Citation
Speleogenesis

Material Information

Title:
Speleogenesis
Series Title:
Speleogenesis
Creator:
Klimchouk, Alexander B. (Aleksandr Borisovich)
Ukrainian Institute of Speleology and Karstology
Publisher:
Commission on Karst Hydrogeology and Speleogenesis of the Union International of Speleology
Publication Date:
Language:
Multiple Languages

Subjects

Subjects / Keywords:
Geology ( local )
Genre:
serial ( sobekcm )

Notes

Restriction:
Open Access
Original Version:
No. 10 (2011)

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
K26-04225 ( USFLDC DOI )
k26.4225 ( USFLDC Handle )
9001 ( karstportal - original NodeID )
1814-294X ( ISSN )

USFLDC Membership

Aggregations:
Karst Information Portal

Postcard Information

Format:
Serial

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. Introduction1.1. Speleology in the high arcticAlthough there is a proli“c literature on speleogenesis in alpine settings, relatively little is published on cave descriptions or speleogenetic studies in high arctic, permafrozen areas e.g. (Ford and Williams, 1989) pp. 489496, (Schroeder, 1979; Juday, 1989; Dixon et al., 1997; Lauriol et al., 2001). Speleogenesis in this climatic zone is enigmatic, as water circulation is restricted to the seasonally active layer, to deep zones beneath the permafrost, or through taliks of various kinds. There is a wide literature on subpermafrost hydrology, describing numerous observations of active water circulation, like open-system pingos, but also many cases of thermal, mineralised water. Water recharge to the sub-permafrost circulation system may take place underneath polythermal glaciers. Since the age of many karst landforms may be much greater than the duration of speci“c climatic conditions during the Quaternary, it is necessary to consider permafrost dynamics over longer timespans. Relict caves in a permafrozen setting, like those presented here or for instance caves in northern Greenland (Davies, 1960; Loubiere, 1987), may either be ascribed to a totally different climatic regime, where permafrost was absent, or to conditions where the caves were part of a talik system i.e. (Lauritzen, 1998). This paper presents new observations on relict and active caves in NW Spitsbergen and discusses their speleogenetic history.1.2. Permafrost and groundwater Permafrost is a thermal condition, where ground temperatures beneath a sur“cial, seasonally thawed zone (the active layerŽ) is permanently below 0 C (Brown, 1970; Price, 1972). Permafrozen areas may be subdivided according to the vertical and lateral extent of frozen zones. In areas of continuous permafrost frozen ground is regionally continuous and may reach a depth of 500 m (Brown, 1970). However, it is not present beneath the sea, freshwater lakes or beneath glaciers that are wider than about 400 m (W erenskiold, 1953). Discontinuous Caves and speleogenesis at Blomstrandsya, Kongsfjord, W. SpitsbergenStein-Erik LauritzenDepartment of Earth Science, University of Bergen, Allegaten 41, N-5007, Bergen, NorwayRe-published from: International Journal of Speleology 35 (1), 2006. … P. 3758. Abstract: Blomstrandsya, at Kongsfjord (780 57N), Spitsbergen, is within the high arctic, a completely permafrozen zone. The bedrock consists of Paleozoic marbles and has yielded a surprising amount of karst features. Early phases of hydrothermal, possibly Caledonian, speleogenesis and subsequent Devonian karsti“cation with redbed deposits is well documented. 62 active seacaves, and more than 30 relict karst caves were found in the coastal cliffs and in escarpment faces around the island. All caves have very limited extent; they are either quite short, like most of the active sea caves, or they are soon choked by froz en sediments and ground ice after a few meters. The deepest penetration was some 34 m into the surface cliff. Many of the relict caves are scalloped and display well-de“ned paragenetic wall and ceiling half-tubes, implying that they are indeed conduits, leading further into the rock mass, beyond their present permafrozen terminations. Most of the speleogenetic volume of the reli ct caves is ascribed to sub-glacial conditions during stadials, when the site was covered beneath thick ice sheets. In many cases, the present caves were formed by reactivation of pre-existing paleokarst voids. Due to the present intense gelifraction and erosion in the littoral zone, and the relatively constant sea level during the past 9.5 kyr, most of the volume of the sea cave s can be explained by processes acting during the Holocene. Keywords: karst; speleogenesis; arctic; permafrost; Quaternary; subglacial Corresponding author: stein.lauritzen@geo.uib.no by the authors. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license

PAGE 2

Lauritzen S.-E. 2online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENpermafrost is limited to depths up to tens of meters and occurs in patches, often restricted to frost-susceptible materials. It occurs at lower latitudes than the continuous variant, and may also exist in the high alpine environment in sub-arctic regions. Taliks are local holes in otherwise continuous permafrost. They may act as sites of groundwater recharge or discharge. Karsti“cation is dependent on ”owing, corrosive water, and would therefore be inhibited by permafrost. Exceptions would occur in the active layer and in taliks of various kinds. In a permafrozen area, two principally different aquifers could exist, the seasonally active supra-permafrost aquifer (within the active layer), and the perennial sub-permafrost aquifer that is con“ned beneath the permafrozen ground. Sub-permafrost groundwater may go very deep, and therefore also transport geothermal heat and brines (Deming et al., 1992). Meteoric recharge of the sub-permafrost system is only possible through taliks. This makes subpolar glaciers in the upland the most likely source of groundwater recharge. Artesian conditions may result when glaciers occupy elevated positions in the surrounding landscape, thus giving rise to pingos and springs that are well documented from arctic areas e.g. (Liestl, 1976), Fig. 1. Subpolar (polythermal) glaciers are partially cold-based, and frozen only at the lower parts, below the equilibrium line, where it is free from snow. Due to the insulating effect of the glacier ice and snow cover, geothermal heat may cause the upper parts of the glacier to become warm-based, i.e. reach the pressuremelting point of ice. Therefore, this unfrozen zone becomes the most important area of groundwater recharge. Consequently, the extent of warm-based glacier cover would be expected to affect the amount of groundwater recharge and therefore be sensitive to climatic change. Glacial retreat during modern time (for instance, as we have seen on Spitsbergen) may therefore explain numerous historical observations of decreasing ”ow from many springs (Haldorsen and Lauritzen, 1993). Paradoxically, glacier retreat in this climate increases the snow-free ablation area, where the glaciers become cold-based, in turn shrinking the warm-based upper reaches where aquifer recharge takes place.1.3. Glacier hydrology and karstGlaciers may be viewed as aquifers, which convey water through discrete conduits, and to a lesser extent through the intercrystalline vein system. Under normalŽ conditions, we may regard a glacier as a perched aquifer resting on an impervious bed (aquiclude), Fig. 2a. Water transport is intensi“ed along the aquiclude contact through R-channels (Rthlisberger, 1972) and N-channels (Nye, 1965), or through a more irregular, linked cavity system (Walder and Hallet, 1979)1. Basal sliding is largely dictated by water storage and hydrostatic pressures along the aquiclude contact, i.e. the basal water “lm (Paterson, 1981). Characteristic features of the glacier aquifer are that the water budget is strongly seasonal, and that the interior channel geometry is thought to vary seasonally in concert with the hydrological conditions. Most interesting and complex cases occur when the glacier aquifer leaks into and interacts with karst aquifers below (Lauritzen, 1991), Fig. 2b. We may expect at least two different effects from this interaction. First, the ice would either erase or even preserve karst forms, depending on their size and permeability prior to glacierization. As mentioned, the sliding velocity of a temperate glacier is directly dependent on the basal water pressure. If the magnitude of leakage from the glacier bed into the karst is large relative to the subglacial water budget, we should expect it to affect the basal sliding velocity of the glacier. Removal of liquid water (i.e. heat) would in turn cause the ice to freeze to its bed. Consequently, plucking and quarrying becomes the prominent erosional effect, as glacial scouring is inhibited. However, if the mechanical strength of the karst forms is greater than the deformation strength of the ice (i.e. larger mogotes or dolines), these landforms may become “lled or surrounded with stagnant ice and till. Ice ”ow is then restricted to zones of deformation higher up in the glacier. The net effect is that smaller karst forms tend to become erased by glacier action, whilst large forms may, in fact, be preserved and in“lled with glacial drift. Second, superposition of glaciers on karst would enhance karsti“cation by increasing hydraulic gradients, particularly in alpine areas. Such an increase in water supply and pressure gradients may stimulate speleogenesis and create caves underneath the snout of alpine glaciers occupying plateaus above deep valleys. Many cases of pro-glacial, invasion-type 1 R-channels are subglacial conduits where walls and ceiling is formed in ice; they are therefore dynamic features, their size and position in the glacier bed will change with (seasonal) hydraulic conditions. N-channels are subglacial conduit that utilize some depression or fracture in the bedrock so that ”oor and at least one wall is bedrock; they tend to be more stable towards changes in subglacial hydrology. Fig. 1. In areas of permafrost (shaded), groundwater recharge may take place from subpolar glaciers in the upland. Groundwater ”ow is con“ned beneath permafrost; artesian conditions may force the groundwater up to the surface and form pingo-type springs. After Liestl (1976).

PAGE 3

3Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENcaves are known (Glazek et al., 1977; Ford and Williams, 1989). In a more generalized situation, the spread of ice sheets during glacial periods made large areas available for subglacial, thawed ground, and karst aquifers could become activated at any altitude within a valley topography, so that phreatic caves could be formed in apparently impossibleŽ positions. We may call this subglacial speleogenesis The mechanism was “rst recognized in Norway by G. Horn (1935, 1947), based on much earlier observations on sub-polar glaciers by Werenskiold (1953); see also the English translation by McCrady Horn, (1978). This mechanism is now generally recognized as a speleogenetic process (Ford, 1984; Ford and Williams, 1989).2. Background2.1. Geographic and geomorphic settingBlomstrandsya, situated vis-a-vis the settlement Nylesund in Kongsfjord (780 57N), covers an area of some 32 km2, Fig. 3. The summits reach around 350 m a.s.l. with steep sides on the NW and N sides. In this respect, the entire island may be viewed as a huge roche moutonne with relatively gentle slopes up-ice (SE) and steep lee-side escarpments (N and NW). There is also a distinct platform at 10 30 m a.s.l. around the coastline. This surface is developed in bedrock and may be viewed as a strand”at, e.g. (Holtedahl, 1996). It is best developed on the SE facing coast, where it is about 1 km wide. Of the ca. 32 km circumference, some 2/3 (20 km) is a rocky shoreline with steep coastal cliffs, the remaining consist of beaches of reworked glacial sediments. Only the central basin and the northern and eastern slopes have any thickness of glacial drift, otherwise the island is barren and rocky. With the high arctic latitude (79 N) and a mean annual temperature of -4 to -5 degrees, Svalbard is within the area of continuous permafrost. Based on observations in coal mines (e.g. the Kings Bay mines in Ny-lesund), permafrost has a general depth of 200 … 400 m (Liestl, 1976, 1980). 2.2. Bedrock and tectonic settingThe bedrock consists predominantly of medium to high-grade metamorphic marbles of paleozoic age belonging to the Generalfjella Formation (Hjelle, 1979). This is part of the basement system (Hecla Hoek) with a deformation phase most probably corresponding to the Caledonian deformations in Greenland and Scandinavia. In Blomstrandsya, the general strike is N-S with two syn-metamorphic phases of isoclinal folding, followed by post-metamorphic crenulation and kink-folds associated with west-directed thrusts and imbrications (Thiedig and Manby, 1992). Small patches of unmetamorphosed, post-Caledonian (possibly Devonian) sediments are widely scattered over the area. These redbed deposits represent the eroded remnants of a previously more widespread blanket of Devonian sediments. The sediments are either preserved in wedges associated with faulting and brecciation of the underlying marbles, or they occur in fan-shaped bodies “lling channels and tubes in the marbles (Thiedig and Manby, 1992). Often, redbed material was also injected into the mesh of calcitic veins that penetrate parts of the marbles. Clastic, in situ redbed in“lling in channels and conduits clearly indicate a karsti“cation phase at this time, Fig. 4.Fig. 2. Interaction between glaciers and karst, from Lauritzen (1991). A) Normal situation of a glacier (which may be regarded as an aquifer) situated on an impervious bed. Practically all supraglacial and englacial drainage emerges through the spring at the snout of the glacier. The hydrostatic pressure of the subglacial water, which ”ows through R-, Nchannels, linked cavity systems, or is stored in the subglacial regulation “lm, plays a major factor (as lubricant) for basal sliding in temperatebased glaciers. See text for further discussion. B) The same glacier situated on a well-developed karst, where subglacial water is pirated into the karst conduits. This may considerably reduce the basal water pressure, thereby impeding basal sliding. See text for further discussion.

PAGE 4

Lauritzen S.-E. 4online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN2.3. Quaternary geologyThe glacial history of western Spitsbergen is quite well known through the last interglacial-glacial cycle. During major glaciations (MIS 6, 5d, 4 and 2), the ice sheets advanced onto to the continental shelf, some 50 km offshore from the mouth of Kongsfjorden (Mangerud et al., 1998), implying that the Kongsfjord area in those periods was covered beneath several hundered meters of ice. Since the Saalian glaciation (MIS 6), there have been two major interglacials the Eemian (ca 130 … 118 kyr) and the Holocene (10 kyr present ) and two interstadials Phantomodden (ca 108 … 75 kyr) and Kapp Ekholm (ca. 5025 kyr), (Mangerud and Svendsen, 1992). The chronology of the last deglaciation and shoreline displacement in the Kongsfjord area has been reconstructed from shorelines, glacial deposits and 14C dates on associated marine materials (Lehman and Forman, 1992; Mangerud and Svendsen, 1992), Fig. 3. After the Late Weichselian Maximum (1918 kyr; Glaciation G by the workers cited above), deglaciation commenced around 16 kyr and reached the present coastline at 12.5 kyr BP, when a late glacial maximum marine shoreline was formed (MG in Fig. 5). At 12.5 14C kyr BP Kongsfjorden was still “lled with ice, but mollusc dates of 9.44 14C kyr BP on an interior location in the fjord imply that Blomstrandsya had emerged from the ice at this point. The corresponding shoreline displacement curve is shown in Fig. 5. At about 9.5 kyr 14C BP, sea level fell rapidly and has stayed constant ( 5 m) since. At the same time, the Fig. 3. Location of Kongsfjorden (Kings Bay) and Blomstrandsya in NW Spitsbergen. White: ocean with depth contours in grey. Dark shade: open land. Light shade: Glaciers and ice-“elds. Dashed lines: extent of the ice sheet during the late Weichselian deglaciation. Numbers in rectangular labels are dates derived from shells in 14C years BP; these dates are minimum ages for icefree conditions at the site, implying that the fjord was completely deglaciated at 9400 BP. Deglaciation and age data from Lehman & Forman (1992). Geography from the Norwegian Polar Institute (NP). Fig. 4. Various types of redbed (Devonian?) inliers. a) Tectonic wedge of boulder breccio-conglomerate, northern shore of Blomstrandsya. b) Vertical paleokarst conduits “lled with redbed sediments, west of Ny-London. (Photo a: S.E Lauritzen; b : Narve Ringset). Fig. 5. Shoreline displacement curve for Kongsfjorden, after Lehman & Forman (1992). At about 9.5 kyr 14C BP, sea level fell rapidly and has stayed constant ( 5 m) since. At the same time, the entire Kongsfjord became icefree (Fig. 3), implying that the coastline of Blomstrandsya has been exposed to open water at a fairly constant sea level during the last 9 kyr. MG: late Weichselian marine limit, IG: interstadial/interglacial.

PAGE 5

5Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENentire Kongsfjord became icefree (Fig. 3), implying that the coastline of Blomstrandsya has been exposed to open water at a fairly constant sea level during the last 9 kyr. In historical time, Blomstrandsya was a peninsula, connected to the mainland by the Blomstrand glacier, which is evident from older maps. Today, the ice front has retreated, leaving a small sound between the island and the ice wall.2.4. Previous workJean Corbel visited Blomstrandsya in the 1950ies (Corbel, 1957) pp. 38 43 which is perhaps the only previous speleological description of the area:  ƒ normalement les grottesŽ de Blomstrand sont soit des failles ouvertes, soit des reinsures de dissolution fossils ƒ. dorigine marine. Elles nont pas de profondeur notableƒ Corbel recognised the paleokarst with redbed in“llings, but he was unable to date them ( op. cit. p. 41). Corbel also noted the large sea caves along the coast, visited at least one of them, for which he suggested a paleokarst origin, i.e. a pre-existing karst cave, modi“ed by the thawing and eroding action of seawater. He also noted fresh water ice at its terminal, permafrozen end. 2.5. Material and methodsExisting maps (1:100 000, contour interval 50 m) issued by Norsk Polarinstitutt (NPI) are of little use for locating karst forms in the “eld. A new “eld map of Blomstrandsya was constructed from existing contour data and details from aerial photographs. In particular, a new coastline was constructed with minute details, allowing the exact location of individual sea caves. In the absence of GPS (1992), objects were located by Fig. 6. Geology and speleology of Blomstrandsya. Geological features adapted from Thiedig & Manby (1992). Nameplaces are in part of“cial (Norsk Polarinstitutt) and unof“cial.

PAGE 6

Lauritzen S.-E. 6online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENusing an aneroid altimeter and compass bearings to prominent landmarks. Seacaves were located by boat off-shore; some could be explored by boat or by wading using survival suits. Secondary calcite was dated by the U-series method, using conventional laboratory techniques3. Results and discussionThe observed karst landforms (caves) may be divided into three main categories, according to morphology and in“lling material: pre-glacial or paleokarst caves, subglacial caves and active, post-glacial (marine abrasion) caves. The classi“cation only refers to the most obvious genetic mode and does not exclude the possibility of a polygenetic origin, which is certainly the case for several of them. Closed depressions and bogaz forms are also described and discussed here, because caves are in some cases associated with them. Topography, geology and karst landforms are shown in Fig. 6; a key to speci“c objects discussed in detail is shown in Fig. 7.3.1. Closed depressions and bogaz forms.Dolines of moderate size (up to about 10 m diameter) are only found close to the seashore, where coastal caves provide local taliks. Here, collapse and suffusion dolines have developed in conjunction with vertical “ssures and blow-holes (Fig. 11a,d & e). Similar, thermal talik (suffusion) dolines are also known from shingle beds on elevated beach deposits elsewhere in Spitsbergen (Salvigsen and Elgersmaa, 1985). Much larger and possibly older karst depressions also exist, of which lake Irgenstjernet is the best example (Lauritzen, 1998). The small lake is hosted inside a closed depression, some 500 m in diameter, Fig. 6 & 8. Most of the circumference is bedrock, although the southern edge was covered in snow and probably has a till blanket. The depression is situated on the summit of a major hill in the ”anks of Irgensfjellet, a peculiar topographic situation that is not easily compatible with a glacial origin. One would expect glacial overdeepening to operate in topographic lows, where an ice stream is accelerated; not on the summit of a topographic high, which would tend to divert ”ow rather than concentrate it. The fact that it is not “lled up with drift from overriding ice sheets, is compatible with the raised topographic position, presumably above sediment-laden basal ice. Moreover, according to the anticipated mechanism of interaction between large karstforms and glaciers discussed above, we should expect the depression to become “lled with (relatively clean) stagnant ice so that ice ”ow would pass over it, not into it. The depression is therefore almost certainly of karstic origin; the large size and the fact that it supports a lake today points to its relict nature and great age. The present lake drains underground through a shallow, suprapermafrost system to a small spring some hundred meters to the north. The in“ltration is therefore inhibited by permafrost because the lake itself is not wide or deep enough to support a talik, also it is far too large to have been formed by post-glacial Fig. 7. Locations of objects that are discussed in detail in the text. Fig. 8. Irgenstjernet, a large closed depression on the eastern summit of the island. Persons for scale in circles on the snowpatch, Blomstrandbreen in the background, looking east.

PAGE 7

7Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENcorrosion. Hence, the depression is relict and of preor interglacial origin. Bogazforms (i.e. karst streets or corridors) occur in three places on the island. The largest and most complex feature ( Bogaz I ) is situated north of NyLondon. This is a complex melange of collapsed and frost-shattered amphitheatreshaped depressions and karst streets. Small caves are found in hanging positions in these escarpments. The locality appears to be guided by a series of imbricated faults with intensive fracturing (Thiedig & Manby op cit ). Bogaz II situated in the slopes between Grotteveggen and Grottevika, comprises a 150 m long, linear karst street that cuts across a ridge in the landscape, Fig. 9. It is oriented N-S, i.e parallel with the shoreline and cliffs. This orientation is also compatible with the major ice ”ow through Kongsfjorden. The bogaz is about 20 m wide, with relatively steep walls, the ”oor is covered with cryoclastic regolith. Soli”uction lobes in its southern end indicate that there is “ne-grained glacigenic material beneath the regolith. The apex comprises a pronounced knick-point where the bedrock ”oor is also exposed. Several small caves and rock-bridges are exposed here. These caves display smooth walls and tube-shaped cross sections where they are not damaged by frost shattering. They either become too tight or choked with frozen material. Bogaz III is situated in Rundsvaryggen, east of Hansneset, Fig. 6 & 7. It is much less pronounced than the other two, in part transformed into a small canyon carrying a meltwater stream. It is oriented E-W, in the direction of the ice movement on the northern part of the island. One major, relict cave Dobbeltgrotta G-18, has a similar orientation and is situated nearby ( vide infra) In summary, all three landforms are oriented parallel with known or anticipated ice movement directions around the island, as judged from glacial striae and general topography. This is compatible with the widely-held view that bogaz-like forms or shallow corridor karst may have accommodated and been formed by subglacial drainage routes (Ford, 1984; Ford and Williams, 1989).3.2. Active caves and sea cavesDue to the permafrost, active caves, i.e. caves that are in contact with (aggressive), liquid water, are only found within the shallow active layer and together with marine taliks. Hence, such caves are almost exclusively sea caves situated in the steep coastal cliffs around the island. The only signi“cant supra-permafrost cave found, is the small resurgence cave named Jakobskelda (Fig. 11b). This is a small tubular passage, situated some 5 m above sea level, discharging a small brooklet of meltwater of relatively low electrical conductivity. The cave is only a few meters deep and ends in cryoclastic debris, under which the little stream emerges. Only the outer (nonfrozen?) part of the cave display a small vadose trench. 62 active sea caves were located around the rocky coastline, Fig. 10 & 11. Their size varies from relatively tight “ssures to large, arched features (< 10 m wide), which could be explored by boat. Regular, arched entrances are found in relatively homogeneous rock, suggesting that such pro“les are a result of spalling and dissipation of stress, similar to a glacier mouth (Fig. 10e). Where prominent tectonic structures occur, like thrustplanes and normal faults, the cross sections become irregular and dictated by these structures (Fig. 10c, Fig. 9. Bogaz with small caves on the western side of Blomstrandsya. Plan map (top) and longitudinal section (bottom), arrow indicates inferred ice and water ”ow. Legend: 1: cryoclastic regolith. 2: Soli”ucted mix of soil, glacial drift and cryoclastics. 3: steep edge. 4: slope. 5: cave.

PAGE 8

Lauritzen S.-E. 8online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN Fig. 10. Various sea caves along the rocky shores of Blomstrandya. a) Arched entrance to a chamber with fairly round cross section, southern shore. b) Interior of same. c) Large seacave with irregular outline, guided by faults and thrustplane, northern shore. d) Interior of seacave with beach, Grottevika landing site. Guiding fracture is a sheared bedding plane. e) Mouth of Blomstrandsbreen, illustrating an arched sea-cave entrance, the shape being dictated by spalling in a relatively homogenous material, cfr. a) and b). f) Fissure-guided sea-cave, guided by faults. The cave has exploited the damage zone between the two faults. g) Small alcoves and a through cave, Ny-London Bay. h) interior of a deep seacave, frozen seawater and ground ice. All photographs: S.E. Lauritzen.

PAGE 9

9Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN Fig. 11. Various sea caves along the rocky shores of Blomstrandya. a) Large seacave guided by vertical fractures, with well-developed blowhole. Southern shore, this cave was also depicted by Corbel (1957). b) Jakobskilda, a small karst spring emerging in the coastal cliff face. The spring conveys supra-permafrost water. c) Small seacaves, Ny-London bay, guided by thrustplane. d) Suffosion doline developed in glacial drift through a blowhole. Southern shore. e) Large arched seacave, with cenote-like blowhole, developed in contact with a Devonian inlier, northern shore. f) Arched sea-stack and remnants of cave, northern shore. Note the abrasion notch at normal tide level. All photographs: S.E. Lauritzen.

