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. 5 (2004)

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:
K26-04220 ( USFLDC DOI )
k26.4220 ( USFLDC Handle )
8996 ( 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

Speleogenesis and Evolution of Karst Aquifers The Virtual Scientific Journal www.speleogenesis.info Towards defining, delimiting and classifying epikarst: Its origin, processes and variants of geomorphic evolution Alexander Klimchouk Institute of Geological Sciences, Nati onal Academy of Science of Ukraine P.O.Box 136, Kiev-30, 01030 Ukraine. E-mail: klim@speleogenesis.info Re-published (modified) from: Jones, W.K., Culver, D.C. and Herman, J. (Eds.). 2004. Epikarst. Proc. of the symposium held October 1 through 4, 2003 Sheperdstown, West Virginia, USA. Karst Water Institute special publication 9, 23-35. Abstract Epikarst is the uppermost weathered zone of carbonate ro cks with substantially enha nced and more homogeneously distributed porosity and permeability, as compared to the bulk rock mass below; a regulative subsystem that functions to store, split into several components a nd temporally distribute authogenic infiltration recharge to the vadose zone. Permeability organization in the epikarst dynamically develops to facilitate convergence of infiltrating water towards deeply penetrating collector structures such as prominent fissures that drain the epikarstic zone. This is manifested by epikarstic morphogenesis that tends to transform dispersed appearance of surface karst landforms into focused appearance adapted to the permeability structure at the base of epikarst. Epikarst is the result of combined action of several agencies including stress release, weathe ring and dissolution. It is a dynamic system which main characteristics are time-variant, ch anging in a regular way during the epikarst evolution. This paper examines the main characteristics of epikar st in the light of its origin and evolution. Keywords: Epikarst, Origin of epikarst, Karst Evol ution, Karst hydrology, Karst morphogenesis Introduction Appreciation of epikarst as a sub-system (structure) that bears specific functions in a karst system (Fig. 1) emerged during 1970s from various kinds of evidence independently obtained within different disciplines. Cave biologists found specific aquatic fauna in drips and seeps from the cave ceilings, suggesting the existence of saturated zones between the surface and caves. Karst hydrogeologists realized that the water budget of karst aquifers and spring hydrograph interpretation suggests the existence of an important storage at the top of vadose zone (Mangin, 1973, 1975). Moreover, such storage was also evidenced by hydrochemical and isotopic studies, which demonstrated strong reduction of the input signal variations, hence an efficient mixing of the infiltrated water with pre-storm water (Bakalowicz et al., 1974). Early finite element modeling of a karst system by Kiraly and Morel (1976) demonstrated the need to impose a high proportion of concentrated infiltration to generate the typical "karstic" storm hydrographs. It was supposed that concentration of originally diffuse infiltration would occur at shallow depth in a thin nearsurface high conductivity layer. Speleological investigations in arid mountains of the Central Asia demonstrated a substantial shaft flow in the vadose zone after long periods without any precipitation and revealed the existence of numerous "hidden" shafts beneath karren fields (Klimchouk et al., 1979, 1981; Klimchouk, 1987, 1989). These works suggested that the near-surface weathered zone of exposed carbonates functions as a "recharge" zone for a karst system, and that concentration of originally diffuse infiltration within it accounts for the formation of hidden shafts at the base of such zone. Mangin (1973, 1975) introduced the term "epikarst" to denote this zone and a perched aquifer within it, at the top of vadose zone.

PAGE 2

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.2 Fig.1. Diagram illustrating principal structural and hydrologic features of epikarst, and its relationship with the vadose zone. SF = shaft flow, VF = vadose flow, VS = vadose seepage. However, it was the paper by Williams (1983) that brought the topic to international appreciation. During 20 years passed since that publication, hundreds of works were published, which highlighted enormous importance of the epikarst to karst hydrology and morphogenesis. In spite of the fact that many important aspects of the role of the epikarst are now widely acknowledged and well accepted, the scope of the concept is still debatable and numerous definitions vary considerably. This is partly the result of enormous spatial variability of the epikarst in local, regional and global scales, but also of its still poorly recognized evolutionary variability. In attempting to refine the concept and definition of epikarst we should demarcate a respective natural system by addressing its origin and inherent characteristics, and reveal its place and functions in the context of a more general system (karst). It is also important to examine its characteristics from the evolutionary perspective. In particular, this paper attempts to elucidate how the epikarst evolve in the course of karst evolution. Essential characteristics of epikarst: criteria used to delineate the epikarst concept The list below summarizes criteria used to delineate the epikarst concept, derived from twelve individual definitions and numerous relevant discussions found in the literature. They fall into the following groups: !" Structural features of the epikarstic zone !" Location of the epikarst !" Origin of the epikarst !" Hydrologic functions and roles in the overall karst system !" Morphogenetic role of the epikarst Respective characteristics of the epikarst are briefly overviewed below. Structural features of the epikarstic zone Structural characteristics are vital in delimiting the epikarst. Most of works stress on high and relatively homogeneously distributed fissure and solution porosity in the uppermost zone of exposed carbonates, with which epikarst is associated. This characteristic is comparative and scale-dependent; it holds true with reference to the underlying bulk rock mass which consist of low permeable blocks separated by much less densely packed prominent vertical fissures. Fissure networks in the shallow subsurface are commonly closely spaced (decimeters to a few meters) and continuous. The spacing of fissures increases with depth obeying a hyperbolic low (Chernyshev, 1983). The typical spacing of prominent vertical fractures with a significant penetration depth is estimated to be on the order of 30–50 m. Estimates of overall epikarst porosity range between 1% (Smart and Friederich, 1986) and 10% (2-10% in Gouisset, 1981; 5-10% in Williams, 1985). They are one to four orders of magnitude more than fracture and solutional

PAGE 3

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.3 porosity in the bulk rock below (0.005 0.5%; Worthington et al., 2000). The contrast in porosity and permeability in vertical direction between the epikarstic zone and the bulk rock mass below is of primary importance as it accounts for major hydrologic functions of epikarst. The lower boundary of the epikarstic zone is commonly highly irregular. It depends on relief, lit hostratigraphy and geological structure. Fissure spacing thickening is more pronounced and penetrate to greater depths along prominent discontinuities. The boundary can be abrupt and very contrasting if coincides with sub-horizontal lithostratigraphic boundaries, or it can be gentle in homogenous sequences. The thickness of the epikarstic zone varies considerably but it is most commonly estimated to be between few meters to 10-15 m. Location of epikarst Most of definitions emphasize that epikarst is the uppermost zone of karstified rocks. Some of them refer to the direct exposure of the rock to the surface but others point out that epikarstic zone extends from the base of the soil. Strictly speaking, there is an apparent contradiction between these two criteria. There are many examples of epikarst without any soil, but in fact the epikarst/soil relationship is a special topic addressed in one of the following sections. These relationships should be viewed as genetic and stage-depe ndent. More generally, this issue is relevant to the problem of origin of epikarst. Origin of epikarst Almost all definitions state that epikarst forms due to enhanced solution in the uppermost zone of the bedrock. This certainly is partly true, but the matter is not so straightforward. It appears to be more adequate to consider the combined effect of several agencies in the formation of epikarst, including stress release, weathering and dissolution. Origin of epikarst in the context of general karst evolution Before discussing the origin of epikarst any further, it is necessary to look at various general evolutionary scenarios of the karst development in order to infer about starting points for the formation of epikarst. The evolutionary classification of karst (Klimchouk and Ford, 2000a; see Fig. 3.1-3, p.50), which views types of karst as variants and stages of general geological/hydrogeological evolution, provides a useful framework. Epikarst can commence in young, diagenetically immature carbonates that have never been buried by other rocks. This is one of the variants of open karst, which evolves from syngenetic karst. Epikarst development and characteristics have pronounced specifics on eogenetic rocks not considered in this paper (for details see Mylroie and Carew, 1995, 2000). The second variant of open karst can be envisaged when a soluble formation remained untouched by karstification during burial/reemergence cycle, and karst commenced only after complete re-exposure (the "pure" line of open karst). In fact, this scenario does not seem quite realistic, as differential linear allogenic entrenchment and point breaching nearly always precedes substantial exposure of carbonate bedrock. This means that at least some inputoutput connections woul d establish and develop before complete exposure, which conforms to the subjacent or entrenched types of karst. Most commonly, karst development commences beneath the cover at some stage en route to complete re-exposure. This is the evolutionary line of intrastratal karsts, which includes deep-seated, s ubjacent and entrenched karst types. All of them cannot have epikarst although some hydrologic functions performed by an insoluble but permeable cover can be somewhat similar to that of epikarst. Complete removal of the cover will bring the former intrastratal karst into the category of denuded karst, which belongs to the class of exposed karst types and is dominated by authogenic recharge. Denuded karst is the most common situation to be considered as the starting point of epikarst evolution (Fig. 2) Origin of epikarst in denuded karst type Two aspects deserve particular attention in case of epikarst evolving in denuded karst type: the way of caprock retreat and the presence of the vadose zone. In denuded karst type epikarst evolves in areas where bedrocks get exposed after retreating caprock. Hence, unloading effects

PAGE 4

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.4 apply. The rate and "completeness" of caprock retreat depends on the nature of denudation agency, the uplift rate and relief. One can envision two starting situations for the epikarst development: 1. Quick and complete removal of caprock; barren carbonate surface is directly exposed to weathering. Soil is originally absent, but it forms in paragenesis with the formation of epikarst as solution residual material. 2. Slow incomplete removal of the cover; regolith is present, providing the base for soil. The presence of regolith cushions the bedrock from physical weathering but may enhance considerably chemical weathering (dissolution). The epikarst formation in these variants can differ substantially because of different effects of unloading and weathering, and different infiltration conditions. Another aspect of utmost importance is that in denuded karst epikarst evolves above a ready made vadose zone, a free-draining percolation zone with predominant vertical percolation, where prominent conduits are inherited from an earlier stage of karstification (Fig.2). Separation of the epikarst from the phreatic zone by the established vadose zone is one of the primary points of the epikarst concept. Otherwise we would deal with a simple hydrographic profile with incipient vadose zone and a single phreatic aquifer body. Role of stress release and weathering The relationship between the dissolution process on one hand, and stress release and weathering processes on the other hand is the key issue of the epikarst origin. Most researchers point to the primary role of dissolution in the formation of epikarst when explaining enhanced porosity in this zone. Although dissolution certainly contributes considerably to porosity enhancement in the epikarstic zone, it relies on availability of initial pathways for percolation and the surface area for reaction. Structural prerequisites for the formation of epikarst evolve largely due to non-solutional processes such as stress release and physical weathering. These processes form the incipient epikarstic zone, a ready-made structure for diffuse infiltration and disperse solution in the uppermost zone of the bedrock. The role of Fig. 2. Denuded karst that evolves from the intrastratal karst is the most common starting situation for epikarst development. solution increases on the farther stages of the epikarst development, and it is particularly important in bringing a specific organization to the epikarst structure. That dissolution is not the primarily factor generating the porosity and permeability contrast between the top layer of the exposed rock and its bulk mass at depth is well illustrated by the fact that some kind of stress release and weathering profiles develop on most types of rocks, not only on soluble rocks. It is well appreciated in the geological engineering literature that stress release and weathering account for considerable near-surface porosity changes that occur when the rock is exposed in natural outcrops after burial. These effects include (Chernyshev, 1983; Klimchouk and Ford, 2000b): !" extension and opening of existing joints and formation of new joints; !" accentuation and opening of bedding planes and "micro-fissures", splitting of beds; !" enhancement of fissure frequency and connectivity of fissure networks. On the other hand, weathering is also responsible for in situ deep chemical and

PAGE 5

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.5 mineralogical alteration of parent rocks, and for generation of fines that may choke fissure network and hence decrease permeability. These effects depend primarily on: !" composition and structure of parent rocks !" climate !" unloading rate (which in turn depends on the balance between uplift rate and denudation rate) !" topography Karst vs. weathering; Epikarst vs. weathering profile Here we come to the important general problem concerning the origin of epikarst, which can be subdivided into two aspects: !" The relationship between the weathering concept and the karst concept. !" The distinction between "non-karstic" weathering profiles (weathering mantles) and epikarst. One potential problem arises from the fact that the concept of weathering in its broadest form tends to embrace karstification, as well as many other individual exogenic processes. It is more reasonable to keep with the narrower notion of weathering, which stresses on the near-surface processes of in situ physical and chemical breakdown and alteration of parent rocks to products that are more in equilibrium with newly imposed physico-chemical conditions (Ollier, 1969). In contrast, under karstification chemical alteration of parent rocks is minor, as is the amount of insoluble residue, and most of material is removed in the dissolved form through internal drainage system of conduits. The non-karstic weathering profiles tend to develop into weathering mantles (or weathering crusts), which are mantles of chemically and mineralogically deeply altered weathering products containing much inert components. The net result is decreasing porosity and permeability of a weathering mantle in most cases. In contrast, the development of weathering profiles on carbonates is characterized by increasing dominance of dissolutional removal of material and increasing capacity of the uppermost zone to transmit even the minor amount of residue that could form on carbonates. Such capacity evolves with the increasing organization of permeability in this zone towards major vertical drains in the underlying vadose zone. The net result is increasing porosity and permeability in the uppermost zone and the formation of epikarst. The formation of a weathering mantle on insoluble rocks and the formation of the epikarstic zone on carbonates should be regarded as different types of hypergenesis, with largely opposite effects to the structure and hydrologic role of the uppermost zone of bedrocks exposed to weathering. Factors in the formation of epikarst The literature on weathering and karst provides an extensive discussion of major factors that exert important guidance on nearsurface changes of structure and porosity of the rock. Although the factors in the following list are commonly recognized to guide dissolutional processes in karst, their effects on the epikarst should be viewed from the standpoint of the combined action of unloading, weathering and dissolution upon exposed rocks: !"Parent rock composition. Susceptibility to mechanical breakdown and dissolution. !" Parent rock structure and texture. Susceptibility to mechanical breakdown; effects on infiltration and solution processes, etc. !" Tectonic structure, and lithostratigraphy in the upper section. General arrangement of structural features, fissure permeability structure in the bulk rock, lateral and vertical variability of rock units within the near-surface zone. !" Local topography. Effects on drainage, shatter movement, stress-release fracturing, etc. !" Presence and thickness of soil. Biogenic effects including CO2 and organic acids production; infiltration and hydrochemistry controls. !" Climate. Influences in a major way the character of weathering (physical/chemical weathering relationship) that will take place in any region, as well as solution rate. hemical weathering is at a maximum in a warm moist tropical climate, while in polar and arid regions physical weathering predominates. Subsidiary factors are precipitation, temperature, vegetation and biological activity. !" Microclimate. Although the macroclimate determines the main character of the weathering in any given region, the

PAGE 6

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.6 microclimate influences in many ways the soil and epikarst pattern in the local scale. !" Tectonic regime. Rate of uplift or subsidence with its effects on denudation and stress release. !" Nature of denudation agency that exposes the bedrock to the surface (high-rate/low-rate denudation). !" History of development (single-phase or multi-phase). !" Time Most of these factors are variable in space and/or in time and are in the complex interplay. There are many feedback loops in their relationships. Fig. 3 is an attempt to visualize their influence and relationships in the formation of epikarst. Fracturing and dissolution are the main processes occurring in the uppermost zone of exposed bedrock to form epikarst, while in situ chemical transformations being of minor importance on carbonates. Fracturing is generated by weathering and stress release, and dissolution is driven by water circulation. The guiding factors can be subdivided into two sets, one being endogenic and the other being exogenic in nature. Rocks in the near-surface zone vary in composition, texture and structure and therefore in their susceptibility to alteration by the epikarst-forming processes. On the other hand, the factors that guide these processes vary with position on the surface and with local conditions, so that these processes at any given locality vary in composition and intensity, and therefore in their capability to generate epikarst. These complex relationships, as well as evolutionary aspects, account for great variability of epikarst characteristics on local, regional and global scales. Hydrologic functions and roles of epikarst in the overall karst system The structural and permeability distinctions between the uppermost zone and the bulk rock mass below account for specific hydrologic functions which this zone performs in the overall karst system. Epikarst hydrology received much attention during last 30 years and is generally well understood. Fig. 1 illustrates the main features. Most of them arise from the fact that infiltration to the epikarstic zone is easier than drainage out of it. The relatively homogeneous hydraulic conductivity field at the top of the epikarst, which allows for diffuse infiltration, becomes increasingly heterogeneous towards its "bottom". Hydraulic conductivity in the epikarst is believed to be two to three orders of magnitude greater that in the underlying vadose zone but heterogeneity in its distribution is even more important. Drainage of the epikarst through prominent deeply penetrating fissures provides for flow concentration at its base. According to Kiraly (2002; see also Kiraly and Morel, 1976), more than 50% of the infiltration arrives to the vadose zone from the epikarst in "concentrated" form, directly into the high conductivity channels. Epikarst is recognized as an important storage subsystem. Some studies (e.g. Perrin et al., 2003) suggest that storage in epikarst can be more significant than storage in the phreatic zone. Decrease in permeability with depth causes considerable lateral component in the flow within epikarst, which converges towards major deeply penetrating fissures. In conveying water down to the vadose zone, the epikarst splits infiltration into several components: conduit or shaft flow, vadose flow and vadose seepage (Fig.1). Epikarst distributes recharge to the vadose zone as quick flow and slow flow. Both hydraulic and transport responses of epikarst to rainfall events depend on its maturity and links with the vadose zone, as well as on rainfall intensity and the pre-event precipitation history. Generally, epikarst accounts for retardation of through flow and considerable mixing, although it provides for quick hydraulic response at shaft flow (and hence at springs) in most cases. The feedback of the flow field on the hydraulic conductivity field, a primary distinct feature of a karst system in general, produces it's effect in epikarst by developing an important organization in the epikarst structure. This organization evolves through dissolution in the course of the epikarst evolution to facilitate convergence of infiltrating water towards collector structures intercepting at the base of the zone. A well-developed epikarst differs from an initial stress release/weathering profile by a degree of such organization. Epikarst hydrologic mechanisms and organization manifests itself through karst morphogenesis.

PAGE 7

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.7 Fig.3. Principal factors of the epikarst formation, and their relationships. The role of epikarst in karst morphogenesis Specific hydrological processes operating in the epikarst, coupled with solution effects, account for important morphogenetic effects. The role of epikarst in karst morphogenesis was specifically addressed in Williams (1983, 1985), Ford and Williams (1989) and Klimchouk (1987, 1989, 1995, 2000). There were two conceptual models proposed for epikarstic morphogenesis. The Williams' model emphasizes the focused dissolution within drawdown cones in the epikarstic water table to generate solution dolines by gradual lowering of the surface above such foci. The Klimchouk model, while appreciating focused dissolution within a 3D volume of drawdown cones, stresses on the enlarging of vertical leakage paths to form "hi dden" shafts at the base of epikarst. This model implies collapsing of the partly discontinued "plug" above a growing shaft, and subsequent rapid mass wasting and enlargement at the shaft mouth prepared by focused dissolution within the epikarst drawdown cone, to form a doline (see Klimchouk, 2000 and Fig.4). Both models rely on recognition of specific hydrologic processes in the epikarst, acknowledge specific morphogenetic mechanisms in this zone and envisage doline-dominated landscape as the result of the epikarst evolution.

PAGE 8

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.8 Fig.4. A "plug" of discontinued boulders (karren field) above a 3 m wide shaft, Kyrktau plateau, Tien-Shan (Uzbekistan). Photo by A.Klimchouk. Fig.5 (right). The photograph illustrating the soil loss in the mature epikarst, the Kyrktau plateau, Tien-Shan (Uzbekistan). The soil loss is most intense above karren fields (encircled with dashed lines), which are projections to the surface of epikarst "catchment" structures of sh afts at the base of the epikarst, as shown in the inset. Photo by A.Klimchouk. There is a view that dolines do not represent epikarst because they are landforms penetrating into it (Bakalowicz, 2004). In contrast, I do consider dolines to be a part of epikarst because they are the morphogenetic result of the epikarstic hydrologic processes, manifested at a certain stage of the epikarst evolution. However, it should be acknowledged that some types of dolines have nothing to do with epikarst morphogenesis (point recharge dolines, true collapse sinkholes, etc.) and hence they can be considered as "holes" in the epikarst. More generally, epikarstic morphogenetic mechanisms act to adjust authogenic surface karstic morphology to the permeability structure at depth, at the level in the base of the epikarstic zone. Hence, they act to transform predominately diffuse appearance of karst landforms in relatively young karst landscapes (karren-dominated) into predominantly focused appearance in mature karst landscapes (dolinedominated). This adaptation function of epikarst continues in mature karsts as denudation continuously lowers the surface and brings still deeper sections of a massif under the action of these mechanisms. In this sense, the epikarst is a kind of a "reaction zone" for karst morphogenesis, which works until complete "consumption" of an exposed carbonate massif. This suggests that the general problem of karst morphogenesis and its variability (with respect to the type of exposed authogenic karst) should be approached from the standpoint of epikarstic morphogenesis. In the view of the discussion in previous sections about the origin of epikarst and factors in its formation, such an approach seems to be broader and more adequate than traditional consideration of superficial solution processes alone. The soil-epikarst relationship Whether the epikarst develops with or without the soil (regolith) cover depends on a coverbeds composition and a mode of its removal (the nature of a denudation agency which exposes the bedrock). However, the

PAGE 9

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.9 presence of the soil is also a function of the development of the epikarst itself. Where the complete exposure is a starting point for the epikarst development, the soil forms during the incipient and young stages of the epikarst development which are characterized by poor links of this zone with the drainage structures at the vadose zone below. Where the soil is present, it enhances solutional enlargement of fissures in the epikarst. With the development of effective drainage organization in the epikarst and collector structures below, the improved conditions for transport of fines through the epikarst bring about damage of the soil cover, and the soil loss will progressively increase. The soil loss occurs first above drawdown cones expressed as distinct karren fields (Fig. 5) and progresses through the area with further maturation of the epikarst and improvement of its links with the vadose zone below. The main natural reason for the soil loss is believed to be ecological crisis, as that produced by climatic changes. The above consideration suggests that the soil loss can also be related to the particular (mature) stage of the epikarst development. Evolution of epikarst Continuous evolution Epikarst is a dynamic system which main characteristics are time-variant, changing in a regular way during the epikarst evolution. Several distinct stages can be envisaged in the continuous epikarst evolution. Fig. 6 illustrates this evolution, and the bar chart in the left side indicates relative changes in the intensity of the main characteristics of the epikarst. Epikarst icons depicting particular stages, and the stage names, correspond to the evolution starting from the complete exposure of the bedrock (without regolith). However, as stated above, the epikarst evolution can also commence under the regolith cover. In this case the epikarst will go through the incipient and young stages being the soilcovered, with respective effects on some characteristics. Discontinuous evolution Buried epikarst presents a specific type of paleokarst. Preserved paleo-epikarst horizons are frequently recognized within carbonate sequences (Osborn, 2002). Re-exposure of paleo-epikarst horizons can result in their exhumation and re-establishment of their hydrologic functions (Table 1). Interruption of epikarst evolution by glacial stripping is the most common, especially in mountain regions. Glaciers can strip away completely the epikarstic zone. The result is the loss of functional relationship of conduit systems with the newly formed relief (Fig.7). The removal of epikarst changes drastically hydrological behavior of the post-glacial karst system. The epikarstic zone tends to re-establish after glaciations, and its evolution follows largely the same pattern as discussed above. However, some differences on the stages of incipient and young epikarst can be imposed by: (i) differences in stress release effects imposed by glacial unloading and those from the original post-burial unloading; (ii) peculiarities of weathering processes in the periglacial zone; (iii) presence of well developed although hydrologically functionless conduits (shafts) in the vadose zone. Most of alpine karst massifs that experienced glaciations during the last glacial maximum (2514 ka) have the epikarst re-establishing, presently on incipient or young stages. The Table 1 presents the evolutionary classification of epikarst, in which the principal categories (in bold) are distinguished on the basis of the evolution continuity and actuality. The types of epikarst (in italic) correspond to main starting scenarios that differ by the principal factors of the epikarst formation. The types in the continuous epikarst evolution can be further subdivided according to their relative age (maturity) as discussed above (see Fig.6.) Definition of epikarst and final remarks Conceptually, epikarst was viewed as either an aquifer, or as a zone in the vertical section of a karst massif. The latter notion seems to be more appropriate for karstology as it allows considering various func tions and properties of the epikarst subsystem in the overall karst system. The epikarstic zone can be defined from various perspectives, although a general karstological definition should attempt to emphasize several principal characteristics, namely: origin, structural distinction, hydrologic functions and the morphogenetic role.

