figure 22: Soil Potential in High Root Zones 72 Figure 23: January SP Contours through Time (E W Electrode Line) 73 Figure 24: Jan 29 30 Rain Event 74 Figure 25: Feb 2 Rain Event 75 Figure 26: February Rain Eve nts 76 Figure 27: March Rain Events 77 Figure 28: April Rain Events 78 Figure 29: May Rain Events 79 Figure 30: High Frequency Anomaly 80 Figure 31: June Soil Potential Profiles 81 Figure 32: July Soil Potential Profiles 82 Figure 33: August Soil Po tential Profiles 83 Figure 34: September Soil Potential Profiles 84 Figure 35: October Soil Potential Profiles 85 Figure 36: November Soil Potential Profiles 86 Figure 37: December Soil Potential Profiles 87 Figure 38: January SP Contours through Tim e (N S Electrode Line) 88 Figure 39: February SP Contours through Time (N S Electrode Lines) 89 Figure 40: February SP Contours through Time (E W Electrode Lines) 90 Figure 41: March April May SP Contours through Time (N S Electrode Lines) 91 Figure 42 : March April May SP Contours through Time (E W Electrode Lines) 92

PAGE 8

vi Figure 43: June July August SP Contours through Time (E W Electrode Lines) 93 Figure 44: June July August SP Contours through Time (N S Electrode Lines) 94 Figure 45: September October November SP through Time (N S Electrode Lines) 95 Figure 46: September October November SP through Time (E W Electrode Lines) 96 Figure 47: December January February SP through Time (N S Electrode Lines) 97 Figure 48: December January February SP throu gh Time (E W Electrode Lines) 98 Figure 49: Typical SP after Rain Events 99 Figure 50: Typical SP after Rain Events 100 Figure 51: Wetting Fronts 101 Figure 52: Curve Fitting for Potential Trends after Rain Events 102 Figure 53: Modeling SP after Rai n Event 103 Figure 54: Anomalous Electrode Profiles 104 Figure 55: PVC Casing for Electrode 105 Figure 56: May Electrodes 106 Figure 57: Profile after February 2010 Electrodes Replaced 107 Figure 58: Matric Potential 108 Figure59: Schematic of a Cond uit Draining to a Confined Aquifer 109 Figure 60: Schematic of a Plugged Conduit 110 Figure 61: Schematic of an Inactive Conduit 111

PAGE 9

vii Self Potential Response to Rainfall Over Plugged and Unplugged Sinkholes in a Covered Karst Terrain Peter B. Bumpus ABSTRACT For the protection of wetland and water resources it would be beneficial to understand when collapse conduits function as recharge points to the underlying aquifer. Inexpensive, noninvasive methods to detect recharge are desirable. Previou s studies show negative self potential (SP) anomalies over sinkholes that correspond to the expected electrokinetic effects of groundwater flowing downward through a conduit. SP surveys are less labor intensive than high resolution 3D GPR and resistivity, and continuous long term monitoring is possible. However, before SP surveys can be reliable indicators of flow, SP contributions from ET, conductivity changes, redox reactions, thermoelectric effects, cultural noise, and lateral flow must be understood. A year of continuous SP monitoring was combined with high resolution 3 D GPR surveys, and intermittent water table monitoring over two small covered karst sinkholes in Tampa, Florida. Positive and negative SP anomalies episodically manifested over conduits, suggesting that conduit flow is dynamic, not static. Three distinct SP flow regimes in the conduits are postulated: fast flow open to the aquifer, slow flow open to the confining layer through the collapse conduit walls, and a conduit, plugged high enough to behave like the rest of the confining layer. SP responses after rain events also appear to measure the effects of two moving Gaussian wetting front curves, one striking the monitoring electrode, one the reference.

PAGE 10

viii The wetting front volumes are different ly dispersed by traveling different distances. By comparing curve shapes for all possible pairs of electrodes, it may be possible to establish surficial infiltration and flow patterns. Temporal SP response clearly shows SP is also affected by soil conduct ivity, rainfall history, and cultural noise. Ultimately, SP changes too frequently to rely on measurements many hours or days apart. Over the course of the year, the electrodes became less responsive to rainfall and more erratic. Extremely wet and dry cond itions seemed to affect responses. The porous faces of the electrodes or the bentonite clay gel used to enhance contact may decline. It appears a better design for electrodes and electrode contact needs to be developed. To test the intermittent behavior h ypothesis, more conduits need to be studied, and moisture and SP must be studied concurrently. Several reference electrodes placed in various topographic, vegetative, geologic, and climatic settings could help distinguish groundwater flow from other SP sou rces. SP is a valuable research tool; however external complexities such as cultural noise, sinkhole lithology, and the state of the unsaturated zone make SP data difficult to interpret without ancillary information.

PAGE 11

1 Chapter 1: Introduction In the c overed karst terrain of west central Florida, a surficial water table aquifer overlies a layer of sandy clays that derived from altered limestone and aeolian sand. This clay layer serves as a semi confining unit over the Floridan aquifer, a deep limestone dolomite region up to 1000 meters deep. Recharge water dissolves the limestone, creating voids. The conduits formed in the dissolved carbonates fill with the overlying fines and silty sands, and a dissolution channel can evolve downward until it reaches th e Floridan. When the surficial and the Floridan aquifers are connected with high permeability sands, surficial water can be transported to the confined aquifer. By this process, recharge to the Floridan is very local, and 1 to 2% of the land surface can pr ovide most of the recharge to the confined aquifer. Conversely, when the Floridan is pumped, surficial levels are drawn down as the potentiometric surface of the Floridan decreases, pulling water down through the high hydraulic conductivity dissolution pat hways. As a result of pumping, lake levels decline, wetlands shrink or dry up, and reduced water is available to plant cover (Stewart, 1998). Ultimately, it would beneficial for the protection of wetland and other domains to understand the patterns and cir cumstances of flow, as well as to know whether regions of subsidence are leaking. Inexpensive noninvasive methods for determining whether a given sinkhole functions as a recharge point are clearly desirable. Sinkholes themselves are readily

PAGE 12

2 located with g round penetrating radar and resistivity methods in west central Florida. Identifying water flow with non invasive methods is much more difficult. Truss et al. (2007) have successfully mapped water flow in limestone conduits with repeated 3D GPR surveys. Jardani et al. (2006) have shown that s elf potential (SP) can used to recognize sinkholes supporting concentrated flow because strong negative potential anomalies appear over such sites The SP surveys are less labor intensive than high resolution 3D GPR. However, before SP surveys can serve as a reliable indicator of flow at sinkholes, we need to understand the relative influences of other processes on SP measurements in karst. SP reflects not only groundwater flow (electrokinetic effects) but matrix and p ore water conductivities, redox reactions, thermoelectric effects, and cultural noise (e.g. Corwin and Hoover, 1979; Telford et al., 1990). In the covered karst of Florida, in addition to lateral groundwater flow it may be necessary to consider evapotrans piration (ET), water table and mineral surface contributions, or effects from complex flow patterns through irregular karst surfaces that often surround vertical transit conduits. In m ost published studies SP data has been collected intermittently, by mov ing an electrode across a site at different days of the year (Ernstson and Scherer, 1986, Craig, 1991 Jardani et al., 2006 ). Continuous measurements are most commonly reported from within single wells (Suski, 2004). Without both spatial and temporal sampl ing the origin of SP signals may be difficult to discern By deploying SP electrodes in arrays and collecting continuously through the dry and wet seasons, it may be possible to detect signatures that could characterize flow patterns and elucidate causes for those patterns.

PAGE 13

3 This thesis addresses several questions: (1) Are there specific SP responses to rain events that reveal whether a sinkhole is flowing? I s it possible to employ lower cost, non invasive SP techniques to determine whether recharge paths are plugged or unplugged? ) (2) Can we resolve the origin of seasonal variations in SP anomalies previously reported over sinkholes? Craig ( 1991) found that SP anomalies disappear during wetter seasons over active conduits in covered karst in Florida. Why would active conduits stop producing negative self potential (SP) when rain is plentiful? Jardani et al. (2006) also found that SP signals over conduits decreased or disappeared in the dry season in loess over chalk terrain. Is the absence of SP a sign of a no flow condition, or are several facets that cause electrokinetic potential simply canceling? (3) Through continuous recording can evapotranspiration effects, water table and mineral surface contributions, or the effects of complex lateral flow pattern s through irregular karst surfaces be distinguished? To address these questions, continuous SP monitoring was combined with high resolution 3 D GPR surveys and intermittent water table monitoring over two small covered karst sinkholes in Tampa, Florida. T he work described here represents a pilot survey conducted with tools available. Nevertheless, simple SP monitoring suggests that sinkhole flow is a significant contributor to SP measured at the surface, and that flow is much more variable than previously expected. Both positive and negative anomalies are observed at the conduits; we postulate that different flow regimes are triggered by the conduit plugging and unplugging. C hanges in potential did not coincide with times when lateral flow from high stage water would seasonally cancel a negative anomaly. A clear diurnal signal, presumably ET generated, is also present and varies seasonally. Our data suggest that although SP is clearly a valuable research tool, external complexities such as

PAGE 14

4 cultural noise, s inkhole lithology, and the state of the unsaturated zone would make it difficult to use by itself without longer term monitoring as a definitive tool to assess flow through sinkholes. Self Potential Self potential (SP) is a voltage difference between any two points on the Earth's surface caused by currents flowing in the ground. It has been used to detect mineral bodies, to study geothermal systems, especially on volcanoes, and to map seeps or leaks around dams (Revil et al, 2003 and references therein). S P has measured response to the pumping of ground water, and measured the depth to water tables and mineral surfaces (Aubert and Atangana, 1995), and the flow of groundwater (Thony et al, 1997, Doussan et al, 2002). Except for strong redox reactions and aft er correcting for telluric effects, the most prominent signal is the result of moving groundwater (Revil et al., 2003). The voltage caused by groundwater flow, called streaming potential, electrofiltration, or the electrokinetic effect, can be positive or negative, and usually manifests as potential differences of less than 100 mV at the surface. In the case of silicate minerals, the flow of water results in increasing positive potential down gradient as positive ions move with the flow and create a separat ion of charge. As measured at the surface, upward aqueous movement in the ground, such as evapotranspiration, could result in more positive potential (Ernstson and Scherer, 1986, Jardani et al., 2006). On the other hand, over a drain such as a dissolution conduit, where flow moves away from the surface, a negative anomaly could be expected. Streaming potentials are generated in porous media at the water mineral interface where an electrical triple layer (often described as a double layer) extends ~3 to 5  from

PAGE 15

5 the mineral face (Revil et al, 1999 and references within). For silica minerals, the innermost layer is the negative ions that have been created by dissolution at the mineral surface (Figure 1). These ions attract a layer of cations in the fluid, whic h can adsorb to the surface. This positively charged region of sorbed ions adjacent to the mineral surface constitutes the second layer, called the Stern Layer. Negative ions, "the counterions," will naturally be attracted to the positive charge adsorbed t o the mineral face, while positive ions, "the coions," will be repelled, thus leaving a surplus of positive charge farther out in the water. The third layer, where ions are segregated, is called the electrically diffuse or Gouy layer (Revil et al., 1999, F itterman, 1978). Laminar flow is assumed, as is a flat surface, due to the very small scale of these layers. The fluid in the Stern layer at the mineral surface has a flow velocity of zero, but at some distance beyond the Stern layer, in the diffuse layer, the tangential pressure of the flowing pore fluid will drag some of the positively charged fluid of the diffuse layer down gradient. The boundary dividing the mobile from the stationary fluid is referred to as the slipping plane. The electrical potential at this plane is the zeta potential. The fluid beyond the electrically diffuse layer is called the free electrolyte (Revil et al., 2003; Figure 1), which, when moving, displaces unbalanced diffuse charge downstream, creating an electrical field, E k The m ovement of ions depends on the flow velocity, and the flow depends on the pressure gradient. Thus, the electrical field is proportional to the pressure gradient. The constant of proportionality between the pressure gradient and the electrical potential gra dient is called the coupling coefficient, C It depends not only on the zeta potential, but also on the dielectric constant of the pore fluid, the conductivity of the pore fluid, an d the dynamic viscosity of the fluid, (Fitterman, 1978). Thus,

PAGE 16

6 E k = "# $ = C % # P = ( & () ) # P or "# $ # P = C (1) Also, C = "# $ # P = ( %& '( )( F / F 0 ), where F 0 is a formation factor, which is a ra tio between the electrical conductivity of the fluid and the rock, with negligible surface conductivity. F is the formation factor for the fluid under study, which can have surface conductivity (Doussan et al., 2002). Previous researc h has shown that in addition to streaming potential many factors can influence soil potential; not all of them are important at every site. A topographic effect has been observed, apparently as water flows away from the watershed ( Ernstson and Scherer, 198 6 ). In the case of silisticlastic terrain, the highest point will be the most negative, since water and any associated positive ions will flow away from this point. The effect varies with lithology and time after meteoric water influx ( Ernstson and Scherer 1986 ). For example, Ernstson and Scherer found a maximum of 80mV/100 m. Telluric effects, except for the rare magnetic storm, will also be insignificant at small sites, as values tend to vary over a scale of kilometers, not meters. Since SP sources are below the earth's surface, varying ground resistivities could influence a reading at the surface. In the USF Geopark, the sediments above the clay layer are similar enough that resistivity variations in space should not significantly distort potential read ings at the surface. This was confirmed by EM 34 surveys. Over time, however, a change in water content could change resistivity enough to change SP ( Ernstson and Scherer, 1986 ). On a small scale, despite the tortuous clay layer surface, conditions are gen erally homogeneous above the uppermost clay layer, from the electrodes' mostly vertical point of view. Thus, it was assumed that a resistivity change due to saturation change would be laterally consistent throughout the research site.