PAGE 10

Lauritzen S.-E. 10online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENf, 11e), although low dip faultplanes can be seen to guide arched entrances as well (Fig. 10d, 11c). The caves are never very long; the maximum horizontal distance explored from the surface was about 15 m. The caves either terminate in vertical rock walls (deep water, Fig 10a, b), or in small shingle beaches (Fig. 10d, g, 11a, c). Deeper terminations (“ssures) taper out to impenetrable dimensions and are always frozen (Fig. 10h). The seacaves generally lack sediment chokes and ground ice plugs as is found in the relict caves in the high cliffs.3.3. Relict cavesMore than 30 caves or cave entrances have been detected in cliffs around the island, Fig. 6, 7, 12 & 13 and Table 2. None of these caves have previously been reported. Compared with previous reports (e.g. Corbel op. cit. ), they are new and quite unexpected discoveries. Although none of these caves could be explored to any great depth due to ground ice, they may indeed be quite deep and they are certainly karst conduits distinctly different from the marine abrasion caves described above. Caves C1 C4 are truncated counduits exposed in the walls and slopes of Grottebekken streamcourse. In spite of being quite small, they are fracture-guided tubes (Fig. 12a). G1G3 are associated with Bogaz II, as discussed above. G4 and G5 are situated in the cliff face named Grotteveggen (cave wall). G4 (Fig. 14) is the smallest of them. It is “lled with red soil and cryoclastic debris and ends after eight meters in permafrost. The slight break in slope of the cave ”oor as seen in the longitudinal section may be viewed as the front of a small soli”uction lobe. Scalloping on the eastern wall indicates former water ”ow out of the cave. Portalgrotta (G5), at 72 m a.s.l., is the largest and most spectacular of the caves, having a large portal entrance that is visible from a great distance (Fig. 12 & 15). Most of the entrance has smooth, scalloped walls, indicating in”uent (northwards) water movement (Fig. 12f). Scallops further into the cave con“rm this observation. The ”oor is covered with the same mixture of red soil, cryoclastic blocks and gravel. The cave terminates in a permafrozen choke 34 m from the surface cliff. The extensive, N-S trending cliff wall, Grotteveggen displays about 10 entrances that require technical climbing to reach. Underneath many of these entrances, the cliff face is stained red from a bleed of sediments out of the caves, indicating that they (like the accessible G4) have redbed-derived “lls. G17 and G18 (Fig 13a) are located underneath the escarpment through which Bogaz III is cut, i.e. they would be on the lee side of ice movements coming out Srvgen sound. G17 is a straight, 9 m deep passage, ending in frozen regolith. G18 is a complex, partially unroofed feature. The area named Rundsvaryggen (roche moutonne hill) displays several glacially striated whaleback forms, several of which have cave entrances underneath their leeside escarpment (G15 & G16, Fig. 13b), and a few also have a corresponding entry at their up-ice slopes. These holes also bleed redbed “nes. A series of very interesting cave fragments are found in the steep cliffs around Srvgen. The Hole-in-the-wall, or Srvggrotta (G21) is a spectacular opening, Fig. 13g, 16. The cave possesses several short side passages and a small aven and it is distinguished by having small speleothems. These speleothems appear as basically hardened moonmilk, forming rounded drapes and stalactitic forms. Some frost-shattered specimens were collected for dating. As seen in many other places in Svalbard, speleothems are almost absent, but do occur in caves underneath bird colonies. This is also the case here, suggesting that biogenic CO2 and organic acids from the guano may trigger precipitation of secondary carbonate. The inner surfaces of G21 are quite frost-shattered, although scallops of indeterminate direction are found on the inner wall. Perhaps the most signi“cant caves occur in the Brattlikollen face. Here, a series of phreatic caves are situated at high altitude (226 … 280 m a.s.l.). G22, Takrret (the rooftube) is a cave with a well-preserved phreatic morphology, Fig 17. There are several ceiling and wall half-tubes; these are well scalloped and are continous with phreatic side passages that branch out from the main chamber, Fig 13d..f. The phreatic passages are developed asymmetrically around the guiding fracture, so that the tube is widened much more above the guiding fracture than below it. Collectively, these features are characteristics of paragenetic development; the passages evolved under phreatic or epiphreatic conditions in contact with a sediment “ll (Lauritzen and Lundberg, 2000). Possible remains of this sediment “ll can be seen as patches of a brown, clayey diamicton still in contact with the ”oor and walls of the tubes. There are a number of other small scalloped caves (tubes) in the area around G22, of which G27 contains fragmented deposits of carbonate-cemented quartz gravel. The cement was dated with U-series technique ( vide infra ). The main passage of G29, Termalgrotta, situated in the eastern ”anks of Irgensfjellet, intersects a completely “lled calcite vug, Fig. 18. The outline of this vug is quite different from the numerous extentional calcite veins on the island that display parallel (fracture) walls. Here, the sparry calcite has “lled an irregular space with bulbous, semicircular outline. Clearly, the host rock must have been corroded prior to “lling with calcite, suggesting that this feature is the remnant of an early phase of hydrothermal, hypogene karsti“cation. The calcite vug does not display the common red stains seen in some of the tectonic veins on the western shores of the island. In both sea caves and relict caves, permafrost is encountered after less than 10 m from the surface, and no penetration was possible beyond 34 m. This is in good accordance with observations in other high arctic, permafrozen areas. Tsi-tcheHan cave in the Yukons appears to penetrate 40 m horizontally from the outer portal, and 27 m from the “rst constriction (Lauriol et al., 2001), Icedam Cave Alaska penetrates 20+ m (Juday, 1989). Maximum depth of exploration in Yukon is 100 m horizontally and 50 m vertically (B. Lauriol, p.c .).

PAGE 11

11Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENFig. 12. Relict karst caves. a) Small phreatic tube (C2), Grottebekken. b) Steep escarpment of Bratlikollen, adjacent to strand”at (left arrow). From left, the portal of G5 and the smaller entrance of G4. c) North end of Bogaz II seen facing S. Apex knickpoint with caves, left G3a..c, right arrow. G2. d) Interior of G4, showing cryoclastic fragments and red cave soil. e) Entrance pro“le of Portalgrotta (G5), mountains surrounding Ny-lesund in the distance. Note the smooth cross section, little disturbed by gelifraction. f) Scalloped wall in entrance of G5. Water ”ow was towards the camera, i.e. into the cave. A ll Photographs: S.E.Lauritzen.

PAGE 12

Lauritzen S.-E. 12online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN Fig 13. Relict karst caves. a) cave entrances in the lee side of Rundsvaryggen. Left G18, right G17. b) G15 and G16, ef”uent passages beneath the lee side escarpment of roche moutonnes. Both caves bleed red sediments into the soli”uction lobes in front of them. c) terminus of G5, 34 m from the surface cliff. Ground ice and hoarfrost. d) Phreatic tube continuing as paragenetic ceiling half-tube, G22 (Takrrgrotta). Red-brown diamictic sediment in lower front part of the tube. e) entrance silhouette of G22, at 226 m a.s.l. overlooking Srvgen and Blomstrandbreen. f) Ceiling half-tube (with hoarfrosted scallops) continuing into phreatic conduit, G22. g) the hole-in-the-wall, entrance to G21, Srvggrotta in the northern wall of Irgensfjellet. All photographs: S.E. Lauritzen.

PAGE 13

13Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN NameLocation and descriptionLength (m)Depth (m) Alt. (m) Grotteelv C1 Small cave “lled with cryoclastic material in Grotteelva canyon. Plugged with ice after 2 m 21125 Grotteelv C2Sloping passage, 0.3 m2, cryoclastic debris and ice plug41130 Grotteelv C3Phreatic tube in canyon cliff, trends 22g, soil and cryoclastics20.5140 Grotteelv C4Small cave sloping upwards, “lled with cryoclastics. Trends 105g. 20.5140 C5Small cave with water-table and ice1.50.363 G1Cave at SE end of Bogaz20.335 G2Small tunnel cave in Bogaz2.5145 G3aSmall rock-bridge with vertical opening. N wall of Bogaz2162 G3bCave “lled with cryoclastics from bogaz, with scalloped ceiling1052 G3cSmall phreatic tube in bogaz wall, ca 20 cm diameter1052 G4 Fault-controlled tube in S-facing cliff. Ef”uent scallops. Cryoclastic “ll. Ends in ground ice 11.5170 G5 Portalgrotta Huge portal visible from Ny-lesund. In”uent scallops. Ends in ice and frozen debris. 39272 G5aPhreatic tube in cliff wall, NW of G5 72 G6Cave entrance beneath snowpatch G7Small cave next to G4 70 G8Filled conduit, 1 m diameter1.527 G9, G10, G11Cave entraces in cliff face, bleeds old-red derived sediments42 G14Small cave 25 G15 Rundsvagrotte 1 Cave below cliff face on lee side of roche moutonne. Filled with ground ice after 3 m 3172 G16 Rundsvagrotte 2 Cave with weathered scallops, “lled with soli”ucted soil and cryoclastics. Lee side of roche moutonne. 72 G17 Fissurgrotta Quite long straight passage along fracture, ends in hoarfrost and frozen sediment “ll. 9160 G18 Dobbeltgrotta Branched cave formed along veins of hydrothermal calcite, ceiling hole (blowhole?) small veins of palygorskite. 15560 G21 Srvggrotta, The-hole-in-the-wall Spectacular entrance in cliff face, vertical aven, tiny spelothems. Cryoclastic “ll and ground ice. 401577 G22 Takrrgrotta Entrance high in N cliff of Bratlikollen, facing Breyane and Nordvgen. Filled with ice, snow and hoarfrost. Branched phreatic passages and paragenetic ceiling tubes with scallops. Emnants of old red …derived “nes on the walls 184226 G23. Bratlirr IPhreatic tube in cliff wall, 50 … 70 cm diameter20.5229 G24 Bratlirr IIPhreatic tube of simailar dimensions as G23255 G25 Bratlirr III Phreatic cave consisting of several tubes, end in ice chokes. Scalloped, but ambigious ”ow direction 83243 G26 RunderretNetwork of phreatic or paragenetic tubes246 G27 Kronegrotta Short, wide phreatic tube, ending in ice and cryoclastic “ll. Wall and ceiling tubes. Deposits of calcite-semented quartz gravel. 83263 G28 Buegrotta Entrance choked with cryoclastics. No clear solution forms. Curvature of roof arch suggests a very large cave underneath. 2172 G29 TermalgrottaThrough cave formed along hydrothermal paleokarst in“lling.179215 G30 Temptation cave Circular entrance above G27, visible at distance from Srvgen. Not explored--280?Table 1.Karst caves above sea level

PAGE 14

Lauritzen S.-E. 14online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN3.4. Uranium-series dating of speleothemsSamples were collected from G21, Srvggrotta and from G27, Kronegrotta. In G21, a piece of frost-shattered wall ”owstone was taken, and in G27, the cemented gravel was dated. Sample G21 is a massive, dense ”owstone crust, some 3 cm thick, which was dated in bulk using alpha particle spectrometry, Table 2. Samples were prepared according to standard procedures (Ivanovich and Harmon, 1992), and processed by tailored software (Lauritzen, 1993).The G21 sample dated to 34.2 1.9 kyr. The measured 230Th/232Th activity ratio is 11, indicating that the sample was contaminated with non-authigenic 230Th at the time of deposition, resulting in an apparent age that is too high. This contamination can be modeled (Schwarcz, 1980; Richards and Dorale, 2003) using an earth mean initial 230Th/232Th activity ratio (B0 = 1.5), which yields a corrected age of 30.5 2.06 kyr. From the isochron regression ( vide infra ), a local B0 = 0.40 can be worked out for the carbonate cement in G27, using this as a correction factor, yields a corrected age for the G21 ”owstone at 33.2 kyr (large errors), which suggests that the estimate of 31 … 33 kyr is relatively robust. The cemented gravel bank was dated by the isochron technique using successive leaching with acids of increasing strength (Table 3). These results were then plotted as Broecker isochrons, the slope of which yields the uncontaminated values of activity ratios 230Th/234U and 234U/238U (Schwarcz and Blackwell, 1992). These values (Fig. 19) convert to an age of 23.9 5.8 kyr. Both ages fall within a known, relatively warm period in Svalbard, the Kapp-Ekholm interstadial (5025 kyr, Mangerud & Svendsen op. cit. ) and is compatible with the idea that speleothems that are fed by meteoric water would grow preferably under interstadial or interglacial periods when soil respiration and water circulation may take place, e.g. (Lauritzen, 1995). 4. Speleogenesis4.1. Paleokarst phasesPossibly the earliest trace of karsti“cation is the calcite vug in G29. This phase was probably earlier than the later, well-documented Devonian karsti“cation. The Devonian karst phase is associated with many deposits of redbed “nes. Thiedig Fig. 14. Cave 4, located in Grotteveggen adjacent to the Bogaz II of Fig. 9. Note the ef”uent scallops at the western wall. Symbols as in Fig. 9; crossed shading is snow or perennial ground ice. Arrow with hook: scallops and inferred ”ow direction. Fig. 15. Portalgrotta (G5), the most prominent and largest cave in Grotteveggen. This is the deepest cave found so far, penetrating 32 m from the cliff wall, before the passage terminates in frozen sediments and ground ice. Note the in”uent scallop ”ow direction and the arched cross sections.

PAGE 15

15Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGEN& Manby op. cit. also claim these deposits to be arranged radially out from the center of the island, suggesting that it was a topographic high also in the Devonian. These paleokarst channels and in“lls have later been eroded and intersected by Quaternary glaciation. 4.2. CavesIn the following, when we discuss the speleogenesis of the numerous relict caves described, we shall assume that the speleogenetic agents that operated in one site, also could operate in other sites, although not necessarily at the same time. It is tempting … but probably wrong … to assume that the last time favourable conditions and agents were available was also the time when they were actively forming the caves. Only dating and independent evidence may solve this question, so one is often restricted to link speleogenetic facies (Lauritzen and Lundberg, 2000) to conditions and agents rather than timing. It must also be kept in mind that all these objects are incompletely known fragments obscured by frozen sediments. Their total extent is unknown and presently beyond observation. Therefore, the observed cave entrances might either be all there is, or they might be former entrances into much more extensive conduits that are “lled with ice and frozen sediments. Here, wall scalloping, combined with hydraulic continuity, provide crucial independent evidence for continuation. A scalloped passage fragment which displays a de“nite ”ow direction has indeed transported water from one point in space to another and must therefore be regarded as a conduit. If we can reject the possibility that the passage in question was a mere rock-mill (eddy), then it follows that the conduit must have a continuation beyond the frozen “ll, into the rock mass. On the contrary, if there is no apparent continuation, like in the sea caves, we have to invoke Fig. 16. Srvggrotta (G21, or The-Hole-in-the-wall) in the NW face of Irgensfjellet. Fig. 17. Takrrgrotta (G22), a cave with welldeveloped and scalloped half-tubes in walls and ceiling. Northern face of Brattlikollen.

PAGE 16

Lauritzen S.-E. 16online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENspeleogenetic agents acting on the surface with only limited penetration abilities. Another limitation is that scallops … and other speleogenetic facies sculptured into cave wallsonly represent the last time slice of the cavitys evolution. A scallop pattern, which generally has a relief of 1030 mm, would as a minimum, require from some 102 to 103 years to form and erase any previous pattern. 4.3. Formation of sea cavesWith very few exceptions, all observed sea caves are located up to about 5 m a.s.l. in cliff faces that are actively eroding today (10 m if we include all blow-holes). In spite of being in contact with the marine talik, none of them are very deep. As mentioned, they are of two types, either short, broad chambers with arched cross sections, or elongated, “ssureguided passages with tall and narrow cross sections. The difference between the two types seem linked to the attitude of guiding fractures (Figs. 10 & 11). The caves end either in solid rock, sometimes frozen, or in shingle beaches. The altitude distribution of all caves, including sea caves, is depicted in Fig. 20. Of these, the sea caves outnumber all other caves, and they are (of course) clustered in the interval [-5 .. +10] m a.s.l. When compared to the postglacial sea level curve (Fig. 4), there is a striking similarity between the number of sea caves and the constant sea level since 9.5 kyr. The number of caves and the time post-glacial sea level has spent at various levels correlate, and (relict) caves occur at much higher levels than any known former sea level. This does not necessarily mean that all caves are sea caves ( vide infra ), but it strongly suggests that the large number of active sea caves are indeed connected to the present-day sea-level and Fig. 18. Termalgrotta (G29), a “ssure cave with rounded corrosion vugs “lled with hydrothermal calcite. Northern slope of Irgensfjellet. J.no U (ppm)234U/238U230Th/234U230Th/232Th Age (kyr) Corrected age (kyr) 4400.39 2.352 + 0.108 0.277 + 0.01311.77 34.2 + 1.930.5 + 2.06 (The dates were done at the Quaternary U-series dating laboratory at Bergen University. The respective journal numbers are refe rred to in the “rst columns of tables II and III).Table 2. U-series dating of speleothem G-21 Srvggrotta J.no234U/238U230Th/234U230Th/232Th234U/232Th238Th/232Th 4380.996 + 0.0090.470 + 0.0061.022 + 0.0062.173 + 0.0072.182 + 0.020 4501.486 + 0.0490.299 + 0.0121.294 + 0.0114.334 + 0.0162.917 + 0.215 4511.042 + 0.0140.262 + 0.0040.810 + 0.0043.094 + 0.0042.970 + 0.042 4521.060 + 0.0231.272 + 0.0500.483 + 0.0500.380 + 0.0690.358 + 0.073 4581.675 + 0.0680.694 + 0.0280.775 + 0.0281.117 + 0.0290.667 + 0.090 4591.197 + 0.0230.542 + 0.0130.507 + 0.0130.935 + 0.0100.781 + 0.025 4601.220 + 0.0440.784 + 0.0310.353 + 0.0310.450 + 0.0260.369 + 0.038Table 3.U-series Isochron data Svalbard G-27

PAGE 17

17Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENprocesses active in the littoral zone. It is also possible that some of the sea caves may have originated from pre-existing karst conduits or karsti“ed zones of weakness. Most of the sea caves in Blomstrandsya face open, deep water in the fjord, Fig. 21. This trend is also evident when one looks at the number of caves per km of rocky coastline. Also, the largest caves invariably face open deep water. This strongly suggests that wave energy was important in their formation, either in the form of direct breakers (e.g. blowholes), or by disintegrating and moving gelifracted erosion products into deeper water. Sea caves in non-karstic rocks are often formed along zones of weakness (fractures) by frost shattering, combined by wrenching effected by ice rafts and tidal cycles, e.g. (Aarseth and Fossen, 2004a, b). Moreover, wave breakers cause enormous pressures and cavitation as well as moving large boulders forth and back in a rock-mill like fashion. The rate of frost-weathering under arctic coastal conditions can get quite high. Andr (1998) quotes that for strongly fractured sites in Spitsbergen, gelifraction rates may exceed 1 … 3 mm yr -1. Under similar, arctic and moist conditions, like during the Younger Dryas at the northern Norwegian coast and at present-day alpine lakeshores, much higher rates have been reported, e.g. 40 mm yr-1; (Rasmussen, 1981), and 27 Fig. 19. Broecker isochrones for the calcite cement in G22. See text for further discussion. Fig. 20. Frequency distribution of cave levels in Blomstrandsya, divided into active seacaves (black) and relict caves (grey shade). SL : present-day sea level, MG: late-glacial marine limit at Brggerhalvya extrapolated (1.5 m / km) to Blomstrandsya. Inset to the right: the cave distribution is compared with the sea level curve from Lehman & Foreman, i.e. Fig. 5. Fig. 21. Exposure aspects of 62 sea caves at or < 10 m above present-day sea level in Blomstrandsya. 94 % of these caves face open, deep sea in the fjord where wave abrasion would be optimal.

PAGE 18

Lauritzen S.-E. 18online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENmm yr-1; (Matthews et al., 1986). Aarseth & Fossen (2004a) estimated a rate of 1 mm yr-1 for Holocene cryoplanation of a lakeshore. Given the relatively small dimensions of the sea caves of Blomstrandsya, the rates of frost weathering and the time that sea level stayed the same, strongly suggest that most of, if not all, observed sea caves may be explained by coastal gelifraction processes acting in the Holocene. Since the host rock is karstic, solution may play a signi“cant additional role in sea cave formation. The marine talik is a zone of con”uence for supraand sub-permafrost groundwater ”ow. This is seen in Jakobskjelda ( vide supra ) and in the talik dolines described by Salvigsen & Elgersmaa (1985). Moreover, we may also anticipate a saltwater wedge, penetrating the thawed, “ssured rock beneath sub-permafrost water, forming a hypothetical halocline. Theoretically, corrosion processes associated with the mixing of freshwater and seawater may contribute to open fractures in much the same way as halocline caves are formed in carbonate islands (Mylroie and Carew, 1988). Therefore, we may propose a model for sea cave formation in a permafrozen, karstic coast which includes all observations and synergy of the processes discussed above, Fig. 22. 4.4. Formation of relict cavesa) Possible origin from marine erosion or frost-pocketing Provided that the caves terminate just beneath the cover of cryoclastics and red earth, they would all be similar to the active sea caves, i.e. pockets of limited extent where the speleogenetic agents must have worked inwards from the surface of the cliff faces. However, the cave locations seem rather randomly distributed in space and there is no evidence of any additional level-controlled erosion around them. If the relict caves originated as sea caves, developed with the same process intensity as the present ones do, we should expect them to occur at distinct levels, independent of geological structure and they should be associated with remnants of erosional notches, etc. This distinguishes them from the sea caves, which are vertically clustered and located in the cliff faces of the prominent strand”at. The relict caves are less diverse in outline than the sea caves, they are more branched and none of them display the wide, arched features seen in Fig. 10 & 11. Only one (G5) out of 30 cave entrances displays any pronounced widening towards the surface, which may indicate modi“cation by marine eroding agents. However, seacave walls are faceted by frost shattering; occasionally, they may display polished surfaces adjacent to the shingle beaches, but these surfaces lack scalloping. The entrance of G5 is completely different, it is large, but smooth and solutional. It bears scallops and displays only slight gelifraction. There is, of course, a remote possibility which we cannot rule entirely out that the relict caves may be so old that former marine features have been eroded away. It seems much more likely that the relict, dry caves are of a different class with a different origin, being formed by relatively recent processes. It is, for the same reasons, likely that the present sea caves may include some modi“ed versions of this class. b) Origin as subglacial karst conduits Scalloping and the principle of hydraulic continuity strongly suggests that the passages were indeed part of conduits that transported water through the rock mass, beyond the limit of human exploration. This water transport occurred at all observed levels, up to about 290 m a.s.l., and must have operated under partial or totally unfrozen conditions. In order to invoke taliks of this kind, we may either raise sea level up to or above the caves, or we may cover the caves under thick ice. The late glacial marine limit, extrapolated to Blomstrandsya, is slightly above 50 m a.s.l. This is however the highest recorded sea level ; sea level could have been much higher during the stadials without leaving any trace. However, even if glacial sea levels exceeded the level of our caves, they would then be situated beneath a thick ice cover and the marine contact would be tens of kilometres away. Internal ice sheet hydraulics would then determine and dominate the local water table, and the water would be fresh. Older, and presumably higher sea levels therefore seem irrelevant for speleogenesis as long as a thick ice sheet was present. The observed scallop directions may at the “rst glance seem contradictory, as neighbouring caves (G4 and G5) transported water in opposite directions. First, this might not have occurred at the same time, nor is reversal of ”ow contradictory with glacier hydraulics. In a subglacial situation, Blomstrandsya would be situated at the con”uence between Fig. 22. Conceptual model of a karstic seacave formation in conjunction with permafrost. See text for further discussion. Legend: 1: seawater, 2: cryoclastic regolith and glacial drift. 3: unfrozen limestone. 4: Frozen limestone. 5: beach shingle and gravel. 6: hypothetical out”ow of sub-permafrost groundwater. After Lauritzen (1998).

PAGE 19

19Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENseveral ice-streams, of which the two important ones are the main ”ow from the southeast through Kongsfjorden and the stream emerging from the east via Srvgen from the mountains behind Blomstrandsbreen, Fig. 3. Therefore, hydraulic gradients through the northern part of Blomstrandsya would alternate between east to west and west to east, as well as south to north, depending on the relative magnitude of the two ice streams at any time. Subglacial water ”ow is dictated more by the surface slope of the glacier than on the underlying bedrock topography. Forced-”ow, subglacial speleogenesis is well known, e.g. (Lauritzen, 1986; Ford and Williams, 1989). This mechanism could accommodate the observed scallop directions in all caves investigated. Second, the unequivocal paragenetic features observed are also fully compatible with subglacial karsti“cation. The large portal of G5 may also be explained as a paragenetic feature. Third, the bogaz forms are also aligned according to the hydraulic conditions deduced above. They would have functioned as ef“cient N-channels, although at different times. Fourth, the whaleback ridges with associated caves (G15 & 16) accommodated ice ”ow from east to west; G17 and G18 would also be ef”uent under these conditions. The speleothem dates provide direct evidence that the caves existed prior to the last Weichselian glaciation, that they were drained above sea level during the Kapp Ekholm Interstadial, and that at least part of the sub-glacial speleogenesis took place during earlier stadials in the Quaternary. c) Pre-glacial speleogenesis As suggested for Irgenstjernet, products of pre-glacial or Tertiary karst processes may be preserved. Several of the relict caves, and maybe some of the sea caves as well, are located adjacent to redbed paleokarst inliers, and they are guided by the same faults. There is no doubt that many of the relict caves contain redbed-derived sediments, either as a primary deposit, or as secondary deposits, eroded from redbeds elsewhere. Therefore, the present caves could then be viewed as paleokarst conduits that are reactivated and considerably widened (speleogenesis sensu lato ) by Quaternary processes and we cannot exclude that speleogenesis of the presently known relict caves commenced in the Tertiary. In fact, since karsti“cation is a continuous process, lacking intrinsic thresholds, it is unlikely that meteoric speleogenesis did not commence in near-surface locations during the Tertiary.5. ConclusionsContrary to the opinions of previous workers, a surprisingly large number of relict caves have been found and described in addition to an even larger amount of active sea caves. Most of the relict caves can be explained by glacier ice contact and sub-glacial processes. The present author suggests subglacial speleogenesis as the dominant process. U-series dating prove that some of the relict caves are at least older than the last Weichselian glaciation. The rapid uplift and subsequent constant sea-level during the Holocene is exceptional for most karst areas that have experienced glacio-isostatic rebound. Thus, Holocene processes alone can explain most of the volume in the sea caves. Consequently, if the sea caves were activated in previous interstadial or interglacial periods, those periods need to have had a sea level history very similar to the Holocene.6. A speleogenetic model for BlomstrandsyaThe speleogenesis of the caves may be summarised as follows (Table 4):7. Further workIn order to test the hypotheses outlined in this paper, the sea caves need to be fully explored above and under water and surveyed accurately in order to obtain more precise morphological information. In this way, one may perhaps be able to observe and sample the hypothesised sub-permafrost groundwater ”ow. The potential for “nding more and interesting relict caves in the steep cliffs is by far not exhausted, and stratigraphic investigations of cave “lls is also needed. Further work on Blomstrandsya and adjacent areas in the Hecla Hoeck marbles is in progress. PhasePossible timing 1Hypogean, hydrothermal karsti“cation and subsequent in“lling with calcite. Caledonian or later, but preDevonian. 2Devonian (?) karsti“cation and tectonics with redbed in“lling. Post-Caledonian (Devonian) 3Pre-glacial commencement of presentday caves, in part reactivating paleokarst conduits Tertiary? 4Subglacial speleogenesis under various extents of ice cover, formation of scalloped conduits, often under paragenetic conditions Quaternary 5Postand interglacial obstruction and freezing of caves. Quaternary 6Intensive, postglacial marine erosion, in some cases invading pre-existing paleokarst and subglacial caves. Holocene and interglacialTable 4. Speleogenesis of the Blomstrandsya caves.