PAGE 10

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.10 Fig.6. The continuous evolution of epikarst and changes in its characteristics.

PAGE 11

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.11 TABLE 1 Evolutionary classification of epikarst Continuous epikarst evolution Epikarst in open karst (evolved on young carbonate platforms which have never been buried) Epikarst in denuded karst (evolved after denudational removal of the cover, in paragenetic relationships with the draining structures in the vadose zone) Incipient Incipient Young Young Mature Mature Old Old Discontinuous epikarst evolution Epikarst reestablished after mechanical removal of the original epikarst (i.e. by glacial scour) Epikarst exhumed after burial Terminated epikarst evolution Paleo-epikarst (buried) Fig.7. Shafts opened to the surface by glacial scour of the epikarstic zone. Left photo: A shaft on the top of the ridge at the elevation of about 3100m, which was shaped by the last glaciation occurred during Holocene (9.1-7.5 ka), Aladag lar massif, Eastern Taurus, Turkey; note that the epikarstic zone is virtually absent. Right photo: A narrow neck of a large (90m d eep, 3m wide) entrance shaft of the Arabikskaja System (-1110m) at the elevation of 2180m, Arabika massif, Western Caucasus. Note that the incipient epikarstic zone is already presen t, formed since the last glaciation that occurred supposedly during the Last Glacial Maximum (25-14 ka).

PAGE 12

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.12 Epikarst is defined as: The uppermost weathered zone of carbonate rocks with substantially enhanced and more homogeneously distributed porosity and permeability, as compared to the bulk rock mass below; a regulative subsystem that functions to store, split into several components and temporally distribute authogenic infiltration recharge to the vadose zone. Permeability organization in the epikarst dynamically develops to facilitate convergence of infiltrating water towards deeply penetrating collector structures such as prominent fissures that drain the epikarstic zone. This is manifested by epikarstic morphogenesis that tends to transform disperse appearance of surface karst landforms into focused appear ance adapted to the permeability structure at the base of epikarst. Further studies of epikarst should help to develop its more adequate typology. Efforts toward detailed characterization of the hydrologic and transport behavior of the epikarst, as well as its morphogenetic role, should be placed in the context of the typological variability of epikarst. Recognition of the complex nature of the epikarst and of different starting conditions for its development allows more comprehensive approach to the general problem of karst morphogenesis. Acknowledgements I thank to Prof. David Culver and the Organizing Committee for making it possible my attendance of the KWI Epikarst Conference. References Bakalowicz, M., Blavoux, B. and Mangin, A. 1974. Apports du traage isotopique naturel la connaissance du fonctionnement d'un systme karstique-teneurs en oxygne 18 de trois systmes des Pyrnes, France. Journal of Hydrology 23. 141-158. Bakalowicz, M. 2004. The epikarst, the skin of karst. In: Jones, W.K., Culver, D.C. and Herman, J. (Eds.). 2004. Epikarst. Proc. of the symposium held October 1 through 4, 2003 Sheperdstown, West Virginia, USA. Karst Water Institute special publication 9, 16-22. Chernyshev, S.N. 1983. Fissures of rocks. Moscow: Nauka, 240 p. (in Russian). Ford, D.C. and Williams, P.W. 1989. Karst geomorphology and hydrology. London, England: Unwin Hyman. 601 p. Gouisset, Y. 1981. Le karst superficiel: gense, hydrodynamique et caractristiques hydrauliques, Univ. des Sciences et techniques du Languedoc, Montpellier, France, 218 p. Kiraly L. 2002. Karstification and Groundwater. In: Gabrovšek, F. (Ed.), Evolution of karst: from prekarst to cessation. PostojnaLjubljana, Zalozba ZRC, 155-190. Kiraly, L. and Morel, G. 1976. Remarques sur l'hydrogramme des sources karstiques simul par modles mathmatiques. Bulletin du Centre d'Hydrogologie 1, 37-60. Klimchouk, A.B. 1987. Conditions and peculiarities of karstification in the shallow subsurface zone of carbonaceous massifs. Caves of Georgia, v. 11, 54-65. (In Russian, res.Engl.). Klimchouk, A.B. 1989. Significance of the Subsurface Zone in Karst Hydrology and Morphogenesis. Kiev, Inst. Geol. Nauk, 44 p. (In Russian, res.Engl.). Klimchouk, A.B. 1995. Karst morphogenesis in the epikarstic zone. Cave and Karst Science.21 (2), 45-50. Klimchouk, A.B. 2000. The formation of epikarst and its role in vadose speleogenesis. In: A.Klimchouk, D.Ford, A.Palmer, W.Dreybrodt, Eds: Speleogenesis: Evolution of karst aquifers. Huntsville: Natl. Speleol. Soc. 91-99. Klimchouk, A.B. and Ford, D.C. 2000a. Types of karst and evolution of hydrogeologic settings. In: A.Klimchouk, D.Ford, A.Palmer, W.Dreybrodt, Eds: Speleogenesis: Evolution of karst aquifers. Huntsville: Natl. Speleol. Soc., 45-53. Klimchouk, A.B. and Ford, D.C. 2000b. Lithologial and structural controls of dissolutional cave development. In: A.Klimchouk, D.Ford, A.Palmer, W.Dreybrodt, Eds: Speleogenesis: Evolution of karst aquifers. Huntsville: Natl. Speleol. Soc., 54-64. Klimchouk, A.B., Rogozhnikov, V.Ja. and Lomaev, A.A. 1981. Karst of the Kyrktau

PAGE 13

A.B.Klimchouk / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.13 Massif, Zeravshansky Ridge, Tien-Shan. Kiev: Inst. Geol. Nauk. 54 p. (in Russian). Klimchouk, A.B., Stotland, V.Ja. and Lomaev, A.A. 1979. Karst and speleological investigations in the Kyrktau massif, Zeravshansky Ridge. Izvestija VGO, vol.111, n.5. 117-123. (in Russian) Mangin, A. 1973. Sur la dynamique des transferts en aquifer karstique. Proc.of the 6th Intern. Congr. of Speleol., Olomouc, v.4. 157-162. Mangin A. 1975. Contribution l'tude hydrodynamique des aquifres karstiques. Thse, Universit de Dijon, 124 p. Mylroie, J.E. and Carew, J.L. 1995. Karst development on carbonate islands. In: Budd, A.D., Saller, A.H. & Harris, P.M. (eds.), Unconformities and porosity in carbonate strata. AAPG Memoir 63, Tulsa, Oklahoma, 55-102. (also published in Journal of Speleogenesis and Evolution of Karst Aquifers 1 (2), 2003, www.speleogenesis.info). Mylroie, J.E. and Carew, J.L. 2000. Speleogenesis in coastal and oceanic settings. In: A.Klimchouk, D.Ford, A.Palmer, W.Dreybrodt, Eds: Speleogenesis: Evolution of karst aquifers. Huntsville: Natl. Speleol. Soc., 225-233. Osborne R.A.L. 2002. Paleokarst: cessation and rebirth? In: Gabrovšek, F. (Ed.), Evolution of karst: from prekarst to cessation. PostojnaLjubljana: Zalozba ZRC. 43-60. (also published in Journal of Speleogenesis and Evolution of Karst Aquifers 1 (2), 2003, www.speleogenesis.info). Perrin J., Jeannin P.-Y. and Zwahlen F. 2003. Epikarst storage in a karst aquifer: a conceptual model based on isotopic data, Milandre test site, Switzerland. Journal of Hydrology 279 (1-4), 106-124. Smart, P.L. and Friederich, H. 1986. Water movement and storage in the unsaturated zone of a maturely karstified carbonate aquifer, Mendip Hills, England. In: D. National Water Well Association, Ohio, Editor, Proceedings of the Conference on Environmental Problems of Karst Terrains and their Solutions, pp. 59–87. Williams, P.W. 1983. The role of the subcutaneous zone in karst hydrology. Journal of Hydrology 61. 45-67. Williams, P.W. 1985. Subcutaneous hydrology and the development of doline and cockpit karst. Zeitschrift fur Geomorphologie 29. 463-482. Worthington, S.R.H., Ford, D.C. and Beddows, P.A. 2000. Porosity and permeability enhancement in unconfined carbonate aquifers as a result of solution. In: Klimchouk A., Ford D., Palmer A. and Dreybrodt W. (eds.), Speleogenesis: Evolution of Karst Aquifers. Huntsville: National. Speleological Society, 463-472.



PAGE 1

Speleogenesis and Evolution of Karst Aquifers The Virtual Scientific Journal www.speleogenesis.info The Moselle piracy: new chronological data from U/Th dating of speleothems B. Losson(1) and Y. Quinif(2) (1)Centre d'Etudes Geographiques de l'Universite de Metz (CEGUM), Laboratorie de geographie physique, Ile du Saulcy, 57045 Metz Cedex 1 (France); e-mail: benoit.losson@umail.univ-metz.fr (2)Centre d'Etudes et de Recherches Appliquees au Karst (CERAK), Faculte polytechnique de Mons, rue de Houdain, 9, B‚7000 Mons (Belgium); e-mail: Yves.Quinif@fpms.ac.be Yves.Quinif@fpms.ac.be Re-published from: Karstologia 2001, no. 37, p. 29-40 Abstract The Moselle piracy is one of the most important changes of the hydrographic network in Lorraine (France). For a long time, this phenomenon has been presumed to be relatively recen t (at the end of the Middle Pleistocene) because of the wellpreserved fluvial morphologies and deposits. With new relations between the surficial and subterranean data in the piracy area, the capture has been dated from 300ka using U/Th met hod on speleothem. This evaluati on reveals an earlier time for the phenomenon, and is more precise than those proposed up to now. The latter were derived from the North-European glacial chronology and one thermolumines cence date obtained in the downstream valle y of Meuse. In fact, the improvements in absolute dating, thanks to different methods and U/Th in particular, lead ge omorphologists to abandon the simple relation between the glacial-interglacial periods and the accumulation-erosion processes in rivers. Keywords: Moselle piracy, Fluvial deposits, Speleothems, U/Th dating, Haye caves Introduction For a quarter of a century, the unequalled development of the absolute dating (14C, U/Th, K/Ar) has highly contributed to the progresses in paleogeomorphology. The classical chronological outline for the Quaternary, based on the Alpine or North European glaciations, has been better established in time. However, the Quaternary phenomena and deposits are more often set back in an isotopic stratigraphy (isotopic stages of Emiliani, 1955). In the field of the valleys change, the relative chronologies always prevail, because of the lack of chronological information from the alluvial deposits. The regular entrenchment of rivers, recently reassessed by some authors (Macaire, 1990; Quinif, 1999), encourages to review the simple causal relations between glacialinterglacial periods and fluvial terraces (Conchon, 1992). At the same time, the temporal setting of the hydrographic reorganizations by captures is itself concerned with this difficulty to make the correspondence between the alluvial levels and the North European chronological outline. On the occasion of a research about the influence of the karst on the Moselle piracy (Lorraine, France), the dating of speleothems by

PAGE 2

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.2 the U/Th method (Quinif, 1989) was seen as a reliable way to refine the capture time. The determination of the ge netic relations between the alluvial deposits and the cave systems of the Moselle valley yields, by successive interdependencies (Fig. 1), to a chronological framework for the Moselle piracy, based on U/Th dating. The Moselle hydrographic network geosystem is under the influence of geologic and climatic parameters, thus morphological too. The Moselle piracy phenomenon is linked by feedback effects to the fluvial sediments, which themselves have been generated in the relationship with the local valley cave systems and their filling. These successive genetic relations allow to establish some chronological links between the stratigraphic and morphologic elements that make up the system. The main link is caused by the existence of surficial and underground alluvia that we can corroborate. Considering this fact, the absolute dating of speleothems will allow to establish minimal ages for each alluvial deposit. And the same will be true of the hydrographic diversion, set by the relative chronometer constituted by the Moselle alluvial deposits. Fig. 1. Relations between the geomorphologic data of this study, leading to the different interpretations. 1. Main features of the relief of the studied area 1.1. General geomorphologic context Two main rivers govern the drainage pattern of Lorraine: the Moselle and the Meuse. Coming from the Vosges, the first one heads for the second one as far as Toul, where they are only twelve kilometers apart, before diverging to join the Rhine. Formerly, the Moselle was flowing in the Val de l'Ane, west of Toul, before meeting the Meuse at Pagny (Figures 2 and 3; Photos 1 and 2). The actual acute bend made by the river is the result of a capture, well studied for more than one century (Davis, 1895; Blache, 1939-1940, 1943; Tricart, 1952; Harmand, 1992; Gamez et al., 1995; Harmand et al., 1995c; Le Roux and Harmand, 1998; Harmand and Le Roux, 2000). The site of this hydrographic diversion encompasses two cuestas systems (Figures 2 and 3): the Moselle cuesta, on the east, made up of Toarcian marls and Bajocian limestones; the Meuse cuesta, on the west, with Bathonian-Callovian marls and clays, and Oxford limestones.

PAGE 3

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.3 Fig. 2. Simplified cross-section through the Meuse and Moselle cuestas. Fig. 3. Morphostructural sketch of the "Boucle de la Mose lle" (from the geological maps of Toul and Nancy at the scale of 1/50,000). Photo 1. The cataclinal paleovalley of the Moselle in the Oxfordian Escarpment, seen from the old fort of Dommartin-les-Toul to the west (snapshot B.L.). Photo 2. The abandoned Paleo-Moselle valley through the Hauts de Meuse (Val de l'Ane), seen from the Ecrouves plateau (snapshot B.L.).

PAGE 4

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.4 In addition to the alternation of depression scarp face dip slope, the local relief is marked by the Moselle valley, deeply entrenched in the Haye plateau (cataclinal and anaclinal branches, to the south and the north respectively; Photos 3 and 4) and much more widened in the orthoclinal depression of the Woevre. Two geomorphologic features correspond to this fluvial axe: a lot of alluvial deposits all along the Moselle stream and its paleocourse towards the Meuse; the presence, on the valleyside, of "died" cave systems (Fenelon, 1968), which shows a decreasing stepping from east to west, in accordance with the general dip. Some marked genetic interactions exist between these two parameters, as will be shown. But before, let us examine more precisely their characteristics. Photo 3. The cataclinal valley of the Moselle at Maron, seen from the entr ance of the CAF's Quarry Cave to the south; on the right, the wooded plateau, which dominates the river over 100 m high, is covered by old alluvial deposits CF7 to CF10 (snapshot B.L.). Photo 4. The Moselle anaclinal valley, seen from Liverdun to the west. In the foreground, the first loop of the meander has been straightened by the canalization of the river (snapshot B.L.). 1.2. The fluvial deposits of the Moselle In the geomorphologic context of a stream piracy, the fluvial deposits of the Moselle are already studied for a long time. The story of these investigations can be split up into three stages: first, selectively, with the intention of giving arguments for the diversion, i.e. mainly in the Val de l'Ane and along the Meuse valley (Buvignier, 1840; Husson, 1864; Vidal de la Blache, 1908; Nickles, 1911; Martin, 1920); then, from a spatial and altimetric point of view (Gardet, 1928; Theobald and Gardet, 1935; Errard, 1942); last, and more recently, about the river deposits as such (Dangana, 1970; Vaskou, 1979; Taous, 1994; Techer, 1995; Harmand et al., 1995a; Dorniol, 1997). Some works are still under way for the determination and the differentiation of the fluvial levels, but on the whole, we can note the following characteristics (Harmand et al., 1995c; Pissart et al., 1997; Harmand et al., 1998; personal unpublished observations) (Fig. 4): residual old alluvia exist above 250 m high (in the study area): the detritic deposits (CF) have been strongly reworked and do not yet form fluvial terraces s.s.; real terraces, as well as the actual flood plain, can be seen below 250 m high: their alluvial deposits (F5 to F0), still often thick and not reworked, have been the subject of petrographic counting whose results vary from one author to another, in their meaning for the differentiation of the diverse deposits. Note that the names F5 to F0 correspond to Fx1 to Fz of Harmand et al. (1995a), but we will only use the former nomenclature in the following text, so as to free ourselves from the North European chronological notion, defined by the small letters. Let us remind above all that the alluvial deposits F4 date from before the capture, whereas the F3 one are posterior to this event (Harmand et al., 1995a). The hydrographic phenomenon probably occurred at the end of the F4 aggradation.

PAGE 5

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.5 1.3. The cave systems of the Moselle valleyside The second important geomorphologic element, although not very visible in the landscape, is the cave systems found all along the entrenched Moselle valley in the Haye plateau. Most of the known caves open onto the sides of the cataclinal defile and in particular at Pierre-la-Treiche (Fig. 3; Photo 5). The main cave systems come down to five caves (Louis and Lehmuller, 1966) (Figures 4 and 5): Chaos Cave (about 240 to 270 m high above sea-level), 7 Chambers Cave (about 210 to 230 m), SainteReine Cave (about 210 to 225 m), Jacqueline Cave (about 205 to 225 m) and Shafts Cave (about 215 to 235 m). Besides these main caverns many other ones exist with far more limited developments, like the CAF's Quarry Cave, at an elevation of 295 m. Following numerous morphosedimentary subterranean observations, the same genesis can be proposed for all this valleyside endokarst. Fig. 4. Location in elevation of the Moselle alluvial deposits near Pierre-la-Treiche, of part of Sainte-Reine and Shafts Caves (developed cross-sections) and of some other main cave systems in the cataclinal valley. Fig. 5. Pierre-la-Treiche cave systems on the right bank of the Moselle (from Louis and Lehmuller, 1966; Louis, 1988).

PAGE 6

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.6 2. Genesis of the cave systems of the Moselle valley: development and relative temporal aspects 2.1. Conditions of cave systems development The caves in the Haye plateau present all the characteristics of a subfluvial karstification (Gamez and Losson, 1998; Losson, 2000): exclusive location of the cave systems close to the sides of the Moselle valley, and a near absence of penetrable cavities, with no fluvial influence of the Moselle (or the Meurthe); marked prevalence of cave macroor micromorphologies in flooding conditions (maze of bedding plane anastomoses, eroded joint patterns, withdrawing chimneys, pockets in the three dimensions of space, scallops which indicate that the paleoflows came from the Moselle thalweg to the interior of the massif, ceiling channels) (Photos 6 to 13); considerable filling, made up mainly of alluvial deposits of Vosges origin (Photos 13 and 14), and petrographic analogies of these detritic deposits with the surficial alluvia of the fluvial terraces, just above the caves (cf. Photo 5). Photo 5. The Moselle valley at Pierre-la-Treiche (snapshot P. Gamez). Photo 6. A view of the Eastern gallery of SainteReine Cave: this pipe was created at the meeting point between a fracture and a bedding plane (snapshot B.L.). Photo 7. The Passage de la Tete de Renard ("Fox Head Bottleneck ") (proba bly a former siphon), in the Western gallery of Sainte-Reine cave (snapshot B.L.). Photo 8. A view of the Western gallery of SainteReine Cave: more spacious, on the whole, than the Eastern gallery, this one is nevertheless filled with a lot of Moselle alluvial deposits (about 4 m thick here) (snapshot B.L.). Photo 9. Typical pipe of the eastern part of SainteReine Cave: the northern connection gallery (notebook = 17 x 11 cm) (snapshot B.L.).

PAGE 7

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.7 Photo 10. The entrance gallery of Shafts Cave, developed along a joint (snapshot P. Gamez). Photo 11. The Shaft 1 of Shafts Cave; it is a pseudo-shaft (vertical pipe) developed on a fracture in the phreatic zone (snapshot P. Gamez). Photo 12. The ceiling of the western Lower Gallery of Shafts Cave, displaying scallops (southward view) (snapshot B.L.). Photo 13. Karst pipes filled by some coarse alluvial deposits, in the old quarry of Pierre-la-Treiche (snapshot B.L.).

PAGE 8

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.8 Photo 14. The entrance of Sainte-Reine Cave, filled at 90 to 100 % by the Moselle alluvial deposits (probably F4); the tota l infilling thickness exceeds 7.5 m (scale given by the folding double rule on the right) (snapshot B.L.). These observations are borne out at all levels of karst cavities, along the cataclinal Moselle course as well as the anaclinal one, after the capture; the sole "Trou des Fees" ("Hole of Fairies") is appearing on the map of Fig. 3, but others cavities exist, like the "Trou de la Grosse Roche" ("Hole of the Big Rock"), for example. It is interesting to notice that the karst at the lithostratigraphic contact, so well represented elsewhere in Lorraine (Gamez, 1992; Nicod, 1994; Jaillet and Gamez, 1995; Jaillet, 1999, 2000), is curiously lacking here or is not yet known, in spite of favorable geomorphologic conditions. 2.2. Chronological considerations: relative conception. The only available temporal framework for the speleogenesis period resulted until then from an investigation of the relations between underground and surficial alluvial deposits, in order to refer to the relative chronology, which the fluvial deposits set up. Actually, the decreasing stepping in function to the age of the different alluvial levels (residual superficial deposits or real terraces) appears to be indisputable. The petrographic differentiation that exists between anteand postcapture alluvia (Harmand et al., 1995a), has allowed specifying the minimum relative age of many karst cavities, in which some fluvial surficial sediments have been analysed (Losson, 2000). The very great majority of the caverns of the Moselle valley contain alluvia coming from F4 (T 1.2 terrace) or older deposits. Here, the elevation of the cave systems is not correlative to the levels of fluvial terraces at all, because of their subfluvial genesis. So the karstification has taken place mostly before the Moselle diversion (along the cataclinal course at least), and at a significant depth under the thalweg for some caves (10 m at least), since we can see antecapture diluvium at an elevation similar to the lower terrace T 0.2 (cf. Fig. 4). The speleothems of the caves allow to give chronological data to this change, and therefore to the capture. 3. The U/Th dating: a tool for paleogeomorphology Eight stalagmites or flowstone floors, collected at various elevations in the different cave systems of the Moselle valley, have been dated by the U/Th method in the C.E.R.A.K. (Mons, Belgium). The 26 isotopic analyses which have been carried out, have supplied 4 main classes of ages: ages similar to or higher than the maximal range (230Th/234U=1): about 270,000 years old and more than 350,000 years old; about 170,000 years old; in the Upper Pleistocene (about 40-50,000 years old); in the Holocene (less than 15,000 years old). The most interesting speleothems in this study are the ones whose age equal or exceed 170,000 years old (Table 1). Several chronostratigraphic and speleogenetic reasons relate to that. In order to date the Moselle diversion as precisely as possible, it is necessary to be fit with the phenomenon, that is to collect the

PAGE 9

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.9 speleothems on stratigraphic conformability with the underground fluvial deposits which surround the capture (F3 and F4). We did not found any speleothems overlying the postcapture alluvial deposits F3 that would allow to supply a minimum age of the diversion right away. We must content ourselves with speleothems whose location on antecapture alluvia doesn't allow determining the anteriority or the posteriority of the hydrographic phenomenon toward the growth of speleothems. Indeed, the concretions give only a minimum date to the speleogenesis and to the setting up of the underlying filling. So, without further case, the capture can be younger or older than the dated speleothems, according to the time span from their development to setting up of the fluvial deposits. In order to try to solve this problem, it is then necessary to take a new element into consideration: the essential draining of the galleries in which the speleothems have grown. This draining, which corresponds to a fall of the water table in the Bajocian limestones, varies, on the whole, according to the entrenchment of the main valleys (of the Meurthe and of the Moselle, that constitute the local base levels). Thus, we tackle the altimetric aspect of speleothems: the lower a speleothem is in elevation, the more likely it has grown late, when the water table has fell with the entrenchment of the valleys. If we manage to demonstrate that the low-lying concretions have grown after the capture (when the draining has occurred), we can deduce that this hydrographic diversion is older than speleothems. To sum up, the choice of the concretions for our demonstration was motivated by their age (the Moselle piracy we want to date definitely occurs before the Upper Pleistocene) and by their lowest elevation we were able to find in the valley. The other datings, which give much younger ages, will be analysed within the framework of another research issue, more targeted on the paleoclimate. TABLE 1. Results of the dating made by the C.E.R.A.K. (Polytechnic Faculty of Mons, Belgium); founding: C.E.G.U.M. (University of Metz, France). The analyses are made in alpha spectrometry. This method sometimes leads to poor chemical yields, above all in thorium (recovery ratio of the element during the chemical preparation). The use of the isotopic dilution method (addition of an artificia l isotope of the uranium and the thorium before this preparation) allows to free oneself from these losses. Nevertheless, a too important loss makes the analytical mistake very significant and considerably reduces the re liability of the age found. Samples [U]ppm 234U/238U230Th/234U 230T h/232Th [234U/238U]t=0 Age (in ka) SR-GE-1 161.3 0.627 1.663 0.005 0.856 0.011 82 5 2.069 169.9 [+4.8/-4.7] SR-GE-1bis 159.6 0.385 1.664 0.003 0.858 0.007 101 5 2.072 170.6 [+3.1/-3.0] Pui-GIO-1.1 0.185 0.002 1.315 0.014 1.072 0.083 3.6 0,6 2.091 442 [+inf./-166] Pui-GIO-1.2 0.254 0.002 1.272 0.012 1.050 0.079 9.2 2,8 1.830 398 [+inf./-133] Pui-GIO-1.2bis 0.253 0.003 1.279 0.011 3.908 0.366 73 23 Pui-GIO-2 0.253 0.004 1.243 0.020 0.969 0.042 19 3 1.518 269.8 [+64/-41] Note: the samples whose name has a same first figur e, are in stratigraphic conformability; the second figure indicates the stratigraphic location of the sam ple in the speleothem: the youngest (1) to the oldest (>1); the names with "bis" correspond to a same sample analysed a second time as a check.

