Hydrochemistry and solution rates in gypsum karst: case study from the Western Ukraine

Abstract

The gypsum karst in the Western Ukraine spreads through a large territory covering more than 20,000 km² and is represented by a range of stages (evolutionary types), from deep-seated through subjacent to entrenched. Correspondingly, hydrogeological settings of karst development, circulation patterns and chemical characteristics of groundwaters differ substantially between the respective areas. Based on 1,800 analyses, this paper summarises hydrochemistry of the gypsum-hosting Miocene aquifer. The majority of sampling has been performed in conjunction with a study regime of gypsum solution rates by means of standard tablets. In this study, which included 53 tablet stations representing varying conditions of water-rock interaction, 644 weight-loss measurements were made over the period 1984–1992. The highest rates are characteristic of entrenched karst although active dissolution is localised along well-defined sinking streams with short underground courses, rare vertical percolation paths and the water table. Lower but still quite substantial rates are characteristic of subjacent and deep-seated (confined) karst. However, the overall dissolution removal is greater due to higher flow through the gypsum and the larger area of rock-solvent contact. The results are generalised in order to derive the approximate solution rates characterising major situations and to be suitable for modeling purposes.

Introduction

Due to the high solubility of gypsum and fast solution rates, sulphate ions commonly dominate in groundwaters of gypsum karst areas. Hydrochemical studies of karstified and adjacent aquifers shed more light on flow patterns and groundwater behaviour in a system. Such studies are particularly informative where a source of sulphate occurs discretely within the geologic section (i.e. a single bed or a series of distinct beds) so that sulphate ions can be used as an indicator of hydraulic communication between flow system components, flow direction and intensity.

Dissolution, and hence creation of voids and conduits, is unevenly distributed within a karst system. Evaluation of the solutional capacity of groundwaters based on representative chemical data is indispensable for verifying speleogenetic concepts, modeling and predicting karst development and subsidence risk assessment. Even more important is the determination of actual gypsum solution rates.

Solution rates from field studies are usually obtained from mass balance (solute load) data, derived from chemographs and hydrographs, related to the aquifer surface or basin area. However, such an approach gives integrated values for a reference basin, which is usually insufficient for a detailed analysis due to the great spatial and temporal variability of solution rates in a real karst system (Klimchouk et al. 1988, 1996). Solution rates for certain flow components or zones is rarely specified due to difficulties inherent in chemographs and hydrographs for separation and evaluation of respective aquifer surface areas. In this respect, field experimental regime data on gypsum solution rates gained in the Western Ukraine through the 1980s and the 1990s are particularly valuable. These rates represent a number of karst settings, flow components and different situations of water-rock interaction, and are linked with hydrochemical data. Thus the characteristics of gypsum karstification processes in different environments can be compared.

This paper reviews and analyses the extensive hydrochemical and solution rate data gained during 1984–1992, which have been expanded and consolidated recently in a single database and GIS system.

Geologic and hydrogeologic background of the gypsum karst development

The Miocene gypsum sequence is widespread on the southwestern edge of the eastern European platform, along the Carpathian Foredeep, where it occupies over 20,000 km² . Gypsum stretches from the northwest to southeast for more than 300 km as a belt ranging from several kilometers to 40–80 km width. It is the main component of the Miocene evaporite formation that surrounds the Carpathian folded region to the northeast, from Poland across the Western Ukraine and Moldova to Romania (Fig. 1a).

Fig. 1

a Location and evolutionary types of gypsum karst of the Western Ukraine. Zones of different karst types are shown by Roman numbers: I the gypsum is entirely denuded, II entrenched karst, III subjacent karst, IV deep-seated (confined) karst. b Distribution of the gypsum stratum, sulfur deposits, and large caves in the Western Ukraine. 1 easternEuropean platform fringe. Carpathian foredeep: 2 outer zone, 3 inner zone. 4 Carpathian folded region; 5 sulfate rocks on the platform. Tectonic boundaries include: 6 platform/foredeep, 7 outer/inner zone of the foredeep, 8 foredeep/folded region. 9 other major faults; 10 flexures. 11 sulfur mineralization; 12 sulfur deposits; 13 gas deposits; 14 oil deposits; 15 large maze caves in the gypsum

Most Miocene rocks along the platform margin overlie eroded Cretaceous strata which include terrigenous and carbonate sediments, mostly marls and sandstones, together with detrital and argillaceous limestones. The Miocene succession comprises deposits of Badenian and Sarmatian age. The Lower Badenian unit beneath the gypsum includes mainly carbonaceous, argillaceous and sandy beds (30–90 m thick) adjacent to the foredeep, and these grade into rocks of calcareous bioherm and sandy facies (10–30 m thick) towards the platform interior (Fig. 2).