PAGE 20

Lauritzen S.-E. 20online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENAcknowledgementsField work in 1991, 1992 and 1997 was “nancially supported by The Norwegian Hydrological Committee (NHK) and by the Norwegian Research Council (NFR). Norsk Polarinstitutt (NP) is thanked for logistic support. Jan Inge Karlsen assisted in the “eld. Narve Ringset is thanked for providing the photograph in Fig. 4b. Margaret Palmer and Alexander Klimchouk are thanked for refereeing the paper. Bernard Lauriol provided data on exploration limits in the Yukons.ReferencesAarseth, I., Fossen, H. 2004a. A Holocene lacustrine rock platform around Storavatnet, Ostery, western Norway. The Holocene 14 : 589-596. Aarseth, I., Fossen, H. 2004b. Late Quaternary, cryoplanation of rock surfaces in lacustrine environments in the Bergen area, Norway. Norsk Geologisk Tidsskrift 84 : 125-137. Andr, M.F. 1998. Holocene rockwall retreat in Svalbard: A triple-rate evolution. Earth Surface Processes and Landforms 22 : 423-440. Brown, R.J.E. 1970. Permafrost in Canada University of Toronto Press, Toronto. Corbel, J. 1957. Les karts du Nord-oest de lEurope et de quelques regions de comparison. Institute des Etudes Rhodanniennes. Memoires 12 : 1-541. Davies, W.E. 1960. Caves in Northern Greenland. National Speleological Society, Bulletin 22 : 114-116. Deming, D., Sass, J.H., Lachenbruch, A.H., De Rito, R.F. 1992. Heat ”ow and subsurface temperature as evidence for basin-scale groundwater ”ow, North Slope of Alaska. Geological Society of America Bulletin 104 : 528-542. Dixon, E.J., Heaton, T.H., Fi“eld, T., Hamilton, T.D., Putnam, D.E., Grady, F. 1997. Late Quaternary regional geoarchaeology of Southeast Alaska karst: A progress report. Geoarchaeology 12 : 689712. Ford, D.C. 1984: Karst groundwater activity and landform genesis in modern permafrost regions of Canada. pp. 340-350 In: LaFleur, R.G., (Ed.) Groundwater as a Geomorphic Agent Allen & Unwin, New York. Ford, D.C., Williams, P.W. 1989. Karst geomorphology and Hydrology Unwin Hyman, London. 601 pp. Glazek, J., Rudnicki, J., Szynkiewicz, A. 1977. Proglacial cavesa special genetic type of cave in glaciated areas. Proceedings, 7th. International Speleological Congress, Shef“eld 1 : 215-217. Haldorsen, S., Lauritzen, S.E. 1993. Subpermafrost groundwater in Svalbard. Hydrogeology of Hard Rocks. 26th Congress of the International Association of Hydrogeologists 2 : 940-949. Hjelle, A. 1979. Some aspects of the geology of northwest Spitsbergen. Norsk Polarinstitutt skrifter 167 : 37-62. Holtedahl, H. 1996. The Norwegian Strand”at a geomorphological puzzle. Norsk Geologisk Tidsskrift 78 : 47-66. Horn, G. 1935. ber die Bildung von Karsthhlen unter einem Gletcher. Norsk Geogra“sk Tidsskrift 5 : 494-498. Horn, G. 1947. Karsthuler i Nordland. Norges Geologiske Underskelse 165 : 1-77. Horn, G. 1978. Limestone Caves in Nordland. Cave Geology 1 : 124138. Ivanovich, M., Harmon, R.S. 1992. Uranium-series Disequilibrium: Applications to Earth, marine, and Environmental Sciences Clarendon Press, Oxford. 910 pp. Juday, G.P. 1989. Alaska Research Natural Areas. 2: Limestone Jags United States Department of Agriculture. Forest Service, Washington DC. 58 pp. Lauriol, B., Prvost, C., Deschamps, E., Cinq-Mars, J., Labrecque, S. 2001. Faunal and Archaeological Remains as Evidence of Climatic Change in Freezing Caverns, Yukon Territory, Canada. Arctic 54 : 135-141. Lauritzen, S.E. 1986. Kvithola at Fauske; Northern Norway: an example of icecontact speleogenesis. Norsk Geologisk Tidsskrift 66 : 153-161. Lauritzen, S.E. 1991: Groundwater in Cold Climates: Interaction Between Glacier and Karst Aquifers. pp. 139-146 In: Gjessing, Y., Hagen, J.O., Hassel, K.A., Wold, B., (Eds.) Arctic Hydrology. Present and Future Tasks, Hydrology of Svalbard Hydrological problems in cold climates Norwegian National Commitee for Hydrology, Oslo. Lauritzen, S.E. 1993: Age4U2UŽ. Program for reading ADCAM energy spectra, integration peak-correction and calculation of 230Thorium/ 234Uranium ages Computer Program Turbo Pascal Code, 5,000 lines, Department of Geology, Bergen. Lauritzen, S.E. 1995. High-resolution paleotemperature proxy record during the last interglaciation in Norway from speleothems. Quaternary Research 43 : 133-146. Lauritzen, S.E. 1998: Chapter 3: Karst Morphogenesis in the Arctic: Examples From Spitsbergen. pp. 51-72 In: Daoxian, Y., Zaihua, L., (Eds.) Global Karst Correlation Science Press; VSP International Scienti“c Publisher, Bejing & Amsterdam. Lauritzen, S.E., Lundberg, J. 2000: Mesoand Micromorphology of Caves: Solutional and erosional morphology. pp. 407-426 In: Klimchouk, A., Ford, D.C., Palmer, A.N., (Eds.) Speleogenesis: Evolution of Karst Aquifers National Speleological Society, Huntsville, Ala. Lehman, S.J., Forman, S.L. 1992. Late Weichselian lacier Retreat in Kongsfjorden, West Spitsbergen, Svalbard. Quaternary Research 37 : 139-154. Liestl, O. 1976. Pingos, springs and permafrost in Spitsbergen. Norsk Polarinstitutt rbok 1975 : 7-29. Liestl, O. 1980. Permafrost Conditions in Spitsbergen Department of geography, Oslo. 20 pp. Loubiere, J.F. 1987. Observations Preliminaires sur les Cavites de la Region du Lac Centrum (Nord-Oest Groenland). Karstologia 9 : 7-16. Mangerud, J., Svendsen, J.I. 1992. The last interglacial-glacial period on Spitsbergen, Svalbard. Quaternary Science Reviews 11 : 633-664. Mangerud, J., Dokken, T., Hebbelin, D., Heggen, B., Inglfsson, O., Landvik, J.Y., Mejdahl, V., Svendsen, J.I., Vorren, T.O. 1998. Fluctuations of the SvalbardBarents Sea ice sheet during the last 150 000 years. Quaternary Science Reviews 17 : 11-42. Matthews, J., Dawson, A.G., Shakesby, R.A. 1986. Lake shoreline development frost weathering and rock platform erosionin an alpine periglacial environment, Jotunheimen, southern Norway. Boreas 15 : 33-50. Mylroie, J.E., Carew, J.L. 1988. Solution conduits as indicators of late Quaternary sea level position. Quaternary Science Reviews 7 : 55-64. Nye, J.F. 1965. The ”ow of a glacier in a channel of rectangular, elliptic or parabolic cross section. Journal of Glaciology 5 : 661-690. Paterson, W.S.B. 1981. The Physics of Glaciers Pergamon Press, London. 380 pp. Price, L.W. 1972. The Periglacial Environment, Permafrost, and Man Associateion of American Geographers, Washington, D.C. 88 pp. Rasmussen, A. 1981. The Deglaciation of the Coastal Area NW of Svartisen, Northern Norway. Norges Geologiske Underskelse 369 : 1-31. Richards, D.A., Dorale, J.A. 2003. Uranium-series Chronology and Environmental Appplications of Speleothems. Reviews in Mineralogy and Geochemistry 52 : 407-460.Rthlisberger, H. 1972. Water pressure in intraand subglacial channels. Journal of Glaciology 11 : 177-203.

PAGE 21

21Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoLauritzen S.-E.CAVES AND SPELEOGENESIS AT BLOMSTRANDSYA, KONGSFJORD, W. SPITSBERGENSalvigsen, O., Elgersmaa, A. 1985. Large scale karst features and open taliks at Vardeborgsletta, outer Isfjorden, Svalbard. Polar Research 3 : 145-153. Schroeder, J. 1979: Le dvelopment des grottes dans la rgion du premier canyon de la rivire Nahanni sud, T.N.O. Canada Ph.D. Thesis, lcole des tudes Suprieures de lUniversi dOttawa. Ottawa. Schwarcz, H.P. 1980. Absolute age determinations of archaeological sites by uranium dating of travertines. Archaeometry 22 : 3-24. Schwarcz, H.P., Blackwell, B.A. 1992: Archaeological applications. pp. 513-552 In: Ivanovich, M., Harmon, R.S., (Eds.) Uranium-series Disequilibrium: Applications to Earth, marine, and Environmental Sciences Clarendon Press, Oxford. Thiedig, F., Manby, G.M. 1992. Origins and deformation of PostCaledonian sediments sediments on Blomstrandhalvya and Lovnyane, nothwest Spitsbergen. Norsk Geologisk Tidsskrift 72 : 27-33. Walder, J., Hallet, B. 1979. geometry of former subglacial water channels and cavities. Journal of Glaciology 23 : 335-346. Werenskiold, W. 1953. The extent of frozen ground under the sea bottom and glacier beds. Journal of Glaciology 2 : 197-200.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. IntroductionKarst systems begin to develop when a hydraulic head appears, generally following uplift. For this reason, most of the large alpine karst systems are post-Miocene. However, study of these systems is dif“cult because of nearly 5 million years of complex evolution and the lack of reliable dating methods for these periods. The goal of this research was to study karst system development in a more recent environment and to determine the evolutionary characteristics of a simple case that could then be used as a model, taking advantage of current dating methods. The New Britain island karsts are of Quaternary age, and no older than Upper Pliocene. They have been intensely uplifted and harbor huge karst systems, such as Muruk Cave, which is the largest through-cave in the Southern Hemisphere.2. A recent karst, a large cave system2.1. The hyperkarst of Nakanai mountainsThe Nakanai Mountains are located on the island of New Britain, east of Papua-New-Guinea (Fig. 1). The island results from the subduction of the Australian plate moving northward under the Paci“c plate. This crustal movement has generated the New Britain trench, edging the southern coast and an active volcanic arc on the north coast. The Miocene limestone platform lies on an old paleogene volcanic arc located along the south coast. These limestones were themselves covered by a thick Pliocene volcano-sedimentary series (Thery and Thery, 1987; Maire, 1990). The active margin has been subjected to vigorous uplifting since the Upper Pliocene. The violent seismicity and regular explosive eruptions highlight the present tectonic activity. The Nakanai mountains include a large plateau, with a high point of 2185 m above see level (asl.) in the north, which descends to the south coast and is cut by canyons more than 1000 m deep. The area studied, around the Muruk cave entrance, Corresponding author: Email: audra@unice.fr by the authors. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license Speleogenesis in the hyperkarst of the Nakanai Mountains (New Britain, Papua New-Guinea). Evolution model of a juvenile system (Muruk Cave) inferred from U/Th and paleomagnetic datingP. Audra1, S. E. Lauritzen2 and P. Rochette31 UMR 6012 ESPACE CNRS, University of Nice Sophia-Antipolis, 98 boulevard Edouard Herriot, 06000 NICE, FRANCE (audra@unice.fr). 2 University of Bergen, Department of Geology, Allegaten 41, 5007 BERGEN, NORWAY (stein.lauritzen@geo.uib.no).3 UMR 6635 CEREGE CNRS, University of Marseille Saint-Jrme, Europle mditerranen de lArbois, 13545 Aix-en-Provence Cd ex 04, FRANCE (rochette@cerege.fr).Re-published from: Nakanai 1978-1998. 20 years of exploration, Antibes. Hmisphre Sud.Abstract: Muruk is the deepest cave in the Southern Hemisphere (1178 m of depth). It gives an access to go through the Nakanai Mountains and across large galleries, sometimes more than 50 m wide. Considering the important rainfall, the very active uplifting and the presence of a rainforest, Papua can be regarded as a hyperkarst, with large morphological forms evolving very quickly. U/Th and paleomagnetic dating on cave sediments con“rm this point of view, assigning a very recent age to this cave system (100 to 200 kyr). Muruk is a model of juvenile systems with a regularly inclined pro“le and with a monophas e evolution excluding any old perched level, unlike usual cave systems. These characteristics are essential for understanding not only the “rst speleogenetic phases, but also the more evolved systems found throughout the world. Key words: speleogenesis, hyperkarst, Nakanai Mountains, New Britain, Papua New-Guinea, juvenile cave system, Muruk Cave, U/Th and paleomagnetic dating.

PAGE 2

Audra P., Lauritzen S.-E. and Rochette P. 26online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011SPELEOGENESIS IN THE HYPERKARST OF THE NAKANAI MOUNTAINS is at an elevation of about 1400 m asl. on the edge of the Galowe canyon, the base of which is around 200 m asl., which represents an important topographic gradient. Since the beginning of the Upper Pliocene uplift, the volcano-sedimentary deposits have been weathered and eroded, bringing the limestones to the surface and allowing the karst processes to occur. Clay sediments presently exist as detritic cover, only a few meters thick, covering the karst relief. This combination results in the formation of rounded hills and deep depressions joined by small valleys. During heavy rainfall, these valleys are “lled by torrents with discharge reaching several m3/s. The ”ow is absorbed after variable distances into sinkholes feeding the underground system. The mountains are subjected to an oceanic monsoon climate, characterized by considerable rainfall (3 m on the coastline, 10 to 12 m in the mountains). This signi“cant and continuous humidity, combined with warm temperatures all year round, allows a rainforest development, still untouched by industrial exploitation. Karst processes in Muruk cave are directed by exceptional conditions: a host rock composed of very pure limestone subjected to solution; moreover, the lack of compaction and chalky facies make it very susceptible to mechanical erosion; signi“cant rainfall supplies large amounts of water concentrated by the clay-rich valleys before being injected into the karst; a rainforest with a highly productive biomass, providing CO2 and humic acids necessary for solution processes; a particularly great topographic gradient (1000 to 1500 m). For these reasons, karstic denudation, which generally takes into account only solution processes, but to which it would be necessary to add mechanical processes, is estimated at 400 m3/ km2/ yr, a world record (Audra 2001b; Maire, 1990). Finally, Nakanai can be considered a hyperkarstŽ, since so many factors favoring karst processes are pushed to an extreme. One can expect therefore to “nd cave systems characterized by a magnitude and evolutionary dynamic without equal.2.2. Muruk, an exceptional cave systemMuruk Cave is presently the largest cave system in the southern hemisphere (Hache et al., 1995) with a hydrogeologic through-passage 10 km long, and a depth of 1178 m (Fig. 2). The entrance acts as a valley sinkhole during heavy rainfalls. Within the cave system, shafts are uncommon and rarely deep, the system being composed mainly of large gently sloping passages. Morphological analysis shows that the system began as a tube sloping gently from sinkhole to spring. This tube evolved afterwards by entrenchment into a canyon and then in the downstream part into large passages that reach more that 50 m in diameter. It is a case of a monophased system, exceptional in a mountain environment, bearing no perched level except for some dif”uences marking a local reorganization in the vicinity of the spring. Water leaves the system at the Berenice resurgence, which has a porch entrance, perched 50 m above the Galowe river. This shows the lag time between karst processes compared to ”uvial entrenchment, which is the result of the rapid uplift. The morphology is typical to of monophased juvenile karst system. Taking into account the rapidity of karst processes, it suggests that the organization of Muruk system is relatively recent. The sediments ages support this hypothesis.3. Chronological dataSampling of “ne detritic sediments and speleothems was made for paleomagnetic dating. The speleothems were also dated by U/Th method. Recent deposits such as ”ooding clays and active ”owstones were avoided. The selection of the most ancient sediments was made according to stratigraphic location.3.1. Two stratigraphic sequences of cave sedimentsThe older sequence consist of two types of sediments: more or less sandy clays, very compact, with a black coating (n 9, 11 in Cassiquiare; n 19 in -800 Bypass). laminated brownish speleothems, overlying the older black clays (n 8, 10 in Cassiquiare; n 17 -800 Shunt). All of these “rst generation sediments have been intensively eroded by a subsequent resumption of the cave Fig. 1 Location of Nakanai mountains, a New Britain karst in Papua-NewGuinea (after Maire et al., 1981).

PAGE 3

27Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P., Lauritzen S.-E. and Rochette P.SPELEOGENESIS IN THE HYPERKARST OF THE NAKANAI MOUNTAINS stream and subsist only as discontinuous clay veneer and eroded speleothems. This older sequence is followed by a second sedimentation phase: a new detritic phase rich in clays (n 1, 2, 3 in -800 Bypass; n 4, 5 in the 12 m Waterfall tributary, around -500) and sandy clays (n 13, 14, 15 in -800 Bypass), not eroded. a second calcite deposition phase with a sound surface uneroded by the stream (n 18 in -800 Bypass). It appears therefore that there were at least two consecutive sequences of stream activity, both followed by a diminished ”ow with speleothem deposition.3.2. A recent paleomagnetic signatureThe natural remnant magnetizations (NMR) from 25 samples were measured with a rotating remanometer (JR5A Spinner Magnetometer) during demagnetization in an increasing alternating “eld (AF), up to 100 mT (Table 1). Most samples show strong directional stability during demagnetization (specimen 1 and 8 b) while a few samples have either a large secondary component erased by an AF of 20-50 mT (e.g. 11) or poorly de“ned behavior due to low intensity (e.g. 18 c) (Fig. 3). All characteristic directions show normal polarities pointing toward deposition in the Brunhes period (< 780 kyr).Fig. 2 Pro“le of Muruk system (surveyed by Papou 1985, Hmisphre Sud 1995 and Nakanai 1998 expeditions). Fig. 4 Stereoplot of site mean characteristic directions (site de“ned in table 1), excluding poorly de“ned specimen direction ( > 20). The mean direction with its circle of con“dence appears as a large triangle while the predicted mean geomagnetic “eld is shown by a circle. All directions are in the upper hemisphere. Fig. 3 Orthogonal Zijderveld plots showing the NRM vector evolution during AF demagnetization up to 100 mT.

PAGE 4

Audra P., Lauritzen S.-E. and Rochette P. 28online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011SPELEOGENESIS IN THE HYPERKARST OF THE NAKANAI MOUNTAINS Mean direction using all characteristic specimen directions is D = 1; I = -14; 95 = 10; N = 25; and a selection of well de“ned directions ( < 20) grouped by site yields D = 0; I = -17; 95 = 12; N = 9. Both directions show correct declination but higher inclination than the predicted mean geomagnetic “eld for Brunhes period at Muruk (-11), although this direction is within the 95% con“dence (Fig. 4). This deviation, if statistically signi“cant, may be due to a recent 6 degree southward tilting of the area. Alternatively, it may correspond to an insuf“cient averaging of secular variation. Indeed, the present day “eld shows an inclination of -25. This “rst chronological approach con“rms the recent age of the Muruk system. Speleothem U/Th dating con“rms these measurements and produces more accurate chronological data as described below.3.3. Speleothems less than 50 000 years oldSpeleothems were also dated using the radiometric U/Th method by mass-spectrometry (Table 2). Uranium content has an average of around 0.5 ppm. The 230Th / 234U ratio allows us to calculate the sample age. The 234U / 238U ratio is slightly above 1 and the 230Th / 232Th is very high. This reveals the absence of detritic contamination linked to the opening of the system, so the ratios indicate a good reliability of calculated ages. Radioactive elements present correspond only to those trapped during calcite deposition. For all samples, ages are recent, dating between 8 and 50 kyr. The data are useful for understanding the impact of climate change in the past in the warm and wet areas of the Earth, such as Papua. Compared to frequency statistics for cold and temperate ”owstones from the northern hemisphere, the datings correspond to frequency minima linked to the last Table 1.Paleomagnetic results. Specimen NRM intensity, characteristic directions with their con“dence angles obtained from AF demagnetization diagram by principal component analysis (a few samples have only one demagnetization step and therefore no con“dence angle), magnetostratigraphic interpretation (measurements performed in CEREGE). Sample ref.LocationSediment typeInt. (mA/m)Decl. ()Incl. ()Direction PNG 1 box -900 BypassClay with mud-cracks 4,8 E-220-242,7 Normal PNG 2 box 4,9 E-216-204,3Normal PNG 3 box 5 E-25-202,4 Normal PNG 4 box 12 m Waterfall tributaryEntrenched ”ood clay 3,7 E-23303424,9 Normal PNG 5 box 4,3 E-26096,4 Normal PNG 8 a Cassiquiare Brown laminated ”owstone, younger than PNG 9 5,3 E-2332-64,0 Normal PNG 8 b 1,5 E-1330-72,2 Normal PNG 8 c 3 E-23233613,5 Normal PNG 9 a CassiquiareDense brown clay with black coating 1,7 E-1342-155,2 Normal PNG 9 b 2 E-1332115,6Normal PNG 10 a Cassiquiare Brown laminated ”owstone, younger than PNG 11 1,5 E-19-224,0 Normal PNG 10 b 1,5 E-110-74,4 Normal PNG 10 c 2,1 E-17-122,4 Normal PNG 11 a CassiquiareDense brown clay with black coating 2,4 E-1326-314,9Normal PNG 11 b 4,9 E-169-134,2 Normal PNG 11 c 3,9 E-117-483,3 Normal PNG 13 box -900 Bypass Muddy sand, older than PNG 18 3,9 E-2347-287,4 Normal PNG 14 box 4,8 E-25-238,2 Normal PNG 15 box 3,2 E-25-2510,3 Normal PNG 17a -900 Bypass Old calcite ”oor, broken by neotectonic 5,8 E-3359-13-Normal PNG 17b 6,3 E-3358-235,3 Normal PNG 17c 4,2 E-3356-17-Normal PNG 18a -900 Bypass Stalagmite corroded by dripping solution 1,2 E-3268-2311,2Normal PNG 18b 7,6 E-32991224,6 Normal PNG 18c 3,8 E-33291524,3 Normal PNG 19a -900 BypassGrey sandstone with black coating 4 E-13165126,9 Normal PNG 19b 5,6 E-172617,6 Normal

PAGE 5

29Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P., Lauritzen S.-E. and Rochette P.SPELEOGENESIS IN THE HYPERKARST OF THE NAKANAI MOUNTAINS cold periods, such as Boreal (for the sample at 8.78 ka), Older Dryas (for 14.1 ka), cold maximum of recent Wm (for 23 ka) and to a less cold interstage (for 49.7 ka). It appears therefore that these periods, characterized by a clear reduction or break in speleothem deposition in mid latitudes, do not show the same trend in low-latitude speleothem growth. Data from about twenty datings on Borneos speleothems lead to similar conclusions (Fig. 5). The dating provides a precise chronological frame. All the deposits are younger than 780 kyr, with the possibility that their age could be very recent within this window. None of the speleothems dated is older than 50 kyr. Though these are not the oldest sediments in the cave because they succeed an argillaceous sedimentation phase. These “rst known deposits could have been introduced into the cave some time after cave system initiation. However, it seems acceptable to suppose that the initiation of the karst system dates to a maximum of some hundreds of thousands of years. Very few in contrast to northern hemisphere alpine karst systems that initiate at the Mio-Pliocene limit, or even before (Audra, 1994).4. ConclusionThe Muruk cave system probably initiated around 100 to 200 kyr, according to a base level located 50 m above the present Galowe River, as shown by the Berenice perched spring (Fig. 2). The ef“ciency of the karst processes, which led to such extraordinary dimensions in both vertical and horizontal extension, and in size can only be explained by the particular conditions discussed above. The most important are the high tectonic activity and the abundant rainfall. Can also be added the role of the rainforest and of clay covers concentrating the water. Finally, another originality comes from the poorly compacted host rock, creating galleries that enlarge quickly by a combination of chemical and mechanical processes from underground streams but also by vault collapse, as the cave is shaken regularly by seismic movements (Audra, 2001a; Maire, 1990). Cave genesis can be described as follows: water ”ows over the clay cap and enters the limestone through the sinkholes. It then follows a nearly straight line to the springs into the limestone mass, not karsti“ed in its deeper parts. Subsequently the cave system evolve by entrenchment and enlargement caused by torrential ”ows in the underground stream, this occurs mainly in the vadose zone, with only local ”ooding in the sections during high waters. The initial general disposition of the cave system was not greatly changed by subsequent evolution. This stretched pro“le differs greatly from the conventional karst model, where shafts dip vertically into the vadose zone and horizontal galleries are located near the water table. Muruk can be regarded as a juvenile system model. Its case can help us understand other more ancient and generally more complex systems in the world. Taking into account the example of Muruk could renew the understanding of some old levels showing similar stretched pro“les.ReferencesAudra, P. 1994. Karsts alpins. Gense de grands rseaux souterrains. Exemples : le Tennengebirge, Autriche ; lIle de Crmieu, la Chartreuse et le Vercors, France (Alpine karsts. Genesis of Fig. 5 Distribution of low-latitude speleothem growth periods during the last 100,000 years, compared to the oceanic paleotemperature curve. Data from Borneo (Farrant, 1995) and Papua. Table 2.Geochemical data obtained by U/Th method and ages of Muruk speleothems (dating by S.-E. Lauritzen, Bergen). Location Sample ref. U (ppm)234U / 238U230Th / 234U230Th / 232Th Age (ka) CassiquiarePNG 80,791,5940,0778618,75 ( 0,48)CassiquiarePNG 10 0,341,210,37187549,7 ( 1,12)Shunt -900PNG 17 0,681,370,193749223,13 ( 0,54)Shunt -900PNG 18 0,741,340,122530114,1 ( 0,31)