PAGE 10

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.10 3.1. The SR‚GE stalagmite This speleothem (Photo 15), collected in Sainte-Reine Cave at an elevation of about 214 m, was unfortunately not in place, but has probably undergone limited displacement (Figures 6 and 7). Its isotopic analysis is very good and gives a great reliability to the obtained age, thanks to the strong uranium and low detritic thorium content. Photo 15. The SR‚GE‚1 stalagmite taken from Sainte-Reine Cave (snapshot B.L.). According to its existence and its location, we can draw two conclusions: the gallery where it stood, was drained at the time of its growth, that is 170,000 years ago; the underlying sediments were set up at an even more former period; it can be assigned to F4 or older alluvia according to sedimentologic studies realised elsewhere in the cave (Gamez and Losson, 1998; unpublished observations). Let us notice that the speleogenesis occurred, on the whole, earlier than the filling deposition (Losson, 1999). Many interpretations about the date of the Moselle capture may be set out on the basis of these findings, that is according firstly to the type of the karst alluvial deposits and secondly to the elevation of the water table under the thalweg. Let us consider first a cave filling made of F4 alluvia. The growth of the SR‚GE stalagmite, 170,000 years ago, could be (Fig. 8) most early at the end of the F4 surficial aggradation (the time of the capture). In that case, the capture would have happened a bout 170,000 years ago; or else at a later stage of surficial incision or aggradation. The capture would have happened more, or even decidedly more, than 170,000 years ago. Fig. 6. Topographic and altimetric location of the SR‚GE stalagmite in Sainte -Reine Cave (Pierre-la-Treiche).

PAGE 11

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.11 Fig. 7. Detail of the SR‚GE stalagmite sampling area in Sainte-Reine Cave (Pierre-la-Treiche). The second hypothesis appears to be the most reliable. Indeed, in the first case, the water table would have been -28 m high under the thalweg, at the highest elevation. On the other hand, in the second case, it could reach a lower relative elevation under the thalweg. Now, a water table lower than -20 m below the thalweg, at that time, seems to be inconceivable to us, considering the supposed regional geomorphologic context, that governs the existence of exsurgen ces of the Bajocian limestones aquifer. Let's now consider a cave filling made of F5 alluvia. The valley was less entrenched at the time of their surficial aggradation. So, if we keep the hypothesis of a water table that is not more than -20 m under the thalweg, we come to the conclusion that the SR‚GE stalagmite has grown after the end of the F4 surficial deposition, that is after the capture. Fig. 8. Simplified representation of the geomorphologic and hydrogeological context in which the SR‚GE stalagmite has grown, acco rding to two hypotheses.

PAGE 12

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.12 Thus, in every case, we get back to an age older than 170,000 years for the hydrographic diversion. In addition, it is interesting to note that the speleothem growth period corresponds to a short warming stage (18O isotopic stage 6.5; Martinson et al., 1987). So, it is possible that the Moselle fluvial landscape has not kept any geomorphologic witness correlative to this period. In these conditions, the development of the stalagmite would relate to any surficial known parameter. 3.2. The Pui-GIO flowstone floor This speleothem (Photo 16) collected in Shafts Cave, was at an elevation of 226 m (Fig. 9). Pui-GIO-1 and Pui-GIO-2 are two pieces of a same flowstone floor whose respective position in the gallery differed: the first one lay on the floor (silty clay filling), whereas the second one was still perched on the wall, a few centimetres above the filling (Fig. 10). More precisely, Pui-GIO-2 is the equivalent of PuiGIO-1.2. Some metres southwards in the gallery, we find this same flowstone floor again, hanged more than 1 m above the withdrawn filling (Photo 17). Photo 16. The Pui-GIO-1 flowstone floor taken from Shafts Cave; on the left, note the calcite infilling inside a desiccation crack (snapshot B.L.). Fig. 9. Topographic and altimetric location of the Pui-GIO fl owstone floor in Shafts Cave (Pierre-la-Treiche).

PAGE 13

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.13 Fig. 10. Detail of the Pui-GIO flowstone floor sampling area in Shafts Cave (Pierre-la-Treiche). Photo 17. A hanging flowstone floor in the western Lower Gallery of Shafts Cave, probably a lateral equivalent of Pui-GIO-1 and Pui-GIO-2; the small calcite cylinder, just above the felt-tip, reveals a former underlying sandy inf illing (snapshot B.L.). The analyses of the first two samples (cf. Table 1), in a stratigraphic succession, are good but they have some quite small isotopic ratios 230Th/232Th. It must be known that this ratio indicates the possible contamination of the speleothem by unfamiliar substances to the carbonated system. We experimentally estimate that the value of this ratio must exceed 20 for the result to be correct. Here, the value of 9 and, all the more so, of 4, is small. The 270,000 years old Pui-GIO-2 is more reliable, because of a higher isotopic ratio 230Th/232Th. The consideration of these three results leads us to a position this flowstone floor is about 300,000 years old. The sample Pui-GIO-1.2bis is characterized by a very bad yield on the thorium: the result must be ruled out. Some identical conclusions to those of Sainte-Reine's stalagmite can be deduced for this speleothem: drained gallery at the time of the calcite deposition, at least 300,000 years ago; a former deposition of the underlying sediments; these ones have unfortunately not been determined near the concretion; probably even former speleogenesis. This old flowstone floor allows some corrections about the previous interpretations. Indeed, Shafts Cave and Sainte-Reine cave systems are only 400 m apart (cf. Fig. 5). Now, if we take the morphosedimentary study of Sainte-Reine Cave E Entrance into account (Gamez and Losson, 1998), where the F4 (or ante-F4) alluvial deposits are up to 225 m high, it can be reasonably thought that the gallery which contains Pui-GIO has been drained after the accumulation of these Moselle detritic deposits in Sainte-Reine Cave. As a consequence, it would be possible to replace the age of 170,000 years old of the previous interpretations with this of 300,000 years old. So, the Moselle piracy would have occurred more, or even more, than 300,000 years ago. Taking up the North European nomenclature, we can note that the capture would not relate to the Saalian glaciation, as admitted since J. Tricart works (1952), but to a stage of the Holsteinian interglacial at least, if not to the Elsterian glaciation (cf. Sibrava et al., 1986, in Conchon, 1992; Renault-Miskovsky, 1992). However, to be true, this remark would require to know the real temporal period for these terms of Saalian, Holsteinian and Elsterian. Some divergences of interpretation exist between different authors, in particular for the limit Holsteinian/Elsterian. Conclusion Like many researches have already emphasized (Maire, 1990; Audra, 1994; Delannoy, 1997; Vanara, 2000; Jaillet, 2000), the karst environment has been proved to be of an unequalled wealth for all kinds of paleogeographic reconstruc tions. As regards to caves of the Moselle valley, of which a little

PAGE 14

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.14 part of the morphosedimentary information only have been mentioned here, some new chronological data have been revealed. The speleogenesis of the concerned cavities by the U/Th dating goes obviously back to a former period than 300,000 years old, like the Moselle piracy phenomenon, and we couldn't be more specific. This result relies on established interdependencies between different parameters of the physical environment: elevation, superposition, morphologic and sedimentologic indications concerning the alluvial deposits and the endokarst. Chronological and geomorphologic correlations with the results of S. Jaillet (2000) about the entrenchment of the streams of Barrois (western Lorraine), are still impossible to establish, in the present state of our knowledge. On the other hand, the proposed interpretations seem to be confirmed by other U/Th datings obtained in Belgium in the Meuse watershed. Indeed, many speleothems close to the present streams in elevation, have supplied ages older than 400,000 years (Quinif, 1999). This means the Meuse and its tributary rivers have practically halted their entrenchment since that time, which really happened on the upper part of the stream after the Upper Moselle piracy (Harmand, 1992; Pissart et al., 1997). As the last stage of notable entrenchment of the Meuse valley goes back to the immediately antecapture period, this major hydrographic diversion would have occurred between 300,000 and 400,000 years old at least. In addition, these results confirm and support the only absolute dating related to the Moselle piracy available until then. It was a 250,000 20,000 years old minimal age obtained by thermoluminescence on burned flints, in the Meuse valley in Maastricht (Huxtable and Aitken, 1985, in Harmand et al., 1995b). The accurate setting of the capture towards the fluvial terraces system in this area was established thanks to heavy minerals studies (Paulissen, 1973, in Harmand et al., 1995b). The flints lay, as for them, in a younger stratigraphic order than the mineralogical break (Vandenberghe et al., 1985). As a result, the Moselle diversion had been dated more than 250,000 years old, without possible further detail. Finally, we can notice that the chronological framework based on the North European glaciations and applied to the fluvial sediments, is less valid with the increase in absolute dating. So, it seems to be desirable to adopt in a first time a specific nomenclature to each main river basin (terraces T 0.1, T 1.2, T 1.1, and alluvial deposits F3, F4, F5, for example), and then to refer preferably to the 18O isotopic stages (Shackleton and Opdyke, 1973; Martinson et al., 1987) as long as absolute dating is available. We dedicate this article to Patrice Gamez, originator of the dating project. His memory led us all the way through this work and we regret he didn't assess the results. Acknowledgements We are grateful to Jean-Jacques Delannoy, Richard Maire and Jean Nicod for the improvements they made to the initial version of this article. In addition, we have received essential information about the researches in the Maastricht site, thanks to Messrs Harmand and Pissart good offices. References Audra, P. 1994. Karsts alpins; genese des grands reseaux souterrains. Exemples: le Tennengebirge (Autriche), l'Ile de Cremieu, la Chartreuse et le Vercors (France). Karstologia Mem. 5, 280 p. Blache, J. 1939-1940. Le probleme des meandres encaisses et les rivieres lorraines. J. of Geomorphology 2(3), 201-212 and 3(4), 311-331. Blache, J. 1943. Captures comparees: la vallee morte de la Bar et les cas voisins. Rev. Geogr. Alpine 31(1), 1-37. Buvignier, A. 1840. Note sur les alluvions de la Moselle dans la vallee de la Meuse. Mem. Soc. Philomathique de Verdun (Meuse) 1, 255-258. Conchon, O. 1992. Que sont Gunz et Mindel devenus? Approches recentes de la stratigraphie du Quaternaire. In: A propos du Quaternaire en Europe, Geochronique 44, 16-18. Dangana, L.B. 1970. Les terrasses de la Moselle entre Neuves-Maisons et Toul. Master's Univ. Nancy 2, 131 p. Davis, W.M. 1895. La Seine, la Meuse et la Moselle. Ann. Geogr., 25-49. Delannoy, J.-J. 1997. Recherches geomorphologiques sur les massifs karstiques du Vercors et de la Transversale de Ronda (Andalousie). Les apports morphogeniques du karst. Doctoral

PAGE 15

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.15 Thesis Univ. Grenoble 1, Villeneuve d'Ascq, P.U.Septentrion, 678 p. Dorniol, Y. 1997. Etude morphologique de la vallee de la Moselle entre Neuves-Maisons et Chaudeney-sur-Moselle. Master's Univ. Nancy 2, 108 p. Emiliani, C. 1955. Pleistocene temperatures. J. of Geology 63, 538-578. Errard, S. 1942. Le probleme de la capture de la Moselle. DES hist.-geo. Univ. Nancy, 217p. Fenelon, P. 1968. Vocabulaire francais des phenomenes karstiques. Mem. et Doc. 4, 1368. Gamez, P. 1992. Hydrologie et karstologie du bassin du Loison (Woevre septentrionale Lorraine). Doctoral Thesis Univ. Metz, Mosella 21, publ. 1995, 453 p. Gamez, P., Wehrli, A., Fizaine, J.-P., Scapoli, J. 1995. Limplication du karst dans la capture de la Moselle. Rev. Geogr. de l'Est 35(3-4), 297-308. Gamez, P., Losson, B. 1998. Premiers resultats de l'etude des remplissages dans le karst de Pierre-la-Treiche (54); l'entree E du reseau Sainte-Reine. Mosella 23(3-4), 41-59. Gardet, G. 1928. Les systemes de terrasses de la trouee Pont-Saint-Vincent, Toul, Foug, Commercy. Bull. Soc. Sci. Nancy 4(3)(3), 237-280. Harmand, D. 1992. Histoire de la vallee de la Meuse lorraine. P.U.Nancy, 146 p. Harmand, D., Kartit, A., Occhietti, S., Weisrock, A. 1995a. Lage de la capture: correlations entre les formations fluviatiles saaliennes de la Haute Moselle et de la Meuse. Rev. Geogr. de l'Est 35(3-4), 269290. Harmand, D., Krook, L., Pissart, A. 1995b. Lenregistrement de la capture de la Haute Moselle dans les alluvions de la basse vallee de la Meuse. Rev. Geogr. de l'Est 35(3-4), 291-296. Harmand, D., Weisrock, A., Gamez, P., Le Roux, J., Occhietti, S., Deshaies, M., Bonnefont, J.-C., Sary, M. 1995c. Nouvelles donnees relatives a la capture de la Moselle. Rev. Geogr. de l'Est 35(3-4), 321-343. Harmand, D., Pissart, A., Krook, L. 1998. L'evolution du paleo-bassin de la Meuse: les enseignements des captures et leurs implications environnementales. In: Arbeiten aus dem Geographischen Institut der Universitat des Saarlandes. Symposium Problemes de l'environnement en Saar-LorLux , Sarrebruck, 157-173. Harmand, D., Le Roux, J. 2000. La capture de la Haute Moselle. Bull. Inf. Geol. Bassin Paris 37(3), 4-14. Husson, N. 1864. Origine de lespece humaine dans les environs de Toul par rapport au diluvium alpin. Pont-a-Mousson, print. P. Toussaint, 63 p. Jaillet, S. 1999. Recul de couverture et karstification dans un karst couvert de bas plateaux: le Barrois (Lorraine / Champagne France). Actes du Colloque europeen Karst 99, Etudes de geographie physique suppl. 28, 123-128. Jaillet, S. 2000. Un karst couvert de bas-plateau: le Barrois (Lorraine / Champagne, France). Structure Fonctionnement Evolution. Doctoral Thesis Univ. Bordeaux 3, 710 p. Jaillet, S., Gamez, P. 1995. Observations morphologiques sur le geosysteme karstique du Rupt du Puits (Meuse, Lorraine). Karstologia 26, 27-38. Le Roux, J., Harmand, D. 1998. Controle morphostructural de l'histoire d'un r eseau hydrographique: le site de la capture de la Moselle. Geodinamica Acta 11(4), 149-162. Losson, B. 1999. Apercu karstogenetique de la grotte Sainte-Reine (Lorraine, France). Regards 37, 29-32. Losson, B. 2000. Modalites des defluviations partielles souterraines de la Moselle avant sa capture. Bull. Inf. Geol. Bassin Paris 37(3), 15-22. Louis, M. 1988. Retrospective d'une decouverte: grotte des 7 Salles, Pierre-la-Treiche. Speleo L 14, 31-35. Louis, M., Lehmuller, D. 1966. Contribution a l'avancement du catalogue des cavites de Meurthe-et-Moselle. Roneo, 137 p.+34 pl. Macaire, J.-J. 1990. Lenregistrement du temps dans les depots fluviatiles superficiels: de la geodynamique a la chronostratigraphie. Quaternaire 1(1), 41-49. Maire, R. 1990. La haute montagne calcaire: karsts, cavites, remplissages, quaternaire, paleoclimats. Karstologia Mem. 3, 732 p.

PAGE 16

B. Losson and Y. Quinif / Speleogenesis and Evoluti on of Karst Aquifers, 2 (1) October 2004, p.16 Martin, P. 1920. Sur la jonction ancienne de la Moselle et de la Meuse par le Val de l'Ane. Bull. Soc. Sci. Nancy 4(1), 181-189. Martinson, D.G., Pisias, N.G., Hays, J.D., Imbrie, J., Moore, T.C., Shackleton, N.J. 1987. Age dating and the orbital theory of ice ages: development of a high resolution 0 to 300,000 year chronostratigraphy. Quaternary Research 27(1), 1-29. Nickles, R. 1911. Contribution a la connaissance de la jonction ancienne de la Moselle et de la Meuse par le Val de l'Ane. Bull. Seances Soc. Sci. Nancy 3(12)(1-4), 282-287. Nicod, J. 1994. Plateaux karstiques sous couverture en France, dapres des travaux recents. Caracteres geomorphologiques et problemes denvironnement. Ann. Geogr. 576, 170-194. Pissart, A., Harmand, D., Krook, L. 1997. Levolution de la Meuse de Toul a Maastricht de puis le Miocene: correlations chronologiques et traces des captures de la Meuse lorraine dapres les mineraux denses. Geogr. Phys. et Quat. 51(3), 267-284. Quinif, Y. 1989. La datation uranium-thorium. Speleochronos 1, 3-22. Quinif, Y. 1999. Karst et evolution des rivieres: le cas de lArdenne. Geodinamica Acta 12(3-4), 267277. Renault-Miskovsky, J. 1992. La palynologie du Quaternaire europeen: chronostratigraphie paleoclimatologie et paleoenvironnement vegetal de lhomme fossile. In: A propos du Quaternaire en Europe, Geochronique 44, 21-24. Shackleton, N.J., Opdyke, N.D. 1973. Oxygen isotope and palaeomagnetic stratigraphy of equatorial Pacific core V28-238: oxygen isotope temperatures and ices volumes on a 105 year and 106 year scale. Quaternary Research 3(1), 39-55. Taous, A. 1994. Le systeme alluvial de la moyenne terrasse de la Moselle en Lorraine meridionale (approche morphosedimentaire et petrographique). Doctoral Thesis Univ. Nancy 2, 201 p. Techer, P. 1995. Etude morphologique et petrographique du complexe alluvial de la moyenne terrasse sur le site de capture de la Moselle. Master's Univ. Nancy 2, 113 p. Theobald, N., Gardet, G. 1935. Les alluvions anciennes de la Moselle et de la Meurthe en amont de Sierck. Bull. Centenaire Soc. Hist. Nat. Moselle 34, 69-100. Tricart, J. 1952. La partie orientale du bassin de Paris. Etude morphologique. Paris, S.E.D.E.S., 474 p. Vanara, N. 2000. Le karst des Arbailles. Karstologia Mem. 8, 320 p. Vandenberghe, J., Mucher, H. J., Roebroeks, W., Gemke, D. 1985. Lithostratigraphy and palaeoenvironment of the Pleistocene deposits at MaastrichtBelvedere, southern Limburg, the Netherlands: stratigraphy, palaeoenvironment and archaeology of the middle and late Pleistocene deposits. Mededelingen rijks geologische dienst 39(1), 7-18. Vaskou, P. 1979. Contribution a la classification des formations alluviales de la feuille de Toul au 1/50 000eme. D.E.A. Univ. Nancy 1, 57 p.+12 pl. Vidal de la Blache, J. 1908. Etude sur la vallee lorraine de la Meuse. Paris, A. Colin, 189 p.



PAGE 1

Speleogenesis and Evolution of Karst Aquifers The Virtual Scientific Journal www.speleogenesis.info Contribution to the speleology of Sterkfontein Cave, Gauteng Province, South Africa J.E.J.Martini(1), P.E.Wipplinger(2), H.F.G.Moen(2) and A.Keyser(3) (1) Center St Remze 07700, France (2) Council for Geoscience, Private bag X112, Pretoria, South Africa (3) 207 Erasmus Rd, Meyerspark 0184, Pretoria, South Africa Re-published from: International Journal of Speleology 32(1/4), 2003, 43-69. Abstract The authors present more data about the speleological aspect of the Sterkfontein Cave, famous for its bone breccia which yielded abundant hominid remains. They al so briefly review the previous volumi nous studies by numerous authors, which are mainly dealing with the paleontology, stratigraphy and sedi mentology of the breccia. The present investigations were oriented to hitherto poorly investigated aspects such as detail mapping of the cave, its country rock stratigraphy and recording the underground extension of the basal part of th e breccia body. The cave consists of a complex network of phreatic channels, developed along joints in Neoarchaean cherty dolostone over a restricted surface of 250x250m. The combined length of all passages within this area amounts to 5, 23km. The system extends over a height of about 50m and the dry part of it is limited downwards by the water-table app earing as numerous static pools. The fossiliferous breccia (= Sterkfontein Formation) forms an irregular lenticular mass 75x25m horizontally by 40m vertically, which is included within the passage network. It crops out at surface and in the cave, and resulted from the filling of a collapse chamber, which was de-roofed by erosion. The present investigation confirmed that the cave and the Ster kfontein Formation are part of a single speleogenetic event. The breccia resulted from cavity filling by sediments introduced from a pit entrance, whereas many of the phreatic passages around it, which are developed at the same elevation, were only partly filled or re mained entirely open up to present. This filling took place mainly in a vadose environment. Taking into account the age of the Sterkfontein Formation (>3, 3-1,5 My, from base to top), the geomorphic evolution of the landscape and the context of other caves in the region, it seem s that the cave might have starte d to form 5 My ago. It has been continuously developing up to present as a result of a slow drop of the water-table. Keywords: Speleogenesis, Sterkfontein Cave Introduction Sterkfontein Cave is situated 35km to the NW of Johannesburg (Fig. 1) and is well known internationally for the hominid fauna found in a paleokarst filling associated with it. It is the most important site among other comparable deposits clustered in the Kromdraai area, which represents a small portion of the regional karst. The numerous publications on Sterkfontein deal mainly with the paleontology, archaeology and sedimentology of this filling, and less with the cave itself. In this paper only the titles most relevant to the former topics are referenced. The present study focussed on a detailed topographic survey of the cave, its geology and speleogenesis. Indeed the previous map produced by the University of the Witwatersrand, although accurate, did not record finer details and many passages have not been plotted. The new survey consisted of theodolite measurements along the tourist route between the two entrances and along a few side passages, in particular the one leading to Ravjee Lake (Fig. 2). The remainder of the passages were surveyed by the usual "compass, tape and clinometer" method performed by the cavers. The map and the sections depicted in Figures 2 and 3 represent the main data gathered during this investigation. Due to the complexity of the system, passage overlap in plan projection and due to a lack of definite cave levels, it was unfortunately not possible to produce a map that is easy to read.