Fig. 2

Generalized litho- and hydro-stratigraphy of the upper part of the sedimentary sequence in the southwest edge of the eastern European platform

The gypsum sequence, 10–40 m thick, is variable in structure and texture but almost everywhere occurs as a single bed. A layer of evaporitic and epigenetic limestone, locally called “Ratynsky”, commonly overlies the gypsum, ranging from half a meter to more than 25 m in thickness. The gypsum and the Ratynsky limestone comprise the Tyrassky Formation.

The Tyrassky Formation is overlain by the Upper Badenian unit which begins with argillaceous and marly lithothamnion limestones and sandstone beds. Above this is a succession of clays and marls with its lower part in the Upper Badenian (the Kosovsky Formation) and its upper part in the Lower Sarmatian, with the total thickness ranging from 40–50 m in the Podol‘sky area to 80–100 m in the areas adjacent to the foredeep. The clay cover thickens to several hundred metres close to the regional faults that separate the platform edge from the foredeep.

The Miocene succession is overlain by the late-Pliocene and Pleistocene glacio-fluvial sands and loess-like loams in the north-west section of the gypsum belt (Sites 1 and 2 on the Fig. 1a), and by sand and gravel alluvial terrace deposits left by the Dniester and Prut rivers through the late Pliocene–Pleistocene in the Podol’sky and Bukovinsky areas (Sites 3 through 6). Many buried valleys, of early to mid Pleistocene age, are entrenched 30–50 m into the Kosovsky and Sarmatian clays and, locally, into the upper part of the Tyrassky Formation.

The present distribution of the Miocene formations and the levels of their denudation exposure vary in a regular way from the platform interior towards the foredeep. The Tyrassky Formation dips 1–3° towards the foredeep and is disrupted by block faults in the transition zone. To the south and south-west of the major Dniester Valley, large tectonic blocks drop down as a series of steps, the thickness of clay overburden increases, and the depth of erosional entrenchment decreases. Along the tectonic boundary with the foredeep the Tyrassky Formation drops down to a depth of more than 1,000 m. This variation, the result of differential neotectonic movement, played an important role in the hydrogeological evolution of the Miocene aquifer system and resulted in the differentiation of the platform edge into four zones (Fig. 1a) three of which represent the distinct types of gypsum karst: entrenched, subjacent and deep-seated (Andrejchuk 1984, 1988; Klimchouk et al. 1985; Klimchouk and Andrejchuk 1988; Klimchouk 1996, 2000b). The gypsum bed is largely drained in the entrenched karst zone, partly inundated in the subjacent karst zone and remains under artesian confinement in the deep-seated karst zone.

In hydrogeologic terms the region represents the southwestern portion of the Volyno-Podolsky artesian basin (Shestopalov 1989). Sarmatian and Kosovsky clays and marls provide an upper confining sequence. The lower part of the Kosovsky Formation and the limestone bed of the Tyrassky Formation form the original upper aquifer (above the gypsum) and the Lower Badenian sandy carbonate beds, in places along with Carboniferous sediments, form the lower aquifer (below the gypsum), the latter being the major regional one. The hydrogeologic role of the gypsum unit has changed with time, from initially being an aquiclude, intervening between two aquifers, to a karstified aquifer with well-developed conduit permeability (Klimchouk 1997a, 2000a, 2000b).

Regional flow is from the platform interior, where confining clays and the gypsum are denuded, towards the large and deep Dniester Valley and the Carpathian foredeep. In the narrow northwest part of the gypsum belt (Sites 1 and 2) the artesian monoclinal slope is most clearly developed (Fig. 3). Infiltration recharge occurs on Rostochje and Opolje hills where both the Upper Badenian and Lower Badenian sediments are exposed. Flow becomes confined towards the foredeep, being “separated” by the gypsum into the upper and lower aquifers. On the opposite flank of the confined flow area, along the very platform margin, regional faulting has brought the Miocene aquifers into lateral contact with the thick Kosovsky clay sequence. Further flow in this direction is prevented and upward discharge occurs locally, focused upon areas where the confining properties of capping sediments are weakened by stratigraphic or tectonic discontinuities, or incised erosional valleys. In the confined flow area the groundwater flow pattern includes a lateral component in the lower aquifer (and in the upper aquifer to a lesser extent) and an upward component through the gypsum in areas of potentiometric lows where extensive gypsum cave systems are developing.