PAGE 17

7 Other potential sour ces of SP must be considered. In the nineteenth century the first SP measurements were used to locate discrete ore bodies that manifest potential from redox reactions (Corwin and Hoover, 1979, Corwin, 1990). In an urban setting, in Florida, where ore bodie s are unlikely, redox can nevertheless manifest as a result of buried metal objects. Vertical well casings, for example, typically generate a negative SP signal from the upper end of the casing (Corwin, 1990). Buried metal can be discovered with a magnetic survey. In addition, SP potentials of up to 150 mV may be produced by vegetation ( Ernstson and Scherer, 1986 ). SP signals from vegetation may be due to water movement from evapotranspiration or to bioelectric effects (Corwin, 1990). The uppermost sediment s create an unconfined aquifer. The water table in the aquifer will contribute an SP signal at its surface as will the aquifer interfaces (Fournier, 1989). The surface of the water table is like a plane of dipoles, and the strength of these dipoles is prop ortional to the piezometric head (Revil et al., 2003). The SP signal reflects the inverted shape of the water table (Fournier, 1989). In some settings, SP values can be used to calculate the depth to water table using inversion (Fournier, 1989, Revil et al ., 2004). When SP is measured over a pumping well, positive SP manifests over the cone of depression. Similarly, in the absence of a shallow water table, Jardani et al. (2006) showed that outside of some sinkhole regions the SP signal was directly proporti onal to the depth of the sediments over a clay layer. However, over dissolution conduits negative potential is attributed to the vertical downward percolation in the conduit (Ogilvy, 1967, Jardani et al., 2006). Aubert and Atangana (1996) call the interfac e responsible for the SP signals the SPS interface. This surface can be either piezometric or the first less permeable

PAGE 18

8 formation (Jardani et al., 2006 from Aubert and Atangana, 1996). The coupling coefficient is greatly reduced in clay of high surface cond uctivity; as a result, strong SP signals are not produced in clayey aquitards (Revil et al, 2003). Jardani et al. (2006) suggest that positive SP could be generated during times of greater evapotranspiration, such as summer. These evaporation effects coul d counterbalance negative signals from downward percolation and cause SP signals to decrease or disappear (Jardani et al., 2006). Temperature variations can change SP because electrodes are sensitive to temperature change and because resistivity changes wi th temperature. The electrode sensitivity may be reflected in data when a reference electrode is placed where it doesn't receive the same insolation as measuring electrodes. These effects can manifest over periods of minutes or hours, or seasonally over pe riods of months (Corwin, 1990).

PAGE 19

9 Chapter 2: Study Site The study was conducted within the University of South Florida Geopark. The university is located in northern Hillsborough County, Florida, in the Hillsborough River watershed, a region of covered karst terrain. The g eopark consists of about 22 acres on the southwest side of the Tampa campus, bounded to the south by Alumni Drive, to the east by Magnolia Street, to the west by Pine Drive, and to the north by a drainage ditch and retention pond (Figur e 2). The study site covered about 420 sq m in the northeast corner of the park. Within that space, SP grids covered about 150 sq m over 2 known sinkholes: one presumed to be an active drain, the other presumed to be plugged. The USF Geopark is covered wit h grass, and the 2 sinkholes chosen for study are 5 to 10 m away from trees on the east side and 20 to 30 m away on the west. A number of the electrodes at the extremities of the arrays are under tree canopy, especially on the east side of the south array (Figures 2 & 3). The study site is relatively flat with about 1 to 1.5 m elevation change between the highest point 10 m west of the site and the lowest point several meters to the east. Over a small site, less than 12 x 25 m, any topographic SP changes wi ll be insignificant. The study site was chosen because (a) near surface soil parameters were largely known (Tables I & II), (b) well borings and cores had delineated numerous sinks filled

PAGE 20

10 with surficial sands and silt, (c) accessibility permitted low cost monitoring, and (d) the cultural noise is comparable to the noise in the sinkhole laden urban environments where SP could potentially be an investigative tool. Hydrostratigraphy In west central Florida a layer of Miocene and post Miocene silist iclastic de posits covers early Tertiary carbonates. On the site the surface sands are fine and well sorted. Beneath these, silty sands and clay silt sand layers are irregularly placed over a very uneven, dimpled, subterranean clayey layer (Figure 4). These sandy depo sits form the matrix of a surficial water table aquifer that consists of well sorted sands uppermost, then silty sands, then clayey sands, and finally a sandy clay, which is often expressed as a reddish and yellow marl. The well sorted sands may be from 1 to 7 meters deep (Parker, 1991), but were 1 2 m in the study area, as can be seen on GPR images that show the depth to the uppermost clay of red yellow marl, about 1 to 1.5 meters below the surface (Figures 5, 6, 7 & 8). The clay content increases with dep th. The permeability of the cover decreases as water moves downward, as the sediment changes from well sorted sand to silty or clayey sand to sandy clay, then to a clay semi confining layer, directly above the limestone dolostone. The semi confining layer is very dense bluish or greenish clay. The clay and the marl above it are insoluble residues derived from the carbonate layer. The clay layer confines a thick sequence of limestones that comprise the Floridan Aquifer. Immediately below the clay layer is a white, sandy Miocene Tampa Limestone; below this is Oligocene Suwannee limestone; finally, the Eocene Ocala Group overlies the Avon Park Limestone (Table III).

PAGE 21

11 The cover of Qua ternary aeolian deposits and earlier shallow marine deposits mask all but the l argest features of the buried karstic clay limestone dolostone domain. Closed depressions many meters across and small depressions perhaps a m eter across are evident in the g eopark. The topography of the karst surfaces, both the clay layer and the limeston e, is far more irregular than the surface. Water flow is not well integrated and may be tortuous in the lateral direction as surface of the clay layer rises and falls several meters horizontally in a chaotic fashion (Stewart & Parker, 1992). Thus, the perm eability in a lateral direction (Figure 4b) could be both anisotropic and heterogeneous in any locale. Hillsborough County is a semi tropical area that is driest in late fall, winter, and into spring, and wettest during June to September (Table IV). Water may not flow laterally during lower water table stages of the dry seasons because it can be impeded by the irregular height of less permeable sediment (Parker and Stewart, 1992). Thus, recharge water may be limited to vertical flow during lower stages. Wh en the height of the water table rises enough to connect regions of highly permeable, well sorted sands during the rainy months, lateral flow may be preferred (Craig, 1991, Parker and Stewart, 1992). The semi confining layer does not entirely isolate the Floridan Aquifer from the overlying sediments. Overlying, unconsolidated, more permeable sediments move down into karst dissolution cavities and fill them. Collapse conduits are filled mainly with sands and silty sands with less than 1 2% clay (Craig, 1991 ). As soluble carbonate material dissolves, sediments collapse and subside, creating a raveled stratigraphy (Parker and Stewart, 1992). As sinkholes and vertical conduits form, they can completely

PAGE 22

12 perforate the clay confining layers and transport water fro m the surficial aquifer to the confined limestone dolostone Floridan aquifer. With hydraulic conductivities several orders of magnitude greater than the semi confining layer, these columns of surficial sands become important channels for Floridan recharge. Terrain varies widely in Florida's karst country, but in the geopark 1 2% of the surface area is cover collapse conduit (Parker and Stewart, 1992). The percentage of recharge passing to the Floridan is far greater than this: As much as 80 to 90% of the re charge water may move downward through 1 to 3% of the surface that is active conduits (Stewart, 1998), making recharge very localized. Cover collapse conduits are believed to be the most important conveyor of recharge to the Floridan. Cultural Factors The site is located at the far east side of the USF Geopark, 15 m from Magnolia Ave to the east, and 50 m south of Medical Center Drive. Both roads are lined with streetlights that turn on a dusk and off at dawn. A large medical complex is located about 200 meters to the west and 300 m to the north. A residential building is located across Magnolia about 200 m to the southeast. A large culvert is known to run about 15 m to the north of the electrode grid.

PAGE 23

13 Chapter 3: Methods To investigate flow through dis solution columns for this study, time surveys of SP were recorded with a data logger over two previously mapped collapse conduits during wet and dry seasons, and these structures were imaged with high resolution 3 D GPR surveys. A magnetic survey was run t o identify buried metal objects; a LiDAR scan was run to measure s urface topography ; an EM survey measured spatial variations in soils resistance; soil resistance was measured in the laboratory at different saturations. Ground Penetrating Radar (GPR) Three GPR surveys were run. The first, in October, 2008, was used to verify the positions of previously mapped regions of collapse, one presumed an active drain and one presumed plugged. A 3 D data cube was collected with a ProEx system from Mala Geoscience, In c. using a 250 MHz antenna (Figures 2 & 9). Lines were spaced every 10 cm on a 12 by 35 m plot with a trace interval of 0.05 m. At this time, during the dry season, a single SP line was also run over the conduits to confirm that there was a negative anomal y over the north conduit, but not over the south one. In October 2009, following SP electrode installation, two sets of higher resolution GPR data were collected with a 500 MHz antenna. Electrodes were left in the ground, and connecting wires were moved o ff the grid as much as possible. A 10 m x 12 m GPR

PAGE 24

14 survey centered over the north conduit was run with a .05 m line spacing and a trace interval of 0.025 m. A 9 m x 16 m survey over the south conduit was conducted with a line spacing of 0.05 m and a trace interval of 0.025 m. For the south grid, positioning was established by pulling the antenna along a guiding string. For the north, positioning was derived from the combined use of the Mala GPR with a laser positioning system, using the guidance and fusing package from 3DGPR Research, Inc. Positions in both cases should be accurate to less than .025 m. Data were processed using both ReflexW from Sandmeier Software and the 3DGPR Research processing package. Processing steps include a dewow filter, time zero corrections, static shift, gain (uniform across the grid), background removal, and 3 D diffraction stack migration. Self Potential Once the sinkhole locations were secured, 2 SP electrode arrays were laid out centered over the collapse regions (Figures 3 & 9). Initially the grids were crosses with north south and east west lines (Figure 3). The area is an open field, but is surrounded by clusters of trees to the east and the west. The western ends of the electrode lines extended into zones heavily invaded by roots, especially around the south sinkhole (Figure 10). The reference electrode was placed 10 m S SE of the south grid in a shaded area devoid of grass cover, at a topographic low point compared to the surrounding terrain and the grids. The electrodes were fabricated as Petiau second generation Pb/PbCl 2 /KCl electrodes using kaolinite as the absorbent and a porous wooden plug to allow contact with the soil (Petiau, 2000) (Figure 11a). Electrodes of this design are non polarizable and have demonstrated st ability better than 0.1 mV per year (Petiau, 2000; Perrier and

PAGE 25

15 Raj Pant, 2005). The thermal sensitivity is ~20 to 30 !V/¡C (Petiau, 2000). After manufacture, electrodes were placed in a KCl bath; measured potentials were compared and were found to be withi n 2 mV of each other. Electrodes were buried completely below the surface, so the bottoms reached a depth of ~30 4 cm. At this depth SP potentials resulting from varying chemicals and moisture in the topsoil are minimized (Fournier, 1989). This under ground installation permitted GPR antennae to be pulled over them by disconnecting the wire leads, but without having to dislodge the electrodes from the ground. The electrodes were set into KCl salted bentonite gel, to ensure good contact with the earth. Spreading the bentonite gel over the bottom of the hole averaged the potential over a 75 100 sq cm area, instead of the much smaller porous electrode face. The gel also reduces variations due to moisture and improves contact with the ground. Before plantin g, the electrodes were wrapped in plastic bags to protect the taped contact with the wire to the data logger. Once in the ground, the top of the electrode was covered with a capped section of 2 in ID PVC pipe, which keeps sediment away from the contact. Fi nally, the assembly was buried in the original sediment, which was packed to resemble the original sediment density (Figure 11b). The ground level above the electrode was restored to match the surface grade to prevent preferential flow to the electrodes. A Campbell 800 series data logger collected potentials every 2 minutes as an average of the previous four readings collected every 30 seconds. This logger was a high impedance device that could record in the mV range. Collection started on January 10, 2009 and was continuous except for removing and replacing electrodes for some GPR surveys, and replacing a drowned data logger after a 25 year rain.