PAGE 2

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.2 Fig. 1. Situation, basic geological and geomorphic data: 1) Kalahari Basin (Upper Cretaceous to Recent sandy sediments); 2) Karoo Basin (Permian shale, sandstone and minor discon tinuous tillite at base); 3) Malmani Subgroup (Neoarchean dolostone and chert); 4) other older or younger terrains; 5) escarpment. History of the discovery, explorations and scientific highlights of the Sterkfontein Cave The fossiliferous deposits of cave-fill breccia and flowstone formation deposits in the “Transvaal Dolomite” were first brought to scientific notice at the inaugural meeting of the newly formed Geological Society of South Africa by its first Secretary, David Draper on 8 April 1895 (Draper, 1896). C.K. Brain (1981) is of the opinion that Draper was describing the exposed cave sediments on Sterkfontein Hill. The actual cave opening was first discovered by Mr. G.Martignalia after a blast. He was mining the Sterkfontein Hill calcite flowstone outcrop in 1896. He termed the opening a “wondergat” (marvelous hole). The cave was immediately explored and was found to be very beautifully decorated by flowstone formations. The beauty of the caves attracted wide interest and articles appeared in both South African and overseas journals. The spectacular beauty of the cave had come to the attention of several geologists in the former Transvaal Republic. In 1898 Draper stated at a meeting of the Geological Society that he had taken steps to preserve the cave for the benefit of the public by approaching the owners, who appeared to be cooperative. However, the protection of the cave was not to last long and the cave was irreparably damaged between 1918 and 1920. At this time the owner, Mr. E.P. Binet was leasing the cave to a Mr. Nolan. Binet was not prepared to extend the lease on expiry. This greatly infuriated Nolan, who then proceeded to blast the flowstone formations with explosives. During the 1920s the cave was exploited for calcite flowstone by the Glencairn Limestone Company. The operations were managed by Mr. G.W. Barlow, who sent specimens of fossils recovered to various scientists. In 1935 the already famous Dr. Robert Broom was appointed as paleontologist at the Transvaal Museum in Pretoria. He immediately proceeded with his search for adult hominid remains, which would accord with the famous Taung Child discovered in 1924 by R. A.Dart. He was conducting excavations in the same area (Skurweberg and Gladysvale sites), when students of the University of the Witwatersrand showed him some fossil baboons that had been found at Sterkfontein. Broom immediately made arrangements and visited Sterkfontein on 17th of August 1936 where he discovered a large portion of a skull and intact upper dentition of an adult Australopithecus (Broom, 1936). Subsequent regular visits by Broom to Sterkfontein returned a number of other australopithecine fossils as well as a wealth of other mammalian material. Shortly before World War II in 1939 mining of calcite was curtailed at Sterkfontein due to a drop in the price of lime and with that the paleontological work also came to an end for the duration of the war. In 1947 Broom, assisted by Dr. J.T Robinson resumed excavations at Sterkfontein using funding arranged by the then Prime Minister of the Union of South Africa, General Jan C. Smuts, who was greatly fascinated by Broom’s remarkable discoveries. Soon afterwards the most complete australopithecene skull was blasted out. This skull subsequently became known as Mrs. Ples.

PAGE 3

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.3 Fig. 2. Map of Sterkfontein Cave. Legend : 1) cavity outline; 2) cavity ou tline under single overlay; 3) cavity outline under double overlay; 4) scarp and edge of pit in cav e; 5) scarp and edge of pit at surface (=entrance); 6) blocks and scree; 7) walls of breccia and boulde r blockages; 8) Australopithecus skeleton; 9) masonry wall; 10) gates and fences; 11) flight of stairs.

PAGE 4

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.4 Operations at Sterkfontein were again curtailed in 1949 when the digging team was moved to Swartkraans, another hominid deposit nearby. In 1959 the owners of the farm Sterkfontein donated the 20 morgen containing the Sterkfontein cave to the University of the Witwatersrand and the area became known as the Isaac Edwin Stegman Nature Reserve. Excavations at Sterkfontein resumed in 1966 under the direction of P.V.Tobias and A.R. Hughes and continued with great success under the direction of Tobias and R.J. Clarke up to the present. More than 500 Australopithecus specimens have been discovered to date. The two most spectacular discoveries were: 1. 1. A specimen of Homo habilis (Stm. 53) found by Hughes in 1976. 2. Little Foot some hominid foot bones which were discovered in a box marked baboon postcranial by R.J.Clarke in 1994. He found that the bones fitted together to produce the most complete australopithecene foot known to date. The bones originated in the so-called Silberberg Grotto. In 1997 Clarke decided to search for the remainder of the specimen and sent two technicians with casts of the foot for possible fit to search for the remainder of the fossil in the grotto. Within a day they located the site where the original foot bones had been found. Clarke then commenced to excavate the bones and to everybodyÂ’s surprise found a near-complete skeleton with a complete skull. This excavation is currently still on-going and the find may represent a new taxon. Sterkfontein is as yet the richest hominid site in South Africa and has the potential to yield a wealth of fossils to scientists for many years to come. As far as caving exploration is concerned, it seems that most sections were already known shortly after discovery, as they are easily accessible. This represented a cumulated length of passages amounting to about 2km. In 1984 a diver (P.Verhulsel) disappeared in the Main Lake and was found dead of starvation 6 weeks later, in a chamber hitherto unexplored into which he had emerged (see details in Sefton et al., 1985). During the rescue attempt, the members of the South African Spelaeological Association discovered and surveyed 892m of new passages. Only minor new sections were found during the survey by the authors, like for instance the Otto Maze (Fig. 2). Two other caves occur immediately north of Sterkfontein, partly overlapping it: Lincoln and Fault Caves. They were discovered and explored at the same time as Sterkfontein and have been mapped by the University of the Witwatersrand, probably in the early seventies (Wilkinson, 1973). In the same period and later, in 1984, the members of the South African Spelaeological Association discovered new passages in Lincoln Cave. In 1989-90, both Lincoln and Fault Caves were re-surveyed in more detail and visually connected through a narrow, but impenetrable tube (Boshoff et al., 1990). Linking with Sterkfontein had been attempted several times in the past, albeit without success. In April 2001, however, after completion of the project it appeared from the survey that two passages in both caves, overlapping each others, were separated by a floor only a few metres thick. By digging from below in a narrow chimney filled with rubble and then from above, a connection could be established, thus integrating the LincolnFault cave system with Sterkfontein Cave (Fig. 3 and 4). Geological context The Sterkfontein Cave is developed in the Malmani Subgroup of late Archaean age (2,52,6 billion years), which has been deposited in the intracratonic Transvaal Basin (Button 1973). The lithology consists essentially of shallow marine stromatolitic dolostone with a variable amount of chert. The dolomite mineral is typically rich in Fe and Mn (up to 3% combined). Its thickness reaches 1450m in the Sterkfontein area (Eriksson and Truswell, 1974). Based on the abundance of chert, the subgroup has been subdivided into 6 formations. The Oaktree Formation (180m thick) represents the basal unit, characterised by its very chertpoor nature. The overlying unit, the Monte Christo Formation (700m thick) is rich in chert. It has thin but spectacular oolitic beds at its base. Sterkfontein Cave straddles the boundary between the two formations, but Lincoln and Fault caves are entirely hosted by the Monte Christo Formation (Fig. 3). Two stratigraphic markers have been identified in the cave. The first one is a tuff seam, up to 30cm thick,

PAGE 5

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.5 Fig. 3 Sections of Sterkfontein Cave. Positions indicated by capital letters on map Fig. 2. Legend: 1) dolostone; 2) chert; 3) calcified old breccia and silt; 4) unconsolidated scree, silt and sand; 5) outline of side passage; 6) flight of stairs; 7) fence or gate.

PAGE 6

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.6 Fig. 4 Simplified map of Sterkfontein Cave (grey, overlays and details omitted) and Lincoln-Fault System (black). 1= Lincoln-Fault link; 2= Lincoln-Sterkfontein link. Map of Lincoln-Fault after Boshoff et al (1990).interstratified in the Oaktree dolostone 8-10m below the top of the formation (Fig. 3 and 6). Macroscopically it appears as a pale greenish grey shale with ghosts of glass shards visible in thin sections. It is widespread throughout the Transvaal Basin and has been successfully used for geochronometric dating (Walraven and Martini, 1995). Another tuff seam, thinner and more sporadically developed, has been indentified a few metres lower down. The second marker is a 4-5m thick chert bed, with minor interstratified dolostone seams, 3-4m above the base of the Monte Christo Formation (Fig. 3). The strata dip about 30 to the NW. The only significant tectonic feature is represented by a long, subvertical, silicified fault running N-S, skirting the cave system to the East (Wilkinson, 1973). Dolerite dykes and sills are numerous in the area; a sill is developed immediately below Sterkfontein, but has not been observed in the cave. Geomorphic setting The cave entrances are located on top of a small hill (1491m) and 45-50m above a broad valley with a stream flowing to the NE. The scenery is hilly rolling country dissected by valleys, situated in the upper reaches of a zone forming the escarpment separating two natural regions: the Highveld (1500-1600m) in the south, from the Bushveld in the North (10001100). These two regions represent surfaces peneplaned by erosion cycles. The Highveld corresponds more or less to the African Surface, which started to develop after a continental uplift during the Cretaceous, whereas the

PAGE 7

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.7 Bushveld coincides in part with the PostAfrican I Surface generated after Early Miocene uplift and south-westwards tilting of the African Surface in the area under consideration (Partridge and Maud, 1987). Due to the long existence of the African Cycle, the African Surface had been well peneplaned and covered with a thick weathering crust. During the development of the PostAfrican I Cycle, the African Surface underwent erosion by river systems flowing both to the southwest (to the Atlantic) and to the north (to the Indian Ocean). On the northern system the erosion was more aggressive, with the development of an escarpment incising the Highveld. The southwestern erosion was more moderate resulting in the formation of a lowrelief landscape not associated with an escarpment as spectacular as in the northern case. This suggests that the renewal of erosion was mainly the result of an increase of the talweg gradient of the African Surface by tilting of the peneplain. This led to widespread degradation of the latter surface. 200km to the west, both surfaces merge into the Kalahari Basin. Here sedimentation dominates over erosion (Fig. 1). About 1km to the north of the cave, the flat bottom of the valley has been incised by a stream down to a depth of 8m (Robinson, 1962: Wilkinson, 1973), leaving gravel terraces on both sides. This entrenchment is possibly due to a recent erosion renewal. Generalities on the karst of the Transvaal basin It is necessary to summarize the nature of the regional karst as it differs from the classic models. One characteristic is the deficiency in surface features like doline, polje, swallowholes and disappearance of the surficial fluvial network. The latter morphology is only observed, although it is not spectacular, on the very flat plateaus of the southwestern quarter of the basin (Martini and Kavalieris 1976, Marker 1980). Elsewhere the dendritic network of streams is always well developed. The residual cover can be very thick, in places exceeding 100m. This is as a result of the large percentage of insoluble impurities in the dolostone (Brink and Partridge, 1965; Brink 1979). In contrast with the deficiency of surface karst morphology, caves are well developed. They are of the hyperphreatic type, forming generally labyrinthic networks of passages which are controlled by joints, or broad flat chambers, the result of the dissolution of chertfree beds sandwiched between cherty dolostone (Martini and Kavalieris 1976). Chambers formed by ceiling breakdown after excessive dissolution are also common. The caves do not seem to form well integrated systems, but rather develop open channels only in zones where the dissolution for some reason was more intense. The latter characteristic is suggested among others by the restricted extension of the cave systems: the most remote places are not more than a few hundreds metres distant from the entrances, whereas the cumulated lengths of all passages may be over 10km. The flow of ground-water through the karst is very slow and where the water-table is reached it forms pools that are apparently static. Perennial underground streams are practically absent. According to measurements of the karst porosity, speleogenesis is maximal just under the water-table, amounting to several per cent, and decreases rapidly deeper (Enslin and Kriel, 1969). Penetrable caves still occur occasionally at greater depth, however, for instance at 79m below the water-table, as observed by exploration after artificial de-watering of aquifers (Moen and Martini, 1996). An important and well documented characteristic of the karst aquifer of the Transvaal Basin is its subdivision into compartments, separated by impervious subvertical dykes of dolerite and syenite, as well as by silicified faults. Two types of detrital cave sediments may be distinguished. In the deep parts, not directly influenced by the surface, sedimentation is minimal and limited to autochtonous material. The reason is that the phreatic flow is too slow to transport particulate material. The sediments consist of chert debris and dark-brown wad pellets detached from the ceiling and the walls, and having accumulated on the floor. The material known as wad is the residue left in a quiet phreatic environment after incongruent dissolution of dolomite, which releases Mn and Fe oxides. During this process, these oxides concentrate on the crystal junctions. As a result, after complete disappearance of the carbonate, these oxides form a micro-boxwork pseudomorphing the dolostone texture (Martini and Kavalieris, 1976). When a cavity "bursts" up to the surface, by ceiling breakdown or

PAGE 8

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.8 Fig. 5 Tourist Exit and bust of Broom contemplating an Australopithecus skull. Fig. 6 25cm thick tuff seam, dipping to the left, base on top hard hat. In Tuff Chamber. suffosion through the residual cover, alluvials are introduced into the caves by occasional floods during storms. Most often, these alluvials consist of chert fragments, reddish silt and sand eroded from surface soils. Description of the cave Sterkfontein is a three-dimensional hyperphreatic maze of fissure passages typical of the karst of the Transvaal basin. Its 25 entrances are the result of the intersection of phreatic channels by the surface. These passages retain the same morphology down to below the water-table, i.e. over an elevation of about 50m (Fig. 2 to 4). The cave system is restricted to a 200x250x50m volume. However, the cumulative passage development of Sterkfontein, including Lincoln Cave, amounts to 4,73km. This figure increases to 5,23km if Fault Cave is added, although it is not yet conventionally linked to the system, as only visual connection was achieved. Mining of calcite did not considerably alter the morphology of the passages, except in the Silberberg Grotto. Here a particularly voluminous stalagmite and a flowstone sheet (the "Boss") have been removed. This suggests

PAGE 9

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.9 Fig. 7 Elephant Chamber. Note long roof pendants. Person to the left gives scale. Fig. 8 Open joint at chert ceiling in Lesser Canyon, Lincoln Cave. Scale in centimetres. that the chamber is mostly man-made. The map indicates that the passages of the whole system follows two main directions of subvertical joints: the dominant one is mostly W to WNW and the subordinate one varies from N to NNE (Fig. 2). The passage morphology is also controlled by the lithology of the country rock. In the chert-poor Oaktree Formation, in which the greater part of the cave is developed, the passages belong dominantly to the fissure type. Here passages reach heights of 15m, while the widths are in the order of a few metres only. Large chambers can form by dissolution of partitions separating swarms of tightly spaced passages. This is typically the case for the Elephant Chamber (Fig. 7), where the remnants of these partitions are left as long roof pendants reminiscent of trunks. Passages are often superimposed, adding more complexity to the map. This splitting into several levels in a same joint is often due to "false" floors composed of rubble and flowstone blockages, rather than undisturbed country rock. Compared to other regional caves, the development by ceiling breakdown is relatively minor. The main voids of this latter type are Terror (alias Jacovec) and Fossil Chambers. In the northern part of Sterkfontein Cave, i.e. in the P. Verhulsel Section, the area to the east of the latter, and in part of the Lincoln-Fault Cave System, the passages display a different morphology. They are still joint-controlled, but as they are developed in the Monte Christo Formation, chert seams up to 30cm thick separate "stacks" of low (0,5-1m) but broad (25m) passages. For instance in the P.Verhulsel Section, up to 4 crawlways levels are in places superposed over a stratigraphic interval of only 3-4m (Fig. 3). These levels often merge into two or more passages by breakdown of the chert floors and develop into "canyons". In this section, and further eastwards, it appears from the map that most passages stop against a same WSW-ENE line marked by pools. The latter represents the position where the 5m thick chert marker dips under the water-table. This

PAGE 10

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.10 prominent chert bed seems to have acted as a barrier, impeding passage development between Sterkfontein Cave and the Lincoln-Fault System. Only in the eastern part of the cave, on account of an unusual, complete breakdown of the chert ceiling, a connection could be established (April 2001 connection). In the Lincoln-Fault Cave System, the splitting of passages by chert layers is still observed, but is less characteristic than in Sterkfontein. The reason for this is that the dolostone is less siliceous. From the survey it appears that the volume of voids remains approximately constant from the surface down to the water-table, that is over an elevation difference of about 50 m. It is also obvious, that the longitudinal axes of the passages, even in the Oa ktree Formation, are generally inclined sub-pa rallel with the apparent dip of the country rock. These observations preclude the presence of preferential dissolution levels. This is not the rule everywhere in the karst of the Transvaal Basin. Cases are known where phreatic mazes strictly developed at specific elevations: for instance in Wonderfontein Cave (Fig. 1), a 9,4km passage network which is restricted to an elevation range of only 3-4m (Kent et al., 1978). The nature of the original tensional joints controlling passage development has been observed in several places on the chert ceilings, for instance in the P.Verhulsel Section. Since chert is practically unaffected by karst dissolution, it is particularly favourable for the preservation of these cracks. The recorded width of these cracks varies from fractions of a millimetre to one centimetre (Fig. 8). The walls of the joints are coated with tiny secondary quartz crystals which have grown before speleogenesis. Open joints in dolostone, which are unaffected by dissolution are rare. Where these have been observed their surfaces are also coated with minute quartz crystals. The latter mineral sometimes impregnates the dolostone over a few millimetres on both sides of the joint. Such open joints have been observed in most other caves in the region and will be the subject of another article. It was proposed that they have formed during the important tensional event associated with the flood basalt volcanism,which affected the entire subcontinent during the Early Jurassic (Kavalieris and Martini, 1976). Fig. 9 Fissure passage, half-flooded at water-table level. Western end of Main Lake. In rare cases, passages are not controlled by subvertical joints, but by moderately inclined fractures, like the eastern part of the long passage linking the tourist route with Ravjee Lake. This fracture is possibly a very minor compression fault, along which the initial cavities developed as primal voids having rhombic cross-sections. Hydrology About 30 static pools have been reported in the cave system, the most important by far being the Main Lake (Fig. 9). With the exception of a small perched pool at the eastern end of the P.Verhulsel Section, they all mark the watertable. The depth of the pools has not been thoroughly investigated, but diving in the Main Lake and in the pools of the P.Verhulsel Section, indicates maxima of not more than 4m. Obviously greater depths have been estimated in Lincoln Cave. The water level of the lakes fluctuates slowly according to the preceding rainfall pattern. At present this is within a range of about 2m. Apparently the water flows to a spring situated 900m to the north of the cave. This spring is situated in the talweg of the broad

PAGE 11

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.11 valley mentioned previously. The position of the resurgence seems to be controlled by the damming effect of the N-S silicified fault, which was mentioned in the geology section and which prevents ground-water to flow eastwards. The spring elevation is very close to that of the Main Lake (Wilkinson, 1973). Against this apparently simple and logic hydrological model, for previous levelling of other pools has indicated great differences in elevation (Wilkinson, 1973). For instance Ravjee Lake would be 4m lower than the Main Lake. The pool in Fault Cave, the most distant, would be 9m lower. According to the present levelling, the respective discrepancy for Ravjee Lake amount to –0.5m. The exploration by diving (Wits Diving Club) also proved that the Main Lake and the pool situated immediately to the north of the P.Verhulsel memorial must be on the same level, since they have been connected by diving (Sefton et al 1985). Careful levelling between the pool at the end of the "Lesser Canyon" in Lincoln Cave, and the pool under the linking chimney in Sterkfontein (April 2001 connection) indicated that the latter is 18cm lower than the former. This difference is probably within the margin of error. Both pools are therefore probably on same the level, although separated by the barrier formed by the chert marker. It seems that the elevation differences between the other pools are not more than decimetric rather than metric. A more accurate survey would be desirable to confirm this conclusion. Speleothems Although the most massive speleothems have been removed by mining, remnants of flowstone or dripstone of calcite or aragonite are still present. The latter mineral has generally reverted to calcite. Evaporite-type speleothems are represented by popcorn and aragonite frost on the walls. Sometimes these are associated with chalky microcrystalline hydromagnesite. These three carbonate minerals are very common in other caves in the Transvaal Basin. The frequency of aragonite is related to the Mgrich nature of the ground-water (ex: Martini and Kavalieris, 1978, Hill and Forti, 1997). Flowstone and stalactites are corroded up to a height of about 6m above the water-table, especially in the vicinity of the Main Lake (Fig. 10). The intensity of this dissolution increases with depth. No speleothems are observed below 2-3m above the lake surface. Some inactive, "old" speleothems in the upper reaches of the cave, for instance in Foss il Chamber, also show signs of re-solution. In these cases the origin of the corrosion is not known. Various factors have been suggested in other caves (Martini and Kavalieris 1976 and 1978): rise of the watertable, dripping solutions undersaturated in calcite, condensation, bat guano and ammonia oxidation by Nitrobacter (Martini and Kavalieris, 1976 and 1978) Fig. 10 Corroded flowstone adherent to the wall at western extremity of Milner Hall, near Main Lake. The main fossiliferous body The fossiliferous body is exposed at surface over an area of 70m and a maximum northsouth width of 25m (Fig. 4). Excavation in this body yielded Upper Pliocene to Pleistocene fauna including hominid remains. It seems to extend at least to a depth of 40m below the surface (Fig. 3). Its elongation is conformable with the orientation of the cave passages surrounding it. The fossiliferous breccia has been opened and stripped of its overburden by