PAGE 6

Audra P., Lauritzen S.-E. and Rochette P. 30online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011SPELEOGENESIS IN THE HYPERKARST OF THE NAKANAI MOUNTAINS large cave systems. Examples: the Tennengebirge, Austria; the Ile de Crmieu, the Chartreuse and the Vercors,France). Karstologia Mmoires 5. University J. Fourier Grenoble 1 Thesis. Fdration franaise de splologie, Paris & Association franaise de karstologie, Bordeaux. 280 p. Audra, P. 2001a. Origine des grands vides souterrains du rseau de Muruk et rle des sismes, montagnes Nakana (The role of seismic activity in the formation of large underground cavities in the Muruk system, Nakanai Mountains) In: Audra P., de Coninck P. and Sounier J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 101-107. Audra, P. 2001b. Valeur et rpartition de la dissolution spci“que dans les karsts des montagnes Nakanai, Nouvelle-Bretagne, Papouasie-Nouvelle-Guine (Karst solution and denudation assessment in Nakanai Mountains). In: Audra P., de Coninck P. and Sounier J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 77-86 Farrant, A. R. 1995. Long-term Quaternary chronologies from cave deposits. Bristol, Unpublished Ph.D. Thesis. Hache, P., Hobla, F., Philips, M., Sessegolo, D. and Sounier, J.-P. 1995. Muruk, hmisphre sud, premier -1000. Spelunca 60, 35-54. Maire, R. 1990. La haute montagne calcaire. Karstologia Mmoires, 3. University of Nice Sophia-Antipolis Thesis. Fdration franaise de splologie, Paris & Association franaise de karstologie, Grenoble. 731 p. Maire, R., Pernette, J.-F., Rigaldie, C., Sounier, J.-P. et al., 1981. Papouasie-Nouvelle Guine. Spelunca suppl. 3, 48 p. Thery, J.-M. and Thery, B. 1987. volution gologique de lle de Nouvelle-Bretagne et hypothses de formation des dolines-avens gantes des Monts Nakanai. Antipodes 85, Spelunca Mmoires 15, 80-88.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info active horizontal caves, although at least one of these shafts has reached what seems to be a fossil conduit. There is very little information regarding the epikarst in this area, beside the existence of several caves like the one described in this paper. In such karst shafts there is usually water dripping and “lm ”ows along the walls, active during weeks and in some cases even months (in the late summer) after any signi“cant rain, pointing to the existence of a suspended groundwater body that supplies the water. To the south of the plateau, a major spring, the Alviela Spring, at the altitude of around 60 m, is located (Fig. 1). The Alviela spring, according to Manuppella et al. (2000), is the largest spring that drains the St. Antonio Plateau. Corresponding author: Email: paulor2005@yahoo.com by the author. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license Speleogenesis of Alecrineiros Shaft Paulo RodriguesAssociao de Espelelogos de Sintra, Rua da Primavera, 11, 2710-994, Abrunheira, Sintra. Portugal Ncleo dos Amigos das Lapas Grutas e Algares Scienti“c Commission of the Portuguese Speleological Federation, Estrada Calhariz de Ben“ca, 187, 1500 … 124 Lisboa Abstract: The Alecrineiros Shaft is a cave developed in the St. Antonio Plateau, Portugal. This cave develops on a gently dipping monocline structure, along two major fracture families. These discontinuities are sub-vertical and exhibit the followin g directions: E-W to N70W and N-S to N30E. The shaft presents evidences of several speleogenetic processes compatible with a genesis and development in a vadose regime at the base of the epikarstic zone. Keywords: Shafts; St, Antonio Plateau; Speleogenesis; Vadose regime; Epikarst. Fig. 1. St. Antonio plateau location in the Macio Calcrio Estremenho. The massif area is represented by dark colours. (Figure adapted from Fernandes Martins, 1949). 1. IntroductionThe Alecrineiros Shaft is a cavity located in the St. Antnio Plateau, Portugal. The St. Antnio Plateau is a geomorphologic unit of several square kilometres in area, of triangular shape, whose vertex direct to the north (Fig. 1). This geomorphologic unit is a part of Portugals largest and more important karst massif, the Macio Calcrio Estremenho, such as it was de“ned by Fernandes Martins (1949). According to Manuppela et al. (2000), the plateau consists of elevated surfaces surrounded at west and east by steep cliffs. The southern surface of the plateau drops more gently to the southern border, from elevations above 500 m to little above 200 m. The entire perimeter of the plateau is bordered by faults, along witch the plateau has been elevated in relation the surrounding area. The surface of the St. Antnio Plateau is almost ”at, slightly tilting to the south, presenting some traces of an ancient ”uvial applanation surface. The later is rearranged by karst and also normal erosion in a limited extent (Fernandes Martins, 1949). The plateau surface shows several typical karst forms such as karren “elds, uvalas, and a considerable number of dolines according to Manuppella et al. (2000). At the St. Antonio Plateau there is a large number of caves (at the order of some hundreds) which are represented mainly of shafts with depths usually on the order of some tenths of meters but reaching 200 m in some cases. The deepest known cave is 220 m depth. These karst shafts have no access to

PAGE 2

23Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoRodrigues P.SPELEOGENESIS OF ALECRINEIROS SHAFT 2. Geological and geomorphologic setting Based on the analysis of the Geological map of Portugal, sheet 27-A-Vila Nova de Ourm at the scale 1:50,000, one can observe that the cavity develops along the formation of Calcrios mcriticos da Serra de AireŽ, composed mostly of micrite limestone. Regarding the structural geology, and based Fig. 2. Location (a, b) and the character (c) of the cave entrance. upon an expedite analysis of the above mentioned geological map, the St. Antnio Plateau is a monocline with some ”exures, where the bedding present a regional direction that ranges from approximately WNW-ESE to NW-SE, gently dipping to the southwest. This monocline is crossed by a series of faults with directions approximate WNW-ESE NW-SE. 3. Cave characteristicsThe cave entrance (540 m high) opens at the base of a karren “eld developed on a ”at surface (Fig.2, a-b). Within the karren “eld the depth of grooves reaches 2-5 m. This karren “eld, according to Rodrigues (1998), holds a large number of kart depressions, some of which are incipient. The cave entrance is located on a ”at surface between karst depressions (Fig. 2, c). The cave is structurally controlled by two major fracture families intersecting each other. One of them has a direction approximately EW to N70W, the other is about N-S to N30E, both are subvertical fractures (Fig. 3). The cave zones controlled by discontinuities with directions N-S to N30E, present a lesser development than the zones controlled by discontinuities which directions range from EW to N70W. The depth of the shafts ranges generally between ten to twenty meters and the maximum shaft width is around 4-5 m. The connections between the various shafts often occur through passages of relatively small section, located near the bottom of the shafts or at higher positions. The shafts bottoms are usually covered of debris resulting mostly from the ceiling and walls breakdown. The western pit is different from the other shafts as it does not has a bottom covered with blocks; the pits width suddenly shrink at about 10 m above its bottom, so that the pit ends on an impassable vertical opening that develops along a sub-vertical discontinuity (Fig. 3). As a result of the ceiling breakdown along an almost horizontal bedding surface, the ceilings are usually ”at. The cave presents a very simple organization. Where there was no breakdown, the morphology of the cave is high and narrow, keeping in a rough way the form of the discontinuities along witch it was developed. It should be noted that virtually all sections of the cave follow sub-vertical discontinuities. These characteristics were de“ned by Bgli (1980) as typical of cavities of primary vadose origin. The shafts present vertical troughs on the walls, developed along the entire depths of the shafts, with sections reaching several decimetres. These troughs are very similar to those described by Baro (2003). According to this author, the troughs are formed by the corrosive and erosive effect of water that ”ows and drips along the shaft walls. According to Lauritzen and Lundberg (2000), the troughs can also be formed by the sprinkling of water, along the walls of sub-vertical fractures, which is also typical of caves developed in the vadose zone.

PAGE 3

Rodrigues P. 24online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011SPELEOGENESIS OF ALECRINEIROS SHAFT 4. SpeleogenesisThe development of the cave along sub-vertical discontinuities, as well as the existence of several shafts developed along the same discontinuity families, raises the possibility of a separate evolution of the pits before they intersected each other. The development in the upper part of the vadose zone, and the termination of the cave shafts either on impenetrable cave passages or in block chaos, indicate that this cave should have been developed in the base of the epikarst zone as de“ned by Klimchouk (2000). The morphology of this cave and its setting has many similarities to the Karst ShaftsŽ described by Baro (2003). The shaft genesis also seems to “t the proposed model of shaft development. According to the above mentioned model the cave should be classi“ed on the d stage of Baro (2003), corresponding to a fully developed shaft. Fig. 3. Alecrineiros Shaft topography with geological survey.5. Conclusions The Alecrineiros Shaft is a cavity that seems to have been developed at the base of the St. Antnio Plateau epikarst, under a vadose hydrogeological regime. Regarding the development stage, as de“ned by Baro (2003), the cavity is a fully developed shaft. Some other cavities of the plateau and surrounding karst areas present shafts with very similar characteristics.Acknowledgements This work was only possible due to the support and hard “eldwork of AES … Associao de Espelelogos de Sintra and Ncleo de Amigos das Lapas, Grutas e Algares. I would also like to thank the contribution given by NEUA Ncleo de Espeleologia da Universidade de Aveiro, NEC … Ncleo de Espeleologia de Condeixa, by members of ECVT … Espeleoclube de Torres Vedras and AESDA … Associao de Estudos Subterrneos e Defesa do Ambiente, and by ARCM … Alto Relevo Clube de Montanhismo in the “eldwork. For last I would like to thank Pedro Robalo that drew the cave map and to Dr. Jos Brando and Elisabete Dias, who revised the text. References Baro, I. 2003. Speleogenesis along subvertical joints: A model of plateau karst shaft development: A case study: the Doln Vrch Plateau (Slovak Republic), Cave & Karst Science 29 (1), 2002, 5-12. Also available at: http://www.speleogenesis.net Bogli, A. 1980. Karst Hydrology and Physical Speleology SpringerVerlag, Berlin Heildelberg New York, 286 p. Manupella, G., Telles Antunes, M., Costa Almeida, C.A., Azerdo, A.C., Barbosa, B., Cardoso, J.L., Crispim, J.A., Duarte, L.V., Henriques, M.H., Martins, L.T., Ramalho, M.M.; Santos, V.F.; Terrinha. P. (2000). Carta Geolgica de Portugal Vila Nova de Ourm, Folha 27-A, scale 1:50000, and Explanation note, Instituto Geolgico e Mineiro, Lisboa. Klimchouk, A. 2000. The Formation of Epikarst and its role in Vadose Speleogenesis, in Klimchouk A.B., Ford, D.C., Palmer, A.N., Dreybrodt W. (eds.) Speleogenesis: Evolution of karst aquifers. Huntsville: National Speleological Society, pp 91-99. Lauritzen, S. and Lundberg, J.2000. Solutional and Erosional Morphology, in Klimchouk A.B., Ford, D.C., Palmer, A.N., Dreybrodt W. (eds.) Speleogenesis: Evolution of karst aquifers. Huntsville: National Speleological Society, pp 408-426. Martins, A. F.1949. Macio Calcrio Estremenho … Contribuio para um estudo de Geogra“a Fsica. Doctoral Thesis, Coimbra University, Coimbra. Osborne, R. Amstrong L. 2003. Halls and Narrows: Network caves in dipping limestone, examples from eastern Australia. Cave & Karst Science 28 (1), 2001, 3-14. Also available at: http://www. speleogenesis.netŽ. Rodrigues, Maria Lusa Estevo. 1998. Evoluo geomorfolgica quaternria e dinmica actual … Aplicaes ao ordenamento do territrio … Exemplos no macio calcrio estremenho. Universidade de Lisboa, Lisboa



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. IntroductionA study of the characteristics of the water cycle (rainfall, ”ows) usually starts with a statistical approach to data collected for a long period and emerges from the mass of data highlighting average values and extreme events, etc. The data presented here were collected in a particular context. The region has no meteorological station in the high altitude forest. The few existing stations are scattered along the coast and their data is intermittent at best. No measure of the ”ow of the two rivers that border the region has ever been made. When more detailed studies of rainfall, runoff and evaporation were made, they proved to be far from what was envisaged. An understanding of this was most important as much for conducting exploration under constant threat of ”oods as for understanding the mode of karsti“cation in one of the most intense karsts in the world. In the absence of data we had no choice but to collect our own. In this we had to choose the parameters to measure, and suitable instruments to measure with, all within the constraints of a short duration expedition (approx. one month in the “eld), which didnt necessarily stay in the same place. We opted to collect the most signi“cant data as easily and as fast as possible. The initial problem centered on several questions from the preceding expeditions: Is it possible to predict heavy rainfall and therefore underground ”oods with the aid of rudimentary meteorological measurements (notably a barometer) considering that weather forecasts are unavailable? Moreover, how much rain is required to cause enough runoff to activate the dry streambeds that feed the caves? Is there a global way to look at the regional water”ow by studying the principal streams in the area? Of course we were unable to completely respond to all these questions, but analysis of the data collected has nevertheless provided us with some interesting information.2. The hyper-humid equatorial mountain climateKnowledge of the climatic data is fundamental to our understanding of the transit of water through the karst. All the data from the Nakanai Mountains is very imprecise, as the climatic conditions are only known from the meteorological station on the coast at Pomio (Fig. 1). We have tried in part to lift the mists that almost permanently mask these mountains and control everything further downstream: rain on the gardens, turquoise stream fed by the karst plateaus around which all village life is organized. Corresponding author: Email: audra@unice.fr by the author. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license Rainfall, ”ows and percolation in Nakanai Mountains karstP. AudraUMR 6012 ESPACE CNRS, University of Nice Sophia-Antipolis, 98 boulevard Edouard Herriot, 06000 NICE, FRANCE (audra@unice.fr).Re-published from: Nakanai 1978-1998. 20 years of exploration, Antibes. Hemisphere Sud. Abstract: Rainfall and runoff data do not exist in Nakanai Mountains, the main range of New Britain Island, Papua New Guinea. We collected “eld data during a 4 weeks expedition in the rainforest. In the mountain, the annual rainfall may be about 12 m; the soil is able to absorb a 50 mm rainfall, but a 20 mm shower starts runoff if the soil is saturated by a previous rain and g ullies sinking in caves are ”ooded. Flooding propagates at about 1 km/h inside Muruk Cave. The discharge of the main rivers, Matali and Galowe, correspond respectively in low water to 27 and 33 m3 /s. The ratios between low, medium, ”ood and exceptional ”ows are respectively 1, 5, 10 and 50. Keywords: Rainfall; runoff; Nakanai Mountains; Papua New-Guinea; soil saturation; discharge; Muruk Cave; Matali River; Galowe River.

PAGE 2

Audra P. 32online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARST2.1. Recreating meteorological dataRainfall data is available from the neighbouring station of Pomio on the coast at the foot of the Nakanai. These cover the period from 1956 to 1970 (Fig. 2). We have tried to “nd out the intensity and distribution of rainfall in the mountains during our stay at Muruk in order to correlate them with the coastal data in an attempt to extrapolate to reconstruct the missing data. The unique conditions in the primary forest necessitated the construction of special equipment.2.1.1. A pluviometer adapted to local conditionsA normal pluviometer is not very suited to the speci“c conditions in the forest. It must be able to measure large falls of rain and still be suf“ciently closed so that the splashes from the large drops of water do not escape from the container. Such an instrument is known, even taking the constraints into account;, all that was left was to lighten and compact it (Fig. 3). It takes the form of a funnel composed to four triangular sides made of polythene joined by strong adhesive tape. The assembly is mounted in a cleared area, suspended from branches by strings attached to metal eyelets. The top of each triangle measures 1 m long so that the funnel covers an area of 1 m2. When folded it “ts into a 20 cm x 8 cm cylinder weighing only 300 g. The water is collected in a 35 L plastic drum graduated on the inside to allow a direct reading (with a 1 m2 surface, 1 mm of rainfall adds 1 L of water. Each graduation corresponded with 5 mm of precipitation). Its capacity was therefore limited to 35 mm of precipitation. For especially heavy rainfall, another barrel, graduated in the same way was placed along side. This receptacle had an opening 30-cm in diameter (1 mm of precipitation over a surface of 700 cm2 gives 0.7 L). Its capacity of 500 mm was fortunately never reachedƒ This system was completely satisfactory; the only reserve was the poor resistance to wind gusts, which fortunately never occurred during our stay.2.1.2. Other meteorological parameters: temperature, atmospheric pressure, and humidityConsidering our other problems, these three parameters were secondary, but an understanding of them allowed us to better understand the genesis of the rainfall. These data were much easier to collect than the rainfall. A mercury thermometer was suspended under a roof out of direct sunlight. The mean daily temperature was calculated as the average between the minimum, just before dawn, and the maximum. A maximum/ minimum thermometer would without doubt have simpli“ed these measurements. Finally, a recording barometer with a thermometer and hygrometer was installed in the main hut. As much as possible, the four parameters were recorded every two hours during the day (Table 1). The nocturnal readings did not really suffer from a lack of automatic recording equipment because the temperature dropped smoothly until dawn when the pressure began to climb once again. The generally clear night skies meant that nighttime precipitation was rare. Nevertheless, an automatic data logger would have removed these constraints as well as allowing us more precise recording of data while we were away from camp for several days. Of course, one could question the reliability of automatic instruments in such a humid environment and it is always advisable the take simple instruments in case you have an electronic failure.Fig. 1. Location of the Nakanai, with generalized climatic data (after Ford, 1973, modi“ed). Fig. 2. Monthly precipitation for the period 1956-1970, Pomio station, altitude 5 m, annual total 6195 mm (after Mac Alpine et al., 1975).

PAGE 3

33Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARST2.2. A meteorological record for close to a monthWe now have meteorological data for practically a month, from 16th January to 13th February 1998 (Table 1, Fig. 4). Admittedly, they appear ridiculous, but they are nevertheless fundamental data, as there are no weather stations inland from the coast in New Britain. Thanks to this data we have a chance to understand the climate of the Nakanai Mountains a little better.2.2.1 Estimate of the annual precipitation at MurukDuring our stay in January and February, 451 mm of rain fell in Muruk. By comparing these with the existing data from Pomio on the coast (Fig. 3) it is possible to extrapolate in order to get an idea of the annual precipitation … keeping the following reservations in mind: the irregularity of the climate (Pomio has a variation of 3 to 10 m between dry and wet years!) implies that our data may not be representative of average values for the same period, the only comparative data that we have is for Pomio on the coast at the foot of the mountains and this data was collected in 1970, the proportionality established for this period between Pomio and Muruk is not especially the same as for an average year. The values calculated from this certainly include a serious margin of error. But in the absence of other data they are the most correct, even if they are the only one available! We can approach it in the following manner knowing that we have the precipitation level for Muruk from mid-January to mid-February. The precipitation level for Pomio for the same period: P = (222 + 151) / 2 = 186,5 mm. Compared to 481 mm at Muruk, there is ratio of 2.57 : 1 in favour of Muruk, which corresponds to a pluviometric gradient of 20 mm/100 m. This factor allows us to extrapolate a rainfall value at Muruk for the dry season (1). In the absence of any data, we can use a pluviometric gradient of exposure to the wind of 50 mm/100 m (minimal probable value) for the calculation of the wet months at Muruk based on the monthly values on the coast (2). (1) Rainfall values at Muruk in the dryŽ season (November-April) = (462 + 856 + 1291 + 1187 + 723 + 447) + (6 months x 50 mm x 14.5 hundred meters) = 3090 mm (2) Rainfall values at Muruk in the wet season (May-October) = (462 + 856 + 1291 + 1187 + 723 + 447) + (6 months x 50 mm x 14.5 hundred meters) = 9316 mm We get close to 12.5 m of annual rainfall at Muruk. As well as this, the spatial distribution of the precipitation is highly variable. One location can be in full sun while only a few hundred meters away it can be pouring rain. The location of the downpours is quite random generally localized. The measurements we took are in no way the result of a long series of annual rainfall recordings.2.2.2. Other climatic parametersThe average temperature during the period was 19.5 C, with a daily variation of 2 C. The diurnal range was 5.5 C, with a variation of only 3 C under cloud and 9 C under clear skies. The daily pressure variation was limited to 5 hPa but normally never passed 2 to 3 hPa. Finally, the humidity was very high, as expected. It never descended below 80% even during sunny periods. While the morning fog was burning off and during showers it would regularly pass 90% up to a maximum of 97%.2.3. Rainfall in the dryŽ seasonWhile not exactly producing weather forecasts, it was important for us to understand the signs preceding a major rainfall, so as to safely continue our exploration of the caves. From December to April the climate of the New Britain is dominated by the northwest monsoon, a steady, moderate wind that brings humidity from the ocean to the opposite side of the island (Fig. 1). The plateaus of the Galowe highlands, on the south side, are sheltered from these winds. Most of the rain falls on the north side, while the south only receives rain from clouds that cross the crest of the range. Out of the wind, the rainfall is moderated and it is possible to have two consecutive days without rain … the reason why this quali“es as the dryŽ (in quotation marks!). As the air descends, it dries out a little Fig. 3. Pluviometer made from polythene sheet with 1 m2 surface and 35 L barrel.

PAGE 4

Audra P. 34online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTand the coast has a generally sunny sky. Clouds and rain are not standard every day. In the opposite season, the direction of the monsoon is reversed and the south coast of New Britain receives the full force of the humid ocean wind as it hits the abrupt mountains. The orographic ascent causes veritable daily deluges. During our stay in the dryŽ season we could see the extreme regularity of the weather that produced an almost immutable succession of daily events. In general terms, a day progressed like this (Fig. 5): At sunrise, the sky was clear and the temperature fresh ( 15 C), the atmospheric pressure at its maximum. With the “rst rays of the sun, the sun rapidly warmed the air producing thermals. After a variable period of sunshine (30 min to 3 hrs) maximal temperature is reached (24 C by 8 h, 26 C by midday if the sky was suf“ciently clear), the clouds would grow quickly and totally cover the sky, pressure would fall 2 to 3 hPa and the temperature would fall back to 20 C.Table 1. Meteorological data for Muruk Camp, 1445 m asl., from 16 January to 13 February 1998.