PAGE 26

16 Magnetics In urban settings, such as the USF Geopark, magnetic surveys can be employed to discover buried metal objects that could influence SP. A magnetic survey was collected over the 2 conduits in March of 2009, and an additional survey was collected October of 2009 west of the sinkholes to cover ground where the first survey suggested a large magnetic anomaly w as located. Data was collected with a Geometrics, Inc. G 858 cesium vapor magnetometer. The March, 2009 survey covered the same 12m x 35 m terrain as the October, 2008 GPR survey; the line spacing was 1 m. The October, 2009 survey started from the 11m poin t to the 36 m point of the earlier survey's westernmost line and extended 18.5 m farther west; line spacing was every one meter. Base station drift was negligible compared to the amplitude of the signals of interest. The discrepancies in magnetic value alo ng the line similar to both surveys was used to normalize the values recorded on 2 different days and create one magnetic map (Figure 12). The source of the magnetic anomaly is unknown, but redox from abandoned construction metal is suspected (verbal commu nication, Mark T. Stewart, January, 2009). Terrestrial LiDAR Scanning A topographical laser tomography survey was conducted in October, 2008 to verify subsidence over the presumed active conduit (Figure 13). The Terrestrial Laser Scanner survey was conduct ed with a phase based FARO Laser Scanner LS880 (terrestrial LiDAR). The LS880 is a dual access compensated high speed laser scanner capable of survey grade accuracy.

PAGE 27

17 EM EM surveys had been done in October, 2008 over the SP grid terrain with an EM 31 hel d at hip height. Both horizontal and vertical dipoles were measured. For the vertical dipole, the ground above 1.8 m depth and the ground below 1.8 m contributed equally to the signal. For the horizontal dipole ground above and below 2.9 m contribute equal ly. Ground near the surface appeared about half as conductive as deeper ground (Figures 14 & 15). Resistivity A soil resistance test was completed with a Campus resistivity meter connected to a 4.05 x 3.65 x 21.90 cm plastic resistivity testing box. Curre nt was supplied to the ends and resistance probes were separated by 14.25 cm on the long side. Unconsolidated gray sand from 30 to 50 cm below the surface, a depth immediately below electrodes, was dried in an oven for 2.5 hours at 80¡C until completely dr y. Water was added in 10 ml increments, and resistivity was measured at each moisture level until saturation was reached (Figure 16 and Table V). Soil porosity was estimated to be 35%. Soil Moisture Soil moistures were recorded from October 2009 to Februa ry 2010 with 3 ECH2O 20 cm "paddle style" moisture sensors and ECH2O data logger. Sensors were installed 3 m north and 5 m west of electrode E12. Data from sensors installed at 33 cm and 94 cm depth are described here.

PAGE 28

18 Chapter 4: Results Rainfall B efore Data Collection Winter is the dry season in Florida; nevertheless, when SP measurements were initiated on Jan 10, 2009, the soil was moist and soft. The previous major rain had occurred on Dec 11, 2008 (1.68 cm), followed by two small rain events on December 27 and January 7, which together produced less than .25 cm of rain. Only 8.48 cm of rain had fallen in the previous 100 days, as measured at the USF Geography Department weather station, located about 1 km from the electrode grids in the USF Geopa rk (rainfall records in Tables IV & VI). This amount of rain is less than half of normal for these 100 days. Voltage Responses in Electrodes Cultural Noise The USF Geopark is in a developed area, so soil potentials recorded by the grid electrodes expressed some changes in potential (relati ve to the reference electrode) that were not related to natural phenomena (We note that all subsequent discussion, unless otherwise stated, "potential" specifically means the grid electrode potential minus the reference electrode potential.) A diurnal, up and down step pattern of 3 to 15 mV in the electrode profiles coincides with streetlights turning on at dusk and off at dawn,

PAGE 29

19 controlled by photoelectric cells. The grid soil potential is raised when the lights are on (Figure 17) In addition, small spikes, about 30 to 70 minutes in duration, simultaneously appear on the voltage profiles of all electrodes. The spikes manifest at irregular intervals 1 to 10 days apart. The abruptness of the spikes' onset suggests they ar e likely caused by electrical noise from an unknown source (Figure 17). Natural Diurnal Variability Although the daytime and nighttime potentials differ in magnitude due to the nighttime electrical field of the streetlights, a natural diurnal source of ch anging soil potential appears to also be present. The potential curve usually reaches a maximum in late afternoon or evening and a minimum at dawn or a few hours later (Figure 18 ). It varies in intensity from day to day and becomes more pronounced into the spring and summer. When the electrical noise is removed, a shallow sine like wave overlays larger changes due to rainfall in January data (Figures 19, 20 & 21). Other episodic anomalies consistently occurred in the 4 electrodes placed close s t to the tree s to the east of both grids (Figure 22). Erratic periods and sudden spikes occur more frequently in the south grid where roots are denser than in the north. Buried Metal A prominent, extensive negative anomaly exists next to the northwest corner of the re search site, and resulting negative potential extends to the region of the north collapse conduit. Magnetic surveys located the source of this negative anomaly, believed to be buried construction metal undergoing redox reactions in the soil (Figures 12a & 12b). Voltage profiles for electrodes on the west end of the E W line of the north grid

PAGE 30

20 demonstrate large, continuous negative values, the highest negativity occurs for electrodes placed closest to the magnetic anomaly (Figure 23a). Response to Rain Event s Dry Season: January to May The beginning of data collection took place under dry season drought conditions. In January the water table was below the level of wells extending to ~5m depth; in February the level was hand measured at 4.74 m below surface; even by June after three weeks of heavy rain the water table was still 4.06 m below surface. S everal fairly consistent trends after rain events were seen, especially for the first five months of the year when the soil was dry, the water table was low, and groundwater flow was presumably limited. First, after larger rainfalls (e.g. Figure 17a, Jan 13 & 29), a sharp increase in potential occurred, which took up to several hours to manifest and reach a maximum. This was a short term response, lasting for sever al hours. Second, after all rain events, even small events where the sharp, short term rises were not evident, the soil potential became increasingly negative for 2 to 5 days (Figure 17a, Jan 20). This decay in potential was non linear. Third, following th is fall in potential, the potential rose at varying rates until the next rain. The first major rain since the middle of September, 2008 occurred on the 29th and 30th of January, 2009, a 2 day rain event of 5.56 cm. The response at all the electrodes was v ery similar, except near heavy root zones. The rain on the 29th produced 1.02 cm rain in 2 batches (Figure 24). No response to the new water ensued for about 30 minutes after rain started; then the potential increased far more dramatically than it did in r esponse to the .5 cm Jan 13 event. The increase was about 55 mV (including a 5 to 10

PAGE 31

21 mV contribution from streetlights) compared to less than 2 mV on the 13th. The rain subsided and restarted five hours later ramping up potential another 10mV (Figure 24). Several hours later, on the 30th, a very large 4.55 cm rain event started that lasted over 10 hours. The change in potential was very small, about 3 mV; moreover, the potential started to decline as after previous rain events, despite continued rain. Befo re the January 29 rain, when the streetlights turned on at dusk, a 5 mV potential increase resulted. When the streetlights turned off at dawn however, the potential decreased by about 20 mV (Figure 24). Surprisingly, the very substantial 4.55 cm Jan 30 rai n produced a change in potential about the same as the .5 cm Jan 13 rain, an order of magnitude less than the Jan 29 rain. The next large rain event occurred on February 2 (Figure 25), only 3 days after the January 30 event. The electrodes responded as ea rlier in January: Soil potential increased 15 to 20 mV, a response similar to the January 29 rain, much greater than the January 30 rain. The typical non linear decrease in potential followed; then, as before, the trend reversed after 4 or 5 days, and the potential trended positive. Although the electrodes were slightly more erratic than in January, about 25 of the 31 electrodes showed a similar profile. As before, most of the anomalous profiles occurred under the tree canopy where roots are densest. The o nly heavy rain occurred on February 2. For the rest of the month, except a .51 cm rain on the 19th, the subsurface was presumably drying. Concurrently, the strength of the electrode response to the streetlight voltage decreased throughout the month, as evi denced by the decreasing voltage response to the streetlights switching on and off as the month progressed (Figure 26).

PAGE 32

22 In March, April, and May the pattern of increasingly negative soil potential 2 to 5 days after rain events continued, as did the subseq uent rise in potential after this initial decay (Figures 27a, 28a & 29a). On the other hand, the short term increase immediately after rain did not manifest. The active conduit electrode, E12, and the electrode 1 m west of it, E5, started to vary extreme ly in March. E12 was remarkably consistent in moving up and down in the same rather large 70 mV potential increment every 2 to 5 days. Over the 2 to 5 day period the potential was higher at first and tended to decrease slightly. This longer period variatio n continued through most of April, but varied daily again at the end of April and during May. The daily potential change remained high however. The soil became much dryer throughout these months as the sun moved higher and the temperature increased. By Ap ril the top 16 20 inches of soil had to be broken with a pickax instead of a shovel; the soil chunks were rock hard and dry. Most of the electrodes exhibited smaller response to the streetlight voltage as soil dried, only 2 to 3 mV or less. In addition som e electrodes became very erratic. Two types of anomalous responses become stronger from March through May. Large amplitude (up to 250 mV), high frequency responses appeared on 5 or 6 electrodes (Figure 30). At the same time, and much more so later in the y ear, higher amplitude, large, rounded maximums at night and spiked minimums (up to 70 mV range) appeared at many electrodes, intermittently in some cases (Figure 29 b & c). Wet Season On May 13, unusually high rainfall of 9.42 cm created a lake that drowne d the data logger. The May 13 rain event was the beginning of rainy season, and it rained 16 of

PAGE 33

23 the 21 days from May 12 when the rains began until the data logger was restarted on June 4, a total of 25.54 cm rain in three weeks, more than the total rain fo r the previous 7.5 months. When data collection resumed, conditions had changed vastly. In June the ground was moist and soft again. Even though the temperature was much higher in summer, the continual rain appeared to have overwhelmed the evaporation that drastically reduced the soil moisture during April and May. A few electrodes were moved from the edges in June to create increased spatial density around the active sinkhole It rained 13 days in June, a total of 15.06 cm. The electrodes became more errat ic, with electrodes E6 and E18 representing the electrode response most typically seen (Figure 31). Most erratic electrodes read higher voltages with much larger diurnal changes. In general these erratic periods appeared and disappeared and persisted at on ly a few electrodes. In June, many rain events large enough to have caused a voltage response in the dry season did not manifest on the graphs, or the response was minimal. The very large 6.58 cm rain on June 26 27 was the only event that triggered the ty pical post rain decline in potential (Figure 31). This large rain was also the only June event that posted the short term increase in potential like those seen earlier in the year. A few days later the substantial June 30 rain of 6.83 cm did not register i n the short or long term (Figure 31a). Another large downpour occurred immediately on July 1, however, and there was an uneven positive response across the electrodes. After this rain the soil potential rose instead of falling at most electrodes, but the n orth conduit location declined (Figure 31). In July the electrode potential was more irregular than in June (Figure 32), and rainfall continued to be very high: 16.61 cm for the month. In general, electrodes did not

PAGE 34

24 react to rainfall. Only a few electrod es slightly reacted to a few larger rain events, such as E9 (Figure 32b). The non linear decline after rain was not evident, nor was the typical ensuing increase in potential a few days later. No short term potential responses were seen during the entire m onth. Only remnants of the sharp steps created by the streetlights remained on most electrodes. Most electrodes manifested inconsistent, large mV, rounded peaks, which had always occurred, but less frequently than in June. The large spike seen on Figure 32 (a, c, d, and e) was caused by disconnecting the north grid electrodes to refurbish the connections in the extremely wet conditions. A shift in potential occurred in some cases when electrodes were replaced. August was similar to July. Many electrodes had unique potential profiles and only weakly manifested changes after rain events (Figure 33). Rainfall events were well reflected in only a few electrodes, such as E9 (Figure 33a). Rainfall was 12.67 cm for the month, with 5.46 cm (43%) of the rain falling in the first 5 days of the month. The large anomaly appearing at the August 5 rain started before the rain and is of unknown origin. No sharp increases in potential in response to rainfall occurred in August like those seen during the first five months of the year. September was a very wet month with 17.75 cm of rain. In September the electrodes tended to have more unique signatures. Reaction to rain increased somewhat after some drier weeks at the end of August. A few electrodes, such as E9 (Figure 34) rev ealed a short term positive response to rain, a non linear decline, and then a slow rise in potential as in the early part of the year. Many electrodes had sudden jumps or switched to large magnitude rounded peaks with sharply pointed minimums (Figures 34)