PAGE 12

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.12 extensive diggings, first during calcite mining and later for paleontological purposes. This breccia body is widely accepted as the filling of a cave of which the roof has been largely removed by erosion. It has been extensively studied by a number of authors, mainly from the surface diggings, but also from an exposure within the cave (Silberberg Grotto) and from 5 boreholes drilled for stratigraphic purposes. In general the breccia consists of dolostone, chert and flowstone clasts of variable size and abundance, set in a calcite and loam-silt matrix. Induration affects all lithological types and is due to the development of calcite, which often constitutes more than 50% of the matrix. A stratigraphic succession has been established, which has been termed the Sterkfontein Formation and is subdivided into 6 members (Partridge, 1978; Partridge and Watt, 1991). As these members result from the deposition of cave sediments, extreme irregularity in thickness and rapid lateral facies changes are the rule. From base to top, the lithological units are as follows: Member 1 0-20m thick, consisting of occasionally voluminous blocks set in a darkbrown manganiferous matrix. It also contains stalagmitic and lenticular flowstone masses, and is paleontologically practically barren. This megabreccia represents the conical accumulation of debris from ceiling breakdown at a stage when the cave had no entrance, or only a distant one. It is exposed in the Silberberg Grotto, in the upper parts of the Name Chamber and at the eastern end of the Milner Hall. It was intersected in the 5 boreholes. Member 2 0-8m thick, consisting of stratified pale reddish-brown, silty loam with rare rock debris and flowstone lenses. Bones are locally abundant. The detrital sediments have been introduced from a pit entrance above and deposited by stream action, sometimes in a subaqueous environment. The member is exposed in the Silberberg Grotto (Fig. 11) and has been intersected by 3 of the 5 boreholes. Member 3 5-11m thick. It starts with a flowstone sheet, which is relatively regular and widespread across the breccia body. The flowstone is overlain by reddish-brown, clayey, silty sand with scattered angular rock fragments and in places abundant bones. These sediments represent colluvial material, possibly with an aeolian component, and were introduced from Fig. 11 Base Sterkfontein Fm at the eastern extremity of Silberberg Grotto. Note dolostone to the right with a pocket of wad pellets (black) covered by flowstone (white), representing the very reduced Member 1 at the southern edge of the breccia body. The flowstone is overlain by reddish and brown silt of Member 2. the surface. It is mainly exposed in the Silberberg Grotto, where remnants left after mining can be observed at mainly inaccessible height on the northern wall. At surface the top of this member is visible in the eastern portion of the breccia body. It has been intersected in the 4 boreholes. Member 4 Up to 9m thick. It contains abundant rock debris, reaching large boulder size, in a dominantly calcitic matrix at the base and reddish-brown to yellowish loam in the upper part, with local pockets of loam and calcite lenses containing minor rock fragments at the top. The sediment originates from colluvium introduced by floods, and blocks detached from the ceiling. An increase in 13C in calcite suggests better ventilation of the cave, due to enlargement of the entrance subsequent to more ceiling collapse. The member is mainly exposed at surface and has been intersected by 4 boreholes. This is the main fossiliferous horizon, having produced abundant hominid remnants. Member 5 Up to 5m thick, this unit rests unconformably on Member 4, which was already calcified sin ce the latter has been observed as boulders included in Member 5 (Robinson 1962). It consists of reddish-brown sandy loam with rock debris, occasional bones and stone tools. The mode of deposition is the same as that of Member 4. At surface it is exposed in the western part of the breccia body. Member 6 Up to 1,5m thick. It is comparable with the previous unit, but the matrix is dark

PAGE 13

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.13 reddish-brown. It also contains some bones and stone artifacts, probably introduced by early man. This member forms only a small outlier in the western part of the breccia body. In the surface diggings, Members 3 to 6 display a gentle WNW dip, assumed to represent the sedimentation slope. This suggests a detrital origin, from an entrance towards the east (Robinson 1962, Partridge 1978) and also indicates that this filled paleocave was inclined westwards, like many present-day passages in the cave below. A similar inclination is also suggested by ceiling remnants observed only in the western side of the breccia body. The morphology of the ancient ceiling is typical of an origin by breakdown (Robinson, 1962). As the breccia is calcite-rich, it was subsequently subjected to karst dissolution at the surface. This resulted in the formation of narrow vertical tubular pits ("makondos") and de-calcification associated with subsidence of the residuals into a "swallow hole". The latter contains mainly soft residuals from Member 4 and 5 and was extensively excavated. Extraction of paleontological and archeological material was greatly facilitated by the soft nature of the material (Clark 1994). By digging deeper during the paleontological excavation, the "swallow hole" broke into the Name Chamber (Fig. 1 and 2). Other paleo-fills in the cave Apart from the exposures in the Silberberg Grotto, the Name Chamber and the eastern end of the Milner Hall, which are directly and obviously part of the breccia body (= Sterkfontein Formation) described previously, comparable occurrences of indurated cave fill have been observed in many other places in the Sterkfontein Cave System. Stratified siliciclastics consisting of gravel and silt, plus flowstone layers, are sticking to the side wall and to the ceiling of the southern side of Terror Chamber. In this chamber breccia also fills fissure passages, forming clastic veins exposed in the ceiling. As these occurrences are situated practically under the main breccia body and at the same elevation as the lower part of this body, it is very likely that they represent more distal sediments of the Sterkfontein Formation. It is not possible, however, to speculate which member they might belong to. Three occurrences of cave filling to the west of the Elephant Chamber are located mainly at the ends of passages. Access to this filling is difficult due to obstruction by large boulders, the result of ceiling breakdown. By inspection of the map (Fig. 2) it seems possible that these boulder chokes are part of the same large, E-W elongated zone of collapse, which does not have an obvious surface expression. Calcification is well developed only in the eastern site. Twelve metres to the northwest of "G" (Fig. 2), calcite mining exposed the bottom of an ancient fissure passage (Fig. 12). Here one observes wad pellets (see generalities about the karst of the Transvaal Basin), which were detached from the wall and gently accumulated onto the floor in a quiet phreatic environment. Later, after lowering of the water-table, the pellets were cemented by carbonate in a subaerial environment. This type of fossil sedimentation, contemporaneous with the speleogenesis, has been frequently observed by the authors in other prospecting pits of the Kromdraai area. Fig. 12 Base of filled passage, revealed after calcite mining. Note wad fragments (black) detached from the wall representing phreatic residual, in calcite matrix subsequently deposited in the vadose zone, overlain by flowstone. Note disconformity with dolostone. Pen as scale. In passages to the SW of Elephant Chamber. Ten metres to the northwest of the previous occurrence, in a small maze, remnants of a flowstone floor form ledges on the walls and bridges splitting the passages in two levels. In places relics of breccia in a silty brown matrix are sticking to the underside of the flowstone (Fig. 13). This suggests that the detrital material was friable. It could therefore be drawn down

PAGE 14

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.14 Fig. 13 Flowstone floor (white) covered with some Recent rubble, splitting fissure passage in 2 levels, with remnant of breccia sticking underneath. In maze at the western end of Elephant Chamber. Fig. 14 Re-dissolved calcified silt (Member 1 or 2). Note dolostone to the left, calcified silt to the right and wad band (dark) marking the contact. Above person, note arched ceiling continuous with the dolostone as evidence of re -solution. In maze south of eastern end of Milner Hall. by further cave development at a lower level. The observed remnents are the result of calcification. Moreover it seems that the flowstone marks the end of detrital deposition since it is not overlain by other sediments. Similar occurrences are observed in Fossil Chamber, the Graveyard, and at the northwest side and the eastern end of Milner Hall. An informative occurrence of the previous type is observed in the central part of the maze of fissure passages developed south of the eastern end of Milner Hall, 10m to the SSW of point -27,3 (Fig. 2). Calcified silty sediments containing minor rock fragments which are irregularly covered by a flowstone floor, separate the passage in two levels. At the ceiling of the lower level, it appears that the removal of the calcified silt was not only mechanical, but accompanied by dissolution (Fig. 14). The floor of the upper level above this point represents more or less the top of the cave fill. Beyond a sharp turn to the left, this upper level can be followed eastwards along a short passage (15m) giving access to the lower part of the Silberberg Grotto where an Australopithecus skeleton has been found (Fig. 2). As the slope is even and upwards, this suggests that the sediments merge into Member 1 or 2. This conclusion is further supported, if after returning to the starting point, the upper level is followed to the northeast. After 10m the passage leads to a broad balcony dominating the eastern extension of Milner Hall. The edge of the balcony is made of a 50cm thick flowstone sheet. Under the balcony, it appears that the flowstone sheet rests on coarse breccia, which merges eastwards into the large boulder breccia of Member 1 (Fig. 3 and 3). As no sediments younger than Member 2 entered the upper level, it appears therefore that in the upstream direction, towards the main breccia body, the debris slope reached the ceiling during deposition, thus blocking further detrital influx.

PAGE 15

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.15 This is the "hopper" effect well described by Wilkinson (1974). Practically no typical calcified breccia and finer sediments, as described up to now, have been observed in the P.Verhulsel Section, in the series between Ravjee Lake and Fairy Chamber, and in Lincoln Cave. In Fault Cave, however, an important body of such sediments is visible at the ceiling of Bone Passage (Boshoff, et al 1990). With the exception of the deposits in Terror Chamber and in the maze of fissure passages south of the eastern extremity of Milner Hall, which are likely linked with the Sterkfontein Formation, it seems obvious that the other occurrences reviewed in this section are not connected to the latter formation. A common characteristic of all the calcified breccias, however, is that they do not occur less than 10m above the water-table. This means that the possibility remains open that they are contemporaneous with the main breccia body. Attention was focused on the possibility of rejuvenation of the passage joints into old flowstone and well calcified breccia. In most places no extension of the joints could be observed, or only irregular very thin fractures with no measurable opening. The latter may perhaps be due to blasting. In the upper part of the Lesser Canyon, in Lincoln Cave, an old corroded flowstone, untouched by the miners, completely chokes a passage and was apparently cracked by joint reactivation, producing a 2-3mm wide opening. It is not certain, however, that the fracture is not due to sagging of the chert seam forming the ceiling of a cavity immediately underneath and supporting the flowstone. In conclusion, it seems probable that the joints have not been substantially rejuvenated after the deposition of the Sterkfontein Formation. Post-Sterkfontein Formation detritals Detrital sediments obviously younger than the calcified breccia are represented by unconsolidated and unsorted rubble forming the floor of most passages down to below the watertable. They mostly represent, on the one hand, residual chert plates and wad, and on the other hand, boulders, gravel and silt reworked from older, poorly consolidated breccia described previously. To a variable extent this material has been transported by colluvial creep and by ephemeral streams after storms. A deposit of this type is the steep fan of debris in the northern side of the Name Chamber, which consists of boulders, grav el and silt originating from under the pit linking the cave to the paleontological diggings at surface (Fig. 2 and 3). It is mainly the result of reworking of the decalcified breccia (Clark, 1994). Another debris fan originates from approximately the same spot, but spread into the eastern side of Milner Hall, extending to the west and splitting in two lobes separated by a large block (Fig. 2). In the middle of Milner Hall and the Elephant Chamber, the slope of both lobes progressively flattens into a plateau at 5 to 7m above the Main Lake (see sections in Fig. 3). A comparable terrace-like surface is observed to the southwest of Ravjee Lake, but its edge is only at 4m above the water-table. At this place, however, water lines, well marked on the wall, are visible 6m above the present lake surface. Age of the Sterkfontein Formation The oldest fossiliferous unit, Member 2, has not been extensively worked yet. Nevertheless it yielded a hyena, suggesting a Pliocene age "substantially" older than Member 4, paleontologically documented by Tobias 1979, Partridge et al 1999. In the lower reaches of the Silberberg Grotto, the remarkable discovery of a near complete australopithecine skeleton embedded between flowstone floors in Member 2, is adding more clues (Clarke, 1998). As the fossil is not completely extracted yet, the results are only preliminary. Nevertheless it indicates a species other than Australopithecus africanus which is common in Member 4. Four inclined calcite flowstone layers, two above the skeleton and two below it, have been sampled for paleomagnetic study (Partridge et al., 1999). It was concluded that the Australopithecus was deposited during the Mammoth Reversal and an age of 3,3 My was proposed. This age is a subject of controversy, however, as it was suggested in a recent publication (Berger Lacruz and de Ruiter, 2002), that these sediments might be younger than 3 My. Member 3, although fossiliferous, has not been investigated yet and no date can be proposed. Member 4 has been extensively excavated and produced a rich fauna, including the great majority of the 600 australopthecine remains found at Sterkfontein. An age of 2,6-2,8 My was estimated by comparison with faunal

PAGE 16

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.16 assemblages of East Africa (McKee, 1993). In a recent publication Berger et al (2002) reported a maximum age of 2,5 My for Member 4, also based on faunal correlation. Member 5 yielded remains of Paranthropus robustus and Homo habilis plus stone tools of the Oldowan culture (2,0-1,7 My) in the base and of the Early Acheulean (1,5 My) above (Clarke, 1994). Speleogenesis As accepted by all authors, Sterkfontein Cave formed by dissolution along joints in a phreatic environment, a process leading in places to ceiling breakdown when the cavity reached a certain size. Only the details have to be discussed here. Since the passage sizes remained practically constant over the entire vertical span, it seems that during the secular drop of the water-table the dissolution intensity remained approximately the same at any elevation above the present phreatic zone. Therefore the model of a spasmodically falling water-table (Partridge, 1978) can be accepted only if it occurred in increments small enough to have no sensible effect on the passage morphology. The restricted extent of the easily penetrable passages, forming a dense network, is a characteristic shared by the majority of caves of the Transvaal Basin, but is difficult to explain. It might be due to zones of initial joints wider than usual, but this hypothesis needs to be demonstrated. Another possibility would be that the cave systems developed where deep water wells up and mixes with ordinary ground-water close to surface. This model is suggested by comparison with the caves of northern Namibia, which also show restricted surface extension, but where there is evidence of such upwelling for some of them (Martini and Marais 1996, Martini et al., 1999). The age of the initial speleogenesis is also difficult to estimate. This process is still active today since no Ca-carbonate is deposited at present under the water-table. For the oldest, most elevated passages, like the Silberberg Grotto and the upper part of Lincoln cave, the presence of Member 2 suggests that they might be older than 3,3 My, provided that the age found by Partridge et al (1999) is accepted. Partridge (1973) calculated the age of dewatering of the cave after the time necessary for the nick point of the Post-African I erosion cycle to reach it. He determined an age of 3,26 My, a figure which would fit well with the age of the oldest fossiliferous sediments (Member 2). At that time, however, at least half of the cave was already dry, which suggests that the de-watering and speleogenesis of the top levels must be even older. Up to now the nine cave breccias investigated in the Malmani Subgroup of the Transvaal Basin have not revealed faunas older than Upper Pliocene (ex. Bamford, 1999). The absence of pre-Pliocene ages seems to indicate that older caves favourable for trapping mammals did not form, or had been eroded. Indeed on the karst of the Transvaal Basin, the African Surface has been lowered to variable degrees (Partridge and Maud, 1987). Another possibility to explain the lack of paleocave fillings older than Upper Pliocene, would be that during the African peneplanation most of the dolostone was protected from dissolution by a cover of impervious shale and sandstone of the Ecca Formation (Permian) of the Karoo Basin (Fig. 1).There is evidence that a widespread Permian cover existed over the Malmani Subgroup not far above its present exposures. Remnants of this formation are widely distributed over the lowered African Surface in the southwestern quarter of the Transvaal Basin. In other words, in this area the African Surface nearly co incides with the prePermian erosion surface. On the Malmani dolostone they form relatively large patches indicated on the geological maps, numerous discrete remnants are hidden under red soil and have been discovered by drilling (ex. Wilkins et al 1987), and a few paleocave fillings have been observed in quarries (Marker 1974) and in caves (observations by the authors). Keeping uncertainty in mind, a possible cave development scenario may be proposed. Erosion was renewed after continental uplift and tilting of the African Surface, an event which started about 18 My ago (Partridge and Maud, 1987) and led to reactivation of karst development. After these considerations and the age of the breccia, the beginning of the speleogenesis might date to the end of the Miocene (~5 My). A future possibility to directly date the speleogenesis of Sterkfontein, would be to use the 40Ar/39Ar method on cryptomelane in wad, provided that the latter contains this mineral. Future investigations should thus focus on wad forming delicate micro-boxwork directly

PAGE 17

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.17 released from dolomite dissolution (see previously). About the younger speleogenetic phases, the Sterkfontein Formation is informative, as it indicates that 3,3 My ago the cave was already de-watered at 20-25m above the present watertable. That the secular drop of the water-table was irregular, comprising temporary rises, is evidenced by re-solution of calcified silt and breccia about 12m above the water-table (Fig. 14). These oscillations might be linked to climatic variations or to more local "accidents" in the evolution of the surficial drainage system. Similarly, speleothems are corroded up to 6m above the Main Lake (Fig. 10). This oscillation was perhaps controlled by the resurgence, when at one stage the stream channel might have been choked by alluvium, thus forcing a localised rise of the water-table, as suggested by the 8m gravel terrace. Conclusions and comparisons The mapping conducted during this project provides more data about the configuration of the cave, added new passages and was an opportunity to clarify the details of the local stratigraphy of the Malmani Subgroup. The most significant contribution, however, is a better understanding of the basal part of the Sterkfontein Formation. Indeed at this level the main breccia body, a filled chamber generated by ceiling collapse, partly splits into clastic veins, which are infillings of phreatic fissure passages. These observations also confirmed that the latter formation is not a paleokarst filling independent from the actual cave. This misleading impression has been induced by the general calcification of both breccia and siltstone by dripping water, also depositing stalactites, stalagmites and flowstone floors, thus rendering it resistant to undermining by continuing speleogenesis and vadose erosion. This also explains the absence of soft sediments contemporaneous with the Sterkfontein Formation. It also appeared that the collapse chamber hosting this formation is contemporaneous with the phreatic passages of the upper part of the cave, although they remained open up to now. These parts of the cave were not filled up, because of blockages ("hoppers") preventing introduction of detritals from surface. A model of cave development is presented at Figure 15. Sterkfontein is an example of a cave still open and in formation after probably more than 5 My and having escaped complete removal by surface ablation. One factor in favour of such a long preservation is a slow rate of karst evolution, a condition which is still met today: although the rainfall is about 80cm/y, the groundwater recharge amounts to only 15% due to high evapo-transpiration (Enslin and Kriel, 1967). Old caves which remained open up to present are known elsewhere in the world, like for instance the highest ones in the Guadalupe Mountains, New Mexico. These caves are located in a semi-arid environment and formed by sulphuric speleogenesis, an event dated at 12 My by 40Ar/39Ar method on alunite (Polyac et al 1998). Another factor favouring long preservation is an entrance on the top of a hill, where ablation is weak and where only minimal residual material can enter the cave. In contrast, where the topography is flat, larger volumes of alluvium are engulfed and the caves are rapidly filled up as soon as entrances form by suffosion in residual cover or by rock ceiling collapse. An informative and well dated example is the paleokarst of the Quercy, in southern France (Plissier et al,. 1999, and references therein). This paleokarst reached a mature stage, with a thick soil cover, during the Lower Eocene and was eventually fossilised under Middle Miocene basin sediments. During this 30 My span, karst activity was reduced, but potholes opened periodically and were filled by bone-rich sediments almost instantly, geologically speaking. Indeed the paleontological content of more than 100 of these potholes indicated that each of them contains a geochronologically punctual faunal assemblage, although the overall ages vary from Lower Eocene to Lower Miocene. Where the African Surface developed on the Malmani dolostone, with regard to its protracted existence, a similar mature karst might have formed, but would probably have been eroded in the Sterkfontein area. Only 200250km to the west of the cave, where sedimentation supersedes erosion, paleokarst deposits related to the African Surface might be discovered. They should yield Upper Cretaceous to Miocene faunas and thus indirectly contribute to a better understanding of the evolution of the karst of the Kromdraai area, including Sterkfontein Cave.

PAGE 18

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.18 Fig. 15 Idealised speleogenetic model of Sterkfontein Cave Passages shown as cross-sections, phreatic residuals and surface soils not marked Elevation of paleo-surfaces assume d after ablation rate of ~5m / My (Gams 1989). Legend : 1) indurated detrital fill from ceiling breakdow n and reworking from surficial soils; 2) cf 1, but unconsolidated; 3) flowstone; 4) Australopithecus skeleton; 5) water-table. Evolution phases : 1) at about 5 My or older, formation of cave in phreatic environment with no penetrable entrance; note dissolution, roof collapse, blocks accumulation; 2) at 3,3 My, after water-table drop a nd opening to surface, introduction of detritals and animal remains; note formation of speleothems due to better ventilation, and calcification of detritals above water-table (deposition of Member 2 and basal Member 3); 3 ) at about 2 My, after deposition of Members 3 and 4, due to continuing drop of water-table, dissolution and re working of older calcified breccia, often left as bridges and relic le dges; 4) actual setting; de-roofing of breccia body, karstification and decalcification of breccia. Co mments for letters A to F : A) sedimentation in Name Chamber; reworking of loose detritals, as a result of continuous speleogenesis and late r of suffosion in de-calcifi ed breccia; B) deposition of Member 2 and 3, followed by re-solution and mechani cal reworking (cf. eastern part of Milner Hall); termination of sedimentation due to "hopper" effect; C) filling of Terro r Chamber by detritals arbitrarily attributed to Member 3, then after induration, continuous speleogenesis by solution-collapse leaving "clastic dykes" exposed at ceiling; D) in M ilner Hall and Elephant Chamber, recen t sedimentation mainly originating from reworking of older detritals; form ation of "alluvial terrace"; E) old phreatic passage, unfilled, intersected by surface; F) formation of the "Swallow Ho le" by subsidence of decalcified breccia. Acknowledgments The authors are indebted to Dr R.J.Clark of the University of the Witwatersrand, who gave them permission to conduct the investigations in the cave. During the survey and exploration they have been assisted by C.Menter, A. and O.Wipplinger, who were often instrumental in climbing and squeezing through narrow holes.

PAGE 19

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.19 References Bamford M. 1999. Pliocene fossil wood from an early hominid cave deposit, Sterkfontein, South Africa. South African J. Sc. 95 231-237. Berger L.R., Lacruz R., and de Ruiter D.J. 2002. Brief communication: Revised age estimates of Australopithecus -bearing deposits at Sterkfontein, South Africa. Amer. J. phys. anthropol. 119 192-197. Boshoff P., Durrheim R., Fraser A., Martini J.E.J., Ralph A. and Sefton M. 1990. The Lincoln-Fault Cave System. South African Spelaeo. Ascn. 31 60-64. Brain C.K. 1981. The hunters or the hunted? Chicago: University of Chicago Press, 365 p. Brink A.B.A. 1979. Engineering geology of Southern Africa. Vol. 1 Building Publications, Pretoria, 319 p. Brink A.B.A. and Partridge T.C. 1965. Transvaal karst: some considerati ons of development and morphology, with special reference to sinkholes and subsidences on the Far West Rand. South African Geogr. J. 47 ,11-34. Broom R. 1936. A new fossil anthropoid skull from South Africa. Nature 138 486-488. Button A. 1973. The stratigraphic history of the Malmani Dolomite in the eastern and northeastern Transvaal. Trans. Geol. Soc. South Africa 76 229-247. Clarke R.J. 1994. On some new interpretations of Sterkfontein stratigraphy. South African J. Sc. 90 211-214. Clarke R.J. 1998. First ever discovery of a wellpreserved skull and associated skeleton of Australopithecus. South African J.Sc. 94 460463. Draper D. 1896. Report of meeting, 8th April 1895. Geological Society of South Africa. Trans. Geol. Soc. South Africa 1 p1. Draper D. 1898. Meeting of the geological Society of South Africa, 12th July 1897. Trans. geol. Soc. South Africa 3 p 63. Enslin J.F. and Kriel J.P. 1967. The assessment and possible future uses of the dolomite ground water resources of the Far West Rand, Transvaal, South Africa. Water for peace conference, Washington 1967 12 p. Eriksson K.A. and Truswell J.F. 1974. Stratotypes from the Malmani Subgroup north-west of Johannesburg, South Africa. Trans. Geol. Soc. South Africa 77 211-222. Gams I. 1989. International measurements of solution by means of limestone tablets. In: Kosa A. (Ed). Proc. Xth Internat. Speleological Congress. 2 Budapest, 473-475. Hill C. and Forti P. 1997. Cave minerals of the world Huntsville (Ala): Nat. speleo. Soc., 463p. Kavalieris I. and Martini J. 1976. Structural control of some Western Transvaal caves. South African J. Sc 72 308-309. Kent L.E., Kavalieris I., Martini J. and Hugo P.L.V. 1978. Wonderfontein Cave. Annals Geol. Surv. South Africa 11 303-308. Marker M.E. 1974. A note on the occurrence of Karoo sediments near Pretoria and its relevance to the dating of karst weathering. Trans. Geol. Soc. South Africa 77 69-70. Marker M.E. 1980. A systems model for karst development with relevance for Southern Africa. South African Geogr. J. 62 151-163. Martini J. and Kavalieris I. 1976. The karst of the Transvaal (South Africa). Intern. J. Speleology. 8 229-251. Martini J. and Kavalieris I. 1978. Mineralogy of the Transvaal caves. Trans. Geol. Soc. South Africa 81 47-54. Martini J.E.J. and Marais J.C.E. 1996. Grottes hydrothermales dans le Nord-Ouest de la Namibie. Karstologia 28 (2), 13-18. Martini J.E.J., Marais J.C.E. and Irish J. 1999. Contribution l'tude du karst et grottes du Kaokoland (Namibie). Karstologia 34 (2), 1-8. McKee J.K. 1993. Faunal dating of the Taung hominid fossil deposit. J. Human Evolution 25 363-376. Moen H.F.G. and Martini J.E.J. 1996. The exploration of Swartgat Sinkhole. South African Spelaeo. Ascn.Bull. 36 ,45-49. Partridge T.C. 1973. Geomorphological dating of cave opening at Makapansgat, Sterkfontein, Swartkrans and Taung. Nature 246 75-79. Partridge T.C. 1978. Re-appraisal of lithostratigraphy of Sterkfontein hominid site. Nature 275 282-286. Partridge T.C. and Maud R.R. 1987. Geomorphic evolution of Southern Africa since the Mesozoic. South African J. Geol. 90 179-208. Partridge T.C., Shaw J., Heslop D. and Clarke R.J. 1999. The new hominid skeleton from Sterkfontein, South Africa: age and preliminary assessment. J. Quarternary Sc. 14 293-298.