PAGE 5

35Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTAt the beginning of the afternoon the mist would invade the forest and the rain would begin. Generally in the form of short but violent showers with several repeat performances during the afternoon. Finally the cloud cover would become denser, the showers would stop, and the pressure would gradually rise. At the beginning of the evening the pressure would once again return to its maximum and the sky would gradually clear until nightfall. During the night would once again clear and the temperature would drop smoothly to 15 C. Some variations to the standard day should be noted: If the cloud cover remains overnight, cooling is lessened and the morning temperature can pass 20 C, instead of the usual 15-16 C (5 February). The faster the clouds clear in the morning, the faster it warms up and produces more intense thermalsƒ and the showers arrive sooner and heavier (20 January). Conversely, Table 1. Meteorological data for Muruk Camp, 1445 m asl., from 16 January to 13 February 1998. (continued)

PAGE 6

Audra P. 36online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTa cloudy morning leads to a more stable but dull day without major rainfall. In this case, the temperature will oscillate between 20-21 C (18 January). The weather, while being controlled by convectional showers, is equally controlled by cycles of rainy days. In general, periods of intense rain occur over 3 consecutive days a week, separated by calmer weather without rain, but they never last longer than 2-3 days. The main causes of these weather variations are without doubt connected with the circulation of high altitude air masses and characteristics of the thermal gradient that generate instability to a greater or lesser extent. However, this cycle is superimposed on that of the humidity variations caused by the general sunshine of each day. Very sunny days are generally accompanied by large thermals that rapidly become showers. Humidity from the forest evaporates next morning, the sky clouds over earlier and earlier, and the showers also arrive earlier. After a few days the sky is cloudy “rst thing in the morning and does not allow the sun to get through. After one or two cloudy days, the sky clears rapidly one morning and the cycle begins again. The weather in the forest is dominated by this double cycle: daily and multi-day, that moderates the daily precipitation with damp fogs in between heavy showers, separated by sunny breaks which are always too short to allow the forest (and our washing) to dry out. While the mild temperatures help keep everything damp, the creeping mould slowly but surely covers everything foreign to the forest, humans included. In the light of our observations and measurements, the rainfall development became clear, even though we did not know the mechanism in precise terms. We did not have elements of the total picture to predict the days most at risk of heavy rain. We were unable to isolate the signi“cant indicator parameter, so, for instance the atmospheric pressure at ground level had no direct relationship to the abundance of precipitation.3. Streamsinks and soft ground: a ”uviokarst in an equatorial forestFrom rain to surface ”ow to underground streams, it is important to understand the successive relationships they have to each other and especially to de“ne a single factor that relates rainfall to stream ”ow. There is no known permanent stream of any size on the Galowe plateau, even though the area is covered in small branching valleys up to several kilometers long, which apparently drain vast areas. It is an example of ”uviokarst where the concentration of surface drainage plays an important role in the form and dynamics of the underground drainage. Due to the binary in”ows, diffuse and concentrated, the karst hydrologic behaviour adopts a dual mode, depending on the hydraulic conditions.3.1. Temporary ravines with spasmodic modeMost of the time the vegetation cover and the soil absorb the precipitation. Some of this is returned to the atmosphere either by direct evaporation or indirectly as evapotranspiration from the vegetation. The rest percolates through the heavy soil to provide a diffuse drainage to the subjacent karst. As we will see later, this complex system regulates the ”ow such that there is normally very little variation in the ”ow with exceptional violent ”ows. In general terms, the appearance to ”owing water is in part due to the nature of the soil, and in part due to the severity of the downpour. In the scale of our study, the “rst parameters can be considered as invariable. It is therefore the distribution of the rain episodes that determine the appearance of ”owing water. We will consider the case of the Muruk gully where activity in the gully can be observed in parallel with the Fig. 4. Ombrothermic diagram from Muruk Camp station, 1445 m asl., 16th January to 13th February 1998. Fig. 5. Evolution of meteorological parameters during the course of a typical day (25 January 1998).

PAGE 7

37Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTrainfall recording (Fig. 4). After the biopedologic cover has dried out due to some days without major rainfall it can absorb large quantities of water without the slightest runoff: 55 mm in the afternoon of January 20, 50 mm on January 30. In the “rst case, the four preceding days had only produced light showers and no rain the day before that. In the second case, a dry day was suf“cient to dry the soil as the day before that 19 mm fell. On the other hand, after several days of moderate falls, the soil becomes saturated. If a heavy rainfall occurs, in“ltration is not possible and runoff is apparent immediately. The stream”ow of January 27 followed a week of continuous moderate rainfall (average 10 mm/day). A fall of 46 mm triggered the ”ow. The period January 31 to February 3 was very different. The soil was already saturated after the 50 mm that fell on January 30. During the following three days, downpours of 35 to 40 mm a day kept the gully running virtually constantly. The gully stopped ”owing on February 3, even with 15 to 20 mm a day for the next three days. It is obviously a threshold below which no stream”ow occurs, at least in this season, as con“rmed by the following example. On February 10 a 22 mm shower saturated the soil and a 34 mm shower the following day triggered a ”ow. However, these few days of stream”ow in the Muruk gully do not directly correspond to major ”oods. The 46 mm on January 27 and the 34 mm on February 11 only provoked a minor ”ow of a few L/s at the end of the afternoon. This amount of water is minimal when compared to the amount ”owing underground and was unable to cause any underground ”ooding. The January 30 … February 3 event caused a moderate ”ood. I was the “rst time when several successive days signi“cantly passed to 20 mm threshold with 50, 35, 40 and 40 mm successively. Observers on the surface saw the gully ”ow varying between a few dozen to a few hundred L/s. Even so, when the rain stopped overnight, the gully stopped ”owing during the night until the middle of the next day, showing that it is not possible for the stream to ”ow for very long after the rain stops. It is interesting to note that the ”ood corresponded to 35 mm at Muruk, then the following two days 40 mm fell each day but only produced a ”ow of a few L/s. This clearly illustrates the spatial variability of the rainfall and it is highly probably that the showers of January 31 were considerably more intense higher up the catchment. A single recording site can only partially describe a system, which is in reality quite complex and diverse. Finally, the members who stayed until the end of the expedition observed a particularly violent ”ood with a pulse of 2 to 3 m3/s. More detailed observation were not taken, which was fortunate in a way, as anyone in some sections of the cave at the time would have been in serious danger.3.2. A relatively balanced underground water”owDiffuse feeders mainly in”uence the underground water ”ow regime with exceptional but powerful additions from surface gullies. When this happens, distribution of water from the gullies on the surface is closely related to the underground drains. Aside from purely scienti“c considerations, this is one of the major concerns of cavers, both in general terms, but especially in these regions of abundant rainfall. Rises in water level accompany practically every exploration; violent ”ood pulses (considerably less frequent), fortunately caused no serious consequences as cavers were never surprised in restrictions or waterfalls. The observation of the regimes of different rivers in Muruk during the course of our exploration, combined with the rainfall records, allows us to establish some notes.3.2.1. Exploration January 30 to February 1 to the Gruyere (-950 m)At 4.00 p.m. at …750 m, the water appeared milky, this increase in ”ow was not signi“cant. The ”ood pulse arrived at around 8.00 p.m. without us noticing it … we were in the fossil Bypass at …900 m. The ”ow rose from 1.5 to 3 m3/s, which forced us to turn back as we could not afford to risk being swept away by the rapids in the Swiss CheeseŽ (Fig. 6). On the surface, 50 mm was registered at 6.00 p.m. The ”ood lasted until morning when the ”ow returned to normal.Fig. 6. Muruk cave. The Swiss CheeseŽ series. On a 120-m distance, around 35-m width and 10-m thickness, the river follows small juvenile passages located at different levels. The ”ood just arrived while we took this picture (photo. J.-P. Sounier).

PAGE 8

Audra P. 38online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTThe second day was calm until 8.00 p.m. On returning to the bivouac (-750 m) we saw the River Thurecht rise rapidly and become turbid. At 8.30 p.m., the ”ow rose from 0.5 to 1.5 m3/s. The beach where we had left our sleeping bags was no more that 5 cm above the water. During the night, the water passed through various shades of brown. On the surface it rained 35 mm and Muruk swallowed a muddy torrent of several hundred L/s from 3 … 4 p.m. Over the 5 km route, the speed of the ”ood was a little less than 1 km/h. This is corroborated by the speed estimated for the dye tracing in 1995, which was of the same order (Hobla, 1997). One could reasonably consider it to be representative of the speed of average torrential ”oods. The abundant rainfall of the third day kept the ”ow level up and the water had a chance to clear a little, while the gully did not stop ”owing. The biggest downpour recorded during the month caused no more that in increase in ”ow by a factor of 2-3 at the main drain. We have no idea of the consequences deep in the cave of the large ”ood at the end of the expedition where Muruk swallowed a ”ow of 2 to 3 m3/s.3.2.2. Exploration February 11 of the …537 sump.On this occasion we were able to observe the behaviour of a variety of in”ows to the cave. The day before, 22 mm of rain had swelled most of the streams and there were still ”ooding during our descent. The 35 mm that fell during the afternoon did not substantially modify their behaviour … it did nothing more than prolong the ”ood. The gully ”owed several L/s during the afternoon. Water appeared from the base of the entrance pitch, undoubtedly coming from some sink higher up the gully. The in”ow at …200 m had a considerably higher ”ow (100 L/s instead of 30 L/s), and the water was earthy (Fig. 7). Our explorations showed that this water also probably came from higher up the gully. At …250 m, this stream disappears into a sump, the re-appears at …480 m after a detour via the River Elmedir, also turbid and in ”ood. On the other hand, three other in”ows were not in ”ood. The main one enters at …80 m while the other two are the large in”ows of the Puits du Visconte, which total 100 L/s in low ”ow. Certainly, their ”ows were slightly higher than normal, their water was perfectly clear. These factors lead us to believe that we have two modes of water entry to the system: direct in”ow, rapid and concentrated by the karst in the gully sinks, such as the in”ow at …200m. The capacity of this sink is limited and in ”ood the excess water ”ows down the gully and into the Muruk entrance. On the other hand, the other in”ows represent diffuse drainage from the water in the soil. Their ”ow is sustained throughout drier periods and ”oods in them are moderated. The pedological cover attenuates the heavier rainfalls and acts as a “lter limiting the transfer of solids in the karst so their ”ows remain clear even in times of heavy rainfall.4. Two coastal streams with approximately 30 m3/s in low ”owDue to the lack of surface runoff, the Nakanai Mountains drain to the coast via two distinct rivers. A few days at Pomio allows for collection of data not only from the Galowe River, but also the nearby Matali. When assessing the quantities of water and erosion, a knowledge of the ”ow rates downstream of the cave systems is vital (Audra, 2001). The Kavakuna area feeds the Matali while the Galowe owes two thirds of its ”ow to the Mayang resurgence. Amongst the secondary resurgences, Berenice is the largest. There is of course no precise data on the ”ow of the local rivers. All we have is some estimates, one from the rainy season. 40 m3/s in low ”ow for the Galowe, a quick estimate (Hobla, 1997). 20 m3/s in low ”ow for the Matali, with an estimate of 90 m3/s for the annual average and 200 m3/s in high water (Maire et al., 1981). The ”ood of 1977 that removed the old bridge was estimated at 1000 m3/s (Favre, in: Maire et al., 1981). This gives ratios of 1 to 10 between minimum and maximum with 50 for the extremes.Fig. 7. Muruk cave. In Miriel River near -220, before reaching P 13, at low water (photo. J.-P. Sounier).

PAGE 9

39Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTIn our study, we estimated the ”ow of each of the two rivers. We did this by taking a transverse section and measuring the depth with a graduated pole at 3 m intervals using a topo“l with one end attached to the bank of the river (Fig. 8). At each of these points the velocity of the current on the surface was calculated by measuring the time it took for a leaf to travel the length of a 3 m bamboo pole. By integrating the successive velocities and surfaces of each section it was possible to calculate the amount of water passing the transverse section (Table 2, Fig. 9). The water ”ows over a bed of medium sized gravel (5-10 cm), without whirlpools or a great deal of turbulence. We used a factor of 0.8 to account for the slowing of the water due to the roughness of the river bed. These measurements necessitate the presence of at least three people. The velocity of the water (more than 1 m/s in a depth of 1 m of water) causes stability problems for the measurers. Because of this we chose measurement points where the depth of the water did not exceed 1 m for the entire section. That is, 30 m upstream of the bridge over the Matali and at the location of the old bridge at the Galowe. It must be noted that the cyclone in March 1997 not only removed the old bridge completely, but also washed down a considerable quantity of cobbles, which greatly facilitated our work. The two measurements were taken during a pronounced dry period … that is to say, after two weeks without major rainfall. So we can consider that these “gures represent minimal values for the ”ow. We measure 27 m3/s for the Matali and 33 m3/s for the Galowe.5. ConclusionsThe results of the observations and measurements have been very useful in our understanding of the local karst. The annual rainfall on the massif could be as high as 12.5 m, and at the very least an enormous rainfall. Of course, Fig. 8. A pole, a survey kit and a vegetable ”oat are used to calculate the discharge of the Matali River, close to its mouth. That day, the discharge was 27 m3/s (photo: Ph. Audra). Table 2.Flow calculations for the Matali (12 January 1998). Fig. 9. Transverse section of the Matali.

PAGE 10

Audra P. 40online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011RAINFALL, FLOWS AND PERCOLATION IN NAKANAI MOUNTAINS KARSTthis value is derived from the extrapolation of one months observations and possible error is very high. Analysis of the weather and rainfall patterns did not allow us to establish any correlation between weather and atmospheric pressure at ground level. Therefore, any weather prediction using local ground-based instrumentation is not feasible. The only possibility of reasonable weather prediction is to use a portable computer connected to meteorological services via satellite. Such a system is currently too expensive, but in view of the rapid development of such electronic devices, they may well be standard equipment on expeditions in the near future. The forest/soil system plays a remarkable role in this extremely wet area. Just one or two days without rain are enough to drain these soils which are then able to absorb up to 50 mm of rain without causing any runoff. On the other hand, if the soil is saturated from previous rainfall, a shower of 20 mm is suf“cient to produce runoff. This demonstrates that in the dry season the karst is essentially fed by diffuse drainage with the occasional episode of turbid runoff entering the caves via the surface gully system, but these are limited in both time and volume. These ”oods only lead to moderate ”ooding underground, even considering the speed of propagation (1 km/h) and the torrential nature of the underground conduits. The small sediment load in these characteristically turquoise waters is the best illustration of this. While we may ask how these characteristics have changed with the partial destruction of the forest by the cyclone, it appears that the effects have been quite small owing to the dense layer of fallen timber that was quickly covered by regrowth. The result may not be the same after the seemingly inevitable advance of the forest minersŽ. Finally, the estimates of the ”ow in Matali and Galowe rivers in low ”ow conditions allows us to establish the ratios between low (1), medium (5), ”ood (10) and exceptional (50) ”ows. Even though numerous important questions remain unanswered, we are convinced that we made a maximum of observations during our limited stay of only a few weeks. Substantial progress cannot really be made without a much longer study of at least one year, but such long expeditions are no longer the order of the day.AcknowledgementsI am grateful to P. Carrega for his useful suggestions concerning climatology.ReferencesAudra, P. 2001. Valeur et rpartition de la dissolution spci“que dans les karsts des montagnes Nakanai, Nouvelle-Bretagne, PapouasieNouvelle-Guine (Karst solution and denudation assessment in Nakanai Mountains). In: Audra, P., de Coninck, P. and Sounier, J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 77-86 Ford, E. 1973. Papua New Guinea: The land and the people Jacaranda Press, National Library of Australia. Hobla, F. 1997. Rapport scienti“que. Hmisphre Sud, Objectif Premier -1000. Rapport de lexpdition splo-plonge en Papouasie Nouvelle-Guine, janvier-mars 1995 Hmisphre Sud, Antibes. 1942. Mac Alpine, J. R., Keig, G. and Short, K. 1975. Climatic Tables for Papua New-Guinea Camberra, Commonwealth Scienti“c and Industrial Research Organization. Maire, R. 1990. La haute montagne calcaire. Karstologia Mmoires, 3. University of Nice Sophia-Antipolis Thesis. Fdration franaise de splologie, Paris & Association franaise de karstologie, Grenoble. 731 p. Maire, R., Pernette, J.-F., Rigaldie, C., Sounier, J.-P. et al., 1981. Papouasie-Nouvelle Guine. Spelunca suppl. 3, 48 p.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. IntroductionIts only when you appreciate the dif“culty of water tracing in the Nakanai that you truly appreciate the resultsƒ The dif“culty of moving around the forest, plus having to descend to the bottom of deep canyons made it impossible to collect regular samples or to get automatic sampling equipment to sample sites. The quantitative aspect of our results is unknown, apart from some transfer speeds when we had observers in the correct place at the correct time, and even so the results are very approximate. Still, these essentially qualitative results do give us evidence of the relationships between different points in the karst system and have helped de“ne the limits of the catchment and given us a better idea of where to search for more caves. Three traces that were to de“ne the relationship between Muruk, Arcturus and Andromeda caves were done in the basin that drains to Berenice resurgence in 1995 and 1998.2. The 1995 tracesIn 1995 the junction between Muruk and Berenice had not yet been explored and the objective of the expedition was in fact to make that hypothetical connection a reality, although at the time such a connection was regarded by some as wishful thinking (Sounier, 1992). However, the connection was con“rmed by the “rst trace (Hache et al., 1995; Hobla, 1997), and later by the joining of the two caves in 1998 (Sounier, 2000). Later traces within Muruk added to the understanding of internal hydrology within the cave.2.1. Muruk-Berenice trace (n 1)The “rst and by far the most important trace was to verify the hypothesis that the underground river in Muruk actually resurged in the Berenices Hair spring and not from the Mayang resurgence as was supposed by the original explorers in 1985. Both resurgences are about 4 km from the …597 m sump in Muruk (Fig. 1). It was this possibility that was the initiative for the South Hemisphere, 1st … 1000 mŽ expedition. The trace was started during the second push beyond the sump: 2 kg of ”uorescein was put into the river just downstream from the Milky Way at …750 m at 2:12 on the 30th of January 1995. We arrived at the Berenice resurgence at 10 a.m. on the same day … 7 hrs. 50 min. after the ”uorescein was injected and the plunge pool at the foot of the waterfall outside the entrance of the cave was already green as was the river within Berenice during the next few hours as we proceeded upstream. The activated charcoal bags we placed 150 m in from the entrance at the “rst traverse were removed at 11 a.m. and showed a strongly positive result with alcohol (Fig. 2). The dye had travelled a little over 2000 m horizontally and 400 m vertically between the injection point and the resurgence in something less that 7 hrs. 50 min. This gives us an apparent maximum velocity of 250 m/h on a slope in the order of 20%. But if we take into account the true distance through the cave and the fact that the inhabitants of Galowe told us they saw a Corresponding author: Email: audra@unice.fr by the authors. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license Water tracing in the Muruk-Berenice system (Nakanai Mountains, New Britain, Papua New Guinea)P. Audra1* and F. Hobla21 UMR 6012 ESPACE CNRS, University of Nice Sophia-Antipolis, 98 boulevard Edouard Herriot, 06000 NICE, FRANCE (audra@unice.fr). 2 University of Savoy, EDYTEM, Campus, 73376 LE BOURGET Cdex (Fabien.Hoblea@univ-savoie.fr)Re-published from: Nakanai 1978-1998. 20 years of exploration, Antibes. Hemisphere Sud.Abstract: Three dye tracing have been carried out in the Nakanai Mountains to determine Berenice spring catchment area and some internal connections inside Muruk Cave. Andromede Cave belongs to Muruk-Berenice system, via Arcturus main drain. Due to special conditions in rainforest with dif“cult of moving, we made qualitative tracing using charcoal bags. Key words: tracing; Muruk-Berenice system; Nakanai; Papua New Guinea; charcoal bags.

PAGE 2

Audra P. and Hobla F. 42online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011WATER TRACING IN THE MURUK-BERENICE SYSTEM (NAKANAI MOUNTAINS, NEW BRITAIN, PAPUA NEW GUINEA)noticeable change in the river colour “rst thing in the morning (at around 9:00 a.m.?) … they had no idea we were dye tracing … we can estimate the ”ow to be in the order of at least 1 km/h. The dye had to be ”owing out of Berenice within 3 hours, by 5 a.m., in order to travel the 12 or so km down the river to Galowe. Apart from proving the Muruk-Berenice connection, this trace also showed us that the transit time was very fast and probably with no major obstacles such as large ”ooded zones in the way. Their junction 1998 con“rmed this. The connection to Mayang was disproved even though it was not watched in any way. It would have been interesting to observe the Pleiades resurgence and although we considered it, we had to remove the ropes for use in Berenice. This resurgence is probably fed by water leaking from the Berenice stream, which is suspended some way above the ”oor of the canyon. This does however pose the general problem that the hydrology of the system is somewhat more complex than we had “rst thought.2.2. Internal tracing in Muruk (n 2)A second dye trace removed a big question mark over the water ”ows in the upper part of the cave well before the …597 sump. Its objective was to ascertain the destination of the waters of the Miriel River that disappears in the sump at the entrance of the Cassiquiare (-233). The theory arrived at from the mapping in 1985 was that it reappeared from the waterfall Fig. 1. Hydrology of the Muruk sector. Cave locations after Sounier (1992), with additions. Fig. 2. Fabien Hobla in Muruk Camp, 1995, extracting the dye traces in the water collected at the Muruk resurgence (photo. J.-P. Sounier)

PAGE 3

43Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P. and Hobla F.WATER TRACING IN THE MURUK-BERENICE SYSTEM (NAKANAI MOUNTAINS, NEW BRITAIN, PAPUA NEW GUINEA)in the wall at the lower end of the Cassiquiare (Genuite, 1987). But the problem is that this river is perched in relation to the Cassiquiare, a semi-fossil over”ow, which is itself perched above the Miriel River. The ”ow rate do not match very well which throws the hypothesis further into doubt in favour of the idea that the Miriel is heading towards the Elmedir. The trace was done over the 6th and 7th of February 1995 as we were derigging Muruk. The “lm team, which we had been travelling “lming with, put about 500 g of ”uorescein into the lake just before the Miriel sump at 10 p.m. on the 6th of February 1995. While we put the charcoal bags in key locations on our way to the -597 sump: two bags in the plunge pool below the end of the Cassiquiare (the supposed resurgence in 1985); two bags in the Elmedir a few meters upstream of its junction with the Muruk stream; two bags in the Muruk a few meters upstream of its junction with the Elmedir stream; and just to be sure, two bags in the tiny sump at the top of the rockfall cone in the puits du Visconte. As we passed the Elmedir junction at 11:45 p.m. the river show no signs of coloration, however, as we arrived at the …597 m sump at 2:00 a.m., it seemed to us that the river was changing colour, although we could not be sure. As we began the climb out at 9:45 a.m. after a bivouac in the Elmira chamber, the water was undeniably green. We removed the capture bags very carefully (we had been travelling in a bright green river!) as we went up at 11:10 a.m. for the Elmedir/Muruk junction and 2:15 p.m. at the Cassiquiare. At the “rst junction, the water was no longer suf“ciently coloured to determine which the coloured branch was. At the Cassiquiare there was no sign of coloration, but this could have meant that the dye had completely passed. There was a chance of contamination here as the diving team had passed through on their way up and walked through the green lake just before passing the charcoal bagsƒ Later examination of the charcoal bags on the 8th of February proved without a doubt that the Miriel river that sinks just above the Cassiquiare rejoins the Elmedir river. This trace also eliminated the theory that the Elmedir comes from Arcturus in favour of the theory that Arcturus connects at the Milky Way, as was later demonstrated.3. The Andromeda trace 1998 (n 3)We hoped to do this trace from Southern Cross cave as it is close to the expected boundary between Berenice and Mayang catchments (Fig. 1). Unfortunately, the massive cyclone damage meant that we didnt get to the cave until close to the end of the expedition, and even then after considerable effort. Under these conditions it was not even feasible to get Table 1.Results of traces 1 and 2 in Muruk in 1995 Muruk … Berenice trace (1)Internal Muruk trace (2) Injection point Downstream of Milky Way junction (-750)Miriel river sink at Cassiquiare (-233) Injection time 30 January 1995, 2 h126 February 1995, 22 h Dye tracer Fluoresceine (uranine)Fluoresceine (uranine) Amount of tracer 2 kg 0,5 kg Observed points BereniceVisconte tributary, Con”uent -491 (VisconteElmedir) Appearance points BereniceElmedir River Appearance time Before 9 am, as early as 5 amBefore 9 am, possibly 2 am, Feb 7 (Elmira) Survey methods Activated charcoal bags and visualActivated charcoal bags and visual Observation time 9 h16 h Length 2 100 m420 m (unknown section) 2,4 km (to sump at …597) Height difference 420 m70 m (unknown section)364 m (to sump at …597) Slope 20 %17 % (unknown section) 15 % (to sump at …597) Charcoal bag analysis KOH in alcoholKOH in alcohol Qualitative results Positive at all points => Junction between Muruk-Berenice Positive at Elmedir => Junction between MirielElmedir Transfert time 7 h maxi, 3 h probable11 h max, as little as 4 h Maximal apparent velocity 300 m/h to more than 1 km/h220 m/h, as fast as 600 m/h

PAGE 4

Audra P. and Hobla F. 44online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011WATER TRACING IN THE MURUK-BERENICE SYSTEM (NAKANAI MOUNTAINS, NEW BRITAIN, PAPUA NEW GUINEA)to Mayang. So the trace was restricted to something closer to Muruk. From the maps, it was obvious that Arcturus would connect to the Muruk system and probably via the Voie Lacte Milky Way (Fig. 1). The trace could at least prove the connection, which we were unable to accomplish due to several sumps along the way. Due to timing between rigging in Arcturus and watching the river inside Muruk, we put the dye into Andromeda, a newly discovered cave close to Arcturus (Table 2, Fig. 1). It is 150 m deep with a small stream with suf“cient ”ow to take the dye that sumps right above Arcturus … there is no doubt about a connection. One kilo of ”uorescein was injected on February 2nd during one of the wettest periods of the expedition. Berenice was watched for 60 hours but the charcoal bags, which were placed in ”ood conditions, were not even in the water when we went to collect them. Inside Muruk, the two charcoal bags stayed undamaged for a week in the plunge pool at the foot of the waterfall where the Milky Way joins Muruk. Unfortunately, the two control bags put into the Galadriel River above the …750 junction were not collected due to high water. Analysis of the charcoal bags in alcohol did not give satisfactory results, on passing them through the spectro”uorometer in the Laboratoire de gologie structurale et applique de Besanon (J. Mudry), some ”uorescein was noted (Table 3, Fig. 3). It is interesting to note that the two bags, which were immersed in the same pool only a few meters apart, did not receive the same dose, which highlights the point that such methods do not give reliable quantitative results. The weak result from Berenice is without doubt more to do with the bag not being in the water for long enough than from dilution of the dye. The trace proved that Andromeda and the caves of the Arcturus area contribute to the Milky Way, as is suggested by their maps. It also serves to delimit the catchment a little better at its western edge. Nevertheless, Arcturus has a ”ow of Table 2.Results of the third trace, from Andromeda Cave in 1998.Table 3.Quantitative results of spectro”uorometric analysis of activated charcoal bags. Injection point Terminal sump, Andromede (-150m, alt. 1220 m) Injection time 2 February 1998, 12 h Dye tracer Fluoresceine (uranine) Tracer amount 1 kg Hydrologic conditions Normal (0,25 l / s), evening ”ood Injection staff Al, Mark, Greg Observed points Berenice (3 m3/s, alt. 265 m) Con”. -750 (riv. Galadr. 500 L/s and Milky Way 1.5 m3/s) Appearance points Milky Way, Berenice Survey methods Activated charcoal bags Observation time 60 h Berenice, 1 week Muruk Length => Berenice 5.2 km Vertical diff. => Berenice 955 m Slope => Berenice 18 % Active charcoal analysis. KOH in alcohol, spectro”uorimetry Qualitative results => connection AndromedeBerenice via Milky Way in Muruk Transfert time Much < 60 h Maximal velocity Much >100 m/h Charcoal bag location C (g/l) pic (nm) Result Milky Way 750 (1) 23476Positive Milky Way -750 (2) 160490Positive Berenice1490Positive (weak)Fig. 3. Spectro”uorometry curves of charcoal bags.