PAGE 35

25 Return to Dry Season In October the electrodes were reset in new KCl bentonite after a GPR survey. The behavior remained similar to the previous months, but electrodes showed even less response to rain events than in the summer. A jump in potential occu rred at many electrodes when they were replaced (Figure 35). Erratic behavior increased in some electrodes and decreased in others when they were reset. Although far less rain fell in October than September, October was twice as wet as the year before. Nov ember and December followed the erratic October pattern with rounded peaks and pointed minimums appearing at almost every electrode (Figures 36 & 37) as well as vague or minimal reaction to rain events. The moisture sensors installed in October show evide nce of surface water infiltration associated with rainfall events (Figure 51). Moisture content spikes following rainfall and decays over ~2 days (33 cm depth) and ~ 4 days (94 cm depth). Conduit Potentials Through Time The changes described above were relative to the reference electrode, and potential responses were fairly consistent across the grids, including the conduit electrodes, as far as reaction to rain events and cultural factors. To discover whether patterns in conduit flow could be connected to SP measurements, changes in potential over the conduits had to be evaluated relative to surrounding grid electrodes, instead of the reference. To find potential changes unique to the conduits, contours through time of both the N S and E W lines at each grid were created. Relative to surrounding electrodes, the electrode over the north conduit manifested positive potential through most of January. After the substantial rain at the

PAGE 36

26 end of January, the potential increased at north conduit electrode, E12, as elsewhere, relative to the reference. Relative to neighboring grid electrodes, however, E12, had a negative spike, then remained more negative than the rest of the grid (Figures 23a and 38a). This negative anomaly at E12 persisted for the first weeks of F ebruary, but gradually trended toward background levels. This can seen most clearly on the N S electrode line (Figure 39a) because the redox activity to the west and root activity to the east confuse the E W line contour (Figure 40a). In March, intermitten t behavior commenced at E12; potential varied radically sometimes potential was similar to the surrounding electrodes, but at other times it was distinctly positive (Figures 41a & 42a). This alternating potential continued on a 2 to 5 day cycle throughou t April and into May. On May 12, a large 2.87 cm rain resulted in an extremely positive spike for the conduit electrode relative to the surrounding terrain. The spike was short lived, less than a day; then data collection was interrupted for three weeks to replace the data logger. When data collection resumed in June, the potential was still positive over the north conduit, but returned to background levels by the end of August (Figures 43a and 44a). In September and October the north conduit electrode regi stered potential similar to its surrounds; in November it became relatively positive again. After November the electrodes became inconsistent, and clear patterns were not seen (Figure 47a & 48a). Ultimately, the north conduit showed a negative anomaly in 2 009 only during February, as it had the on the previous October reconnaissance SP measurement; most of the year the north conduit manifested a positive anomaly. The electrode over the south conduit exhibited the same potential as the surrounding electrode s during the dry season, January to May (Figures 23b, 38b, 41b, &

PAGE 37

27 42b). This changed dramatically at the 25 year rain in mid May. At this time, the south conduit electrode registered potential more negative than its surrounds, and this continued throughout the rainy season (Figure 43b & 44b), except a period of several weeks at the end of July when potential suddenly became positive. The potential returned to negative relative to background in August and remained so through November except for several weeks in September where it became positive (Figure 45b & 46b). From December to February N S and E W lines were inconsistent with each other in regard to relative potential over the conduit electrodes (Figures 47 & 48).

PAGE 38

28 Chapter 5: Discussion Rain and Flow After a rain, changes in SP could be expected as new water flows past each electrode differently. The electrode responses to rain were distinct in the beginning part of the year. Later in the year, during continual rain in the summer and fall, response to rain was much harder to see. Since changes in soil potential were not proportional to the amount of rain; other factors besides groundwater flow or increase in water table, which would be expected to reflect rainfall amounts in some fashion, were clearly affecting differences in potential between the grid electrodes and the reference. Early in the year, three distinct tendencies resulted from the sporadic, isolated rain events of the dry season: A short lived sharp increase in potential appeared within a f ew hours after rain started, followed by a non linear decay in potential, then a slow increase in potential (Figures 49 and 50). In the dry season, most grid electrodes exhibited similar changes in potential relative to the reference electrode, whereas lat er in the year, greater variability prevailed, and rain events often produced no consistent response. The reference electrode in this study was at a low point in the terrain, hydrologically down gradient from the grids over the conduits. In addition, the reference was under more canopy than most of the grid electrodes. Presumably, the canopy would

PAGE 39

29 shade the reference electrode from rain, at least initial rain, reducing groundwater volume or delaying groundwater arrival at the reference. Most rain events we re short, lasting only a few minutes or a few hours. In simplest terms, rain can be considered an impulse. The impulse solution to the diffusion equation is Gaussian (e.g. Revil et al, 2004). Theoretically, if diffusion shapes the wetting front, the volu me of water passing a point would be a bell shaped curve. Since groundwater moves toward the forward side of the diffusion curve, it is logical that the front side of the bell shaped curve would be compressed. The back side of the diffusion curve could be spread out as diffusion, capillary forces, and friction hold water back, making the curve asymmetrical. Measured wetting front curves can be seen in Doussant et al. (2002) and Byrdina et al. (2003) as well as in geopark data (Figure 51). Both the decay cur ves and the ensuing increase in potential in this study were often fit well with a n inverse quadratic (Figure 52), but no consistent function (exponential, Gaussian) best fit all rain events. Initial Spike Several hours after a significant rain, the pote ntial starts to rise sharply and continues to rise for a number of hours (Figures 24 & 25). This increase in potential is delayed after rain begins, presumably to allow enough time for water to flow from the surface past the bottom of the electrodes. Since some flow is downward, some question arises as to why the potential wouldn't be negative. For streaming potential, an increasing positive potential would reflect a steadily increasing amount of positive ions passing the grid electrodes, more than at the r eference. Two possibilities arise: More groundwater could be flowing laterally past the grid electrodes than the reference, or downward

PAGE 40

30 flowing water could be bringing increasing numbers of positive ions past the electrode from above, due to the passing of the increasing volume. The flow at the grid electrodes must have a larger amplitude than flow at the reference to produce the positive potential spike (Figure 53, compare b and c with a). Non Linear Decay After the sharp rise, the potential at the grid electrodes started to decay toward more negative values for 2 to 5 days. Increasingly negative potential could be downward flow, but it could also be the result of diffused wetting fronts reaching the grid electrodes before a wetting front reached the shad ed, topographically lower reference electrode at the bottom of the slope. As the volume of flow at the reference and grid electrodes equalized over time, the difference in SP would decrease; the decay curve reflects the difference in the two Gaussian diffu sion curves. Diffusion curves could have different amplitudes, or they could be differently dispersed. Similar peaks and decay were found in Doussan et al. (2002) after rain events. That study measured the difference between 2 electrodes, one placed abov e the other. In that case it would have been a descending wetting front, not a lateral flow, that reached the upper electrode before the lower. In the geopark, there is no way to distinguish between downward or lateral flow without wells or moisture sensor s near the grids and the reference electrode, since groundwater could diffuse whether it flows down or sideways. Two Gaussian curves were subtracted to see whether the shapes seen in the data after rain events could be modeled as the difference of asymmetr ical Gaussian curves. If the second curve was more dispersed, so that it had less amplitude and a broader base, the

PAGE 41

31 rain response curves seen in the data can be created (Figure 53d & 53e). This models the same amount of flow reaching the grid and reference electrodes, but a higher volume of water reaches the grid electrodes first. It is logical that groundwater flowing farther to the reference could have a more dispersed wetting front. Rise in Potential Combining the two Gaussian diffusion curves produces not only the non linear decline and the initial spike, but also the slow rise afterwards. The degree of dispersion can create a slow, nearly flat increase or a non linear rise, both seen in the data. More disperse flow or less flow to the reference electro de than the grid electrodes creates a lower, flatter rise. During the drought in April to May the rise was shallow and flat, possibly because the entire volume of groundwater flow did not reach the reference during this time of high ET. Soil Conductivity A t the same time that water flux is moving ions to affect SP (Equation 1), the SP can also vary because the soil resistivity changes (Equation 2). For a given water flux, the greater the soil conductivity, the lower the potential difference. Differences in geopark moisture content led to soil conductivity differences that vary by a factor of 10 (Figure 16). We noted that the voltage response to the streetlights turning on and off decreased during the long dry period in April. The response to streetlight volt age increased immediately after a rain. It can be inferred that in the dry season subterranean currents associated with streetlights and cultural noise are focused in deeper wetter zones. Low current density near the surface renders potential differences b etween the grid and reference electrodes smaller. This observation indicates that soil conductivity variations

PAGE 42

32 associated with degree of saturation can be significant in the geopark. Thus, potential changes associated with rainfall events are partially due to changing soil conductivity. Distinguishing between the effects of flow and conductivity require soil moisture measurements. Anomalous Electrode Behavior Heavy Rain The cycle of SP spiking, decay, and increase did not occur when a rain event immediate ly followed a previous event. This seems reasonable according to the above scenario because groundwater would already be flowing past the reference after the earlier rain, so no difference in flow would exist between the grid and reference electrodes with a new rain. This same process could explain the disappearance of response to rain events during the rainy season, especially June and July. For most of this period rain fell 2 out of every 3 days; groundwater was flowing ubiquitously; thus, no difference i n flow between grid and reference and no change in SP could be expected, unless a rain was very large, as at the large event of June 26 to 27. In August there was a 3 week period when rain subsided. A modicum of rain response returned, presumably because g roundwater flow decreased or soil dried enough that meteoric events could once again significantly increase flow differences between grid and reference electrodes. Drought In March and April, two types of erratic response started to appear. The first was the high frequency, high amplitude changes at night that were continuous at some electrodes and intermittent at others (Figure 30). A similar pattern was seen on the blank channel of the data logger where the streetlight E field passed through a small conn ection

PAGE 43

33 screw at night. The screw is small, nearly a point receiver, whereas the soil would respond to the E field as an infinite plane. The metal screw seemed to register every voltage fluctuation emitted by the streetlights, and the resultant potentials w ere recorded in the data logger (Figure 54c). During March three electrodes were removed from the ground and replaced in the ground inside closed PVC containers filled with local sediment (Figure 55). The response of these isolated electrodes was very sim ilar to the blank channel. In addition, aberrant, high frequency, high amplitude responses were increasingly seen at grid electrodes from March through May (Figure 30). This was a very dry period of the year, during which the upper ~40 cm of soil became ex tremely dry and almost as hard as rock. Perhaps the soil was so dry that the connection with the ground was lost, or water and ions could not exchange with the electrodes. As a result, many electrodes became point receivers like the blank data channel or t he electrodes isolated in PVC (Figure 54b & c). The potentials were far higher in the screw, the PVC encased electrodes, and the electrodes presumably isolated from the soil plane by extremely dry sediment; the polarity of the potential switched constantly in these locations. The response to the streetlights is uniquely positive, not the result of an alternating E field, which would manifest positive and negative polarities. The streetlights work by creating a current of electrons; these must induce consta nt positive charge in the soil while they are on. Evidently, the vast plane of the soil does not respond quickly enough to register the AC E field, or it disperses or cancels its rapid field effects. In small objects such as isolated electrodes or a metal screw, the alternating E field is captured in both polarities every time a measurement is taken. It is possible that drying

PAGE 44

34 caused poor contact between the clay gel and the soil for some grid electrodes, but the clay gel was always tightly bonded to the e lectrodes when they were changed. Doussan et al. (2002) suggested that cracks at the electrode clay interface or saturation variations in the soil near the electrode or variations in the electrode's porous interface could all disturb measurement of SP. El ectrode Design The second anomalous pattern that manifest in February was spiked minimums, and by May, some locations also started exhibiting rounded, higher amplitude maximums, which varied in amplitude (Figures 29b & c). The diurnal pattern persisted thr ough out the year and was assumed to be a response to vegetation moving water through the soil, ET, or temperature changes (Figures 19, 20, & 21), but the erratic, intermittent appearance of higher amplitude, rounded responses at some electrodes likely had other unknown causes. Doussan (2002) noted that the soil mud used for contact showed high variation in electrical conductivity. At the USF Geopark the KCl contact mud was exposed to daily voltage and high water fluxes. Doussan et al. (2002) also noted sin gle electrodes manifested jumps not related to rain or evapotranspiration. Random spikes and sudden, sharp, large changes in voltage were seen often in this study. These could be the result of sudden contact or mud conductivity changes. Several times all e lectrodes simultaneously showed a sudden change in voltage, but this was assumed to be an electrical change in cultural noise levels from the nearby hospitals or dormitory. In response to erratic behavior, electrodes and KCl clay were replaced several tim es. In May, after periodic high amplitude responses appeared, E5 and E12 in the north grid were replaced. E5 showed response similar to initial implantation, then behaved as it