Fig. 3

Circulation patterns in the confined Miocene aquifer system (4th zone) under natural (a) and disturbed (b) conditions

In the wide south-east section of the gypsum belt the deeply incised river valleys of Dniester and its sub-parallel tributaries separate the Miocene sequence into a number of isolated, deeply drained interfluves. This is the entrenched karst zone (Podol’sky area, Sites 5 and 6) where most of the explored, presently relict maze caves are located, inherited from the confined karst stage. To the south–southeast of Dniester (Bukovinsky area, Sites 3 and 4), where erosional entrenchment is less deep, the gypsum remains largely intact and is partly inundated (the subjacent karst zone). Further in this direction, with an increase in depth of the gypsum occurrence and decrease in depth of the erosional entrenchment, the Miocene aquifer system becomes confined (the deep-seated karst zone).

In the confined flow area the groundwater flow pattern includes a lateral component in the lower aquifer (and in the upper aquifer to a lesser extent) and an upward component through the gypsum in areas of potentiometric lows where extensive gypsum cave systems are developed.

Methods of study

Water chemistry and saturation Index

Water chemistry in the gypsum karst of the Western Ukraine is characterised from 1,800 samples analysed for major cations and anions. Water temperature and pH were measured in the field. In the resultant database, ion content is expressed in mg/dm³, mg-eq/dm³ and in %-eq. TDS content was calculated in mg/dm³ .

Water chemistry data were used to calculate, for each sample, a Saturation Index (SI_g) characterising a degree of deviation of a natural solute from equilibrium with respect to gypsum:

$$ {\text{SI}}_{\text{g}} = {\text{lg}}\left( {\frac{{{\text{a(Ca}}^{{\text{2}} + } {\text{)}} * {\text{a(SO}}_{\text{4}}^{{\text{2 - }}} {\text{)}} * \mathop \gamma \nolimits_{{\rm Ca}} * \mathop \gamma \nolimits_{{\rm SO}_{4}} }} {{{\text{}}_{\text{g}} }}} \right) $$

(1)

or

$${\text{SI}}_{\text{g}} = {\text{lg a(Ca}}^{{\text{2}} + } {\text{) }} + {\text{ lg a(SO}}_{\text{4}}^{{\text{2 - }}} {\text{) - lg K}}_{\text{g}} = p\left[ {K_{\rm g} - {\text{a(Ca}}^{{\text{2}} + } {\text{) - a(SO}}_{\text{4}}^{{\text{2 - }}} {\text{)}}} \right]$$

(2)

where:

• a(Ca²⁺) and (SO ⁼₄ ) are activity products for the respective ions (determined from the Debye-Hückel equation; Garrels and Christ 1965);

• K_g is the thermodynamic equilibrium constant for the reaction of gypsum dissolution

• γ_Ca and $\gamma_{SO_{4}}$ are coefficients accounting for the ion pairing effect (from Yu. Shutov, personal communication).

The K_g function of temperature for the range 0–50°C was taken from Aksem and Klimchouk (1991):

$$K_{\rm g} = 2.160 + 2.59 \times 10^{-2} \times \hbox{t} - 5. 62 \times 10^{-4} \times \hbox{t}^{2}$$

(3)

$$ pK_{\rm g} = 4. 667 - 5. 197 \times 10^{-3} \times \hbox{t} + 1. 133 \times 10^{-4} \times \hbox{t}^{2} $$

(4)

that closely agrees with the values provided by Wigley (1973). SI_g has a value of zero if water is in equilibrium with gypsum, has negative values for undersaturated solutions and positive values for supersaturated ones.

Sampling strategy and datasets

The original sampling was performed in conjunction with various karst research projects conducted through 1982 to 1992. It was aimed at individual characterisation of different karst hydrogeologic settings (confined, unconfined) and of each distinct component of the respective circulation systems, i.e. various recharge sources, flow within and discharge from individual aquifers, etc. In many localities continuous sampling was performed to reveal chemical variability through different flow regimes. Besides our original analyses, available historic data from literature and technical reports were incorporated into the resultant database to amplify characterisation of some system components or periods poorly covered by our own sampling. Chemical data from other sources were involved if proper identification of samples was possible with regard to time and circulation pattern. The overall database totals 1,800 analyses. Table 1 gives mean TDS and SO₄ contents and averaged SIg and gypsum solution rate values for principal situations studied.