PAGE 5

45Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P. and Hobla F.WATER TRACING IN THE MURUK-BERENICE SYSTEM (NAKANAI MOUNTAINS, NEW BRITAIN, PAPUA NEW GUINEA)around 200 L/s while the Milky Way, only one kilometer away, has a ”ow of 1.5 m3/s. There is a major feeder which drains a large area arriving between these two points. This water is coming from somewhere higher up the plateau, perhaps from the Southern Cross area if it does not go to Mayang, or perhaps further to the southwest … an area unknown to cavers. This possibility was con“rmed by physico-chemical analysis of the underground waters (Audra, 2001). In view of the results, it is unfortunate that the cyclone damage to the forest stopped us from doing a trace from Southern Cross Cave as it would have much better delimited the extent of the Berenice catchment.ConclusionsThe results of the three traces in the Muruk area have allowed us to approximately delimit the Berenice catchment and de“ne some internal connections well before actually making the physical connection. It seems that the Milky Way is fed by the plateau above Arcturus, but unfortunately the state of the forest did not allow us to resolve the question as to the boundary between the Berenice and Mayang catchments which must be in the region of Southern Cross, so we still do not know which resurgence this major cave ”ows to. The travel velocities of the dye were extremely fast as all the streams are mountain torrents rather than slow phreatic ”ows. Finally it must be noted that while the use of charcoal bags may not be the recommended method for a normalŽ trace, it is not practical or sensible to take expensive equipment to a location with such dif“cult access let alone get an automatic sampler into a place like Berenice.AcknowledgementsOur thanks to J. Mudry for the spectro”uorometric analyses.ReferencesAudra, P. 2001b. Valeur et rpartition de la dissolution spci“que dans les karsts des montagnes Nakanai, Nouvelle-Bretagne, Papouasie-Nouvelle-Guine (Karst solution and denudation assessment in Nakanai Mountains). In: Audra P., de Coninck P. and Sounier J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 77-86 Genuite, P. 1987. Perte de Muruk. Antipodes 85. Rapport des expditions nationales Papou 85Ž et Niugini 85Ž en Papouasie Nouvelle-Guine. Spelunca Mmoires 15, 48-51. Hache, P., Hobla, F., Philips, M., Sessegolo, D. and Sounier, J.-P. 1995. Muruk, hmisphre sud, premier -1000. Spelunca 60, 35-54. Hobla F. 1997. Rapport scienti“que. Hmisphre Sud, Objectif Premier -1000. Rapport de lexpdition splo-plonge en Papouasie Nouvelle-Guine, janvier-mars 1995 Hmisphre Sud, Antibes. 1942. Sounier, J.-P. 1992. Deux moins mille en Nouvelle-Bretagne, ou lintret dorganiser une expdition mixte splo-plonge. Spelunca 46, 15-18. Sounier, J.-P. 2000. Muruk : lpilogue ? Spelunca 77, 15-22.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. Seismic activity in Papua New Guinea1.1 A subduction zone in the Ring of FireŽNew Guinea, the Bismarck Archipelago and the Solomon Islands form an arc along the junction between the continental Australian plate in the south and the Paci“c oceanic plate in the north (Fig. 1). The contact between these two plates is the origin of strong seismic and volcanic activity. This area of the famous Paci“c Ring of FireŽ concentrates 5-10% of the worlds seismic activity and is one most mobile and unstable regions of the earths crust (Hobla, 1997). The plate of the Solomon Sea is plunging towards the north beneath the island arc of New Britain at a speed of about 10 cm/year. This subduction is the cause of the New Britain trench (-7880 m) and of the active volcanic arc, which borders the northern coast of the island. Local seismic activity is recorded at Rabaul Volcanic Observatory (RVO), along with other parameters in order to gain better understanding of the seismic activity of the region. These data are then sent as input to the Preliminary Determination of EpicentresŽ of the US Geological Survey. During our stay, 69 seismic events of a magnitude between 3 and 5.8 were recorded. The statistical treatment of these data has allowed the calculation of the frequency of major seismic events (Table 1). Large earthquakes of the type that ravaged Armenia in 1998 (m. 6.9) have a frequency of more than one a century while catastrophic earthquakes of the magnitude of the 1906 San Francisco earthquake (m. 8.25) have a frequency in the order of one every two centuries. If we apply this frequency to the Quaternary Period, the number of catastrophic earthquakes (probably including magnitude 9+ events), is in the order of 10,000.1.2. Earthquakes felt during the expeditionThree earthquakes were felt in various degrees during our time in the Nakanai (Fig. 2).At 7:14 am on the 21 1. st of January (21:14 20/1/98, GMT), a magnitude 3.7 shock with its epicentre 75 km away and at depth of 33 km shook the bottles on shelves of medical equipment in Muruk camp (1445 m asl.). Only the expedition doctor noticed the earthquake (intensity III). At 7:34 on the 27 2. th of January (21:34 27/1/98 GMT), a magnitude 5.1 shock with its epicentre 50 km away at a depth of 83 km shook the camp from side to side for 10 to 15 seconds. The camp ”oor, built on fallen tree trunks, as well as the framework of the hut on stilts was noticeably shaken (intensity IV). At the same time, three members of the expedition were having breakfast in the …750 m bivouac. They felt no movement, but did hear a loud, dull boom. They thought that a violent ”ood was on its way, but nothing arrived. At 8:30 on the 5 3. th of February (the exact time is uncertain) a shock preceded by a rumbling noise was felt at Mara camp (intensity IV), at 700 m asl., above the Berenice resurgence. Nothing was felt at Muruk only 4 km away. Corresponding author: Email: audra@unice.fr by the author. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license The role of seismic activity in the formation of large underground cavities in the Muruk System, Nakanai Mountains, New Britain, Papua New GuineaP. AudraUMR 6012 ESPACE CNRS, University of Nice Sophia-Antipolis, 98 boulevard Edouard Herriot, 06000 NICE, FRANCE (audra@unice.fr)Re-published from: Nakanai 1978-1998. 20 years of exploration, Antibes. Hemisphere Sud.Abstract: Papua New Guinea is one of the worlds most active seismic areas. On the surface, huge landslides are found on mountainsides and steep canyons slopes. Underground large passages and megadolines result mainly from the erosion of soft limestones by underground streams, but seismic movements may accelerate their evolution. Morphological characteristics were derived from statistical data and “eld observations. Key words: seismic activity; caves; Nakanai, Papua New Guinea

PAGE 2

47Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.THE ROLE OF SEISMIC ACTIVITY IN THE FORMATION OF LARGE UNDERGROUND CAVITIES IN THE MURUK SYSTEMTo these we should add another earthquake at 00:54 (GMT) on February 5th, and another at 09:28 (GMT) on February 6th. The “rst had a magnitude of 4.6 with an epicentre 150 km away, and the second had a magnitude of 3.9 with an epicentre close to 250 km away. In general terms, the frequency of shocks felt at Muruk is in the order of one a week, which is quite a lot, especially when compared to the virtually non-seismic regions where we live. Differing effects at various locations is obvious. The rumbling and shock felt at Mara, which were no more than different manifestations of the same event, passed unnoticed at Muruk which is not far away but at a higher altitude. Finally, no shocks were felt underground. The same effect was noticed in 1995 (Hobla, 1997). The dull explosion heard at …750 m could be interpreted as movement along a fault or joint moving suddenly, or more probably a block or part of the ceiling falling due to the vibration. This aspect of an only minor effect underground is because the rock mass is under lithostatic pressure, which limits its movement, and its only at the surface that the waves transmitted by the lithosphere become obvious due to resonance and re”ection. As well as these observations I should add that we observed rock shards, obviously ejected from a fault, and only washed a little by the rainy season. This therefore could have only occurred a few months before.1.3. A considerable morphological impactOn the surface, the effect of seismic activity on the morphology is obvious. While it is insigni“cant on the gentle plateaux, it isnt in the 500 … 1000 m deep canyons such as the Galowe. Large white scars cutting the sub-vertical forested walls attest to the instability of these slopes (Fig. 3). The chalky Yalam limestones, which are porous, weathered and saturated with water, are particularly unstable. One such slide, fortunately a small one, occurred on the Berenice wall that was rigged with ropes. Its highly likely that it followed the shock that was felt so strongly at Mara on the 5th of February. While seismic activity cant explain all of these gravitational movements, it is clear that it contributes to the breakdown of cliffs in the region. The 1985 earthquake (magnitude 7.1) Fig. 1. Structural outline of the west Melanesian archipelago (after Karig, 1972, in Debelmas and Mascle 1993). Table 1.Recurring periodicity of large seismic movements in New Britain (after RVO).Fig. 2. Earthquakes that may have been felt in Nakanai in January and February 1998 (after RVO). Numbers between brackets correspond to earthquakes described in the text. Magnitude (Richter) Local intensity (MMI) Potential effects on landscape Recurring periodicity (year) 67-8Minor landslides and ground subsidence 1 / 30 78-9Large landslides and ground subsidence 1 / 80 810-11Catastrophic landslides and ground subsidence 1 / 200

PAGE 3

Audra P. 48online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011THE ROLE OF SEISMIC ACTIVITY IN THE FORMATION OF LARGE UNDERGROUND CAVITIES IN THE MURUK SYSTEMcaused a large number of landslides including one giant slide in the Bairaman canyon 25 km south of Muruk. This one kilometre long cliff collapse produced a dam 200 m high which had to be blasted to avoid a potential catastrophe for residents downstream should the reservoir “ll, and then break (King et al., 1987, in Maire, 1990). An estimate of the rock mobilised by earthquakes combined with the frequency of events puts the estimated volume of rock moved by seismic activity in the same order as that moved by karst dissolution, though of course the former occurs sporadically in both time and place while dissolution is much more diffuse and generalised (Maire, 1990). At depth, numerous effects can be seen. Tectonic readjustment can cause movement along structural discontinuities. We have seen “ssures a few millimetres long in active calcite formations in nearby Tucana cave (top of crystals). At …900 in Muruk the route through the cave takes a small, perched, inactive passage the BypassŽ, which is developed along a sloping bedding plane. Large cobbles have been jammed in the bedding plane when the passage was an active watercourse, then cemented by calcite. The joint has later moved a few centimetres and the chalky cobbles have been sheared apart as they were weaker than the calcite cementing them (Fig. 4). This evidence is all that weve so far found and is rather moderate when compared to that found in some alpine caves. This can partly be explained by the lack of dry levels which tends to conserve such features better, and also due to the low mechanical strength of the rock which is regularly reworked by ”oods and collapses so that any surface irregularities are quickly removed. It is highly likely that the older markings have long since been removed by erosion. Vibration effects are likely to cause collapses of the walls and ceilings in varying degrees. This seems to correspond with the phenomenon heard at …750 m, and it is probable that seismic activity is one of the factors, which explains the existence of such large passages and megadolines in the area.2. Development of large underground cavities and the role of seismic activityThe formation of large cavities depends on the state and the size of different passages observed. It is possible to reconstitute the evolution of a passage from a normal to a giant passage (Fig. 5).2.1. Successive stages of evolution towards a large passage (based on “eld examples, mainly in Muruk Cave)A (entrance passages of Tucana Cave) Due to a lack of fractures, the majority of passages start at bedding planes. The majority of these are later modi“ed due to tectonic constraints and slippage between layers, as indicated by obvious striations. This reworking is critical in the widening of the discontinuities. The initial passage develops as a sinuous passage along the joint. Owing to extremely contrasting water Fig. 3. Scar on the vegetated cliff of the Galowe canyon, on the left face to Berenice spring, probably linked to a debris ”ow provoked by a seismic movement or a cyclone; the height of the cliff is about 1000 m (photo by Ph. Audra). Fig. 4. Cobble cemented into the ceiling, sheared off by a recent movement of the bedding plane (photo by J.-P. Sounier).

PAGE 4

49Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.THE ROLE OF SEISMIC ACTIVITY IN THE FORMATION OF LARGE UNDERGROUND CAVITIES IN THE MURUK SYSTEM”ow regimes, the passage is either dry, with limited erosion (modest, non-aggressive ”ow in normal conditions; Audra, 2001b), or ”ooded. In the latter case high, aggressive ”ows the passage functions as a tube with enlargement of the ceiling and ”oor on either side of the joint. B (entrance passages of Tucana and Muruk Caves, Cassiquiare passage) Enlargement of the tube allows some of the water to have an air surface in moderate ”ood. This produces a deepening of the canyon as it traces the passages initial sinuous route without exaggerating it. The walls of the canyon are washed by torrential ”ows, while during larger ”oods the passage is totally ”ooded. As the ”ood subsides, suspended clays are deposited on the walls of the tube. These clays stay in place, as the torrential ”ows cant reach them to remove them. C-D-E (extreme upstream of the Milky Way Main Drain) As the passage is enlarged, there is a differentiation of the walls depending on their stability which is linked to the distribution of various mechanical constraints (Renault, 1967). A rounded ceiling is very stable. The walls at the base of the canyon develop due to widening and deepening. Meanwhile, the most unstable zone is at the top of the canyon. With the unloading on one side, the decompression severely affects the walls, which spall off plates, which are then removed by the water below (Gilli, 1984). This eventually produces an inverted water dropŽ passage. Depending on the ”ow regime, large ”oods may still “ll the passage to the roof and deposit more clay. If not, there will be a slope of dusty gravel covering a sloping wall of plates in the process of spalling. F (Galadriel River downstream to its con”uence with the Elmedir River) Continued enlargement of the passage causes instability in the ceiling. Bed collapses at a time forms an inverted ladderŽ, and tends towards a new pro“le with its ceiling in equilibrium (Fig. 6). It is also possible the ceiling collapse could be stalled when it encounters thicker beds. G (Milky Way passage upstream to the con”uence) Continued enlargement causes collapse of the ceiling due to decompression and unhinging of wall plates. The volume of blocks that accumulate in the river limits any increase in depth of the canyon. The waters energy dissipates passing through these rock obstacles as well as against the walls. As the passage is no longer deepening, it enlarges and may reach a diameter of 50 m. The size of the fallen blocks increases as the collapse increase in magnitude and passage development is very active. The presence of a major fault can allow an even greater enlargement and form a large room, although it is in fact no more than a localised section of structurally enlarged passage. H (-780 Con”uence Chamber, Mirror of Galadriel) Once a section of passage has attained a certain equilibrium, collapse of large rock plates ceases. The talus on the ”oor becomes covered with small chalky rocks, which come from wall collapse in the “nal stages of equilibrium. As the rocks get smaller, is river once again able to deepen its course and begin then next phase of collapse and enlargement.2.2 Development of large passagesIt is necessary to remember that the different stages of development do not as a whole have any chronological signi“cance. All of the passages described are contemporaneous, each being at a particular stage of development. A given passage moves from one stage to the next due to erosion, each following an almost identical pattern, Fig. 5 Evolutionary stages of Muruk passages (see comments in the text).

PAGE 5

Audra P. 50online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011THE ROLE OF SEISMIC ACTIVITY IN THE FORMATION OF LARGE UNDERGROUND CAVITIES IN THE MURUK SYSTEMwith differences in detail due to local structure. This erosion is proportional to the force of water running through the passage. It is therefore logical to see the largest passages, those which have developed the most, at the greatest depths where the waters have concentrated into powerful underground rivers. The ”ow observed in low conditions is in the order of 1 to 2 m3/s. In ”ood it has been estimated at 20 m3/s while in extreme ”ood it could reach 100 m3/s (Audra, 2001c). The “nal phase of development is the formation of a megadoline, which forms an open window to the subterranean river below. This is due to the stooping upward of the ceiling of a large passage roof below the ”oor of a doline, as in seen in Nare and Minye (Maire, 1981). In the upper Galowe plateau where Muruk lies megadolines are practically absent, apart from Wunung and Haricot, both of which are blocked with “ll and dont give access to the underground river. This difference is simply explained by the much greater depth below the surface of the major drains. Around Minye the drains are at a maximum depth of around 350 m. With ceilings up to 150 m high and dolines above, the thickness of rock in the roof can be quite thin. At Muruk, where the depth is in the order of 450 m, the thickness of limestone between the highest ceilings and the deepest dolines is never below 350 m, which explains the virtual non-existence of collapses. The difference in structure must also be considered. From the lithological point of view, the chalky limestones of the Yalam provide suf“cient mechanical strength for the development of large voids, helped along by strati“cation that favours equilibrium and cantilever ceilings (White and White 1997). Following their collapse the rocks break easily and are readily removed by the underground river. Moderate fracturing on a local scale allows the passage volume to grow as large as possible without its enlargement being interrupted by any major discontinuity. The principle of undermining-unloading has already been recognised as the major processes in the formation of large underground voids (Gilli, 1984). After studying numerous examples in France and around the world this author has ranked it in front of faulting, the abundance of water ”ow in underground rivers, and the presence of a contact between massive fractured limestone and underlying soft marls or equivalent. In our case, the high ”ow rate is the main force of downcutting. In every case, the large passages develop in the very heart of the limestone with no particularly soft underlying rock that will erode laterally. Certainly, the chalky Yalam limestone favours this process more than does a massive limestone. It is also apparent that the abundance of high-energy water ”ow generates large voids without having to consider the structural context, even though it didnt appear in Gillis work. We also must note that the passages in Muruk dont exceed 50 … 70 m in width, which is in the order of a quarter of the size of the worlds largest known passages. Even though the driving force for the formation of large voids is the torrential ”ows which provokes gravity collapse, it is also clear that seismic activity and tectonic reworking can do nothing but accelerate this process. Fractures are rare, but nevertheless, the bulk of large rooms are aligned along large faults (Fig. 6). Also, the horizontal bedding is broken into individual blocks by well de“ned joints, as can be seen by slide marks, which add to the number of discontinuities and aid in their removal from the ceiling. Tectonic activity plays a dual role: on one hand it helps create the discontinuities and prepare the rock for collapse, on the other hand it actually provokes the collapse by moving the rock along the discontinuity, or simply unbalancing rocks with a sudden jolt. The Muruk system, with an age of a few hundred thousand years (Audra, 2001a; see also Audra, Lauritzen and Rochette in this issue), has, in its development, probably been subjected to more than a thousand very large earthquakes. The importance of seismic activity as a parameter cannot be ignored if we want to understand the size of these caves and the rapidity of their development.AcknowledgementsMy thanks to P. de Saint-Ours et B. Talai, of the Rabaul Volcanic Observatory (RVO), for their help in supplying so much documentation on the seismicity and vulcanology of the region.Fig. 6. Cantilever roof in one of Muruks large passages (photo by J.-P. Sounier).

PAGE 6

51Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoAudra P.THE ROLE OF SEISMIC ACTIVITY IN THE FORMATION OF LARGE UNDERGROUND CAVITIES IN THE MURUK SYSTEMReferencesAudra, Ph. 2001a. Lhyperkarst des Montagnes Nakana, PapouasieNouvelle-Guine, Nouvelle-Bretagne. Modle dvolution dun rseau juvnile, gouffre Muruk, bas sur des datations U/Th et palomagntiques des sdiments, in: Audra, P., de Coninck, P. and Sounier, J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 93-99 Audra, P. 2001b. Valeur et rpartition de la dissolution spci“que dans les karsts des montagnes Nakanai, Nouvelle-Bretagne, Papouasie-Nouvelle-Guine, in: Audra P., de Coninck P. and Sounier J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 77-86 Audra, Ph. 2001c. Prcipitations, ruissellement et in“ltration dans le karst des montagnes Nakana, Nouvelle-Bretagne, PapouasieNouvelle-Guine (Rainfall, ”ows and percolation in Nakanai Mountains karst). In: Audra P., de Coninck P., Sounier J.-P. (Ed.), Nakanai 1978-1998. 20 years of exploration Antibes. Hmisphre Sud. 65-76 Debelmas, J., Mascle, G. 1993. Les grandes structures gologiques Paris, Masson, 300 p. Gilli, . 1984. Recherche sur le creusement et la stabilit des grands volumes karstiques souterrains Doctoral Thesis, University of AixMarseille. 2 t., 180 p. Hobla, F. 1997. Rapport scienti“que. Hmisphre Sud, Objectif Premier -1000. Rapport de lexpdition splo-plonge en Papouasie Nouvelle-Guine, janvier-mars 1995 Association Hmisphre Sud, Antibes. 19-42. King, J. P., Loveday, I., Schuster, R. L. 1987. Failure of a massive earthquake-induced landslide dam in Papua New Guinea. Earthquakes & Volcanoes, 19, 2, 40-47. Karig, 1972. Remnant arcs. Bull. Soc. Geol. Americ. 83, 1057-1068. Maire, R. 1981. Synthse hydrogologique et karstologique, in: Maire, R., Pernette, J.-F., Rigaldie, C., Sounier, J.-P. et al., 1981. Papouasie-Nouvelle Guine. Spelunca suppl. 3, 23-30. Maire, R. 1990. La haute montagne calcaire. Karstologia Mmoires, 3. University of Nice Sophia-Antipolis Thesis. Fdration franaise de splologie, Paris & Association franaise de karstologie, Grenoble. 731 p. Renault, Ph. 1967. Contribution ltude des actions mcaniques et sdimentologiques dans la splognse, Annales de Splologie 22, 1, 5-21 and 22, 2, 209-307. White, E. L., White, W. B. 1997. Mechanics of cave breakdown: Relative importance of shear strength and fracture toughness. Proceedings of the 12th International Congress of Speleology, La Chaux-de-Fonds 1, 155. Union internationale de splologie, Postojna.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifersonline scienti“c journal www.speleogenesis.info 1. IntroductionThe Jurassic Lar limestone with a total area of about 400 km2 belongs to the Alborz chain. Due to tectonics and dissolution processes, this aquifer was karsti“ed and represents a major hydrogeological unit in the north of Iran. The Poshte-Naz (36o 34 to 36o 42 N and 53o 16 to 53o 50 E) of mostly montaneous terrain (elevation 50-1400 m) constitute a part of the Lar formation, and covers an area of about 80 km2 (Fig .1). The local population strongly depends on water from karst areas and karstic springs, the Emarate spring in particular, which supply almost half of the Behshahr City drinking water needs. The mantled Posht-e-Naz karstic terrain with well developed karstic features (sinkholes, ponors, caves, and dry valleys) has a thickness of several hundred meters (250-300), receives over 800 mm/y of rainfall and experiences humid climatic condition. Dense vegetation cover and relatively thick soil cover this karst formation on the top. Impermeable shales and schist of Shamshak formations (lower Jurassic) underlie the aquifer at the base. Many karstic aquifers hold important groundwater resources that are extensively used for different purposes. At the same time, karst aquifers are seriously vulnerable to contamination resulting either from human activity or environment. Contamination can easily reach the groundwater through thin soil, swallow hole or via sinkhole, and are rapidly transported over large distances in the conduit network (Vesper et al., 2001). In general, because of speci“c hydrogeologic characteristics of karstic aquifers, saline water intrusion, microorganisms and turbidity are the most critical contaminants. Thus, karst groundwater requires speci“c protection. Proper management of karst aquifers needs a better knowledge of ”ow and transport mechanisms in these systems (Ozyurt and Bayari, 2007). Tracer tests, hydrochemical and microbiological investigations are suitable methods for providing a scienti“c basis for development of sustainable groundwater protection schemes (Bakalowicz, 2000; Drew and Hotzl, 1999). Water tracing is a well-developed, powerful tool of the karstic hydrologist that enables catchment boundaries to be estimated, areas of recharge to be determined and sources of pollution to be identi“ed (Ford and Williams, 1989). Groundwater tracing studies using ”uorescent dyes are a commonly accepted technique to determine ”ow connection between accessible input and output points, to delineate karst Corresponding author: Email: kalantari_n@scu.ac.ir by the authors. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license A dye-tracing investigation in the Poshte-Naz Karstic aquifer, Alburz Mountain, northern Iran N. Kalantari1 and R. Mohammadi21 Geology Department Shahid Chamran University Ahvaz, Iran2 Water resource organization Qom, IranAbstract: The tracing technique has been recently used in karsti“ed Zagros structural belt in northern Iran. A tracer study (uranine injection) was conducted in Jurassic limestone of the Poshte-Naz area in the Alborz belt to evaluate aquifer parameter s and hydraulic relations between a large (about 100 m in diameter) sinkhole and springs. A main goal of the project was to “nd out the source of turbidity of the Emarate drinking water supply spring (SP4) in rainy seasons. Eight springs, three wells and the Neka River were selected for monitoring and totally 989 samples in 107 days were collected. In order to select reliable samplin g stations, hydrochemical analysis of major ions was carried out and for better interpretation of concentration-time curve, sprin g discharge was also measured. The results of the tracing by sampling water indicated only a hydraulic connection between the injection point and the Sange-Nou spring (SP8) and, whereas the charcoal bags analysis revealed tracer exits also from spring SP1, SP3, SP4, SP5, SP8, in wells W1 and W2, and in the Neka River. This paper discuses concentration/time curves from charcoal bags for qualitative analysis and tracer exit curves for quantitative analysis. Key words: Posht-e-Naz, tracing, spring, sinkhole, charcoal bag

PAGE 2

53Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoKalantari N. and Mohammadi R.A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERdrainage basins, and to investigate the ”ow behavior of karst aquifers (Wan fang et al., 2002). Many ground water problems, such as source water protection investigation, can bene“t from simple point to point groundwater tracing studies (Eckenfelder, 1996). A part of the Posht-e-Naz karstic aquifer is an area extremely susceptible to contamination by suspended particles. Therefore, a tracing test was proposed in the Posht-e-Naz karstic aquifer with aim to “nd out the source of turbidity (sinkhole) and to evaluate aquifer characteristics.2. GeologyThe Posht-e-Naz anticline form a part of limestone highs in the south of Behshahr City and lies almost between the Neka River in the south and the Khazar fault in the north. The study area is dominantly comprised of Jurassic Lar limestone (Fig. 2), while a small part is covered by calcareous sand stone, rock avalanche and loess. In general, the water bearing Lar formation overlies impermeable Shamshak and Gorgan formations (Fig. 3).Fig. 2. Geo-morpho-hydrogeological map of the study area. Fig. 1. Location map of the study area.