PAGE 45

35 had earlier in May. E12 manifest higher amplitude when replaced (Figure 56). Si x days later the data logger was replaced. When data collection resumed in June, E5 was over limit, but underwent a sudden change in voltage, followed by response similar to remaining electrodes. The E12 profile looked like it had before the electrode chan ge (Figure 56 c & d). In July and October electrodes were replaced after geophysical surveys were conducted at the site. In July a change in voltage level occurred after reinstallation. Amplitudes changed from small to large and vice versa, sometimes inte rmittently, as before the change (Figure 32). As the drier month of August arrived, electrodes appeared less erratic than in wetter months and effects from the constant rainfall were suspected (Figure 33). After the reinstallation in October, however, elec trodes no longer correlated with rain events or with each other, even as rain decreased (Figures 36 & 37). Four electrodes were brought in from the field in February, 2010 and placed in a KCl salt bath along with unused electrodes that had been manufacture d at the same time. All electrodes were within 1 mV of each other. Two of the removed electrodes and 2 unused electrodes were returned to the field and the same behavior continued (Figure 57 a f). Since the electrodes seemed functional, doubt was cast on t he consistency of the salted bentonite gel. Doussan et al (2002) concluded that the contact mud beneath electrodes could change during times of high drainage. In the field various amounts of salt were added to bentonite for six electrodes, but the replante d electrodes at all levels of salting behaved as they had before. Florida has extremely high water flux during May through September when it rained 2 out of every 3 days most of the time. Doussan et al (2002) suggested that

PAGE 46

36 a more effective design needs to be created for electrodes remaining in shallow, unsaturated soils for long periods, a conclusion that seems borne out in this study. Dense Roots Throughout the year four electrodes, E20, E21, E31, and E32, were more erratic and differed from the rest. Th ese were in the vicinity of heavy roots, closest to the trees on the east side of the grid (Figure 10). It is possible that roots have their own streaming potential or other electrical signal, different from the surrounding minerals. Roots may move water u p or down while absorbing water from the soil, creating streaming potentials. Perhaps wetting and drying took on more complicated patterns, and electrode contact problems were amplified near roots. Electrodes were replaced in the middle of January at posi tions E32, E31, E27, E21, and E1, where anomalous behavior was seen, to check whether an electrode problem had occurred. The anomalous behavior continued after the change, except anomalous behavior was intermittent at positions E27 and E1. Electrodes 26 an d 27, in the south grid, also had frequent anomalous spiking. A few large roots run near most electrodes in the south grid (Figure 10). Electrodes E20, E21, E31, and E32 behaved unlike the rest of the grids for the entire year. Their erratic behavior was a ttributed to the presence of dense roots, and they were ignored with reference to groundwater movement. Diurnal Changes The site exhibited diurnal changes that were not due to cultural noise, which can be seen when streetlight and other electrical noise is removed from graphs (Figures 19, 20 & 21). Diurnal changes in soil potential can be expected from the water movement due to roots absorbing water from the soil, from temperature changes that affect electrode

PAGE 47

37 resistivity, or from ET, which could cause posi tive SP when upward vapor flow reached a maximum at the hottest part of the day. Vegetation The entire site was covered by grass; the east side of the grids was quite close to trees. A diurnal pattern could result from trees absorbing water. During the da y, especially in the late afternoon when there is direct sunlight, high temperature, and low relative humidity, transpiration lowers pressure in the trees' leaves. Canopy pressure is then more negative than the pressure in the vadose zone, where roots pref erentially reside. Water is thereby drawn into the tree roots (Mark Rains, e mail communication, Jan 13, 2010). Presumably, water is moving down toward the roots, away from the electrodes, causing negative streaming potential. Conversely, leaf transpiratio n decreases at night, so leaf pressure increases and could become higher than vadose zone pressure. Root intake decreases or ceases, or if vadose zone water pressure is extremely low, a positive pressure could exist in the roots (Mark Rains, e mail communi cation, Jan 13, 2010). The minimum potential, however, occurs around dawn or later in the morning. The maximum manifests in the late afternoon or evening. It seems negative potential (minimum) would occur in the late afternoon when trees would pull the mo st water down to their roots. The maximum (most positive) potential could be expected late in the night when root absorption would slow or cease. The presumed root absorption pattern does not match the measured potential. However, the reference electrode m ay be affected by water absorption exactly as the grid electrodes are, and absorption effects may not have

PAGE 48

38 been captured. In the future an additional reference electrode needs to be established in a known zero root zone. Temperature Soil temperatures coul d influence SP in two ways: (1) the electrodes have a temperature sensitivity, and (2) soil conductivity varies with temperature. A diurnal pattern could also result from temperature changes at the near surface, where the electrodes resided. Most grid elec trodes received more direct sun than the reference electrode, which was in a low, flat, shaded region, about 15 meters S SW of the south electrode grid, as far from a magnetic anomaly as possible. Temperature usually peaks in mid afternoon and reaches a mi nimum before dawn. The grid electrodes reached a maximum at the end of daylight and a minimum at or after dawn, which is very similar to temperature variation. T he electrode response to temperature change is 0.2 mV/deg C¡ for Pb PbCl 2 electrodes (Petiau, 2 000). A 15¡C temperature difference due to uneven insolation would produce the 3 mV change in soil potential seen during the hottest part of the year, but this is a very large temperature change to occur 20 cm beneath the surface. Soil temperature changes measured at the three nearest Florida Automated Weather Network stations at 10 cm depth showed an average daily fluctuation of 1.6 to 4.2 ¡C with occasional values reaching 7 ¡C (Florida Automated Weather Network); changes at 20 cm would be smaller. Temper ature driven soil conductivity changes can be estimated using a linear equation from Revil et al, (1998), f / 25 = m ( T # 25 ) + 1 (2)

PAGE 49

39 and an average slope of .021, as found in Hayley et al. (2007) for a sand silt clay soil. Well water in wells on the study site varied less .5¡C on any day or a couple of degrees over four months of monitoring. Even though this water is exposed to air, it is probably more stable than soil closer to the surface, but even a 3¡C change with m = .021 would result in f / 25 = m ( T # 25 ) + 1 = .021 ( 3 ) + 1 = 1.063 only a 6.3 percent change in conductivity. Thus, temperature effects may contribute to diurnal patterns, but could be responsible for only a small fraction of the variation. Temperature effects are probably insignificant for the rain driv en SP signals and intra grid variations. Evapotranspiration (ET) In a warm vegetative region such as Florida's gulf coast, ET would be expected to create positive SP as water vapor moves up towards the surface and vegetation transpires. The rock hard, dry soil that manifest in dry season is likely evidence for high ET at that time. The reference electrode was placed in a shady region that was not covered with grass to the extent that the rest of the site was; therefore, a relative difference in ET between the grid electrodes and the reference would be expected on the potential profiles of all the grid electrodes. One would expect ET to reach a maximum at the end of daylight and a minimum at the end of the night. In general the maximums and minimums occur at these times or a few hours later. Decagon water potential sensors were installed in October. They showed a sine like diurnal natural variation that matched the diurnal pattern of the SP response (Figure 58). Since temperature effects would be minimal and vegetation effects would have a different pattern, it is reasonable to assume the diurnal variations are primarily due to ET.

PAGE 50

40 Flow in Collapse Conduits It is widely accepted that downward flow through sinkholes creates a negative response (Ogilvy, 1967). Nevertheless, previous experimentation in the USF Geopark showed that during times of greater rain, when more meteoric water was available to increase downward flow, the negative anomaly over active collapse conduits disappeared (Craig, 1991). The main obj ective of a time survey of soil potential was to see whether these potentials could illuminate the underlying process. What was observed was more complex than a simple change in signal with season. At different times conduit electrodes manifested more posi tive potential, similar potential, and more negative potential than surrounding electrodes. Changes were often associated with major rains, but not always. We considered several mechanisms to explain the unique surface potentials that appear over collapse conduits compared to neighboring electrodes. In particular, phenomena that could produce positive potential and could explain rapid, irregular, and intermittent switching of polarity must be considered. ET might be suspected because it creates positive SP that could cancel a negative anomaly or create a positive one; however, ET would increase uniformly at all grid electrodes relative to a reference. This would not change the negative anomaly over the conduit relative to surrounding locations. Also, it has been shown that a small amount of clay in sand suppresses SP readings to background levels (Bogoslovsky and Ogilvy, 1972). Perhaps clay is washed in and out of underground subsidence depressions, affecting steaming potential. Last, it has been suggested th at when the water table rises significantly, lateral flow is less constrained by tortuous subterranean karst and can more easily reach nearby collapse conduits. As a result of greater lateral flow, increased influx of positive ions to the

PAGE 51

41 conduit region co uld overwhelm the negative potential due to downward flow into the conduit (Craig, 1991). Lateral Flow Craig observed negative anomaly disappearance over conduits in the wet season. He postulated that lateral flow became the dominant SP source when water levels were high. However, in this study a consistent relationship between high water tables or high rainfall and SP values does not exist over conduits. Relative to surrounding electrodes, positive potential existed at the north conduit, and background le vels existed at the south conduit in January, 2009, when the water table was very low; in fact, a 4.28 m deep well, located 5 meters southwest of the south grid and piercing the confining layer, was dry (Table VIII), and lateral flow likely did not exist. After the significant rainfall of Jan 29 30 (Figures 38a &39a), the north conduit electrode suddenly measured negative readings, which lasted through most of February, after which the soil potential nearly returned to background levels. Thus, the potential turned negative at the north conduit as the water table increased, although the water table was still unusually low. A new well dug between the grids in February revealed that the water table had risen 1.12 m after the late January and early February ra ins, putting the water table at a depth of 3.53 m. The original well had a similar water level. Intermittent positive readings occurred at the north sink through March to May, which cannot be explained by lateral flow, because a long dry period would likel y be accompanied by a receding water table; also, decreasing lateral flow would cause potential changes in only one direction. In May a very large 2 day rain event ended the dry season. By early June the water table

PAGE 52

42 was only 2.25 m below the surface. It sl owly raised another .5 m until the end of September. During this period of rising water table, the potential over the north conduit varied between positive and background levels, then trended positive through the rest of the year, even after the water tabl e started to fall during the last 3 months of 2009. At the south conduit the potential turned negative after the large May rains and remained so for many months as the water table rose through the summer. In the fall the potential at the south conduit alt ernated between positive and negative as the water table was at its maximum. The potential suddenly reversed polarity several times for days and weeks, but the water table does not change that abruptly. Intermittent Flow An alternative hypothesis is propos ed to explain the irregular temporal patterns of the conduit potentials relative to neighboring electrodes. The conduits transition between three states: (1) The conduit carries flow rapidly draining to the aquifer (Figure 59). The negative SP anomaly of d ownward flow in a collapse conduit can be thought of as a large dipole with the positive end down (Jardani et al., 2006). T he negative end of the dipole would be near the surface over the conduit (Figure 59), creating the negative anomaly. In the Jardani ( 2006) model, the downward flow must be fast to create the negative anomaly. Regions of higher porosity and lower resistivity have been found over other sinkholes in the USF Geopark (Kruse et al, 2007). This could allow for higher flow rates that could incr ease the magnitude of SP signal. (2) The conduit is "plugged" in that water is no longer f lowing down into the Floridan. Surficial w ater would continue to flow laterally toward the conduit, however, because water would flow into the confining layer sidew ays through the conduit walls

PAGE 53

43 above a plug (Figure 60). Water would continue to flow, albeit more slowly than in case (1), because the surface area inside the collapse conduit would be much greater than on a horizontal surface of the same diameter The con duit may extend downward many meters above a plug, perhaps up to 7 meters (Figure 4). Thus, positive charge flows laterally into the depressed subsidence region, the terminus of lateral flow. A positive anomaly would exist when the conduit is inactive as a drain to the limestone aquifer, when fast downward flow does not occur, and the high permeability sands filling the conduit have become congested at depth with fines. The case of a plugged conduit would be like a topographical low or like a pump, over whi ch soil potential is positive. Flow moves h orizontally toward the conduit, producing positive potential over the conduit (Figure 60). (3) The conduit is plugged at shallow depth (Figure 61). The subsidence region has downward flow similar to the surroun ding terrain as relatively similar volumes of water percolate downward through the confining sediments (Figure 61). Hence there is no local SP anomaly. For example, the north conduit site appears to be in the second mode in January 2009 After an abnormal ly dry fall, 2008, which delivered only 0.12 m rain from September to December, compared to an average .32 m for that time of year, the north conduit electrode, E12, indicated positive readings compared to surrounding electrodes. A sudden change in potent ial followed the voluminous January 29 30 rain. Possibly, this rain could have washed away silt or clay plugging material, allowing the conduit to drain switching to the first mode The short lived, sudden negative spike could reflect water flowing down a t a faster rate before an equilibrium was reached. After the