PAGE 20

J.E.J.Martini et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.20 Partridge T.C. and Watt I.B. 1991. The stratigraphy of the Sterkfontein hominid deposit and its relationship to the underground cave system. Paleont. Africana 28 35-40. Plissi T., Vianey-Liaud M., Marandat B, Aymard K., Cres G. et Gaffard K. 1999. Les phosphatires du Quercy. Spelunca 73, 23-38. Polyak V.J., McIntosh W.C., Gven N. and Provincio P. 1998. Age and origin of Carlsbad Cavern and related caves of the Guadalupe Mountains based on 40Ar/39Ar dating on alunite. Science 279 1919-1922. Robinson J.T. 1962. Australopithecines and artefacts at Sterkfontein, Part I. Sterkfontein stratigraphy and the significance of the extension site. South African Archaeological Bull. 17 87107. Sefton M., Martini J. and Ellis R. 1985. A report on the events surrounding the search for Mr P.Verhulsel at Sterkfontein Tourist Cave: the involvement of the South African Spelaeological Association. 29th September 1984 to 6th January 1985 South African Spelaeo. Ascn. Bull. 26 3256. Tobias P.V. 1979. The Silberberg Grotto, Sterkfontein, Transvaal, and its importance in palaeoanthropological researches. South African J. Sc. 75 161-164. Wilkins C.B., Eriksson P.G. and Van Schalkwyk A. 1987. Two generations of karst fill sedimentary rocks within Chuniespoort Group dolomites south of Pretoria. South African J. Geol. 90 155167. Walraven F. and Martini J. 1995. Zircon Pbevaporation age determination of the Oaktree Formation, Chuniespoort Group, Transvaal Sequence: implications for Transvaal-Griqualand West basin correlations. South African J. Geol. 98 58-67. Wilkinson M.J. 1973. Sterkfontein Cave System: evolution of a karst form. Unpublished M.A. thesis, University of the Witwatersrand, Johannesburg.



PAGE 1

Speleogenesis and Evolution of Karst Aquifers The Virtual Scientific Journal www.speleogenesis.info Modeling the evolution of karst aquifers and speleogenesis. The step from 1-dimensional to 2-dimensional modeling domains D. Romanov (1), F. Gabrovsek (2) and W. Dreybrodt (3) (1) Karst Processes Research Group, Institute of Ex perimental Physics, University of Bremen, 28359 Bremen, Germany. E-mail: dr omanov@physik.uni-bremen.de (2) Karst Research Institute ZRC SAZU, Postoj na, Slovenia. E-mail: gabrovsek@zrc-sazu.si (3) Karst Processes Research Group, Institute of Experi mental Physics, Univers ity of Bremen, 28359 Bremen, Germany E-mail: dreybrod@physik.uni-bremen.de Abstract First models of karst evolution consider ed a single isolated fracture with no lo ss of flow along its entire length. Under conditions of constant head dissolution of limestone creates a positive feedback-loop of increase of aperture widths and flow until at breakthrough the flow and aperture width are enhanced dramatically. If a second dimension is added to this model domain, in the simplest case by an exit-tube connected to the isolated channel, water loss from the isolated channel occurs. We have investigated the influence of the water loss on the br eakthrough time of the single channe l. In all cases, when water loss is present, more aggressive solution enters at the input. The aggressive solu tional activity penetrates deeper along the conduit. Therefore dissolutional widening at the exit is enhan ced and breakthrough times are re duced. This is discussed in detail by investigating the profiles of hydraulic head, flow rates, aperture widths, and cal cium concentrations along the conduit as they evolve in time and comparing them to those of the isolated 1-D conduit. In a further step the 1-D conduit is embedded into a net of fr actures with smaller aperture widths. The conduit is located in the center of the rectangular domain and connected to the 2-D net at equally spaced nodes. By this way exchange flow from the conduit into the net can arise. But also flow from the net to the conduit is possible. We have studied the evolution of this aquifer considering dissolution also in the network of the narrow fissures. Flow from the main central fracture into th e net again reduces breakthrough times. After breakthrough, however, a complex exit fan e volves in the net, which later on is overprinted by a net of entrance fans propagating down flow. Th ese fans are related to flow from the net into the central fracture. The evolution of these fans resulting finally in a maze-like structure is significant for high hydraulic gradients (i 0.1) as they exist at artificial dam site s. For such situations realistic modeling has to include dissolutional widening in the net. For low hydraulic gradients, i<0.03, the evolution in the ne t is slow compared to that of the central conduit and therefor e the aquifer is dominated by the evolution of the central fracture. Keywords: Limestone, Karst aquifers, Sp eleogenesis, Modeling, Conduit evolution. 1. Introduction The evolution of karst terrains is governed by many factors, such as climate, geological settings, the location and geometry of the catchment areas, the type of the soluble rock, and the distribution of primary fractures in this rock. Karst has been a subject of extensive research since centuries. The articles of Shaw (2000), Lowe (2000), and White (2000) are interesting reviews about the development of the speleogenetic studies from ancient times to present days. People are interested in karst evolution not only because of the beauty of the karst landforms, but also because of their practical importance. Karst aquifers (rock bodies sufficiently permeable to transmit groundwater (Bear and Veruijt, 1987) are the main source of drinking water for about 25% of the world population (Ford and Williams, 1989). Sinkholes, sinking streams, closed depressions and caves characterize the topography of karst terrains. All these different landforms have a common element – the well developed subsurface drainage system. Initially, when the hydraulic conductivity of the soluble rock is small, only a small amount of water can enter. The initial aperture widths of the fractures are in the range of several 100 m # Therefore the flow through them is laminar. CO2 containing water is an aggressive solution capable to dissolve limestone. When it flows through the fissures in the soluble rock their aperture widths increase due to dissolutional widening. Some of the fissures widen faster than others. The flow through these increases and consequently the rate of their widening becomes higher. This positive feedback loop is

PAGE 2

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.2 the reason for the development of secondary porosity and consequently for the development of a complex, extremely heterogeneous aquifer. Flow through some of the widened fractures finally becomes turbulent. Until this time the hydraulic conductivity of the karst aquifer has increased by orders of magnitude. Nowadays, we have the basic knowledge about the properties of karst aquifers, about the hydrodynamics of flow through them, and about the dissolution kinetics of the soluble rocks. Together with the computational power, this enables us to build numerical models of karst aquifers and study them. Dreybrodt (1988, 1990, 1996) and Palmer (1988, 1991) have presented the first numerical models. They are constructed on the basic principles of groundwater chemistry and hydrology, and study the evolution of an isolated onedimensional conduit under various boundary conditions. Using the information about the evolution of a single fracture, we are able to build and understand more complex two-dimensional models. Lauritzen et all. (1992), Groves and Howard (1994), and Howard and Groves (1995) presented models for the evolution of twodimensional networks. Siemers and Dreybrodt (1998), Siemers (1998), and Dreybrodt and Siemers (2000) have investigated the evolution of two dimensional perc olation networks under various lithological and hydraulic conditions. They extended their studies for cases of practical interest, namely the karstification in the vicinity of large hydraulic structures. The karstification below dam sites is investigated by Dreybrodt et al. (2002) and Romanov et al (2003). Clemens et al (1997a; 1997b; 1996), and Bauer (2002) reported on a double porosity model. They couple the large conduit flow with the flow in the surrounding continuum of narrow fissures and calculate the evolution of the conduits. Kaufmann and Brown (1999, 2000) and Kaufmann (2003) used a similar approach. In their model, however, prominent conduits are incorporated into the continuum. Gabrovsek (2000), Gabrovsek and Dreybrodt (2000a, 2000b) have studied the evolution of a single fracture and two dimensional percolation networks under various chemical and hydrological boundary conditions. Together with the numerical results, several analytical estimations for the breakthrough time are given. Gabrovsek and Dreybrodt (2001) have put forward also a model for the evolution of an unconfined aquifer. Any of these different modeling approaches has its advantages and disadvantages. An important result is that all of them give similar results for basic scenarios, specially designed for comparison. The goal of this paper is to discuss the transition from one-dimensional (single fracture) models to two-dimensional (fracture networks) ones, and the new mechanisms arising. Therefore we start with a short overview of the evolution of a single fracture under a constant head boundary condition. We compare these results with the evolution of the same fracture, but embedded into a twodimensional network. 2. Theoretical background The basic element of our models is the single conduit. We use this block to create complex systems and model the processes, which govern the evolution of natural karst aquifers. A single conduit is presented on Fig. 1. It has a rectangular shape, but it could be a cylindrical tube or have any kind of characteristic geometry. The aperture width a0, the width b0, and the length L characterize the rectangular fractures. Fig. 1. Single rectangular conduit, characterized by its length L width b0, and aperture width a0. It is possible also to model the evolution of cylindrical conduits (tubes), but for the rest of this work we will deal only with rectangular fractures. To model the evolution of the fracture in time, we need to know: a) The hydrological laws – governing the flow through the fracture; b) The chemical laws – governing the change of the calcium concentration along the fracture caused by dissolution. Details about these elements are given in the literature: (Dreybrodt, 1988; Dreybrodt, 1996; Siemers and Dreybrodt, 1998; Gabrovsek, 2000).

PAGE 3

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.3 Fig. 2. Mass conservation in the part of the fracture between x and x x $ % To calculate the widening rate F(x) we need to know the concentration c(x) along the fracture. Fig. 2 represents part of a fracture between x and x+$x where x is the distance from the entrance. We use conservation of mass and obtain the following equation: dc Q dc x A x v dx x P x c F & & ) ( ) ( ) ( )) ( ( (1) where A(x) is the cross-sectional area at x P(x) the perimeter v is the velocity of the fluid and Q is the flow rate. The solution of this equation for plane parallel walls is given by (Dreybrodt, 1996; Dreybrodt and Gabrovsek, 2000; Gabrovsek, 2000) (2): ( ( ( ) ( ( ( + , / / 0 1 2 % , / / 0 1 2 3 , / / 0 1 2 , / / 0 1 2 '2 s n n n s n eq s n n s eq inx x x x c c k x F x x x c c k x F x F ; 1 1 ) ( ; exp 1 ) ( ) (1 1 1 14 4 where cin is the concentration at the input of the fracture and: ) 1 ( 11 1 12 & & , / / 0 1 2 & & & & '2n k P c c c Q k P c Qn n eq s eq n eq4 4 (3) P is the perimeter, k1 and kn are kinetic constants, ceq is the equilibrium concentration with respect to calcite, n is the kinetic order, and xs is the position of the switch between the linear and the non-linear rate law (see Eisenlohr et all. 1999). Note the importance of the nonlinear part for the karstification. If the rate law is fully linear then the rates decrease exponentially along the fracture. By this way the exit of the fracture remains practically unaffected. On the other hand, because of the non-linear rate law, dissolution is active along the whole length of the fracture. By this way its exit part is widened, and flow increases. Consequently the dissolution rates increase, the exit is widened faster, and the flow rises faster in time. This positive feedback loop leads to the breakthrough event. The moment of this event is called breakthrough time. The flow increases by several orders of magnitude at this moment, and the concentration at the exit becomes practically equal to cin. After breakthrough the fracture continues to widen evenly along its entire length. It is not possible to give an analytical expression for the evolution of the fracture width with time. But if we assume that the rates at the exit are active along the entire length of the fracture, then its walls remain parallel during the entire evolution, a reasonable approximation by an analytical solution can be obtained. The widening along the conduit, is given by the equation: ) ( 2 t L F dt da & '5 (4), where (5): n n t n n eq s nt a a L c c k t L F2 '( 6 ( 7 8 ( ) ( + % & & , / / 0 1 2 '1 3 0 3 01 ) ( 1 ) (4 a0 is the initial aperture width, 5 converts the dissolution rates from mol & cm-2& s-1 to retreat of bedrock in cm & year-1. 5 is 1.17 & 109 for limestone. For details about solving the equation see (Gabrovsek, 2000). The result is: 1 2 1 01 ) (% 2, / / 0 1 2 'n n BT t a t a (6). The breakthrough time TB is: ) 0 ( 1 2 1 2 10L F a n n TB& % 2 & '5 (7). If we insert Eq. 5, we obtain an upper limit of the breakthrough time TB, and also its dependence on the basic parameters determining karstification (Dreybrodt, 1996; Dreybrodt and Gabrovsek, 2000; Gabrovsek, 2000) (8): 9: 1 1 1 2 1 1 2 0) 1 ( 24 1 1 2 1 2 12 2 2 %, / / 0 1 2 , / / 0 1 & % 2 & 'n n n n eq n n Bk ghc n L a n n T; < 5

PAGE 4

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.4 where ; is the density of water, < its dynamic viscosity. g is the EarthÂ’s acceleration. To obtain exact results one must take into account that the shape of the fractures shows a funnel like profile. Details on this numerical procedure are given by Gabrovsek (2000), Dreybrodt (1996). We use the single conduit to build a 2D network. Fig. 3 depicts an example of a rectangular two-dimensional fracture network. We are able to apply different hydrological and chemical parameters to every single fracture. Varying the spacing ( s ) between the fractures and their initial aperture widths we are able to model different hydraulic conductivities in the domain. The relation between the hydraulic conductivity and the parameters of the fractures is (Lee and Farmer, 1993): s a g K3 012< ;' (9), where s is the spacing between the fissures. A detailed description of the algorithms used to create 2D networks is given by Dreybrodt and Gabrovsek (2002). Fig. 3. Two-dimensional rectangular network, and junction of four fractures. 3. Evolution of a single fracture A single or isolated fracture is one, which cannot gain or loose flow along its length. The only way for it to exchange fluid with other fractures or a porous media is through its entrance and exit. Fig. 4a depicts percolating pathways embedded into a continuous and impervious block of soluble rock. It is obvious, that every straight part of any of the pathways can be defined as a single conduit with an entrance, which is the exit of the preceding fracture. But in the context of the definition above, the whole pathway 1 can be defined as a single conduit. Its length is equal to the sum of the lengths of all straight fractures, which form it. For the pathways 2 and 3 this is not true. They are connected at the points A, B and C, and therefore they can exchange flow. This has a significant influence on their evolution. Imagine now, that the block is not impervious, but has a structure of fine fractures Fig. 4b. Of course any of them has its own width and length, but for simplicity we assume that they are equal. Now the percolating pathways are part of a network. What makes them different from the surrounding fine fractures is their initial aperture width. Pathway 1 cannot be accepted as an isolated conduit anymore. The question is: How can the fine network influence the evolution of the percolating pathways? Fig. 4. Pathways percolating through a block of soluble rock: a) The block is impervious. Pathway 1 cannot exchange flow with Pathways 2 and 3; b) The block carries a system of fine fractures (aperture widths a0). The percolation pathway 1 (aperture width A0) is able to exchange flow with Pathways 2 and 3. We start with the evolution of the single fracture shown on Fig. 5. The input Node is Node 0, and the output one is Node 2. We need one more node for the exchange of flow somewhere in fracture (later in this discussion). Therefore Node 1 is introduced. It splits the initial conduit into two parts with the same initial aperture widths a0=a1=0.03 cm If no flow is exchanged through Node 1, the system

PAGE 5

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.5 remains unaffected. Therefore we threat these two fractures as one with a length equal to the sum of both lengths. The parameters and the boundary conditions for this model are listed in Table 1. We introduce a parameter 1 0 3 3 k. Varying its value, we are able to move the exchange node along the fracture. The length of Fracture 0 is then L k L & 0. The length of Fracture 1 is L k L ) 1 (12 '. The total length L=L0+L1 is kept constant and only the position of the Node 1 is varied. The electrical equivalent of the hydrological set up from Fig. 5a is depicted in Fig. 5b. The resistors R0 and R1 have the same numerical value as the hydraulic resistances of Fractures 0 and 1. The hydraulic resistance of a fracture is given by (Beek and Muttzall, 1975): ) ( ) ( ) ( 12 ) (3t M t b t a L g t R; < (10), M is a geometrical factor defined as b a M / 3 0 6 0 2 ellipsoidal shape and b a M / 6 0 1 2 rectangular shape. Fig. 6 depicts the numerical results for the evolution of the single fracture from Fig. 5. It shows the profiles of the aperture widths (Fig. 6a), the pressure head distribution (Fig. 6b), the concentration profiles (Fig. 6c), and the flow rates (Fig. 6d). The colors represent time, depicted in years. There is no flow out through Node 1. Therefore the flow rates through Fracture 0 and Fracture 1 are equal (Fig. 6d). They are limited by the constriction at the exit of Fracture 1 (see Fig. 6a). The pressure head at t=0 is decreasing linearly along the conduit and then is changing with the change of the aperture width. The main head loss per length is at the tip of the widened zone (see Fig. 6b). The concentration profiles are also typical for the single fracture case (see Fig. 6c). Because of the slow growth of the exit part of Fracture 1, the flow rate remains low for most of the evolution time. The solution entering the fracture is attaining high Ca concentr ations. Shortly before breakthrough, when the rate of widening of the exit is higher, the fresh solution starts to penetrate deeper into the channel. After breakthrough (at 781 years), the concentration becomes low along the entire length of the fracture (Fig. 6c) and therefore widening is even along the entire length. Our next step is to make the one-dimensional model from Fig. 5 – two-dimensional. We use the system depicted in Fig. 7. There is a third fracture starting at Node 1. By this way flow is exchanged through it. The system Fracture 0 + Fracture 1 cannot be regarded as a single fracture anymore. Varying the position of Node 1 and the parameters of Fracture 2 we are able to study the effect on the evolution of the system from Fig. 5. In this paper we concentrate mainly on the changes in the hydrology. Therefore we use a configuration like the one shown on Fig. 7 – one input and two output TABLE 1. Hydrological parameters Value Units Length of the fracture (Node 0 to Node 2) – L=L0+L1 742.5 Meters Number of horizontal sub fractures 2 Head on the input side 100 Meters Head on the output side 0 Meters Initial aperture width of the single fracture a0=a1 0.03 Centimeters Initial width of the fractures – b0 1 Meter Chemical parameters Ca concentration of the inflowing water 0 mol/cm3 Equilibrium concentration with respect to Ca 2e-6 mol/cm3 Temperature of the water 10 = C Switch concentration cs 1.8 10-6 mol/cm3 Diffusion constant D 10-5 cm2/s Nonlinear kinetics order n 4 First order kinetics rate constant – k1 4.10-11 mol cm-2 s-1 Fourth order kinetics rate constant – k4 4.10-8 mol cm-2 s-1

PAGE 6

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.6 Fig. 5. Basic settings for the single fracture: a) Single fracture consisting of tw o isolated fractures connected at Node 1. (Fracture 0 – starting at Node 0 and ending at N ode 1, and Fracture 1 – starting at Node 1 and ending at Node 2). The basic parameters are: H=100m, h2=0m, L=742.5m, a0=a1=0.03 cm The basic relations are shown in the figure; b) Electrical circuit e quivalent to the fracture system of Fig. 5a. The hydraulic resistances of the fractures are replaced with resi stors. The potential difference U between the nodes represents the pressure head differences, and the current – I the flow rate through the fractures. The basic relations are shown in the figure. Fig. 6. Evolution of the aperture widths, hydraulic head, concentration and flow rate for the scenario of a single fracture with length L a) Evolution of the aperture width. The behavior is typical for the single fracture – fast widening close to the entrance and much slower opening of the exit zone; b) Evolution of the pressure head distribution. The highest gradients are at the tip of the opened zone; c) Concentration along the fracture in units of the equilibrium concentration ceq. The concentration is low only in a relatively small area close to the entrance, and after the end of the widened part is close to saturation; d) Evolution of the flow rate. This is a typical breakthrough curve of a single conduit. The flow rate is increasing slowly during most of the evolution time and then jumping with several orders of magnitude at breakthrough time. The color code is the same for a), b) and c) and is depicted in Fig. 6c. BT=781 years.fractures. Furthermore we simplify the problem using boundary conditions, which allow only flow out from the system (Fracture 0+Fracture 1) through Node 1. This is the most simple setup to study. The parameters of Fracture 0 and Fracture 1 are the same as those from Fig. 5a. The settings of the third fracture – Fracture 2, are varied. Therefore we need to introduce two more parameters. These are > 3 3 m 0 and > 3 3 f 0. The first one is used to define by m&L the length of Fracture 2 L m L & '2. The second one is used to relate its initial aperture width a2 to a0 0 2a f a & The boundary conditions are as follows:

PAGE 7

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.7 Fig. 7. Basic settings for the system of three fractures us ed to study the influence of exchange flow on its evolution. a) Three single fractures (Fracture 0 – from Node 0 to Node 1, Fracture 1 – from Node 1 to Node 2, and Fracture 2 – from Node 1 to Node 3) are connected at Node 1. The basic parameters are: H=100m, h2=h3=0m, a0=a1=0.03cm, L=742.5m, b0=1m The different colors of the arrows represent the difference in the Ca concentrations of the inflowing and the outflowing solution; b) Electrical circuit equivalent to th e fracture system of Fig. 7a. The hydr aulic resistances of the fractures are replaced with resistors. The potential difference U between the nodes represents the pressure head differences, and the current – I the flow rate through the fractures. The basic relations are shown on the figure. 1) Constant head H=100 m at the entrance node of fracture 0 (node 0); 2) Constant head h2= h3=0 m at the exit nodes of fractures 1 and 2 (nodes 2, 3); 3) The initial aperture widths of fractures 0 and 1 are the same, a0= a1=0.03 cm The initial aperture width a2 of Fracture 2 is defined as 1 2a f a & ', where 0 f. Dissolution is neglected along Fracture 2 and therefore the value of a2 remains unchanged until the end of the simulations – a2(t)=const ; 4) All fractures have the same width b0=b1=b2; 5) The lengths of the fractures are: Fracture 0 – L k L & '0, where 1 0L L L % ', 0
PAGE 8

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.8 (Fracture 0+Fracture 1). One can easily see that the situation at Node 1 will change with the evolution of Fracture 0 and Fracture 1. It is very difficult to obtain an analytical solution for the breakthrough time of the system Fracture 0 + Fracture 1. Therefore we will solve the problem numerically. Varying the values of f m and k we study the effect of the position of the exchange node, and the amount of the exchanged flow, on the evolution of the single fracture from Fig. 5. 3.1. Influence of exchange flow We start with the variation of f The values for m and k are fixed ( m = k =1/2). The value of f is varied in the range from f almost equal to zero (this scenario is similar to the one presented by Fig. 5), to > @ f. For f =0.33 Fig. 8 depicts the profiles for the aperture width (Fig. 8a), pressure head (Fig. 8b), concentration with respect to ceq (Fig 8c), and the evolution of the flow rate (Fig. 8d). The breakthrough time is only 20 years lower than the one for isolated case from Fig. 5. Profiles are very similar to those of the isolated fracture in Fig. 5. There is only one small detail, which is actually the reason for the reduction of the breakthrough time (see Fig. 8d). The flow rate through Fracture 1 is a little bit lower than the one through Fracture 0. The reason for this is the presence of Fracture 2. Its resistance is so high that it cannot significantly influence the evolution of the system. But even in this case we observe a reduction in the breakthrough time. We expect that this effect will become stronger with increasing values of f Fig. 8. Evolution of the profiles for a scenario with f=0.33, k=1/2, m=1/2 a) Evolution of the aperture width. The behaviour is similar to the one of the isolated case (Fig. 5) – fast widening close to the entrance and much slower opening of the exit zone; b) Evolution of the pressure head distribution. The highest gradients are at the tip of the opened zone; c) Concentration along the fracture in un its of the equilibrium concentration ceq. The concentration is low only in a relatively small area close to the entrance, and after the end of the widened part is close to saturation; d) Evolution of the flow rate. Again a typical breakthrough curve of a single conduit is obtained. The flow rate increases slowly during most of the evolution time and then jumping with several orders of magnitude at breakthrough time. The flow through Fracture 0 and Fracture 1 is almost equal during the entire evolution. The color code is the same for a), b) and c) and is depicted on Fig. 8c. BT=761 years.