PAGE 3

Kalantari N. and Mohammadi R. 54online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERThe minor and major structural features such as fractures and faults occur in the area. The main fault is the Khazar thrust fault, which control the hydrogeological properties of the karstic formation. The most widely distributed forms of the lineaments are fractures and small scale faults. The density of longitudinal, lateral and shear fractures is very signi“cant in the study area and the fracture widening approaches to few centimeters. As shown in rose diagram (Fig. 4a), the longitudinal and lateral fractures are respectively trending N160-170 and N25-30. Two sets of shear fractures were identi“ed, trending N45E and N35W. Apart from fractures, several normal faults extending NE-SW and NW-SE cross the karstic formation. In addition to these small scale faults, one outstanding fault (Kenet) trending NE-SW crosses the area (Fig. 2). Faults and fractures have a crucial in”uence on the hydrogeological properties of the karstic system, and the surface and sub-surface drainages are associated with these faults and related fractures. The Kenet fault and the lateral fracturesin particular play a signi“cant role in connection with in“ltration and leading groundwater movement.3. GeomorphologyThe well developed karstic geomorphological features are mostly concentrated on the northern limb. In addition to dry valleys, ponors, active and inactive caves, the large and deep sinkholes appear to be more frequent features on the crest of the Posht-e-Naz anticline and some of them (Sh1, Sh4 and Sh5) act as swallow holes. Lithological properties, fracture density, and rainfall frequency are controlling factors in sinkhole development in the area. They have been mainly developed by combination of dissolution and subsidence processes. Majority of the large sinkholes occurr along the Posht-e-Naz anticline axis and those located in the vicinity of the main spring (the Emarate spring; SP4) are labeled as Sh1, Sh3 and Sh4. Two to three axes of the sinkholes (Fig. 4b and c) indicate different forces with diverse magnitudes in operation, so their longer axes directed along the main linear features. Sinkholes are entrance point for contamination of groundwater systems that mainly include sedimentary solid particles of various sizes (“ne sand, silt and clay). The altitude, diameter, swallow hole diameter and depth of the injection sinkhole (Sh1) are respectively 800, 200, 1 and 10 m. The largest sinkhole (Sh4) with three swallow holes occupy an area of about 20 hectares. 4. HydrogeologyA major part of the Posht-e-Naz limestone is subjected to intense karsti“cation, and the area under investigation mostly represents a typical doline karst. The dolines are often aligned and some of them act as swallow holes. The favorable climatic condition, lithological characteristics and tectonics features (fractures, faults and the Posht-e-Naz anticline), as well as geomorphological features and dissolution, enhance fracture porosity, impact karst groundwater potential and generate prominent aquifer systems. Diffuse recharge from in“ltration through pore space and concentrated recharge through welldeveloped ponors, sinkholes and shafts recharge the karstic reservoir. In addition, the position of the impermeable rocks in the core of the Posht-e-Naz anticline, that hinder downward groundwater movement, play a crucial role for underground reservoir development. The Posht-e-Naz karstic aquifers are rich in groundwater and form the principal source of water supply for Behshahr City and surrounding areas. Water supply systems have been mainly based on the intake of natural discharges from karstic aquifers at springs. Due to topographic gradients, stratum dip, fault behavior, exposed fracture and coincidence of groundwater with base level erosion, groundwater is emerging in the form of springs with variable discharge in different altitudes (Table 1). Fig. 3. Geological cross section of the study area. Fig. 4. Fractures rose diagram and sinkholes axes.

PAGE 4

55Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoKalantari N. and Mohammadi R.A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERThe spatial distribution of springs has been dominantly determined by small and large structural features. The collected data suggest that the shear fractures were mainly responsible for emergence of springs. The greater number of springs in the northern part of the Posht-e-Naz anticline is the most obvious hydrogeological feature in the area (Fig. 2). Among all springs, SP2, SP3 and SP4 are characterized by higher discharge, and the main spring in the area is SP4 with an average discharge of 60 L/s. The springs with higher discharge, namely SP3 and SP4 (Fig. 2), were assumed to be connected with sinkholes and the fault zone. It was believed prior to tracing that SP4 discharge is governed by recharge from Sh1 (sinkhole) and Kenet fault (Fig. 2). The Emarate spring (SP4) collects water from large parts of karstic aquifer. Hence, the contribution of the swallow hole of Sh1 in dry season is small, but it becomes considerable in the rainy period. Based on discharge ”uctuation, springs SP2, SP3, SP4 were distinguished as conduits, SP5 and SP6 exhibited fracture ”ow and SP1, SP7, SP8 and SP9 demonstrated diffuse type. The time-discharge breakthrough curve analyses of the high discharge springs (SP3 and SP4) have demonstrated micro ”ow regimes. 5. Hydrochemistry Groundwater quality in the Posht-eNaz is generally good and concentration of the major ions has not evoked signi“cant problems. However, a few good yielding springs used for drinking water supply, particularly the Emarate (SP4), are turbid in rainy season due by surface water input into sinkholes. From the point of view of pollution this situation is closer to surface system than to groundwater system. The use of hydrochemical data proven to be an important tool in understanding groundwater hydrology, commonly used to depict groundwater ”ow regime. When detailed analysis of spring water is done, conspicuous differences in water composition indicate catchments area and anomalies in distribution of hydraulic parameters (Christy et al., 1999, Zanini et al., 2000, Nizar and Alaa, 2008). For the same purpose, from most springs and a few wells in the northern part of the study area samples were collected as many as three times in 2006. Physicochemical analysis of samples was carried out (Tables 2 and 3), and changes of groundwater composition in the command area were assessed. In the course of investigation, springs SP2, SP3, SP4, SP5 and SP6 exhibited more changes in electrical conductivity (EC) than SP1, SP7, SP8 and SP9 and higher temperature ”uctuation and dissolved constituents (Table 2). The EC, which depends on the amount of dissolved solids, is a reliable descriptor of karst water residence time. Thus any decline in EC and ”uctuation in temperature of spring water means fast groundwater ”ow in the system whereas slow groundwater movement would exhibit higher EC and ions concentration and low temperature changes. EC and temperature ”uctuation of SP2, SP3, SP4 indicate fast ”ow, SP5 and SP6 are fracture dominated, and SP1, SP7, SP8 and SP9 exhibit diffuse ”ow (Table 2). As shown in table 2 and 3, the EC and ion concentration at springs indicate minor variation and can be attributed to geochemical characteristics of springs local catchments area, residence time and input water composition at swallow holes of the sinkholes. At all springs, Ca+2 and HCO3 are the dominant cations and anions. Another signi“cant cation is Mg+2. Concentration of the other cations and anions are not important. In spite of point source pollution (sinkhole), nitrate concentration is low. The EC of well samples ranges from 998 to 1458 S /cm2 and the reason for relatively wide range of EC is mostly concerned with lithological variations, recharge rates and well positions with respect to fracture zones. The springs water was of the bicarbonate calciummagnesium type, while the well samples indicated more or less mixed type. The concentration of major ions were diverse at wells, and the dominant cations and anions were respectively sodium, calcium, bicarbonate and sulfate (Table 3). As a result, samples are approaching mixing line in Piper diagram (Fig 5). In spite of temperature, EC and minor chemical composition variations at springs in low and high ”ow condition, the overall EC of springs water in high ”ow period demonstrated hydraulic connection between different parts of the Lar karstic aquifer as a result of highly fracture zones (Table 4) During the study period, similarities in water chemistry between Sh1 and SP4, differences between these two and SP8, and similarities between Sh1 and SP8, were more pronounced as Table 5 demonstrates. These include: (1) saturation of water with respect to calcite (SIc) at Sh1 and SP4, probably accounting for their hydraulic relationship, and the positive SIc value of SP8, are attributable to diffuse ”ow; (2) NO3 concentration exhibited a rise at Sh1 and SP4, as compared to other samples, on one hand, and the decrease at SP4 on the other hand, undoubtedly related to dilution; (3) closer CO2 and Table 1.Springs discharge (L/s) and elevation Spring NosMin discharge Max dischargeFlow typeElevation (m) SP158Diffuse150 SP2867Conduit180 SP335120Conduit550 Sp445120Conduit50 Sp51228Fracture90 SP6735Fracture200 SP7410Diffuse140 SP835Diffuse480 SP924Diffuse775 Sh1 ( swallow hole)5 ( input)20 (input)-800

PAGE 5

Kalantari N. and Mohammadi R. 56online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERdissolved oxygen (DO) values at Sh1 and SP4 and similarities between some other physico-chemical parameters of these springs probably evidence their hydraulic connection. 6. Tracer experimentPrior to the tracer test it was assumed that water ”ows from Sh1 towards SP4, so that the objective was to con“rm the underground connection between these localities. These assumptions were based on the following: (1) the position of the Kenet fault ,which extends almost N-S, to which both localities are alligned; (2) the distance between Sh1 and SP4 is shortest as compared to other sinkholes; (3) Sh1 is located in the SP4 topographic catchment area and elevation difference between Sh1 and SP4 is 750 m; (4) turbidity of SP4 with rainfall. Therefore inspection of the area suggested that the best point for tracer injection is Sh1. Springs and places where appearance of the tracer would be probable were also determined. Tracer tests in karstic rocks make it possible to obtain information on the groundwater ”ow path, ”ow type, hydraulic relations and the other aquifer properties. A review of literature (Kogovsek et al., 1997; Pavicic and Jvicic, 1997; Stevanovic et al., 1997; Kass, 1998; Vu and Nico, 2006; Zargham et al., 2007) indicated that uranine tracing has been most widely used in karstic terrain. The observation stations were checked for natural or man made background by passive detectors prior to the dye experiment. On October 18, 2006 at 7 pm 5 kg of uranine in the form of pre-prepared aqueous solution was injected into sinkhole (Sh1). Then, in addition to rivulet discharge into the shaft, after end of tracer injection 6000 L of water was injected into the shaft in order to clean the shafts wall and to ”ush the dye tracer. Sampling was done by direct water samples and by bags with activated charcoal. Tracer experiment was monitored in 12 locations including the Neka River, springs and wells around Sh1. The observation period was extended over 107 days and in total 989 samples were collected. SP4 and SP5 were sampled in most details. Analyses were carried out with a spectro”uorometer model AU 10, with an accuracy of 0.001ppb. Uranine “rst appearance was 158 hours after injection, and it was only detected at SP8 in water samples during the 107 day test. Since water sample results of injection test demonstrated a connection between Sh1 and Sp8, therefore, probably part of the water sinking at Sh1 emerges to the northeast at SP8. On the other hand activated charcoal results indicated connection between Sh1 and SP3, SP4, SP5, SP8, well 1 and 2, and the Neka River. Activated charcoal residence time distribution curves are not commonly in use for qualitative analysis. Lois & Ponta (1999) determined fracture ”ow path by peak dye concentration of charcoal bags in contaminated karstic terrain. Activated Fig. 5. Piper diagram of the water samples. Table 2. Eelectrical conductivity (S /cm2) and temperature (C) of spring waters. Spring Nos Min temp Max temp Temp difference Min EC Max ECEC difference SP11314.51.5331482151 SP21117.66.6212820608 SP31119.58.5350610260 Sp412.520.17.6334660326 Sp513163.0334670336 SP61116.85.8380650270 SP71415.61.630339895 SP81415.51.543652084 SP912.514.41.929537883 Sh113.416.53.1202504302

PAGE 6

57Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoKalantari N. and Mohammadi R.A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERcharcoal results (Figs. 6 and 7) depicted tracer appearance time and underground network. Comparison of Figs. 7 and 8 shows that the concentration of SP8 samples is approximately 2 times greater than the water samples concentration. This happens due to continuous absorbance of dye tracers on activated charcoal bags. The concentration of dye in SP4 samples (Fig. 6), as compared to SP8 samples, indicate more than 80 times reduced concentration. This might be a result of tracer dilution or low concentration of tracer in SP4 direction and could be also affected by uncontrolled exit of tracer. The differences in arrival date of tracer (charcoal bag results) in SP4 and SP8 with respect to their distance to Sh1 con“rm diffuse and conduit ”ow type for SP8 and SP4. This assumption, and samplers result of the other stations, suggest that Sh1 has also a hydraulic connection to the remaining collected sampling points. The almost unimodal SP8 dye breakthrough curve (Fig. 8) indicates that dye appeared 158 hours after injection. Maximal concentration period was 13 days, and about 100 days later the curve returned back to its initial position. The sharp and unimodal characteristic of curve suggests one main watercourse with no large underground reservoir. The tracer test result in the SP8 catchment is shown in Table 6. Applying a tortuosity of 1.5, the actual distance was obtained. The dominant velocity (v) was computed from actual distance (L) and time with maximal concentration (Tp). Smart (1988) equation was taken to calculate mean residence time Table 3.Physico-chemical characteristics of water samples during low ”ow condition (Mg/L).Table 4.Some characteristics of springs in high ”ow period. Sampling NosPHECCa+2Mg+2Na+K+NO3 -HCO3 -SO4 -2Cl-SIcSId SP17.2949150.123.23.72.00.43238.315.88.2-0.20-0.54 SP27.4339840.0815.82.82.69.03183.116.87.6-0.27-0.73 SP37.4149732.0631.63.51.38.53244.114.96.8-0.2.8-0.35 SP47.1559452.129.24.01.712.0287.113.97.0-0.25-0.55 SP57.2152150.124.33.63.23.96251.513.96.9-0.26-0.62 SP67.2857550.130.419.31.76.13262.915.83.5-0.16-0.33 SP77.2850562.1217.03.52.36.13232.416.87.0-0.12-0.60 SP87.6543038.0714.6160.84.0201.44.814.20.03-0.28 SP97.3156684.1624.34.71.24.022615.824.80.03-1.29 W17.68126470.1426.81013.5-457.652.8460.550.97 W27.7105162.1227.989.62.4 -433.276.8250.510.97 W37.898882.1629.255.23.5 -414.943.342.50.520.83 W47.07145898.1925.5149.43.2 -494.2105.71170.08-0.14 Sh16.7150448.0920.7152.316.023815.817.7-0.8-1.73Table 5.Physico-chemical characteristics of samples in meq/L. SpringsTemperature ( C )PHEC( S /cm2) SP1137.35482 SP215.98.13212 SP315.48.39350 SP415.17.30334 SP514.97.28305 SP615.17.20427 SP7147.29423 SP814.27.65303 SP9137.23319 Sh113.48.0303 Rain-6.8220 StationsSh1SP4SP8 1 21 21 2 Temperature C18 16.5 20 1518.316 PH6.77.157.65 TDS mg/l312349159 Ca++2.42.41.9 Mg++1.72.61.2 Na+0.60.180.7 K+0.060.04.0.02 HCO3 -3.95.03.3 SO4 -0.330.30.11 Cl-0. 50.20. 4 NO3mg/l16124.0 Hardness mg/l688152 Sic-0.8-0.250.04 CO2 mg/l 32.234.727.5 DO mg/l 6.76.75 2.4 Ca/Mg1.41.081.6

PAGE 7

Kalantari N. and Mohammadi R. 58online scienti“c journal www.speleogenesis.info Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFER(Tr) and skewness (S). Dispersion coef“cient (Dd) and the “rst dye appearance time were used to determine basin width, where this in turn was applied to estimate watershed boundary. The distance, velocity and dye appearance determines ”ow type. On the basis of injected dye mass (kg) and the area under curve (A), maximum SP8 discharge (Qm) was 12.7 L/s which is roughly in accordance with SP8 discharge in high rainfall years.7. ConclusionsIn the studied area, occurrence of well-developed geomorphological karst features, spring discharge variations and hydrochemical evidences suggest the presence of underground karst channels. The results of the tracer test with the injection into Sh1 do not support well channel connections. Even, in spite of tracer appearance (activated charcoal) in SP4, due to absence of dye in SP4 water samples, Sh1 and SP4 interconnection is unlikely to account for the source of turbidity in SP4. The karst underground ”ow follows privileged directions, most frequently tectonically predisposed. Then, the question is that what is the role of the Kenet fault? The low returned quantity of tracer probably suggests an uncontrolled exit. The other possible reason for not detecting tracer in water samples in SP4 and low recovery is the connectivity between different parts of the aquifer in the catchment area of the Emarate main spring (SP4) which results huge SP4 discharge and tracer dilution. Therefore, the water-tracing test at this stage only provided some basic information regarding aquifer characteristics in limited area. For detailed understanding more tracings have to be done in different hydrologic conditions. As the differences in water chemistry between the sampling points are not pronounced, it could be indicative interconnection among them (Sh1, SP4 and SP8). The quantitative results of the uranine tracer test for Sang-e -Now diffuse spring (SP8) suggested the catchment area of about 10 km2 actual distance of 4050 m from injection point and the dominant velocity of 0.36 cm/s. Fig. 7. Uranine concentration in charcoal bag at spring SP8. Table 6.Calculated results of tracer test at SP8.Fig. 6. Uranine concentration in charcoal bags at spring SP4. Fig. 8. Breakthrough curve results for uranine at spring SP8. Dominant velocity0.36 cm/s Actual distance4050 m Mean residence time465 hour Skewness30% Dispersion coef“cient0.293x10-2 km2/h Mean basin width1.72 km Flow typeDiffuse Catchment boundary6.96 km2

PAGE 8

59Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 online scienti“c journal www.speleogenesis.infoKalantari N. and Mohammadi R.A DYE-TRACING INVESTIGATION IN THE POSHTE-NAZ KARSTIC AQUIFERReferencesBakalowics M. 2005. Karst groundwater : a challenge for new resources. Hydrogeology Journal 13, 148-160. Christy A C., Brian G.K. and Joshua J.H. 1999. hydrochemical evidence for mixing of river water and groundwater during high-”ow condition, lower Suwannee river basin, Florida, USA. Hydrogeology Journal 7, 454-467. Drew D. and Hotzel H. 1999. Karst hydrogeology and human activities: impacts, consequences and implications. Int Contrib hydrogeology 20, 151-159. Ford D. C. and Williams P. W. 1989. Karst Geomorphology and Hydrology. Chapman and Hall London, 601 p. Eckenfelder Inc. 1996. Guidelines for well head and spring head protection area delineation in carbonate rocks. US Environmental protection agency, Region 4, Atlanta, Georgia. Kass W. 1999. Dyes. In W. Kass (ed.), Tracing technique in geohydrology : 19-122, Rotterdam, Balkema. Kogovsek J. Petric. M. and Liu H. 1997. Properties of under ground water ”ow in karst area near Lunan in Yunnan Province. In A. Kranjc (ed.), Tracer hydrology : 255-261, Rotterdam, Balkema. Lois D.G. and Ponta.G. 1999. Dye study tracks historical pathway of VOC-bearing industrial waste water from failed pond at metals coating facility: In B. F. Beck., A .J. Pettit & G. Herring (eds), Hydrogeology and engineering geology of sinkholes and karst : 293299 Rotterdam: Balkema. Nizar A.J and Alaa K. 2008. Groundwater origin and movement in the upper Yarmouk Basin, Northern Jordan. Environmental Geology 54, 1355-1365. Ozyurt, Nur N. and Bayari, Serdar C. 2008. Temporal variation of chemical and isotopic signals in major discharges of an alpine karst aquifer in Turkey: implications with respect to response of karst aquifers to recharge. Hydrogeology Journal 16, 297-309. Pavicic A and Jvicic. D. 1997. Drainage basin boundaries of major karst springs in Croatia determined by means of groundwater tracing in their hinterland: In B. F. Beck., A J.Pettit and G. Herring (eds), Hydrogeology and engineering geology of sinkholes and Karst 273278. Rotterdam Balkema. Smart C.C. 1988. Quantitative tracing of the maligne karst system, Albert, Canada. Jour of hydrology 98,185205. Stevanovic Z., Dragisc.V., Papic P. and Jemco I. 1997. Hydrochemical characteristics of karst groundwater in Serbian CarpathoBalkanides. In G.Gunnay and A.I. Johnson (eds), Karst waters environmental impacts 199-204, Rotterdam Balkema. Vesper, D., Loop, C. M., and White, W. B. 2003. Contamination transport in karst aquifers. Theoretic Appl Karstol 13-14, 101-111. Vu Thi,. M. N. and Nico. G. 2006. Tracer tests, hydrochemical and microbiological investigations as a basis for groundwater protection in a remote tropical mountainous karst area, Vietnam. Hydrogeology Journal 14, 1147-1159. Wanfang Z., Barry F.B., Arthur J.P. and Brad J. S. 2002. A ground water tracing investigation as an aid of locating groundwater monitoring stations on the Mitchell Plain of southern Indiana, Environmental Geology 41, 842-851. Zanini L., Novakowski K.S., Lapcevic P., Brckerton G.S. Voralek, J and Talbot C. 2000. Ground water ”ow in a fractured carbonate aquifer inferred from combined hydrogeological and geochemical measurements, Groundwater 38, 350360. Zargham M., Raeisi .E. and Zare M. 2007. A dyetracing test as an aid to studying karst development at an artesian limestone subaquifer: Zagros Zone, Iran, Environmental Geology 52, 587-594.



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifers 1. Introduction T racing techniques are widely used for proving hydraulic involves using tracers, whose interaction with particular *Corresponding author: E-mail: michael.sinreich@unine.ch licenseComparative tracing experiments to investigate epikarst structural and compositional heterogeneityM. Sinreich1, 2* and R. Flynn1, 31 Centre of Hydrogeology CHYN, University of Neuchtel, Switzerland2 Proc. 8th Conference on Limestone Hydrogeology, Neuchtel (Switzerland) 21-23 sep. 2006, p. 253-258, Presses universitaires de Franche-Comt, Besanon, France.Abstract: comparing reactive tracer response to that of a simultaneously injected non-reactive conservative substance. Conversely, of tests completed in the vadose zone overlying a limestone aquifer employed a cocktail of particles along with reactive and non-reactive solute tracers to investigate transport rates between the ground surface and monitoring points approximately Comparison of particle (microorganisms) and non-reactive solute tracer breakthrough revealed that particle tracers experience pore exclusion resulting in higher peak relative concentrations which arrive earlier than those of the solute. Prolonged tracer injection and monitoring points were inaccessible to particles, but could allow solutes to pass through them. Similarly, the when soil cover is thin to absent. Keywords: tracer; solute and particle transport; microorganisms; vadose zone; epikarst

PAGE 2

61 aquifers as well as an indication of the role of heterogeneity 2. Field site and experimental setup 2 while reactive solutes consisted of a range of dye tracers, R. eutropha percolation points in the gallerys ceiling during irrigation 3. Results and interpretation Injection type Irrigation pattern Flow conditions pulse transient pulse step Table 1.

PAGE 3

62 These tracers were regarded as surrogates for real organic groundwater quality thus allowing attenuation processes in the responses of selected solute tracers relative to the conservative particle response is helpful for interpreting the structure and those fractions of the effective porosity that R. eutropia earlier than those of the conservative solute, conservative solutes, this alone does not necessarily prove R. eutropha Nature of tracer Rel. max. conc. C/C0Relative recovery organic organic inorganic inorganic Kinorganic Srinorganic Table 2.