Table 1 Mean TDS, SO₄, SI_g and solution rates in different aquifers and situations of the Miocene aquifer system in the Western Ukraine

In confined setting, represented by Sites 1 (Jazovsky area) and 2 (Nikolaevsky area), samples were taken from boreholes gouged separately for the lower, gypsum (“Gypsum aquifer-mid” in Table 1 and on diagrams) and upper aquifers by a sampler that enables sampling from specific depths. Samples were taken at the gauge level. In the Jazovsky area samples were also taken from a sump in the large sulphur quarry which bottom cuts to the top of the gypsum bed (“Gypsum aquifer-top”). Samples from the sump represent upward discharge from the gypsum karst systems.

In unconfined setting, represented by Sites 3 (Zoloushka Cave), 4 (Dankivsky Collapse), 5 (Seret-Nichlava interfluve, Ozernaya and Optimisticheskaya caves) and 6 (Atlantida cave and Mylevtsy area), the following system components were characterised by sampling (Table 1 and Fig. 5a): (1) precipitation, (2) spring discharge from the Plio-Quaternary aquifer, (3) focused surface flow (streams) above the gypsum level, (4) focused percolation through the caprocks sampled in vertical dissolution pipes in caves, (5) flowing cave pools (aquifer “windows”), (6) cave pools perched on clay fill, (7) “downward” spring discharge from the sub-gypsum aquifer, (8) spring discharge from the Cretaceous aquifer, and (9) spring discharge from the Devonian aquifer.

Some of these situations were further categorised according to certain criteria significant for chemical evolution of waters. For instance, dripping localities in caves (focused percolation) were classified into three categories according to a degree of contact between dripping water and the gypsum surface. Sampling in cave pools was performed at fixed levels (depths) and the data grouped into three categories (0–5, 5–15 and >20 cm) to characterise chemical stratification of water.

Gypsum solution rate

The method of field measurements of solution rates (SR) based on weight loss of standard samples (usually tablets), was initially employed to study the processes in limestone karst (Gams 1981). It appears to be more effective for gypsum than for limestone because of the much higher characteristic dissolution rates and higher spatial unevenness of dissolution in gypsum. This makes errors inherent in measurements relatively insignificant and allows dissolution dynamics to be monitored even over comparatively short timescales (Klimchouk et al. 1988).

A gypsum solution rate study by means of standard tablets was performed in the Western Ukraine mainly during 1984–1992, with additional measurements made in recent years. In total, 53 stations were organised to characterise different situations of water–rock interaction in the three major intrastratal karst settings: entrenched, subjacent and deep-seated (confined). Standard tablets 40–45 mm in diameter, 7–8 mm thick and weighting 18–25 g, were produced from a single variety of 97% pure massive microcrystalline gypsum from the Kudrintzy quarry (Site 6). Before control weighting, the tablets were dried overnight at 40°C prior to and after exposure. On stations, the tablets were so placed as to avoid possible mechanical damage and direct impact of strong water currents. In cave lakes, the tablets were suspended on a nylon fishing line at different fixed depths. In springs and boreholes, the tablets were suspended inside widely perforated plastic capsules. In boreholes, such capsules were lowered and suspended at a gauge level.

Control weightings were generally made every 3 months, but sometimes at other intervals ranging from 1 month to 6 months, depending upon the actual dissolution dynamics and accessibility of a sample location. Tablets on stations were substituted with new ones each time their original weight and dimension changed more than 20% from the original parameters. In total, 644 measurements were made in the course of the study. As a rule, measurements were accompanied by water sampling for chemical analysis and subsequent determination of SI_g values; respective mean TDS, SO₄ and SI_g values are given in Table 1 apart from the values derived from the entire dataset.

Solution rate values are expressed in units of mg/cm² /day and mm/year. The latter notation is used throughout this work as it is compatible with commonly used units of karst denudation.

All water chemistry and solution rate data were geographically related and incorporated into the GIS “Gypsum Karst of the Western Ukraine”. The GIS also contains other geocoded information about regional geology, tectonics, hydrogeology and karst features that enables various scientific and practical applications.