PAGE 9

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.9 Fig. 9. Evolution of the profiles for a scenario with f=1, k=1/2, m=1/2 a) Evolution of the aperture width. The zone of fast widening has penetrated deeper along the fracture with respect to the isolated case; b) Evolution of the pressure head distribution. There is a jump depicting the decreas e of the pressure head at Node 1; c) Concentration along the fracture in un its of the equilibrium concentration ceq. The zone of low concentration is penetrated deeper with the respect to the isolated case; d) Evolution of the flow rate. The fl ow through the different fractures is not equal. In the initial moment the flow rate through Fractures 1 and 2 is equal. Afterwards with the widening of Fracture 1, the flow rate through Fracture 1 increases faster. The flow rate through Fractur e 2 is increasing relatively fast until, because of the widening of Fracture 0, the pressure head at the Node 1 becomes close to the maximal one at the entrance. From this moment on the flow rate through fracture 2 remains almost constant. The color code is the same for a), b) and c) and is depicted on Fig. 9c. BT=555 years. We increase the value of f to f=1 This is a symmetrical structure, because the initial values of the aperture widths are the same for all fractures, and their lengths also. Of course, this is true only at the beginning of the simulation. Later on, because of the dissolution along Fracture 0 and Fracture 1, their resistances decrease, while that of Fracture 2 remains constant. With respect to the cases already discussed the widening of Fracture 0 is more effective (Fig. 9a). The opening zone progresses faster towards the exit. The differences in the pressure head distribution are even easier to note. There is a clearly visible change at Node 1 (see Fig. 9b). Due to the presence of Fracture 2 the gradient increases along Fracture 0. The resistance of Fracture 2 is already sufficiently low and the flow rate through it is comparable to the flow rate through Fracture 1 (Fig. 9d). The profiles of the concentrations are depicted in Fig. 9c. Because of the deeper penetration of the widened zone (see Fig. 9a) and the higher flow rate through Fracture 0, the zone of lower concentration has penetrated deeper along the fracture. The reduction of breakthrough time for this scenario is already more than 300 years. As we have already seen the influence of Fracture 2 is quite important. Until this moment we have worked with values of fA1 In other words, the initial aperture width of Fracture 2 was always smaller or equal to the one of the single fracture. Our next set of experiments will study the behavior of the system at high values of f

PAGE 10

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.10 Fig. 10. Evolution of the profiles for a scenario with f=4.33, k=1/2, m=1/2 a) Evolution of the aperture width. Fracture 0 grows independently of Fracture 1 until the moment of breakthrough at 90 years. Fracture 1 remains unchanged during this period. After the breakthrough event of Fracture 0, the development of Fracture 1 starts under si milar conditions – highest po ssible hydraulic gradient, and low input concentration; b) Evolution of the pressure head distribution. The pressure head distribution resembles the evolution of two fractures – shifted in time. The second one starts to evolve after the jump in the value of the head at Node 1 – 90 years; c) Concentration along the fracture with re spect to the equilibrium concentration ceq. The evolution of the concentration profiles also shows two independently evolving fractures; d) Evolution of the flow rate. Becau se of the high initial aperture width of fracture 2, the initial flow through Fracture 0 is also high. The contribution from the flow through Fracture 1 is negligible. After the breakthrough of Fracture 0, the flow through Fracture 2 remains constant. The flow through Fracture 0 remains close to it until the moment of breakthrough of Fracture 1, which marks the breakthrough for th e whole system. The color code is the same for a), b) and c) and is depicted in Fig. 10b. BT=221 years. The arrows indicate restricted breakthrough of Fracture 0, where the flow rates are limited by the finite resistance of Fractures 1 and 2. For f=4.33 the results are depicted in Fig. 10a-d. The influence of Fracture 2 on the pressure head at node 1 is significant (Fig. 10b). The value there is close to the minimal value of the exit boundary condition. This is the reason for the separate evolution of Fractures 0 and 1. There is almost no change in the aperture widths of Fracture 1 during the first 90 years of evolution of Fracture 0 (before breakthrough) (Fig. 10a). The jump at 90 years marks the breakthrough of Fracture 0. This is also visible from the concentration profiles (Fig. 10c), the pressure head distribution (Fig. 10b), and the jump in the flow rates in Fig. 10d. After this, the value of the pressure head at Node 1, changes from 0 m to 100 m – the maximal possible value. At that moment dissolution along Fracture 1 becomes effective. The flow through Fracture 2 remains constant, because there is no dissolution along this conduit. The evolution of Fracture 1 takes roughly the same time, and a second breakthrough event for the system is depicted by the abrupt changes in the profiles (see Fig. 10 a-d) at 221 years. As a last example we increase f to 433.33. This case is surprisingly interesting. Normally

PAGE 11

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.11 one should expect that a further increase of the initial aperture width would result in further decrease of the breakthrough time of the system. This is not the case. The breakthrough time increases by almost 100 years. This “simple” system of three fractures (see Fig. 7a) is actually not so simple. The profiles are depicted on Fig. 11a-d. At a first glance the only change is in distribution of for the pressure head (Fig. 11b). A closer look shows that the aperture width of Fracture 0 after its local breakthrough event is several times larger than the one for the previous simulation. This is easy to explain. The limitations for further evolution of Fracture 0, in the previous scenarios, are Fractures 2 and 1. Now, Fracture 2 is so wide, that it does not limit flow through Fracture 0 and allows considerable opening even before the breakthrough of Fracture 1. After breakthrough of Fracture 0 the head at the junction increases and the evolution to breakthrough of Fracture 1 is initiated. Since the increase in head is smooth now, its evolution is delayed compared to the previous case. Fig. 11. Evolution of the profiles for a scenario with f=433.33, k=1/2, m=1/2 a) Evolution of the aperture width. Fracture 0 grows independently of Fracture 1 until the moment of breakthrough at 90 years. Fracture 1 is unchanged during this period. After the breakthrough event of Fracture 0, the development of Fracture 1 starts under similar cond itions – highest possible hy draulic gradient, and low input concentration. The new point with respect to the previous scenario is the further development of Fracture 0 because Fracture 2 is not limiting at this moment; b) Evolution of the pressure head distribution. The pressure head distribution resembles the evolution of two fractures – shifted in time. The second one starts to evolve after the increase in the value of the head at Node 1. In this case this is a slow increase; c) Concentration along the fracture in un its of the equilibrium concentration ceq. The evolution of the concentration profiles shows two independently evolving fractures; d) Evolution of the flow rate. Becau se of the high initial aperture width of fracture 2, the initial flow through Fracture 0 is also high. The contribution from flow through Fracture 1 is negligible. After the breakthrough of Fracture 0, the flow through Fracture 2 continues to increase together with the development of Fracture 0. The flow through Fracture 0 remains close to it until the mome nt of the breakthrough of Fracture 1, which marks the breakthrough for the whole system. The color code is the same for a), b) and c) and is depicted on Fig. 11c. BT=221 years.

PAGE 12

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.12 Fig. 12 depicts the influence of the parameter f on the breakthrough time of the system. As long the aperture width of Fracture 2 is small, f <0.1 little flow is delivered from Fracture 0 and the breakthrough time is close to that of the isolated fracture. With increasing 0.1< f <10 the breakthrough time of the system drops. The minimum breakthrough time would occur when two conditions are valid. First the aperture width of Fracture 2 is sufficiently wide such that initially the total head drop is along Fracture 0. Then breakthrough time of Fracture 0 is that of an isolated fracture with length L/2 Secondly f is sufficiently small to allow a significant head at the junction afer the breakthrough. Then breakthrough of Fracture 1 is also close to that of an isolated fracture with length L/2 The total breakthrough time of the system then is the sum of both and an upper limit can be derived from eqn. 8 as Tb(fmin)/Tbsingle=0.32. This is close to the value at f =10 in Fig. 12. Fig. 12. The influence of the parameter f on the breakthrough time of the system from Fig. 7a. – numerical calculations. See text for discussion. When f increases further the breakthrough time increases. For f @> all flow from Fracture 0 leaves the system through Fracture 2. Therefore Fracture 1 receives no flow and widening along it is excluded, and the breakthrough time of the system becomes infinite. 3.2. The influence of the position of the exchange node Fig. 13 depicts the influence of the parameter k on the breakthrough time of the system. We select several values for f representing interesting regions of the breakthrough curve from Fig. 12. These are: f=0.33, f=1, f=4.33, f=433.33 Changing the value of the parameter k corresponds to a movement of the position of Node 1 along the fracture. It is clear that the maximum of the breakthrough time will be at values of k = 0 or 1 In these cases the system develops like a single conduit. With the movement of the junction along the fracture, the breakthrough times decrease. The position of the minimum depends on the value of f and m For large values of f and m=1/2 the minimum is k=1/2 Then both fractures develop in sequence and independently of each other. The breakthrough time of the system depends only of the lengths of Fractures 0 and 1. The minimum is attained for L0=L1, i. e. k=1/2 Fig. 13. Influence of the position of the junction to the evolution of the system ( m =1/2). The minimum of the curves is changing with the change of the f For values of f much greater than 1, the minimum in the breakthrough time is when the junction is in the middle of the single fracture. With decreasing f the minimum moves towards the entrance of the fracture. In the following we will discuss the case with k=m=1/2 Depending on the value of f two regions of the impact of exchange flow can be envisaged. For small f<<1 exchange flow is small with respect to the initial flow through Fractures 0 and 1, because of the resistance of Fracture 2. Note that this resistance increases with the third power of f In the extreme case f=0 the system behaves like an isolated fracture. When the resistances of all three fractures are of similar magnitude, i.e. f=1 then the existence of exchange flow acts as positive perturbation to reduce breakthrough times. This is the situation we have encountered in the system of Fig. 8. ( f=0.33 ) and Fig. 9 ( f=1 ), where slight reductions of breakthrough times have been observed. Now we turn to the case f>1 If the resistance of Fracture 2 is much smaller than that of

PAGE 13

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.13 Fracture 0, the latter fracture experiences a high drop of the hydraulic head. Under these conditions it behaves like an isolated fracture with a head drop close to H and experiences breakthrough accordingly. After breakthrough the head at node 1 becomes close to H During this time Fracture 1 has increased its aperture width slightly by 1a $ according to the small head, which has acted along it. After breakthrough it experiences the full hydraulic head and breakthrough occurs like for an isolated fracture with aperture width (1 1a a $ %). The total breakthrough time for the system is just the sum of these of the two subsequent events. Such scenarios are shown by Fig. 10 ( f=4.33 ) and Fig. 11 ( f=433.33 ). In both cases the total breakthrough times are equal since R2 is larger by a factor of 81 or a factor of 81 & 106, respectively. In other words, in both cases R2<2 we find an almost constant but significant reduction of total breakthrough time. For f<0.33 breakthrough times are those of the isolated series of Fracture 0 and Fracture 1. The intermediate range of perturbation is between 0.33
PAGE 14

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.14 TABLE 2. Parameter Value Unit Length of the limestone bed 742.5 Meters Width of the limestone bed (longitudinal) 375 Meters Distance between the horizontal fractures 7.5 Meters Distance between the vertical fractures 7.5 Meters Head on the input side of the domain H 100, 10, 3 Meters Head on the output side of the domain h 0 Meters Initial aperture width of the central fracture A0 0.03 Centimeters Initial aperture width of the narrow fractures for scenario A – a0 10-5-0.03 Centimeters Mean of the log – normal distribution of the aperture widths of narrow fractures for scenario B – a0 10-5-0.03 Centimeters 4.1. Evolution of the central channel (Scenario A) We start with the first experiment. The head on the left hand side is fixed at 100 m. The aperture width of the central channel A0 is 0.03 cm, and the aperture widths of the fine net fissures are a0=0.02 cm (uniform net). The results are depicted in Fig. 16. It shows the following profiles along the central fracture: a) Pressure head distribution – Fig. 16a; b) Flow through the central channel – Fig. 16b; c) Aperture widths along the channel – Fig. 16c; d) Concentration along the central conduit, depicted as the ratio between the actual concentration and the equilibrium concentration – Fig. 16d. We compare these results with those for the isolated fracture from Fig. 5. They are depicted in Fig. 17. The colors of the curves on both figures show the regime of the flow through the conduits. Red lines depict the evolution under laminar flow conditions. Green is used to outline the transition between laminar and turbulent flow, and actually depicts the situation shortly before and shortly after breakthrough. The evolution under the turbulent flow regime is depicted by the blue curves. The breakthrough time for the isolated conduit is about 800 years, and only 86 years for our current scenario. There is one very important point. Initially the pressure head is evenly distributed and linearly decreasing along the isolated and the non-isolated channel, and also along the entire fine network (Figs. 16a and 17a). This is the reason to select the uniform initial distribution of the aperture widths in the fractures of the fine network. There is no head difference perpendicular to the impervious boundaries. Consequently there is no exchange of flow between the central conduit and the fine network. This means that the evolution of both fractures starts from exactly the same point, and all the differences developing in the later stages are due to the presence of the fine network connected to the conduit. These differences are clearly visible even in the earliest stages of the evolution of both channels. During the first 80 years for the non-isolated case (Fig. 16c) and the first 700 years for the isolated fracture (Fig. 17c) the dissolutional widening is propagating downhead. For the nonisolated fracture, because of the slower widening of the fractures in the fine network (Eq. 6), the pressure difference in the direction to the network is high and the amount of flow leaving the central fracture is also high. This maximum is at the end of the widened zone. From there on part the loss of water into the continuum decreases and reaches zero at the exit of the fracture (Fig. 16b). In contrast, the profiles of the pressure head for the isolated case (see Fig. 17a) are smoother. The curves of the flow rate are horizontal lines because no flow can be lost. The most narrow part of the fracture is acting like a bottleneck and is limiting the flow along the conduit (see Fig. 17b). As it opens flow increases in time. The profiles of the concentration show a fast exponential increase to a value close to equilibrium, for the isolated case (Fig. 17d). The concentration remains almost constant to the exit. For the non-isolated case the profiles are not so steep. The concentration increases almost linearly along the fracture (Fig. 16d). At the beginning of the unwidened part it reaches a stable value, which remains practically unchanged along the channel, and is lower than

PAGE 15

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.15 Fig. 16. Evolution of the profiles along a nonisolated conduit for: a) pressure head (the numbers denote the sequence in time); b) flow rate; c) aperture width; d) concentration relative to ceq. Red lines – 0 to 80 years every 10 years; Green lines – 82 to 90 years every 2 years; Blue lines – 90 to 185 years every 20 years. Fig. 17. Evolution of the profiles along an isolated conduit for: a) pressure head (the numbers denote the sequence in time); b) flow rate; c) aperture width; d) concentration relative to ceq. Red lines – 0 to 700 years at every 100 years; Green lines – 710 to 740 years at every 10 years; Blue lines – 750 to 850 years at every 10 years.

PAGE 16

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.16 that for the single case. The reason for this difference is the higher inflow of aggressive solution into the entrance of the channel. The transition period and the breakthrough event (the green profiles, 80-90 years for the non-isolated, and 710-740 years for the isolated conduit) start when the widened part reaches the exit (see Figs. 16c and 17c). The pressure head has still the same shape but now the zone of the high head is almost at the end of the fracture for the non-isolated case (Fig. 16a). The profiles for the isolated case show that the zone of the high head has moved towards the exit, but again as in the early stages of the evolution, the profiles are not as steep as in the non-isolated case (Fig. 17a). The flow distribution shows that the zone of the maximal outflow from the fracture to the surrounding environment is now close to the exit of the fracture, following the change in the pressure head distribution (Fig. 16b). Again the concentration profiles along the isolated channel are steeper than the ones along the non-isolated conduit Figs. 16d and 17d. The dissolution rates at the end of the isolated channel are considerably lower than those at the end of the non-isolated one. This is the reason for much faster widening of the exit zone, and earlier breakthrough for the non-isolated case. The moment of breakthrough is clearly visible on all pictures (86 years for the non-isolated and 740 years for the isolated scenario). After the breakthrough the pressure head distribution changes and is more evenly distributed along the fractures. The zone of the high head has moved backwards along the conduits (Figs. 16a and 17a). At the same time the zone of considerable widening has reached the exit. A dramatic increase (by almost two orders of magnitude) of the flow rates is depicted in Fig. 16b and Fig. 17b. The increase of the concentration now is almost linear along the conduits, for both cases, but the values, at their exits are far away from equilibrium, much lower than the switch concentration (Fig. 16d and Fig. 17d). After the breakthrough, the widening along the fracture stays even along its entire length. There is an interesting detail on the curve depicting the flow rate along the fracture in the non-isolated scenario at 88 and 90 years. Both curves are not smooth at the end. This is visible close to the exit for the curve depicting the situation at 88 years, and clearly visible further upstream for the curve depicting the flow rate at 90 years. The reason is the growth of an exit fan, into the fine network. This will be discussed later. After the dramatic increase of the flow rates along the conduits, the flow regime changes from laminar to turbulent. This part of the evolution is depicted by the blue profiles on Fig. 16 and Fig. 17 (100-197 years, for the non – isolated case, and 750 – 850 years, for the single conduit). The zone of the high head continues to move backwards, for the isolated fracture, and the distribution is becoming smoother (Fig. 17a). Because of the even widening with high linear dissolution rates along the entire profile the aperture widths profiles are approaching equal widths and consequently the head distribution becomes linear as at the beginning. The non-isolated conduit shows more complex behavior because of the entrance and the exit fans developing in the fine fractures network. The numbers on the head profiles (green and blue) in Fig. 16 and Fig. 17 indicate their sequence in time. For the isolated conduit, the flow rate increases continuously following the widening of the narrowest part of the fracture (Fig. 17b, c). For the non-isolated case, because of the strong influence of the fine fractures network, the picture is different. There are regions where flow is coming from the net and regions where it is lost to it. As expected the concentration along the fractures, at this stage of the evolution, is close to zero and the widening is even along them. This is depicted by Figs 16c, d and Figs. 17c, d. 4.2. Evolution of the 2D network After having discussed the evolution of the central channel we now turn to the evolution of its environment, the 2D-net. This is discussed for Scenario A where all fractures have equal aperture widths, and for Scenario B, where the net consist of fractures with statistically distributed aperture widths (lognormal, 02 00 a). Whereas this difference has no significant influence to the evolution of the central channel it shows new features in the net. Note that all fractures carrying flow 10000 times smaller than the maximal are not displayed. The evolution of the aperture widths for both scenarios A and B is presented in Fig. 18 and Fig. 20. The color code depicts the fracture widths in centimeters, and the black lines show the head distribution inside the domain. We expect a similar evolution of both during the period before breakthrough. This is clearly visible in Figs. 18a and Fig. 20a. In both cases a

PAGE 17

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.17 Fig. 18. Evolution of fracture aperture widths for the standard scenario A. The color code depicts the fracture aperture widths in centimetres. The head distribution is illustrated by the black lines of constant head in steps of 5 m. The numbers in the lower right corner depict the time. Fig. 19. Evolution of flow rates for standard scenario A. The color code depicts the ratio between the flow rate through the current fracture and the maximal flow rate, which occurs at that time in the net. The value of the maximal flow rate in cm3/s is depicted at the upper right corner. The head distribution is illustrated by the orange lines of constant head in steps of 5m. The numbers in the lower right corner depict the time.

PAGE 18

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.18 conduit propagates downstream along the central channel. The outflow into the surrounding environment can be visualized from the pressure head distribution. In the statistical case some fissures are initially wider than 0.02 cm. They open relatively fast if the conditions at their entrances and exits are suitable (high pressure gradient, and low calcium concentration for example). This is seen for the widened fractures on the left hand side boundary and for the ones around the tip of the central conduit (see Fig. 20). The flow distribution for Scenario A is illustrated by Fig. 19. The color code depicts the ratio between the flow rate through the actual fracture and the maximal flow rate, which occurs in some fracture of the network. The value of this maximal flow rate is depicted at the upper right corner of every figure. All fractures carrying flow 10000 times smaller than the maximal one are omitted from the figure. The orange lines depict the pressure head distribution as in Fig. 18. Fig. 18a depicts the situation at 70 years. The flow out from the central fracture to the surrounding environment starts close to the left hand side boundary. It increases with the distance from the entrance and becomes maximal at the tip of the opened central conduit. After that it decreases. Clo se to the exit there is no exchange between the central conduit and the surrounding network. After the tip in the direction of the right hand side border, there is a zone, where the pressure lines are perpendicular to the central fracture. This means that the flow, which leaves the tip is di ffused into a large area of fractures. Consequently the aggressive water widens the fractures closest to the central one. The more distant fractures remain largely unaffected because of the large area and the relatively high calcium concentration. The flow distribution on the right hand side boundary is even and still unaffected by the penetrating channel. Fig. 18b and Fig. 20b illustrate the situation after 85 years for the uniform case and after 81 years for the statistical one. This is shortly before breakthrough. The significantly widened part of the central conduit is already close to the exit. The main outflow is at the tip of the widened part. This is also visible at Fig. 19b. There is an important difference compared to the situation at 70 years. The zone of the outflow is close to the right hand side boundary. The main consequence is that all the fluid from of the central fracture is not diffused into a large area of unwidened fractures. It is directly connected to the open flow boundary and the water is channelled through the neighboring Fig. 20. Evolution of fracture aperture widths for the standard scenario B. The color code depicts the fracture aperture widths in centimetres. The head distribution is illustrated by the black lines of constant head in steps of 5 m. The numbers in the lower right corner depict the time.