PAGE 54

44 negative spike, the steady, negative streaming potential during February could represent a steady flow draining to the Floridan. This negative potential gradually reduced the next month; througho ut February, illuviation could have transported organic material or mineral fines into the sand pores, decreasing permeability and downward flow, causing a gradual increase in potential until it became similar to background levels. Through most of March a nd April, the north conduit electrode uniquely measured potential that varied between background level and positive readings in 2 to 5 day cycles (Figure 54a). Only 5 modest rain events occurred in these months. Hypothetically, dissolution, raveling, and collapse could have episodically allowed permeability, and therefore streaming potential, to pulsate. A cycle of plugging and filling and slow flow could be followed by collapse and rapid drainage. An excavated sinkhole in similar karst terrain in Pasco Co unty Florida was discovered with banded iron oxide deposits interpreted to represent repeated plugging and draining of the sinkhole (reported in Parker, 1991). Immediately after a huge rain in the middle of May, when data logger failure interrupted soil po tential measurement for three weeks, the potential had a positive spike (Figures 41a and 42a). When data collection resumed in June, the conduit electrode exhibited mostly positive potential alternating with some background levels intermittently through Au gust, although with a far lesser degree of pulsation than in the spring. After May, the north conduit registered mostly positive potential (Figures 43a &44a). This could suggest that it was plugged at the end of 2009 and has remained so. Quite a different potential history occurred at the southern collapse conduit, which was presumed to be plugged at the onset of the survey, because there was no negative SP

PAGE 55

45 anomaly and no visible surface subsidence. For the first 5 months of 2009 the south conduit manifest ed potential matching the surrounding terrain, interpreted as its having the same permeability and streaming potential as the surrounding area, in mode 3 (Figures 23b, 38b, 39b, 40b, 41b, & 42b). After the 25 year rain of May 14 15, the south conduit elect rode started to manifest negative potential compared to nearby terrain. The negativity was strong until the middle of July when the electrode at the south conduit suddenly switched to positive potential for about 18 days; then returned to being relatively negative into the beginning of September. Earlier in the year, data at the north sink suggested that it might be alternately filling and draining. Theoretically, the south conduit, originally presumed plugged, could have started draining to the aquifer aft er the May rain, except for period of plugging and release during July through October (Figures 43b, 44b, 45b, & 46b). The potential changes at the two conduits suggests that flow in these conduits could be turning on and off as they plug with fines, and t hen clear. Switching patterns in SP polarity, such as those seen over the two conduits were not seen anyplace else in the electrode grid, except at E5, which had unusual signals throughout the year. E5 is only one meter from the north conduit. Long Term E lectrode Behavior Periods of positive spiking of grid electrodes relative to reference were seen at several locations, but these episodes were never negative. These spikes also always stopped abruptly instead of decaying gradually, as over the collapse con duits. Spikes were not related to the large rainfall events, as the changes at the conduits were. The cause for the spikes is unknown; some lasted only a few days; most lasted weeks. J umps in potential were seen by Doussan et al. (2002) over months of mon itoring with

PAGE 56

46 electrodes of the same design Doussan et al. (2002) suggest that potential s pikes could be related to contact issues They concluded that a better design for electrodes and electrode contact needs to be developed. In this study, e xtreme wet a nd dry conditions seemed to affect responses. The installation of moisture sensors in October coincided with a period when the nearby electrodes did not show consistent responses to rain events that consistently affected soil moisture. This result suggests that following the end of the wet season, electrodes may no longer have been responding to rainfall, as they clearly did in previous dry season. Over time, it is possible the porous interface of the electrode declines. It is also possible that the bentoni te clay gel needs to be more precisely produced although a preliminary test suggested clay salinity did not significantly influence potential readings Clearly more direct measures of both vadose and saturated zone flow in the vicinity of the conduits are needed to assess the long term stability of the electrode behavior, as well as to assess the conduit hypotheses proposed here. Such an investigation will require new electrodes and instrumentation beyond the scope of this study. Once soil parameters are a vailable to calibrate numerical models (e.g. Jardani et al., 2006), relative amplitudes of SP signatures of various flow regimes can be tested with numerical simulations.

PAGE 57

47 Chapter 6: Conclusion s A year of continuous time series measurements of SP at g rids over small collapse conduits in covered karst showed responses on a variety of time scales. The data complexity suggest that flow in response to rain, ET, and electrical noise are all significant contributors to the self potential anomalies measured o ver sinkhole conduits. The measured potentials on the electrode grid, relative to a reference electrode hydrologically down gradient, follow the form expected if wetting fronts reached the electrode grid before they reach the reference electrode. The spik e in potential followed by the non linear decay and the slow rise can be explained as the results of Gaussian diffusion volumes reaching the grid electrodes and the reference electrodes with different amplitudes, possibly due to dispersion as wetting front s flow different distances. Comparison of potential measured directly over the collapse conduits relative to neighboring electrodes (1 10 m distant) indicates that hydrologic behavior in c onduits studied is dynamic, not static. Three distinct types of cond uit anomalies are observed: conduit potential more positive, conduit potential more negative, and conduit potential equivalent to neighboring grid. Changes in mode occur in association with rainfall events. W e postulate three modes of sinkhole flow to pr oduce the three distinct SP anomalies: fast flow open to the aquifer, slow flow open to the confining layer through the collapse conduit, and an inactive conduit, plugged high enough to behave like the rest of the

PAGE 58

48 confining layer. Our preferred interpreta tion is that c onduits open and close intermittently as migration of clayey and silty sands changes the conduit permeability. The Floridan aquifer recharge patterns in west central Florida covered karst have been documented to be extremely spatially hetero geneous 1 2% of the land surface can provide most of the recharge. This study suggests that the 1 2% providing recharge can be temporally heterogeneous as well. Continuously changing permeability and access to the aquifer could affect how estimates of flo w to the aquifer are made. The complexity of SP signals suggests that it could be difficult to interpret "spot" SP surveys, conducted intermittently in time over sinkholes, as a gauge of whether the sinkhole serves as a point of deeper aquifer recharge. SP signals depend on soil moisture, because of resistivity changes in the ground, and because the amount of groundwater that is already flowing shapes the SP response to a rain event. Distinguishing cultural noise also requires continuous monitoring. Clea rly more study is needed to better understand the SP response and apparent intermittent sinkhole behavior. Many more conduits need to be studied, and instrumentation to sense moisture movement at numerous points must be included in the study. Use of sever al reference electrodes placed at distance from the grids in various topographic, vegetative, and climatic settings would help to distinguish groundwater flow from other SP sources. Numerical modeling of flow and SP for the hypothesized conduit flow regime s is needed. This work could be conducted with the Comsol finite element modeling package. SP is clearly valuable fo r studying overall infiltration, as well as conduit flow. It may be possible to use wetting front curves after rain events to perceive flow patterns, by

PAGE 59

49 comparing Gaussian curve shapes for all possible pairs of electrodes. However, the amplitude of the curves following a rain even is likely influenced by changes in soil conductivity as well as moisture Moisture sensors could be used estimate conductivity changes based on the conductivity versus moisture curve. Finally, a s Doussan et al (2002) concluded, a better design for electrodes and electrode contact needs to be developed. Long term (more than a few months) behavior of the electrodes in w etting and drying ground is not documented in the literature. Future research is planned to simultaneously monitor soil potential and soil moisture and other soil properties Future studies should help distinguish streamin g potentials from other factors, and better understand the behavior of Petiau electrodes.

PAGE 60

50 Figures

PAGE 61

51 Revil et al., 2003 a. Revil et al., 1999 Fi gure 1: Electrical Triple Layer The electrical triple layer, responsible for the streaming potential, consists of the mineral surface with negative charge, a layer of positive ions adsorbed to the surface, called the Stern layer, and an electrically diffu se layer where negative counterions are attracted to the surface and positive coions are repelled. Beyond the Stern layer fluid flow will carry the surplus of positive ions down gradient. Globally the three layers are electrically neutral, but the offset o f positive ions creates an E field. b.

PAGE 62

52 Tampa Bay Tampa St Pete USF Gulf of Mexico Florida Figure 2: Study Site The USF Geopark is located in the southwest portion of the USF campus, which is just northeast of Tampa, Florida. Electrodes were planted in the northeast section of the USF Geopark. They were placed in 2 crosses over 2 known collapse conduits in a grassy clearing surrounded by trees on the south, west, and east sides. N

PAGE 63

53 Reference Figure 3: Grid Layouts Electrodes were laid out in crosses over two known collapse conduits. The north collapse was suspected to be actively draining because a negative anomaly existed at the subsiding region in late 2008. Subsidence could be se en in a 1 meter diameter around electrode E 12, and it changed throughout the year: subsiding and filling with sediment. The south conduit was a region of diffuse settling and no specific region of subsidence could be seen. It was assumed to be inactive as a drain since no negative anomaly appeared over it in late 2008.

PAGE 64

54 From Parker, 1991 From Stewart & Parker, 1992 a. b. Figure 4: Schematic Cross Section ( a.) Two known locations of cover collapse conduits were chosen for the study. The north conduit (~105 m) was presumed to be draining the water table aquifer (sands) to the Floridan (limestone) because of ongoing subsidenc e and negative SP signals. The south region was more diffuse (~ 90 m), with no clear region of subsidence evident at the surface. It was assumed to be an inactive conduit because subsidence was not evident and SP was the same as the background. Depth to Co nfining Clay (b.) The surface of the clay layer, beneath various types of aeolian and marine sands, is tortuous. As a result, lateral permeability can be inhomogenous and anisotropic, and groundwater may flow differently depending on water table stage.

PAGE 65

55 E12 Conduit Electrode Electrode Positions E17 E5 E18 E6 E8 E7 E19 E20 Figure 5: North Conduit E W Electrode Line Electrode E12 is over the north conduit. The significant layer visible about 1 m below the surface is the top of the red yellow marl. The electrodes are buried in yellow sa nds, with grey sands immediately below. W E

PAGE 66

56 Electrode Positions E12 Con duit Electrode Electrode Positions E4 E13 E3 E14 E15 E2 E1 E16 Figure 6: North Conduit N S Electrode Line Electrode E12 is over the north conduit. The significant layer visible abou t 1 m below the surface is the top of the red yellow marl. The electrodes are buried in yellow sands, with grey sands immediately below. S N

PAGE 67

57 E25 Conduit Electrode Ele ctrode Positions E26 E27 E24 E23 Figure 7: South Conduit N S E lectrode Line Electrode E25 is over the south conduit. The significant layer visible about 1 m below the surface is the top of the red yellow marl. The electrodes are buried in yellow sands, with grey sands immediately below.

PAGE 68

58 E25 Conduit Electrode Electrode Positions E30 E29 E31 Figure 8: South Conduit E W Electrode Line Electrode E25 is over the north conduit. The significant layer visible about 1 m below the surface is the top of the red yellow marl. The electrodes are buried in yellow sands, with grey sands immediately below.

PAGE 69

59 E25 Conduit Electrode E12 Conduit Electrode 12 Condui t Electrode Clayey Sand (Marl) Fine Sands above Silty Fine Sands below Figure 9: 250 MHz GPR Vertical Section through North & South Conduits Reconnaissance GPR survey located regions of subsidence that had been mapped by cores and wells by Parker (1991) See Figure 4 for Schematic.

PAGE 70

60 Figure 10: Electrodes Relative to Roots Electrode positions re lative to the horizontal GPR slice at 65 cm depth. Potential at the 2 easternmost electrodes on both grids lie in regions with heavy roots. These locations typically had potential profiles different from the rest of the grid. Electrodes in the southern an d western arms of the south grid also had sporadic spikes that did not manifest anywhere else. These could be the result of the evident roots, but the roots in these locations are not any denser here than at other locations where spikes were not seen. Conduit ! ! ! ! ! ! ! ! ! ! ! ! ! Electrode Reference Electrode N ! ! ! ! !

PAGE 71

61 Connect to Data Logger KCl kaolinite with PbCl 2 PVC Constriction Porous Wood Plug PVC Cap Connect to Data Logger Electrode Soil KCl Bentonite Pb Wire Figure 11: Petiau Electrodes and Burial (a.) Forty eight Petiau electrodes were constructed and 32 were planted in arrays. Several were replaced during the year with spare electrodes. (b.) Electrodes were set in bentonite gel. PVC caps, plastic bags, and electrical tape protected the underground connection to the data logger wire, which was run in underground conduit most of the way to the data logger. a. b.

PAGE 72

62 Figure 12: Magnetic Anomalies Light pink dots indicate electrode locations. Collapse conduits are at the centers of the crossed arrays. Magnetic anomal ies, presumably from buried construction metal, create continuous negative potential to the west of the arrays.

PAGE 73

63 Figure13: LiDAR LiDAR was applied over the region where magnetic and GPR data were collected. The north SP grid crossed a conduit past with topographic surface expression and previous negative anomaly. The south SP grid crossed a collapse region that was evident on GPR, but revealed no surface subsidence. It was presumed plugged. Sinkhole ! ! ! ! ! ! ! ! ! ! Electrode Reference Electrode N Figure and data from Joseph Van Gaalen ! ! ! ! ! ! !

PAGE 74

64 ! Electrode ode Reference Electrode N ! ! ! ! ! ! ! ! Figure 14: EM 31 Survey Horizontal Dipole For the vertic al dipole the ground above and below 2.9 m contribute equally. The horizontal dipole reflects a larger contribution from deeper sediments, which were about half as conductive as shallower sediments. ! ! ! ! ! ! ! !