PAGE 4

63 R. eutropha concentration and stops increasing despite a plateau of constant relative concentration and particulate tracers, whereas others are At the end of the continuous tracer represented the proportion of total solute concentration would continue to rise consequence of older water, present in the

PAGE 5

This is consistent with data presented collision frequency with reactive surfaces over a longer period and with an additional data; the latter acts as a surrogate for the ionic strength of the A strong decrease of phage concentration occurred at with increasing electric conductivity until no phage were

PAGE 6

concentration until a critical deposition point was reached when concentrations declined the application of low ionic strength water These results illustrate, on the one hand, sensitive to ionic strength increase than strongly indicates that all of the particle Solute tracer responses are also surfaces along these later paths contained

PAGE 7

66 4. Conclusions and conceptual model pathways, very high collision frequency of particles in conduit along of this research with respect to groundwater protection, which R. eutropha

PAGE 8

Acknowledgements References 52 43 Colloid & Surface Engineering Series Water Research 44(4) No 21 Meeting, 40(6) Hydrogeology Journal 17(1) Ground Water 39(6)



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifers The Edwards Aquifer is a large karst aquifer located in southcentral Texas USA. The Edwards Aquifer is the primary source of water for approximately two million people in the greater San Antonio area and is noted for some of the largest yielding wells in the USA (www.edwardsaquifer.org). The Edwards Aquifer extends from the boarder of Mexico at Del Rio, Texas, east beneath San Antonio where it then trends northeast to Austin and Waco, Texas a distance of approximately 550 kilometers (350 miles). The Edwards Limestone is approximately 140 meters (450 feet) thick and may extend as much as 1,000 meters (3,000 feet) below the land surface and more than 700 meters (2,100 feet) below sea level. Water levels in Well J-17, located at Ft. Sam Houston in San Antonio, have been monitored continuously since 1934; and J-17 is one of 52 observation wells operated by the Edwards Aquifer Authority. Because of the long continuous record of water level measurements at the well, J-17 is used as an Index Well for the aquifer and water levels are used to trigger a Critical Period (drought) program in the greater San Antonio area. The top of the J-17 well casing is 222.75 meters (730.81 feet) above mean sea level (msl) with the top of the Edwards Limestone at 74.01 meters (242.81 feet) above msl. J-17 is 265.18 meters (870 feet) deep and does not fully penetrate the Edwards Limestone. However, the aquifer is under artesian conditions at J-17 with water levels ranging from a high of 214.37 meters (703.3 feet) msl in 1992 to a low of 186.69 meters (612.5 feet) msl in 1956. In October 1998, San Antonio received 45.90 centimeters (18.07 inches) of rain in two days and water levels in J-17 rose 2.05 meters (6.72 feet) within a 24-hour period. Artesian wells around the world commonly record earthquakes. There is a long record of earthquakes being detected at J-17 including the December 2004 Sumatra earthquake, the January 2010 Haiti earthquake, and the February 2010 Chile earthquake. Because of the importance of J-17, there are a number of different water level measuring devices in the well (Fig. 1, left). However, earthquake signals for this well are best exhibited with an analog recorder (Stevens Type A-71 Chart Recorder; Fig. 1, right). During the March 11, 2011 Honshu, Japan earthquake (9.0 magnitude), the chart recorder at J-17 showed foot) during the initial response and continuing to oscillate for approximately two hours after the event (Fig. 2). The timeline is set to Central Standard Time USA and the elevations are measured in feet msl. Note UTC time in the narrative box for comparison. The earthquake seismic water took approximately 15 minutes to arrive in the San Antonio area. This material was prepared with the help of Mark Hamilton, David Gregory, and Rob Esquilin with the Edwards Aquifer Authority.Honshu, Japan Earthquake of March 11, 2011 9.0 Magnitude recorded in the Edwards Aquifer, San Antonio, TexasGeary M. SchindelEdwards Aquifer Authority, San Antonio, Texas, USA Corresponding author: Email: gschindel@edwardsaquifer.org by the author. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution censeAbstract: The Edwards Aquifer is a large karst aquifer located in south-central Texas USA. The Index Well J-17, in which water levels in the aquifer are continuously monitored since 1934, detected distinctly the March 11, 2011 Honshu, Japan continued to oscillate for approximately two hours after the event.

PAGE 2

69Speleogenesis and Evolution of Karst Aquifers, Issue.10, 2011 Schindel Geary M.JAPAN EARTHQUAKE OF MARCH 11, 2011 RECORDED IN THE EDWARDS AQUIFER, SAN ANTONIO, TEXAS



PAGE 1

Issue 10, 2011Speleogenesisand Evolution of Karst Aquifers IntroductionBell Hole MorphologyBell holes are common in caves from tropical and subtropical environments. Bell holes can be described as cylindrical to conical, vertical voids in the ceilings of caves. Bell holes have the general appearance of the inside of a church bell with the clapper removed, hence the name. The vertical axes lengths of bell holes are much greater than their corresponding horizontal axes lengths, with heights up to 4 m or more, and widths up to in caves from Sarawak as cylindrical, elongate cavities with circular cross sections and near to completely vertical axes. He also noted that the bell holes seemed to have formed with no regard for variations of bedding dip and joint plane attitudes, or changes in the lithology of the limestone. Bell holes rarely appear as isolated features, but more commonly as groups features known as bell pits (Lauritzen and Lundberg, 2000) are shallow depressions in bedrock commonly associated with the bell holes directly above them (when conditions allow the in diameter, and shallower, than the corresponding bell hole above. Bell holes are most often found in clusters as opposed to being isolated features. The clusters can be very tight, less than a meter apart from one another, and sometimes individual bell holes have merged together (lower right side of Fig. 1). Bell holes have been observed that have been intersected by the apex has been completely removed such that cylinder walls other cases some bell holes open as a smaller hole adjacent to the apex of the cylinder because of intersection by a slope (Fig. 3). Because of the ubiquity of bell holes, the mechanism or mechanisms responsible for bell hole formation are important for understanding the development of macroporosity in karst environments. Bell Hole Origin: Constraints on Developmental Mechanisms, Crooked Island, BahamasAndrew N. Birmingham1, Joan R. Mylroie1, John E. Mylroie1* and Michael J. Lace21 Department of Geosciences, Mississippi State University, Mississippi State, MS 397622 Coastal Cave Survey, 313 1/2 West Main Street, West Branch, IA 52358-9704 Martin, J. B., and Siewers, F., D., eds., 2010, Proceedings of the 14-th Symposium on the geology of the Bahamas and other carbonate regions, p. 18-30. Corresponding author: Email: mylroie@geosci.msstate.ed u by the authors. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution licenseAbstract: Bell holes are vertical, cylindrical voids, higher than they are wide, with circular cross sections and smooth walls found in the ceilings of dissolutional caves primarily from tropical and subtropical settings. They range in size from centimeter to meters in height and width. The origin of bell holes has been controversial, with two proposed categories: vadose mechanisms including bat activity, condensation corrosion, and vadose percolation; and phreatic mechanisms including degassing and density convection. Crooked Island, Bahamas has a number of caves with bell holes of unusual morphology (up to 7 m high and 1.5 m in diameter), Surface intersection has little impact on the phreatic mechanisms, which were time limited to cave genesis from 119 to 131 ka ago, but greatly reduces the time window for later vadose mechanisms, which need to have been completed before bell hole intersection by surface denudation.

PAGE 2

71 Birmingham Andrew N. et al analysis of bell hole morphometry and distribution in caves to provide a means for understanding the processes that may be responsible for bell hole genesis and other cave ceiling dissolutional features. However, as an unpublished MSc thesis, this work has not been readily available wide variety of cusps, pockets, and other convex-upward forms in the centimeter to meter scale; bell holes are one such type of feature, albeit one with a very distinctive morphology. He collected ceiling concavity data from both continental caves in North bell holes in shape. The bell holes were measured along their vertical axes at 20

PAGE 3

Birmingham Andrew N. et al cm intervals, and then along their horizontal axes. He found bell holes in Majors Cave on San Salvador, Bahamas to have an average diameter of 30.97 cm and an average height of 60.20 cm. Similar averages were also found in Lighthouse Roppel Cave in Kentucky revealed ceiling pocket dimensions very different from those of the carbonate island karst, with ceiling pocket diameters almost twice that of the vertical axes. bell holes. Quantitative analysis has also been accomplished for Runaway Bay Caves, Jamaica (Lundberg and McFarlane, 2009), where bell holes with similar average width and depth values were reported (35 cm width, 62 cm height). Bell holes have values that ranged into larger dimensions, as much as Bell holes can be found all over the world in both continental and island karst environments. There have been two major dissolutional categories recognized that encompass 2006; Lundberg and McFarlane, 2009), condensation corrosion The bat activity model has a long history, which is reviewed in Lundberg and McFarlane (2009). Miller (1990; 2006) is a recent proponent of this idea, which was based in part on his observation that bell holes were not found in caves beyond obstacles such as breakdown chokes or sumps that precluded bat entry. He felt that bat feces and urine, as well as claw and wing activity, were the excavation mechanism. Lundberg and McFarlane (2009) examined the bat question more closely. They disagreed with Miller (2006) about the mechanism, such activity would not focus dissolution at the bell hole apex, as is necessary for the vertical cylindrical shape, and noted the lack of claw marks on bell hole walls. However, Lundberg and McFarlane (2009) were able to show how small bat colonies occupying shallow roof concavities could create a unique environment from their own body heat, and their exhalations of CO2 and H2O. They measured rock temperatures in bell holes with bat colonies in them, noting that occupied bell holes were 1.1 Co CO2. The CO2 plus H2O mix would create carbonic acid, H2CO3, which would drive CaCO3 dissolution. They estimated the rock dissolution rate to be as much as 0.005 cm3/day. The bell hole temperature elevation was argued to keep the dissolutional focus at the bell hole apex, as the warm moist bat exhalations would rise, to produce the vertical cylindrical shape. This model was made for bell holes with an average height of 62 cm and an average width of 35 cm (Lundberg and McFarlane, 2009, their Table 1), which limited the volume of rock removed to approximately 0.07 m3. The Lundberg and McFarlane (2009) model is a biologic condensation corrosion model, where condensing water, carrying dissolved CO2, attacks the limestone wall rock. The bats help create the environment where their metabolism provides the heat and gases necessary to create a dissolutionally-aggressive water condensate directed upward. corrosion, but without biological mediation as provided for by bats in the Lundberg and McFarlane (2009) model. Using the necessary water condensation. As a result, bell hole size the bell holes closer to entrance sites being larger. A problem with condensation corrosion models is that bell holes lack the CaCO3, but then evaporates, leaving a powdery CaCO3 crust behind that is easily dislodged. Lundberg and McFarlane (2009) have the advantage of working in cave environments that have left the phreatic realm of cave origin, and are now in the vadose realm of cave senescence. Such a setting provides a somewhat open-ended time window for the development of bell holes. The phreatic models assume some sort of circulation in at selected sites in the ceiling. One argument against the vadose models is that they fail to explain bell pits, the shallow pits appear unlikely to be due to dripping vadose water because bell holes show no evidence of dripping water during their formation. Many caves have thick guano deposits on prior to guano deposition, or subsequently under the guano from Bahamian caves, exposing the original dissolutional conditions, where the convective cell is bounded by both the (Mylroie and Mylroie, 2009). The convective model appears margin caves, as well as in traditional turbulent conduits under

PAGE 4

Birmingham Andrew N. et al The Bahamian SituationSubaerial Bahamian caves are abundant. The largest of margin caves are developed in eolian calcarenites less than these dry caves could have formed only when a glacioeustatic sea level higher than todays had elevated the fresh-water 5e) has been high enough (+6 m) and recent enough (131 to 119 ka) to have formed the caves as they are observed in the present, correcting for isostatic subsidence (Carew and Mylroie, 1995). The youthfulness of both the caves, and the limestones containing them, place important constraints on any bell hole model. which means they can form quickly (less than 131,000 years, the maximum age of the caves), in small hills on small islands, mechanism requires that the bell holes are syngenetic with by dissolution within an elevated fresh-water lens during the 12,000-year time window of the last interglacial sea-level The vadose alternative would allow for a 119 ka time window the present, when vadose conditions existed within the caves. The short time windows for both proposed mechanisms of bell hole formation require that a very large amount of bedrock is removed in a very short period of time. The correct mechanism for formation would also remove bedrock vertically more rapidly than horizontally given the average measurements for Wilford (1966), bell holes are unaffected by bedding and jointing attitudes and carbonate facies leading to the conclusion that bell holes must form uniformly upward from a horizontal plane producing the characteristic circular, cylindrical shape. The dissolutional mechanism must be powerful enough to ignore in the distal margin of the fresh-water lens is both rapid and powerful (Mylroie and Mylroie, 2007). The development of bell holes in very young Bahamian limestones has impact on another idea concerning bell holes. Ford and Williams (2007) argued that the development of bell holes is accelerated by the removal of calcite cement between carbonate grains, as the grains are less soluble. Cement removal would allow the grains to fall out of the bell hole, decreasing the volume of rock necessary to remove by dissolution. However, the grains are still aragonite, which means the grains are more soluble than the calcite cement, so the grainfall idea cannot be supported for Bahamian bell holes. been cut by dissolution so as to be smooth with the bell hole surface, indicating dissolution after vadose conditions had allowed calcite precipitation. This evidence was used to support the vadose model, as the phreatic origin of the cave short fall of a few meters for perhaps 1 ka before sea level rose back to its highstand position (Carew and Mylroie, 1999). The cave could have formed and the bell holes developed phreatically, then the vadose speleothems were deposited would have been subject to phreatic conditions during the second highstand pulse, and partially dissolved. Therefore the speleothem material, unless dated as younger than 119 ka, cannot be used as proof of a vadose origin for bell holes in the Bahamas. Miller (1990, 2006) reported bell holes on the underside of breakdown blocks, and argued such features attest to a bat (vadose) origin. This interpretation assumes that breakdown assumes that breakdown occurring in a vadose environment phreatic cycle situation. caves which developed rapidly in relatively youthful carbonate rock, offer constraints on bell hole formation that do not exist in other tropical and subtropical settings. The purpose of this paper is to report simple bell hole observations from Crooked regarding bell hole modeling.Observations and interpretations Of particular note from these visits were the large number of bell holes found in the caves, their large size, and the interaction of some of those bell holes with the land surface above. The observations fall into four important categories.Bell Hole Complexity

PAGE 5

Birmingham Andrew N. et al centimeters in height and width to almost a meter across and 4 meters high as seen in Fig. 5, or larger. The broad range in sizes observed, and the connection of some by lateral dissolution holes, argues against the bat origin theory. Additionally, some bell holes are smaller than the voids bats are known to roost in today, such as the smaller openings in Figure 5. Fig. 6 (as well as Fig. 1 from Rum Cay, Bahamas and Fig. gypsum requires that the dissolutional process has stopped, and that the gypsum subsequently formed by evaporation of sulfate-containing water entering from above the cave as indicate that the 34S in the gypsum crusts is identical to marine sulfur (Bottrell et al., 1993), and has not been biomediated. much of the gypsum, show development as a result of interaction with either sea water ions or bat guano, or both source of the sulfur based on the Bottrell et al. (1993) stable isotope study. The gypsum crusts appear to be evaporative deposits formed by wicking of vadose water contaminated by sulfate-containing sea spray out onto the bell hole walls. Because the bell holes extend upwards into the cave ceiling, bell holes, but also found on cave ceilings and walls (Fig. 7), indicating the crusts form independently of bat activity.Bell Hole Size be 30 cm in diameter and approximately 1 meter in height, in (2009), as noted earlier, report bell hole sizes in Runaway bells holes, up to 5.7 m high, but most below 3 m in height, with basal diameters ranging from 0.25 m to 1.3 m. On Crooked

PAGE 6

Birmingham Andrew N. et al bat model, where dissolution is continually focussed at the bell hole apex.

PAGE 7

Birmingham Andrew N. et al surface (Fig. 2). Both the phreatic and the vadose bat models require that bell holes form with a sealed apex. Therefore the breaching of the bell holes by surface denudation must postdate the formation of the bell holes. The denudation necessary to intercept these bell holes, and reduce their domed apex to a simple exposed cylinder, reduces the available time to make 3, the intersection has been from the side, further indicating the independence of bell hole formation from outside surface effects. Within a cave, missing side walls of bell holes, as model offers no explanation for the dissolution of a partial side of a bell hole as the dissolutional focus in that model is solely on the apex. While it may be possible for a bell hole to open to the surface because of actions within the bell hole, no model offered so far can convert a bell hole from a small apex opening into a circular right cylinder without an apex cone. As all gradations of slight apex cone opening to fully denuded the opening of the bell hole to the surface is a result of surface denudation.Discussion observation without sophisticated measurements can assist in constraining model development and execution. The general setting of the Bahamas provides tight time constraints on any

PAGE 8

77 Birmingham Andrew N. et al 119,000 years or less. The time window can be further constrained for the two leading vadose models of bat activity and condensation the caves, growing upward, then they must predate any rates in eogenetic island limestones are about 70 mm/ka such that only a right cylinder remained, would require at least 50 cm of rock loss at the top of the bell hole. As soon as the apex is penetrated, the internal bell hole conditions for the vadose models disappear. At least a minimum of 7,000 years (500 mm/70 mm/ka) is required to remove a bell hole apex (once a bell hole apex is gone, there is no way to determine how much of the lower cylinder has also been lost). That 7,000 years must be taken away from the bell hole time window. This calculation assumes the bell hole apex was just below (and tangent to) the surface when bell hole development ceased. removed in order for the bell hole apex to be intersected by surface denudation (see Fig. 3B). Therefore the 7,000 years is a minimum, and unlikely, estimate. A meter of extra rock would take an additional 14,000 years. A more reasonable estimate would be ~20,000 years (1 m surface denudation, m of bell hole intersection) to create a bell hole exposed to the surface as a simple cylinder with no apex, as in Fig. 2. which forms as a sealed chamber (Mylroie and Carew, 1990), must be broken into by surface denudation to create an entrance for either macro-organism entry (bat model), or air commonly form from hill slope retreat on the side of the cave, and are therefore independent of the previous vertical denudation argument. Based on existing Bahamian caves, such denudation must be several meters to penetrate the cave wall, and given the denudation rate for these limestones, affects both the initiation (lateral intersection to allow bat and air entry) and the termination (vertical denudation to cause bell hole intersection) of the vadose model time window. Therefore the vadose models do not have close to 120,000 years to act to make bell holes, but perhaps only 70,000 years, (120,000 minus the 30,000 years of lateral denudation to create a cave entrance to start bell hole formation, minus another 20,000m years for vertical intersection to stop bell hole development) or 60% of that time window. Lundberg and McFarlane (2009) estimate that their bat CO2). They estimate this time drops to ~23,000 years if 30% porosity is assumed for the host rock, very similar for Bahamian eolian calcarenite porosity values. For bell holes over 5 m example, a 5 m high, 1 m diameter bell hole requires removal 3 of rock material, or ten times more than for the bell holes described by Lundberg and McFarlane (2009); given the dissolution rates calculated, the age of the bell holes would exceed the age of cave formation for a cave developed on the age of the enclosing rock). Taller bell holes also create a problem for disposal of dissolved CaCO3 dissolution is taking place at the bell hole apex at the top of CaCO3 any CO2 degassing to avoid re-precipitating the CaCO3 on the bell hole wall. When bell pits are considered, the solution must aggressivity to make the bell pit. The walls of bell holes do not show the vertical grooving that such aggressive solutions would necessarily create. The bat model does not successfully scale up to the observed larger bell holes; given the bat model rate proposed by Lundberg and McFarlane (2009), there isnt time in 70,000 years to make a 4 to 5 m high bell hole. The vadose models fail to explain the bell pits associated with bell holes. Bell hole like structures are reported by which eliminates all vadose models. The vadose models are also inadequate to explain why bell holes form regardless of carbonate bedrock structure and facies. The vadose models are admittedly slow compared to phreatic models, and bell hole dissolution would be expected to be sensitive to subtle differences of rock structure and composition. Finally, for bell holes to form by either vadose method, there has to be an initiation point. Neither bat metabolic production nor condensation corrosion would create a domed structure from a to allow unique micro-environments to form and persist. Lundberg and McFarlane (2009) argue that most cave roofs have a cuspate nature from initial phreatic speleogenesis, so that incipient hollows already exist for bat groups to occupy. Once a cylindrical hole has been formed, the proposed models have mechanisms that may work. But there must be a the activation energy required for certain chemical reactions to occur). The presence of bats in bell holes with gypsum crusts on the walls and apex argues against bats being a major contributor to bell hole formation; their exhalation, and its subsequent H2O condensation on the bell hole surfaces, should strip out the gypsum crust by the same dissolution argued to create the bell 3, and is dissolves by simple ionic dissolution, requiring no special acidic pH effects as required for CaCO3 dissolution. The bell hole surfaces cannot be precipitating highly soluble gypsum while at the same time dissolving less soluble CaCO3.

PAGE 9

Birmingham Andrew N. et al The phreatic model also operates under severe time year time window to develop by phreatic dissolution, as their an elevated sea level. Mylroie and Mylroie (2007) demonstrated that the dissolution in the distal margin of the fresh-water lens is extremely potent. These caves have chambers that cut across variations in rock type (eolian versus subtidal in the For the Bahamian examples, this dissolution was complete in a short period of time (12,000 years) to form extremely large chambers (tens of thousands of meters in volume). occur. Mylroie and Mylroie (2009) have demonstrated that density variations can occur in such a hydraulic setting by cave setting can be considered somewhat isothermal, and so thermal convection is not likely. However, the distal margin of water below the lens, the fresh-water lens itself, and descending vadose water reaching the lens from the land surface above. with the phreatic fresh water of the lens resulting in a mixing environment. Both waters could be saturated with respect to CaCO3, but if they had done so at different initial conditions, the mixing would create an aggressive solution that would drive additional CaCO3 dissolution. This dissolution would be localized wherever vadose water was slightly concentrated to create cupolas in the cave ceiling. Such mixed water would contain more solute than either parent water, and so would be denser than either water. That solute-laden mixture would with marine water, and further renewed aggressivity, would be as that cupola grew upward to form the bell hole cylinder and apex, it would trap the density convection cell. The quick saturation of the water at the top of the bell hole would prevent further dissolution along the bell hole sides as the solute-heavy water descended, maintaining the uniform cylindrical nature would create the bell pit, as further mixing dissolution would be available there as a result of mixing sea water with fresh water. The aspect ratio of the bell hole and bell pits as seen today would be the aspect ratio of the convection cell when it was last functioning. Alternatively, Mylroie and Mylroie (2009) have proposed gas production as a result of oxidation of organics, to create a buoyant water package. Water pressure in a fresh-water lens margin is minimal (less than 2 atm), so gas evolution is possible. The base of the fresh-water lens is a density interface known to collect organic material and create dissolutional potential by CO2 production, and under anoxic conditions, H2S (Bottrell et al., 1991). Rising gases would create a convection cell that would localize at the cave ceiling in any indentation already present. The ceiling rock would be dissolutionally attacked by the CO2 (or H2S) rich water to create a bell hole, which would lock in the convection cell position. Bell pits are also explicable The phreatic models require bell hole development when sea level was high, and the fresh-water lens had invaded the the ~120 ka since that time, the land surface above the caves as well as the climate have changed. The absence of drip water entering bell holes today does not indicate what vadose cave. The role of bacteria in the production of bell holes has not been addressed by any studies. Based on 34S studies, bacteria have not participated in the gypsum crust production. However, bacteria might have at some stage been important for original bell hole dissolution, for either the vadose or the phreatic model, or even a combination of them.Conclusions The bat model is compelling because one commonly sees bell holes with bats in them; however, to assume cause and effect is incorrect. Bats clearly like bell holes, but their presence is merely opportunistic. Condensation corrosion is compelling, because the evaporation, condensation and dripping of such water from a cave roof is a simple concept; however, it requires an energy sink to absorb the 539 cal/g tropical environment. The vadose model of bell hole dissolution requires that those dissolutional conditions cease so that gypsum crusts can subsequently line the bell holes. Many bell holes today contain both a gypsum crust and a bat population, which implies the bats have utilized an existing hole. The vadose models cannot explain bell pits. The phreatic convection cell model presented here is purely speculative, but it has none of the many separate and unrelated problems that offered to the reader as a testable hypothesis for future work. the impetus to examine the bell hole problem once again.Acknowledgements

PAGE 10

Birmingham Andrew N. et al References Birmingham, A. N., Lace, M. J., Mylroie, J. R., and Mylroie, J. Program, National Speleological Society Annual Convention 115. of Bahamian blue holes. Applied Geochemistry v. 6, 99-103. Bottrell, S. H., Carew, J. L., and Mylroie, J. E. 1993. Bacterial Proceedings of the 6th Symposium on the Geology of the Bahamas Field Station, 17-21. Carew, J. L., and Mylroie, J. E. 1995. Quaternary tectonic stability of Quaternary Science Reviews 14, 144-153. Carew, J. l., and Mylroie, J. E. 1999. A review of the last interglacial Mylroie, J. E., eds., Proceedings of the Ninth Symposium on the Geology of the Bahamas and other carbonate regions, Bahamian Field Station, San Salvador Island Bahamas, 14-21. MSc thesis, Mississippi State University, 106 p. Hydrology. Chapman and Hall, New York, 601 p. 293 p. morphological perspective. Special Paper No 1, National Cave and Karst Research Institute, Carlsbad, NM, 106 p. Lauritzen, S.-E. and Lundberg, J. 2000. Solutional and erosional Speleogenesis: Evolution of Karst Aquifers. The microclimatic impact of bat roosting, using a case study from Runaway Bay Caves, Jamaica. Geomorphology tropical caves. Geo2 Occasional Paper 4, v. 17, no. 2-3, 76. features of Caribbean caves. Geological Society of America Abstracts with Programs Mylroie, J. E. and Carew, J. L. 1990. The Flank Margin Model for Earth Surface Processes and Landforms 15, 413-424. Journal of Cave and Karst Studies 69, 59-75. Mylroie, J. E. and Mylroie, J. R. 2009. Unique dissolutional Proceedings of the 15th International Congress of Speleology National Speleological Society, Huntsville, Alabama 2, 916-922. Carbonates and Evaporites Minerals of San Study Series, 70 p. bell hole development on Cayman Brac. Cave and Karst Science 25, no. 3, 19-30. Wilford, C. E. 1966. Bell Holes in Sarawak Caves. Bulletin of the National Speleological Society