PAGE 19

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.19 fractures parallel to the central conduit. This is the shortest path out of the domain. The central conduit carries the main outflow at the right hand boundary. The outflow decreases in the neighbouring fractures and after some distance it becomes constant. The result is visible on Fig. 18b. Some fractures starting from the central conduit and perpendicular to it are wider than the rest of the fine network. In the statistical case (Fig. 20b) the number of the widened fractures is higher and their distribution is not symmetric. But the main result is similar: Fractures start to develop between the tip of the widened region of the central conduit and the right hand side open flow boundary condition. They belong to the shortest pathways to the exit. Figs. 18c and 20c depict the situation shortly after breakthrough – for both scenarios. The flow through the central conduit is already turbulent. The narrow part at the end of the central fracture is removed and the head distribution is changed. For the uniform net (Fig. 18c) the pressure lines move backward the gradients are more evenly distributed after 110 years, Fig. 18d. There is a fan at the downstream end of the central fracture. The flow through it is laminar, because the exits of the channels still have relatively small aperture widths. This is the reason for the head difference between the fractures belonging to the fan and the central conduit. A small amount of water between them flows mainly in the direction of the central conduit (Fig. 18c, Fig. 19c, Fig 20b). The exit fractures of the fan are widened until a breakthrough event happens there, and the flow through the whole fan becomes turbulent. This is what is observed in both cases (Fig 18d, Fig. 20c). The flow through the fan becomes turbulent and the pressure difference between its fractures and the central conduit is close to zero. The main inflow to the fan region has moved backwards. Some perpendicular fractures start to develop on both sides of the central conduit. The water entering them is still sufficiently aggressive for relatively fast widening. This increases inflow from the central conduit. The water is not diffused into the rest of the network, because of the exit fan. The influence of the right hand side boundary condition propagates upstream deeper inside the domain and the same mechanism, fast widening of the fractures comprising the shortest pathway to the developed part of the fan, is repeated. Consequently the fan is propagating upstream, on both sides of the central fracture. This is observed in Figs 18d and 20c. The flow through the fan is already turbulent as depicted by the red, thick fractures. There are two regions of the head distribution. The first region starts at the left hand side of the domain and continues to the region of the fan. The flow is mainly directed from the central region into the unwidened fractures surrounding it as can be seen from the pressure lines at their tips. This means that the outflow is from the central fracture and also from the closest ones parallel to it. The water entering the unwidened part of the domain has lost most of its dissolutional power in the area where the fan grows actively. Therefore it cannot significantly alter the net outside. The second region of the pressure distribution starts at the upstream end of the exit fan and propagates to the end of the network. Here the situation is exactly the opposite. The pressure lines are bended backwards, showing inflow from the outer unwidened part into the fan and into the central fracture. The concentration of the water coming from the fine fractures is close to saturation and cannot influence the evolution of the widened part. Consequently, the widening of the fan region is limited, and the only direction of growing is upstream towards the entrance of the domain. This is observed at the later stages of the evolution, as shown for the statistical case in Fig. 20d. The exit fan is wider and not symmetrical, but the same behavior is observed. It stops to grow in the direction perpendicular to the central conduit. Due to more favorable pathways, some channels grow from the entrance (left hand side boundary) downstream towards the exit fan. The situation 30 years later is depicted in Fig. 18e for the uniform and in Fig. 20e for the statistical scenario. In the uniform net the fan is growing further upstream towards the entrance. As mentioned above there is no growth in the direction of the impervious boundaries, because of the saturated inflow coming from the fine network. The outflow is concentrated to the fan and the central fracture, (see Fig. 19e). In contrast to the right hand side, the changes in the left hand region are significant. There is a second fan, propagating down stream. It is formed by the central fracture and the two parallel neighboring fractures. In contrast to the exit fan it is growing from the neighboring fractures in the direction of the central conduit (see Fig. 18e). The growth of the exit fan is supported by the relatively aggressive solution flowing out of the central fracture into the fine network. The situation for the fan at the entrance is different. The horizontal fractures at the left

PAGE 20

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.20 hand side border of the domain are widened. The initial aperture width ( a0) of these fractures is 0.02 cm for this scenario. The breakthrough time of the single fracture is proportional to the 3rd power of (1/a0), for n=4 (see Equation 8). A fracture with 0.02 cm initial aperture width, will take almost 3.4 times longer time for breakthrough to the exit than a fracture with initial aperture width 0.03 cm. A period of 140 years is long enough to observe widening in the fractures forming the entry part of the fine network. During the evolution to breakthrough, the fractures distant from the central conduit (in the direction of the impervious boundaries) have experienced different pressure head at their exits than the fractures close to it during the whole evolution so far (see Figs. 18a-d). This explains the smaller aperture widths of the entry fractures close to the central conduit. The entry fan starts to grow from the first pair of fractures neighboring the central conduit and parallel to it (Fig. 18e). All entry fractures (including the first pair) have enough time to widen considerably. This increases the pressure head at the nodes of their exits. Consequently the pressure distribution along the entire entrance part of the domain is changed. There is a zone of inflow to the central conduit. It starts at the ends of the fractures forming the left boundary of the domain, and ends close to the tip of the entry fan. Part of the water is diverted directly to the central conduit, and the other part continues to flow parallel to it. This water is still aggressive, and causes fast widening. The presence of the exit fan diverts a relatively high amount of water in the direction parallel to the central conduit, and consequently the growth of the entry fan is fast in both directions (forward – parallel to the flow), and in the direction perpendicular to the central channel. Actually there is no mechanism to stop its growth. The water, entering the domain on the left hand side is aggressive, and if there is a wider central conduit, then sooner or later the evolution will come to a phase when the flow will be diverted towards this wider fracture. At this moment the entry fan starts to grow until the whole domain is conquered. The pressure distribution, in the transition zone between the entry and the exit fans, matches the switch between the mechanisms governing their growth. Next to the end of the entrance fan, the flow is already parallel along the entire central fracture, and the exchange is close to zero. This depicts the beginning of the zone, where the exit fan is developing. Fig. 20e depicts the situation for the statistical scenario at 140 years. The widened fractures on the left hand side are forming the entrance fan, which is already connected with the exit fan. The whole central zone exhibiting widened aperture widths (red region) can be considered as equivalent to one huge central fracture. As already discussed, there is no mechanism limiting the growth of the entry fan. It is extending further in horizontal and in vertical direction at both sides of the central zone. The area of the inflow to the central region is extending from the left hand side of the network to the right border. There is no longer flow out from the central channel to the surrounding network. Consequently development of the exit fan is stopped. In contrast to the uniform case, only some of the horizontal entrance fractures are widened (see Fig. 20e). The others remain unchanged. On the other hand, the ones, which are widened extend deeper. The reason is the statistical distribution of the initial aperture widths of the fractures. Any of the widened and deeper penetrating channels, can be considered as a central wider fracture for the region, where it grows. All the conclusions, so far, are true also for such a small sub domain. The widened path continues to grow in the direction of the highest hydraulic gradient. These pathways can reach directly the right hand side boundary or can divert their growth in the direction of the widened area in the central part of the domain, and connect to it. Entrance and exit fans develop on both sides of these pathways. An example for such a place is depicted by the black rectangle on the figure. Actually, when these pathways connect either to the central widened zone or to the right hand side boundary they exhibit their local breakthrough event. The situation after 180 years of evolution is depicted in Fig. 18f for the uniform scenario and in Fig. 20f for the statistical one. As expected, the entry and the exit fans are already connected in the uniform scenario. The entry fan continues to grow. The pressure distribution is similar to the one for the statistical case at 140 years. The flow is directed towards the huge widened zone around the central conduit (Fig. 19f) and there is no area of outflow from the central fracture to the fine network. As already discussed, the consequence of this is the end of the evolution of the exit fan. We expect continuous growth of the entry fan in both directions – horizontal, to

PAGE 21

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.21 the exit of the domain, and vertical, in the direction of the impervious borders. This is depicted in Fig. 20f – the statistical scenario, where almost the whole domain is conquered by the entrance fan. 4.3. Influence of the exchange flow on the breakthrough time Our next goal is to investigate the influence of the hydraulic conductivities of the fine fracture network on the evolution of the central fracture. To this end the initial aperture widths a0 of the fine fractures are varied from 10-5 to 0.03 cm and A0 is fixed at 0.03 cm for the central conduit (see Table 2). Everything else is left unchanged. The dependence of breakthrough times on the initial aperture width a0 for both scenarios A and B is shown on Fig. 21. The two curves look almost identical, which means that for the period until breakthrough, the evolution of the two systems (uniform and statistical), is similar. The breakthrough time is maximal when: a) The aperture widths of the fractures of the surrounding network are negligibly small ( A0>>a0); b) The aperture widths of the fine fractures are the same as the aperture width of the central fracture ( A0=a0). This is possible only for the uniform scenario A. In the first case the resistances of the surrounding fractures are so high, that the flow they can carry is close to zero. Consequently the exchange between the central fracture and the surrounding network is also close to zero and the central fracture can be regarded as an isolated one. In the other extreme case, the head distribution is uniform along the whole network (only for the uniform networks). There is no pressure difference between the central fracture and the surrounding network. All flow lines are parallel to the upper and to the lower boundaries and there is no exchange of water. In both cases, there is no flow out from the central fracture into the continuum and the breakthrough times should be equal to that of an isolated fracture. This is exactly what can be seen on the picture. For finite aperture widths a0<0.03 cm flow from the central fracture into the net exists. Consequently by the higher inflow of aggressive water into this fracture the dissolutional widening at the exit is enhanced and breakthrough times become shorter, with increasing a0. A reasonable question is: What would happen if the initial widths of the surrounding fractures were increased even further, beyond the size of the central one. Then their breakthrough times would be shorter in comparison to the breakthrough time of the central fracture, and the breakthrough event will happen in the surrounding network. Our goal, however, is to show the evolution of the central conduit. Therefore we will not continue the curve in Fig. 21 further. Fig. 21. Breakthrough time for Scenario A (squares) and Scenario B (triangles), as a function of initial aperture widths of the continuum fractures a0.

PAGE 22

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.22 4.4. Extended scenarios In all cases discussed so far, the pressure head on the input site of the domain is extremely high ( H=100 m ). At the same time the initial aperture widths of the fine fractures have values relatively close to the ones of the central conduit: a0=0.02 cm and A0=0.03 cm respectively. We start the next set of simulations with a change of the pressure head applied to the input site. It is fixed at H=10 meters The head at the output site remains unchanged – h=0 meters The other initial and boundary conditions are the same as for the standard scenarios A and B (see Table 2). Fig. 22a-d depicts the evolution of the fracture aperture widths for the uniform scenario. Lowering the hydraulic head by one order of magnitude increases the breakthrough time of a single conduit (see Equation 8). The breakthrough time of the central conduit in the case of a0=10-5 cm when exchange flow is negligible is about 800 years under the pressure head of H=100 meters With a 10 times lower head it has to be in the range of 17000 18000 years. The breakthrough time of our uniform scenario under the pressure head H=10 meters is close to 7500 years. Even under the lower hydraulic gradients, the breakthrough time of the aquifer is reduced due to the presence of exchange flow. Fig. 22a shows that the central fracture starts to develop downstream. The exit fan starts to grow and at 7800 years (shortly after the breakthrough of the central conduit) some of the fractures are considerably wide – Fig. 22b. Fig. 22c depicts the situation after 9100 years. The exit fan has migrated backwards. At the same time an entry fan starts to grow downstream towards the right hand side boundary. After 9400 years, the evolution of the exit fan is stopped, but the entry fan continues to penetrate deeper into the block (Fig. 22d). After this moment the constant head boundary condition cannot be supported anymore and the simulation is terminated. The difference to the standard scenario is in the penetration distances of both fans. These are much shorter. This is an important observation, because it shows that the difference between the time scales for the evolution of both hydraulic systems (the fine network and the central conduit) is important. The fine network evolves slower with respect to the central fracture, than for the standard scenario A with H =100 m. Fig. 22. Evolution of the fracture aperture widths for exte nded scenario A. Everything is the same as in the standard scenario A (see Table 2.1.1), except the hydraulic head at the input side. In this case it is H=10 m The color code depicts the fracture aperture widths in centimetres. The number at the lower right corner depicts the time.

PAGE 23

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.23 Fig. 23. Evolution of the fracture aperture widths for extended scenario B.Fig. 23 a-d depicts the evolution of the statistical case. The results are similar to those for the uniform scenario. Everything is the same as in the standard scenario B (see Table 2), except the hydraulic head at the input side. In this case it is H=10 m The color code depicts the fracture aperture widths in centimetres. The number at the lower right corner depicts the time. In our next experiment we reduce the head at the input to 3 meters. This value is realistic for natural karst aquifers. Everything else is like in the standard scenario A (see Table 2). Figs. 24ad depicts the evolution of the fracture aperture widths for the uniform scenario. The differences in the time scales for the evolution of the fine network and the central fracture are already so large, that there is no time for considerable changes in the network. After the breakthrough event, the central conduit continues to evolve under turbulent flow conditions. The concentration along its entire length is low and the rates of widening are close to the maximal ones. The next experiment has the following initial and boundary conditions. Everything (including the hydraulic head on the entrance) is as for the standard scenario B (see Table 2). The only change is the average initial aperture width of the fractures in the fine network. In this case it is set to 01 00 acm. The result is depicted by Fig. 25 a-d. Because of the smaller initial aperture widths compared to scenario B with a0=0.03 cm, the time for the evolution of the fine network is increased. Similar to the low hydraulic gradient case ( H=3m ), there is no time for considerable changes in the fine fractures. This is a confirmation for the conclusion, that when the time scales of both hydraulic systems are similar, then the changes in the fine network are important for the development of the karst aquifer. On the other hand, if these time scales are sufficiently different, then both systems develop almost independently. The following statements can summarize these few examples of aquifer evolution. The evolution proceeds in two distinct steps. In the first period until breakthrough of the central fracture, flow always is injected from the wider fracture into the net. The main evolution is therefore widening of the fracture, whereas the fissures in the net are not affected. After breakthrough the central fracture widens evenly and continuously about 0.1 cm/year and the hydraulic gradient becomes constant along it.

PAGE 24

Romanov D. et al. / Speleogenesis and Evolution of Karst Aquifers, 2 (1) October 2004, p.24 Fig. 24 Evolution of the fracture aperture widths for exte nded scenario A Everything is the same as in the standard scenario A (see Table 2), except the hydraulic head at the input side. In this case it is H=3 m The bar code depicts the aperture widths of the fractures in cm. All fractures, which are smaller then 0.08 cm, are omitted from the figure The number at the lower right corner depicts the time. Fig. 25. Evolution of the fracture aperture widths for exte nded scenario B. Everything is the same as in the standard scenario B (see Table 2), except the initial aperture widths of the fine fractures. In this case they are statistical distributed with average at a0=0.01 cm. The bar code depicts the aperture widths of the fractures in cm. The number at the lower right corner depicts the time.

PAGE 25

At the same time exit and entrance fans are generated. The first step of entrance fan evolution is a breakthrough event from the input point of the nearest neighbour longitudinal fissure to the next node and then down the vertical fissure connecting to the central conduit. The head difference along this path is h/j where j is the number of nodes in the central fracture. For a significant evolution of a fan system about j/2 of such subsequent events are necessary. By use of eqn. 8 and multiplying the thus obtained breakthrough time by j/2 one obtains a crude estimation of fan evolution time Tfan as: c c fanT a A T j a A T3 0 0 3 1 3 / 4 3 0 06 1 3 4 , / / 0 1 , / / 0 1 B2 where Tc is the breakthrough time of the central channel, j=100 If Tfan is larger than the time elapsed after the breakthrough until the constant head can no longer be supported fans cannot evolve. This is the case for long breakthrough times as in natural karstification with small fracture aperture widths and low hydraulic gradients. For high hydraulic heads, as supported close to dam sites, breakthrough times are short and extended fans can evolve below the dam site as shown recently by Romanov et al. (2003). Conclusion We have discussed the changes of conduit evolution of a single isolated fracture, under constant head boundary conditions with highly undersaturated input solutions, when one allows exchange of flow, either into a tube connected to it, or by embedding the single fracture into a net of narrow fissures with smaller aperture widths, into which water can be transferred. Essentially both means to add a second dimension to the onedimensional single tube. In all cases when exchange flow leaves the conduit more aggressive solution enters into its entrance and penetration length of dissolutional activity increases. By this way dissolution rates at the exit are enhanced and the time needed until breakthrough becomes shorter. When the fracture is embedded into a network of narrow fissures it looses flow into the net during the entire period of initial evolution until breakthrough. After breakthrough the concentration along the entire conduit becomes low and these highly aggressive waters are lost to neighboring fractures, causing small breakthrough events through them. This creates an exit fan, which propagates upstream. With some delay an entrance fan develops from the input propagating downstream and invading the entire net to a finally maze like structure. This is a result of highly aggressive waters flowing from fractures close to the central channel into it, thereby creating breakthrough. The structure extends all over the net, because once a neighboring fracture has become sufficiently wi de it acts like the central one to the outer fissure next to it and so on. The evolution proceeds in two steps. In the first until breakthrough of the central channel, flow is directed into the net and dissolution occurs mainly in the channel. After breakthrough the channel widens evenly with a rate of about 1mm/year. At the same time an entrance fan is created by subsequent breakthrough events from neighboring fissures to the channel. If the time for the evolution of this fan is shorter than the time, when a constant head can no longer be supported, extended maze like structures evolve. In this case correct modeling must consider dissolutional widening in the net. In cases where fans cannot develop the evolution of the aquifer affects mainly the central channels and models neglecting dissolution in the net (Bauer et al., 2003) are a satisfactory approximation. Bibliography Bauer S. 2002. Simulation of the genesis of karst aquifers in carbonate rocks, Tuebinger geowossenschaftliche arbeiten (TGA), C62, 2002. Bauer S., Liedl, R., Sauter, M. 2003. Modeling of karst aquifer genesis: Influence of exchange flow. Water Resources Research., 39 (10), 1285, 2003. Beek W.J. and Muttzall K.M.K. 1975. Transport Phenomena. New York: Wiley. Bear and Verruijt. 1987. Modeling groundwater flow and pollution. Dordrecht, Holland: D. Reidel Publishing company. Clemens T., Hckinghaus D., Sauter M., Liedl R., Teutsch G. 1996. A combined continuum and discrete network reactive transport model for simulation of karst development. In Calibration and realiability in groundwater modelling – proceedings of the ModelCARE 96 conference held at Golden, Colorado, September 1996, IAHS Publ. no. 237, 309-318. Clemens T., Hckinghaus D., Sauter M., Liedl R., Teutsch G. 1997a. Modelling the genesis of karst aquifer systems using a coupled reactive network model. In: Hard Rock Geosciences. IAHS Publ., 241, Colorado: 3-10. Clemens T., Hckinghaus D., Sauter M., Liedl R., Teutsch G. 1997b. Simulation of the evolution of maze caves. In proceedings of the 12th International congress of speleology, La Chaux-de-Fonds, Switzerland, Volume 2, pp65-68. Dreybrodt W. 1988. Processes in karst systems Physics, Chemistry and Geology. Springer Series in Physical Environments 4, Berlin-New York: Springer, 188p. Dreybrodt W. 1990. The role of dissolution kinetics in the development of karstification in limestone: A

PAGE 26

model simulation of karst evolution. Journal of Geology 98, 639-655. Dreybrodt W. 1996. Principles of early development of karst conduits under natural and man-made conditions revealed by mathematical analysis of numerical models: Water Resources Research 32, 2923-2935. Dreybrodt W. and Siemers J. 2000. Cave evolution on two-dimensional networks of primary fractures in limestone: In: Klimchouk, A., Ford, D.C., Palmer, A.N., and Dreybrodt, W., (Eds.), Speleogenesis: Evolution of karst aquifers. Nat. Speleol. Soc., USA, 201-211. Dreybrodt W., Romanov D. and Gabrovšek F. 2001. Karstification below dam sites: a model of increasing leakage. In: Beck, B.F., Herring, J. G., (Eds.). Geotechnical and environmental applications of karst geology and hydr ology: Proceedings of the 8th Multidisciplinary Confer ence on Sinkholes and the Engineering and Environmental Impacts of Karsts, Louisville, Kentucky, Balkema, p. 131-137. Eisenlohr L., Meteva K., Gabrovšek F., Dreybrodt W. 1999. The inhibiting action of intrinsic impurities in natural calcium carbonate minerals to their dissolution kinetics in aqueous H2O-CO2 solutions. Geochimica et Cosmochimica Acta 63, 989-1002. Ford D.C. Williams P.W. 1989. Karst geomorphology and hydrology. Unwin Hyman, London Gabrovšek F. 2000. Evolution of early karst aquifers: From simple principles to complex models: Zalozba ZRC, Ljubljana, Slovenia, 150p. Gabrovsek F. and Dreybrodt W. 2000a. The role of mixing corrosion in calcite aggressive H2O-CO2CaCO3 solutions in the early evolution of karst aquifers. Water Resources Research 36: 1179-1188. Gabrovsek F. and Dreybrodt W. 2000b. A model of early evolution of karst conduits affected by subterranean CO2 sources. Environmental Geology 39:531-543, 2000. Gabrovsek F. and Dreybrodt W. 2001. A model of the early evolution of karst aquifers in limestone in the dimensions of length and depth. Journal of Hydrology 240: 206-224. Groves C., Howard A. 1994.Early development of karst systems 1. preferential path enlargement under laminar flow. Water Resources Research 30, 28372846, Howard A.D. and Groves C.G. 1995. Early development of karst systems 2. Turbulent flow. Water Resources Research 31, 19-26. Kaufmann G. and Braun J. 1999. Karst aquifer evolution in fractured rocks. Water Resources Research 35, 3223-3238. Kaufmann G. and Braun J. 2000. Karst aquifer evolution in fractured, porous rocks. Water Resources Research 36, 1381-1391. Kaufmann G. 2003. A model comparison of karst aquifer evolution for different matrix-flow formulations. Journal of Hydrology 283 (1-4), 281-289., Lauritzen O. N. 1992. Modeling the evolution of channel networks in carbonate rocks. In J. Hudson (Ed), ISRM Symposium: Eurock’92, pages 57-62. Thomas Telford. Lee C. and Farmer I. 1993. Fluid flow in Discontinuous Rocks. London: Chapman & Hall. Lowe D. 2000. Development of Speleogenetic ideas in the 20th century: The early nodern approach. In: Klimchouk A., Ford D.C., Palmer A.N., W. Dreybrodt (Eds), Speleogenesis: Evolution of karst aquifers. Nat. Speleol. Soc.,USA. Palmer A.N. 1988. Solutional enlargement of openings in the vicinity of hydraulic structures in karst regions: In: 2nd Conference on Environmental Problems in Karst Terranes and Their Solutions. Assoc. of Groundwater Scientists and Engineers Proceedings, USA, 3-15. Palmer A. N. 1991. The origin and morphology of limestone caves. Geol. Soc. of America Bull. 103, 121. Romanov D., Gabrovsek F., Dreybrodt W. 2003. Dam sites in soluble rocks: a model of increasing leakage by dissolutional widening of fractures beneath the dam. Engeneering Geology 70. Shaw T. 2000. Views on cave formation before 1900. In: Klimchouk A., Ford D.C., Palmer A.N., W. Dreybrodt (Eds), Speleogenesis: Evolution of karst aquifers. Nat. Speleol. Soc., USA. Siemers J. 1998. Simulation von Karst-Aquiferen. PhD Thesis, Universitaet Bremen. Siemers J. and Dreybrodt, W. 1998. Early development of karst aquifers on percola tion networks of fractures in limestone. Water Resources Research 34, 409-419. White W. B. 2000. Development of speleogenetic ideas in the 20th century: The modern period, 1957 to the present. In: Klimchouk A., Ford D.C., Palmer A.N., W. Dreybrodt (Eds), Speleogenesis: Evolution of karst aquifers. Nat. Speleol. Soc.,USA.