PAGE 75

65 ! Electrode ode Reference Electrode N ! ! ! ! ! ! ! ! Figure 15: EM 31 Survey Vertical Dipole For the horizontal dipole, the ground above 1.8 m depth and the ground below 1.8 m contributed equally to the signal. The horizontal dipole reflects a larger contribution from shallow sediments, which were about twice as conductive as deeper sediments. ! ! ! ! ! ! ! !

PAGE 76

66 Figure 16: Soil Resistance A soil resistance test was applied to unconsolidated gray sand from 30 to 50 cm below the surfac e, a depth immediately below electrodes, Water was added in 10 ml increments, and resistivity was measured at each moisture level until saturation was reached. Sample was oversaturated after last water was added, so 100% saturation was assumed to be halfwa y between the last two increments. Soil porosity was estimated to be 35%.

PAGE 77

67 Jan 20 .18 cm a. b. Streetlights On at Dusk Streetlights Off at Dawn Figure 17: Typical Electrode Responses (a.) After a rain a positive spike occurs when the rain is large enough. Th en, non linear decay occurs for 2 to 5 days. If there is no more rain, potential rises until the next rain event. In some cases the rise is very small; in some cases it starts to level out (b.) Positive potential manifests when streetlights turn on and di sappears when they turn off. (c.) Spikes likely represent intermittent electrical noise from equipment turning on and off. Jan 29 30 5.56 cm Feb 2 1.78 cm

PAGE 78

68 Figure 18: SP Maximums and Minimums versus Time of Day Minimums occur near dawn or early morning. Maximums occur late afternoon or early evening. The most extreme values were chosen for minimums and maximums June data had two minimums: The time of the less extreme minimum (not graphed) coincided more clo sely with other months' minimums. Occasional shifts in maximum during June are unexplained. April shifts are unexplained.

PAGE 79

69 Figure 19: Monthly Potential Profiles with Cultural Noise Removed With streetlight noise removed a natural diurnal variation of 3 m V is quite consistent by March.

PAGE 80

70 Figure 20: Monthly Potential Profiles with Cultural Noise Removed

PAGE 81

71 Figure 21: Monthly Potential Profiles with Cultural Noise Removed

PAGE 82

72 a. b. c. d. Figure 22: Soil Potential in High Root Zones Throughout the year the electrodes nea rest the trees and dense roots on the east side of the grids were erratic and unlike the remaining electrodes.

PAGE 83

73 a. b. North Grid South Grid Time (hours) Di stance along Electrode Line West to East Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) E12 over N collapse conduit E25 over S collapse conduit Figure 23: January SP Contours through Time (E W Electrode Line) The soil potential remains fairly stable through time in January until the very large rain January 29 30 when an oft seen increase immediately after a rain occurs. The large negative values to the west at the north grid are presumed to be redox. The anomalous peaks at the east of the south grid are in a heavy ro ot zone. The east side of the north grid is also in a heavier root zone. Note the sudden relative negativity at E12.

PAGE 84

74 Jan 30 4.55 cm cm" Dawn Dusk Jan 29 .23 cm Jan 29 .79 cm cm" Figure 24: Jan 29 30 Rain Event The initial rainfall caused a large increase in potential after several hours. The next rainfall caused a smaller increase, and the third rainfall, which was much larger than the previous two, caused a very small increase. The potential gain at streetlight onset is much smaller than the potential decrease at streetlight shutdown.

PAGE 85

75 Dusk Dawn wn Feb 2 1.78 cm cm" Figure 25: Feb 2 Rain Event After small responses from later rains during the Jan 29 30 rain event, the Feb 2 rain once again caused a large increase in potential after the rain.

PAGE 86

76 Feb 2 1.78 cm Feb 19 .51 cm Figure 26: Februar y Rain Events Most of the month was dry. It appears that the response to streetlight voltage decreased throughout the month as the soil dried.

PAGE 87

77 March 1 .69 cm March 22 .29 cm March 29 1.07 cm Typical March Profile Figure 27: March Rain Events (a.) The drought continued in March. (b.) Anomalous behavior started at some electrodes: in one case, voltage was higher within diurnal changes and curves had rounded maximums and spiked minimums.

PAGE 88

78 April 6 .06" April 14 1.49" Typical April Profile North Conduit Figure 28: April Rain Events (a.) The drought continued in April. (b.) Anomalous behavior continued at E12, the north conduit electrode, where 2 to 5 day shifts in voltage occurred. (c.): The anomaly with higher voltage rounded maximums and spiked minimums, which appeared at some electrodes in March, comes and goes at E2 in April.

PAGE 89

79 May 12 1.07 cm Typical May Profile Figure 29: May Rain Events (a.) The drought ended on May 12 13 when a 25 year rain event occurred, raining as much in 2 days as it had in the previous 7.5 months. (b.) The anomaly with higher voltage rounded maximums and spiked minimums, which appeared at some electrodes in March, comes and goes at E2 in April. (c.) Anomalous behavior continued at E12, the no rth conduit electrode, but the 2 to 5 day shifts in voltage no longer occurred. North Conduit

PAGE 90

80 Figure 30: High Frequency Anomaly As the soil became very dry in April, some electrodes manifested high frequency, higher voltage res ponses. This pattern matches the blank channel and is attributed to the electrodes becoming isolated from the ground due to extremely dry conditions. The electrodes acted as point sources, like the blank channel screw, unlike the infinite plane of the terr ain, and responded to every change in the streetlight E field when isolated from the ground.

PAGE 91

81 a. b. June 30 3.78 cm Typical June Profile c. d. June 5 7 2.46 cm June 17 .61 cm June 23 .76 cm June 26 27 6.58 cm Figure 31: June Soil Potential Profiles (a.) It rained 13 days in June; larger rain events shown. (b.) (c.) (d.) R esponses to rain are minimal. July 1 6.83 cm North Conduit South Conduit

PAGE 92

82 a. b. North Conduit c. d. July 1 6.83 cm Figure 32: July Soil Potential Profiles It rained 19 days in July; larger rain events are noted. Response to rain events has become very minimal. The anomalous spike of the north grid electrodes reflects an electrode change. SP patterns look more unique across electrodes tha n earlier in the year. July 20 1.70 cm e. f. g. h. July 30 1.27 cm

PAGE 93

83 a. b. North Condui t c. d. e. f. g. h. South Conduit Figure 33: August Soil Potential Profiles August was relativel y dry for that time of year, but it rained many days. The spike at the beginning of the month is of unknown origin. Electrodes are not reacting to rain events and move independently of each other. Aug 25 1.32 cm Aug 20 21 22 1.91 cm Aug 15 16 .56cm Aug 8 .99 cm Aug 5 3.96 cm Aug 3 1.50 cm

PAGE 94

84 a. North Conduit c. d. e. f. g. h. South Conduit Figure 34: September Soil Potential Profiles It rained 15 days in September; larger rain events are noted. Rain res ponse was evident at a few electrodes in September, but in general, electrodes did not respond clearly to rain events, and electrode potentials moved independently of each other. Sept 25 26 27 1.87 cm Sept 17 1.30 cm Sept 15 5.13 cm Sept 11 12 13 2.91 cm Sept 4 2.13 cm Sept 2 .68 cm

PAGE 95

85 a. b. North Conduit c. d. Oct 6 .97 cm Figure 35: October Soil Potential Profiles Electrodes were removed for GPR surveys where there is missing data. After replacement electrodes became more erratic than at any time earlier in the year. Oct 16 1.55 cm e. f. g. h. Oct 27 1.32 cm South Con duit

PAGE 96

86 a. b. North Conduit c. d. Nov 10 2.44 cm Figure 36: November Soil Potential Profiles e. f. g. h. Nov 25 3.05 cm South Conduit

PAGE 97

87 b. North Conduit c. d. Dec 2 1.85 cm Figure 37: December Soil Potential Profiles Dec 4 5 3.17 cm e. f. g. h. Dec 18 .64 cm South Conduit

PAGE 98

88 a. b. North Grid South Grid Time (hours) Distance along Electrode Line South to North Distance along Electrode Line South to North Time (hours) SP (mV) SP (mV) E12 over N collapse conduit E25 o ver S collapse conduit Rain Event Rain Event Figure 38: January SP Contours through Time (N S Electrode Line) The soil potential was positive relative to the surrounding electr odes at the north grid until the large rain at January 29 30. At the south grid the collapse conduit electrode has the same potential as surrounding electrodes. Spikes at the south end of the south grid are of unknown origin.

PAGE 99

89 a. b. North Grid South Grid Time (hours) Distance along Electrode Line South to North Time (hours) SP (mV) SP (mV) E12 over N collapse conduit E25 over S collapse conduit Figure 39: February SP Contou rs through Time (N S Electrode Lines ) The negative soil potential at the north grid slowly returns to slightly positive. At the south grid the collapse conduit has the same potential as surrounding electrodes. Spikes at the south end of the south grid are of unknown origin. Distance along Electrode Line South to North

PAGE 100

90 b. North Grid South Grid Time (hours) Distance along Electrode Line West to East Distance along E lectrode Line West to East Time (hours) SP (mV) SP (mV) E25 over S collapse conduit E25 over N collapse conduit Fi gure 40: February SP Contours through Time (E W Electrode Lines) There is redox affecting the west end of the north grid and roots affecting the east end; nevertheless, the conduit electrode, E12, appears to remain negative relative to its nearest neighb ors. Roots appear to cause strong variations at the east end of the south grid, where the conduit potential is the same as nearby electrodes.

PAGE 101

91 b. North Grid South Grid Time (hours) Distance along Electrode Line South to North Distance along Electrode Line South to North Time (h ours) SP (mV) SP (mV) E12 over N collapse conduit E25 over S collapse conduit Figure 41: March April May SP Contours through Time (N S Electrode Lines) While the south conduit's potential is equal to surrounding region s, the north conduit has intermittent behavior for 2.5 months. When the 25 year rain arrived in the middle of May, a large increase in positive potential was the result.

PAGE 102

92 a. b. North Grid South Grid Time (hours) Distance along Electrode Line West to East Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) E12 over N collapse conduit E25 over S collapse conduit Figure 42: March April May SP Contours through Time (E W Electrode Lines ) (a.) The north conduit switches to posi tive potential at the large rain in the middle of May. (b.) The south conduit has the same potential as its surrounds for the first five months of 2009.

PAGE 103

93 a. b. North Grid South Grid Time (hours) Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) Distance along Electrode Line West to East E12 over N collapse conduit E25 over S collapse conduit Figure 43: June July August SP Contours through Time (E W Electrode Lines ) (a.) The change in potential at the north conduit is uncertain due to many erratic spikes (b.) The south conduit shows the sudden negative potential over the collapse conduit that appeared at the big rain in May continues in June..

PAGE 104

94 a. b. North Grid South Grid Time (hours) Distance along Electrode Line South to North Distance along Electrode Line South to North Time (hours) SP (mV) SP (mV) E12 over N collapse conduit E25 over S collapse conduit Figure 44: June July August SP Contours through Time (N S Electrode Lines ) (a.) When a new data logger was connected on June 4 the north conduit electrode measured positive potential, as it had since the large May rain, A little intermittent behavior continued in June. (b.) The south collapse conduit suddenly exhibited potential after the large May rain. I t remained negative for most of the summer except for 14 days it becomes extremely positive.

PAGE 105

95 a b North Grid South Grid Time (hours) Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) Distance along Electrode Line West to Ea st E12 over N collapse conduit E25 over S collapse conduit Figure 45: September October November SP Contours through Time (N S Electrode Lines)

PAGE 106

96 a. b. North Grid South Grid Time (hours) Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) Distance along Electrode Line West to East E12 over N collapse conduit E25 over S collapse conduit Figure 46: September October November SP Contours through Time (E W Electrode Lines)

PAGE 107

97 a. b. North Grid South Grid Time (hours) Distance along Electrode Line South to North Time (hours) SP (mV) SP (mV) Distance along Electrode Line South to North E12 over N collapse conduit E25 over S collapse conduit Figure 47: December January February SP Contours through Time (N S E lectrode Lines)

PAGE 108

98 a. b. North Grid Time (hours) Distance along Electrode Line West to East Time (hours) SP (mV) SP (mV) Distance along Electrode Line West to East E12 over N collapse conduit E25 ove r S collapse conduit Figure 48: December January February SP Contours through Time (E W Electrode Lines)

PAGE 109

99 Figure 49: Typical SP after Rain Events After rain events, especially larger ones, there is an increase in voltage, followed by a 2 to 5 day non linear decay in voltage, followed by a slow increase in voltage that oft en lasts until the next rain event. a. b. c.

PAGE 110

100 Figure 50: Typical SP after Rain Events After rain events, especially larger ones, there is an increase in voltage, followed by a 2 to 5 day non linear decay in voltage, followed by a slow increase in v oltage that often lasts until the next rain event. a. b.

PAGE 111

101 Figure 51: Wetting Fronts Soil Moisture change with time reveals an abrupt increase in moisture after a rain followed by a non linear decrease in rain content. The decay curve is expected to be Gaussian, as it is presumably caused by diffusion. The measurements for the bottom curve were taken .05 m below those in the top curve, and the curves are starting to spread. Extreme spreading was seen another .05 m deep, but this moisture had entered the sandy marl.

PAGE 112

102 a. b. Figure 52: Curve Fitting for Potential Trends after Rain Events (a.) After the initial, short lived positive spike after a rain, the decay is non linear. (b.) After the decay in potential, the increase is non linear, but is sometimes qui te shallow, unlike (b.).

PAGE 113

103 Figure 53: Modeling SP after Rain Event (a.) An asymmetrical Gaussian curve representing diffusion volume passing an electrode, compressed in the direction of travel. (b.) (c.) Asymmetrical Gaussian curves that have been dispersed. If curve (a) is a water volume passing a grid electrode, and curves (c) and (d) are dispersed volumes passing reference electrode, subtracting (b) or (c) from (a) creates (d) or (e). respectively. This difference is what the data logge r measures, and these curves are similar to data after rain.

PAGE 114

104 a. b Electrode in PVC casing c. blank data logger port March 1 .69 cm March 22 .23 cm March 29 1.07 cm Electrodes 21, 27 & 11 encased in PVC chambers Figure 54: Anomalous Electrode Profiles (a.) March soil potentials over the north conduit were episodic, varying every 2 to 5 days between 2 levels. Voltage shift was high like the voltage changes seen in other electrodes with rounded maximums and spiked minimums. (b.) In an attempt to isolate background noise, 3 electrodes were placed in PVC containers. These point sources reacted differently to background than the infinite plain of the soil. (c.) Blank channel (metal screw) in the data logger records continuous background electrical noise. Electrical noise at night declined throughou t February, but jumped back to January high levels in middle of March. Streetlights return to high voltage response Background voltage Background voltage increases during the day

PAGE 115

105 Figure 55: PVC Cas ing for Electrode In March, 3 electrodes were encased in PVC and buried to see whether background noise could be isolated from soil potential. The background noise was recorded, but it differed from soil measurements in sensitivity and magnitude because i t was a point, not an infinite plane. When assemblies were placed on top of ground, the result was the same. In addition to the streetlights at night, the background in the USF Geopark includes a continuous electrical signal, presumably from surrounding bu ildings. Starting in mid March an elevated signal occurred midday. Wire to Logger Electrode Local Soil PVC Casing

PAGE 116

106 a. Figure 56: May Electrodes Replaced Removing electrodes E5 and E12 in May r esulted in similar behavior after replacement, although behavior was erratic. b. c. d. Electrode Replaced Electrode Replaced

PAGE 117

107 a. c. d. e. f. Cultural Electrical Change Cultural Electrical Ch ange Figure 57: Profile after February 2010 Electrodes Replaced Results from replacing electrodes were ambivalent. The same variable behavior continued.

PAGE 118

108 Figure 58: Matric Potential The water potential sensor at 33 cm depth (port3) manifests a diurnal pattern that matches the pattern for SP potential. Apparently the natural diurnal variation is due to ET changes, as hydrostatic pressure and therefore flow changes in the soil. Port 4 is a sensor at 94 cm depth.

PAGE 119

109 Figure59: Schematic of a Conduit Draining to a Confined Aquifer The conduit carries flow rapidly draining to the aquifer. The negative SP anomaly of downward flow in a collapse conduit can be thought of as a large dipole with the positive end down. T he negative end of the dipole would be near the surface over the conduit, creating the negative anomaly. The downward flow must be fast to creat e the negative anomaly.

PAGE 120

110 Figure 60: Schematic of a Plugged Conduit A positive anomaly could exist when the conduit is inactive as a drain to the limestone aquifer, when fast downward flow does not occur, and the high permeability sands filling the condui t have become congested at depth with fines. Positive charge could flow laterally into the depressed subsidence region, the terminus of lateral flow. Water would continue to flow down the conduit because the surface area inside the collapse conduit is much greater than on a horizontal surface of the same diameter Flow moves h orizontally toward the plugged conduit, producing positive potential over the conduit much like at a topographical low or a pump, over which soil potential is positive.

PAGE 121

111 Figure 61: Sc hematic of an Inactive Conduit When a conduit is plugged at shallow depth,. t he subsidence region has downward flow similar to the surrounding terrain since relatively similar volumes of water percolate downward through the confining sediments everywhere Hence there is no local SP anomaly.

PAGE 122

112 References Au bert, M. and Atangana, Q.Y., 1995. Self potential method in hydrogeological exploration of volcanic areas. Groundwater, 34 (6), 1010 1016. Bogoslovsky and Ogilvy, 1972. The study of streaming potential on fissured media models. Geophysical Prospecting, 20 (1), 109 117. Bogoslovsky, V.A. and Ogilvy, A.A., 1977. Geophysical Methods for the investigation of landslides. Geophysics 42 (3), 562 571. Byrdina et al., 2003. Self potential variations associated with ultra long period seismic signals at Merapi vol cano. Geophysical Research Letters, 30 (22), SDE (6 1 6 4). Corwin, R. F., 1990. The self potential method forenvironmental and engineering applications. Geotechnical and Environmental Geophysics (SEG) Investigations in Geophysics, 5, 127 145. Corwin, R .F., and Hoover, D.B., 1979. The self potential method in geothermal exploration. Geophysics, 44, 226 245. Craig, R., 1991. Self potential mapping of recharging sand columns in a covered karst terrane. University of South Florida master's thesis. Doussan et al, 2002. Variations of self potential and unsaturated water flow with time in sandy loam and clay loam soils. Journal of Hydrology, 267, 173 185. Ernstson, K and Scherer, H.U., 1986. Self potential variations with time and their relation to hydrogeol ogic and meteorological parameters. Geophysics 51 (10), 1967 1977. Fitterman, D.V., 1978. Electrokinetic and magnetic anomalies associated with dilitant regions in a layered earth. Journal of Geophysical Research, 83 (B12), 5923 5928. Florida Automated Weather Stations ftp://if fwn prdw01.osg.ufl.edu/fawnpub/data/15_minute_obs/ Fournier, C., 1989. Spontaneous Potentials and resistivity surveys applied to hydrogeology in a volcanic area: Case history of the Chaine des Puys (Puy de Dome, France). Geophys ical Prospecting, 37, 647 668.

PAGE 123

113 Hayley et al., 2007. Low temperature dependence of electrical resistivity: Implications for near surface geophysical monitoring. Geophysical Research Letters, 34, L18402 (1 5). Jardani et al., 2006. Self potential signals as sociated with preferential groundwater flow pathways in sinkholes. Journal of Geophysical Research, 111, B09204 (1 13). Kruse et al., 2006. Sinkhole structure imaging in covered karst terrain. Geophysical Research Letters, 33, L16405 (1 6). McNeill, J.D. 1980. Technical Note TN 6: Electromagnetic Terrain Conductivity Measurement at Low Induction Numbers. (Geonics Limited: Ontario). Ogilvy, A.A., 1967. Studies of underground water movement. Geological Survey Canada Report 26, 540 543. Parker, J.W., 1991 Surficial aquifer hydrogeology in a covered karst terrane. University of South Florida masters thesis. Perrier, F. and Raj Pant, S., 2005. Noise reduction in long term self potential monitoring with traveling electrode referencing. Pure Applied Geophysi cs, 162, 165 179. Petiau, G., 2000. Second generation of lead lead chloride electrodes for geophysical applications. Pure and Applied Geophysics, 157, 357 382. Revil, A. et al.,1998. Electrical conductivity in shaly sands with geophysical applications, J ournal of Geophysical Research, 103(B10), 23, 925 23, 936. Revil et al., 1999. Streaming potential in porous media: 1. Theory of the zeta potential. Journal of Geophysical Research, 104 (B9), 20021 20031. Revil et al., 2003. Princples of electrography a pplied to self potential electrokinetic sources and hydrogeological applications. Water Resources Research, 39 (5), 1114. Revil et al., 2004. The hydroelectric problem of porous rocks: inversion of the position of the water table from self potential data. Geophysical Journal International, 159, 435 444. Sill, W.R., 1983. Self potential modeling from primary flows. Geophysics 48, 76 86. Stewart, M.T., 1998. The Florida water wars, a geologic perspective. Geotimes [#], 24 27. Stewart, M.T. and Parker, J ., 1992. Localization and seasonal variation of recharge in a covered karst aquifer system, Florida, USA. [ ]

PAGE 124

114 Suski et al., 2004. A sandbox experiment of self potential signals associated with a pumping test. Vadose Zone Journal, 3 1193 1199. Telford, W.M., Geldart, L.P., and Sheriff, R.E., 1990. Applied Geophysics, 2 nd Ed,, Cambridge Univ. Press, 770 pp. Thony et al., 1997. Field characterization of the relationship between electrical potential gradients and soil water flux. Earth and Planetary Scie nces, 325, 317 321. Truss et al., 2007. Imaging rainfall drainage within the Miami oolitic limestone using high resolution time lapse ground penetrating radar. Water Resources Research, 43, W03405 (1 15). University of South Florida Department of Geogra phy Weather Station http://www.weathercenter.usf.edu

PAGE 125

115 Appendix A: Tables

PAGE 126

Appendix A 116 Table I: Soil Layers from Auger Buckets on SP Site From Parker, 1991

PAGE 127

Appendix A (continued) 117 Table II: Soil Laye rs from Rotary Borings on SP Site From Parker, 1991

PAGE 128

Appendix A (continued) 118 Table III: Stratigraphy of Floridan Aquifer From Clasen, 1989

PAGE 129

Appendix A (continued) 119 Table IV: Monthly Average Rainfall for Tampa FL

PAGE 130

Appendix A (continued) 120 Table V: Saturation versus Resistance

PAGE 131

Appendix A (continued) 121 Table VI: Rain Events: Sept 2008 Feb 2010

PAGE 132

Appendix A (continued) 122 Table VI: Rain Events: Sept 2008 Feb 2010

PAGE 133

Appendix A (continued) 123 Table VI: Rain Events: Sept 2008 Feb 2010

PAGE 134

Appendix A (continued) 124 Table VI: Rain Events: Sept 2008 Feb 2010

PAGE 135

Appendix A (continued) 125 Table VII: Water Tables

PAGE 136

Appendix A Peter B.Bumpus graduated from the University of South Florida in 2007 with a degree in Physics. He won a NSF grant to continue his studies in the geophysical study of groundwater in 2008. His area of interest in groundwater and collapse con duit flow. He is currently an environmental geophysics graduate student at the University of South Florida. About the Author


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004581
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Bumpus, Peter.
0 245
Self-potential response to rainfall changes over plugged and unplugged sinkholes in a covered-karst terrain
h [electronic resource] /
by Peter Bumpus.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (MS)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: For the protection of wetland and water resources it would be beneficial to understand when collapse conduits function as recharge points to the underlying aquifer. Inexpensive, noninvasive methods to detect recharge are desirable. Previous studies show negative self-potential (SP) anomalies over sinkholes that correspond to the expected electrokinetic effects of groundwater flowing downward through a conduit. SP surveys are less labor-intensive than high-resolution 3D GPR and resistivity, and continuous long-term monitoring is possible. However, before SP surveys can be reliable indicators of flow, SP contributions from ET, conductivity changes, redox reactions, thermoelectric effects, cultural noise, and lateral flow must be understood. A year of continuous SP monitoring was combined with high-resolution 3-D GPR surveys, and intermittent water table monitoring over two small covered-karst sinkholes in Tampa, Florida. Positive and negative SP anomalies episodically manifested over conduits, suggesting that conduit flow is dynamic, not static. Three distinct SP flow regimes in the conduits are postulated: fast flow open to the aquifer, slow flow open to the confining layer through the collapse conduit walls, and a conduit, plugged high enough to behave like the rest of the confining layer. SP responses after rain events also appear to measure the effects of two moving Gaussian wetting front curves, one striking the monitoring electrode, one the reference. viii The wetting front volumes are differently dispersed by traveling different distances. By comparing curve shapes for all possible pairs of electrodes, it may be possible to establish surficial infiltration and flow patterns. Temporal SP response clearly shows SP is also affected by soil conductivity, rainfall history, and cultural noise. Ultimately, SP changes too frequently to rely on measurements many hours or days apart. Over the course of the year, the electrodes became less responsive to rainfall and more erratic. Extremely wet and dry conditions seemed to affect responses. The porous faces of the electrodes or the bentonite clay gel used to enhance contact may decline. It appears a better design for electrodes and electrode contact needs to be developed. To test the intermittent behavior hypothesis, more conduits need to be studied, and moisture and SP must be studied concurrently. Several reference electrodes placed in various topographic, vegetative, geologic, and climatic settings could help distinguish groundwater flow from other SP sources. SP is a valuable research tool; however external complexities such as cultural noise, sinkhole lithology, and the state of the unsaturated zone make SP data difficult to interpret without ancillary information.
590
Advisor: Sarah Kruse, Ph.D.
653
Collapse conduits
Hydrology
SP
Groundwater
Self-potential
Sinkholes
690
Dissertations, Academic
z USF
x Geology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4581