Hydrochemistry of the Miocene aquifer system and gypsum solution rates in different environments

Confined settings

The confined conditions of the Miocene aquifer system were the primary conditions throughout the whole region prior to the Late Pleistocene. Later, uplifts in the interior parts of the platform and accompanying denudation and deep linear erosion have caused the formation and expansion of subjacent and entrenched karst zones. The neotectonic uplifts were less intense on the platform edge, along the foredeep boundary, where the confined conditions have remained until they now represent the original circulation pattern of the layered artesian system responsible for the specific type of speleogenesis (Klimchouk 1996, 2000a, 2000b). Due to the positive feedback existing between flow pattern and speleogenesis, the latter in turn has played a fundamental role in establishing the resultant structure of the groundwater exchange in the Miocene aquifer system.

The zone of confined karst is shown in dark tint on Fig. 1a. The upper profile on Fig. 3a illustrates the original situation of the artesian monoclinal slope complicated by block faulting, representative of the north-west section of the gypsum belt. On the Rostochje and Opolje uplands, where infiltration recharge occurs, the unconfined aquifer contains HCO₃–Ca waters with TDS of about 0.5 g/dm³ . The system becomes confined towards the gypsum belt and the foredeep, being then “separated” by the gypsum into the upper and lower aquifers. Under natural conditions, the heads in the confined flow area are well above the bottom of the Kosovsky clays, even above the ground surface in places.

Since the flow and hydrochemical pattern in the confined karst zone has been heavily impacted by opencast mining and associated groundwater withdrawal, the natural and disturbed situations should be distinguished. The disturbed situation has developed through many areas due to massive abstraction of sulphur and clay since the early 1970s. Our datasets of the confined zone represent two sites, Site 1 being the area of the large Jazovsky sulphur quarry where groundwater withdrawal continued, until recently, through the whole period of our own studies. The pre-quarrying conditions here are characterised only by historic data. Site 2 is an area of the Nikolaevsky quarry where abstraction of the Kosovsky clays stopped in 1982. For this site our own sampling characterises only the post-quarrying situation but the situation duriing the quarrying period can be inferred from historic data.

The quarries, cut into the caprocks, at the top of the Ratynsky limestone (Nikolaevsky quarry) or at the top of the gypsum (Jazovsky quarry), became the focuses of discharge from the Miocene aquifer system. To maintain the operation, groundwaters were pumped out with rates reaching 112,000 m³ /day in the Jazovsky quarry and 288,000 m³ /day in the Nikolaevsky quarry. Such activities, particularly, have resulted in:

• The fall of the potentiometric surface (up to 90 m in the immediate vicinity of the Jazovsky quarry) and the formation of drawdown cones extending up to 10–12 km around the quarries

• The activation of groundwater circulation and increase of flow velocities through every horizon (the velocities measured by tracing experiments were up to 2.5 km/day in the Jazovsky area and up to 10.2 km/day in the Nikolaevsky area)

• The partial reversal of the circulation pattern; the fall of the head in the upper aquifer gave rise to downward percolation through disturbed zones in the Kosovsky clays and the piracy of the surface Runoff in vicinity of the quarries

• The respective changes in groundwater chemistry and intensification of karst processes.

The lower aquifer is the major aquifer due to its laterally “continuous” high transmissivity. Within the areas of potentiometric lows groundwater from the lower aquifer flows upward into the upper aquifer through conduits in the gypsum (Klimchouk 1997a, 2000a, 2000b). Lateral flow within the gypsum occurs only locally because of the clustered nature of maze conduit systems. Analysis of the exploration drilling data suggests the existence of well-developed multi-storey maze conduit systems in such areas, similar to the well-known caves documented in the entrenched karst zone.

Under natural conditions, the lower aquifer in the Jazovsky area contained groundwaters predominantly of HCO₃–Ca composition, locally of HCO₃–SO₄–Ca or SO₄–Ca composition. TDS contents ranged between 0.15 and 2.9 g/dm³ (average 1.14 g/dm³) and SO₄ contents varied between 0.004 and 2.9 g/dm³ (average 0.58 g/dm³). Locally high sulphate contents indicate backward density-driven circulation loops from the conduits in the gypsum; waters high in SO₄ outflow from the gypsum down to the lower aquifer. Under disturbed conditions, TDS and SO₄ contents in the lower aquifer decreased considerably (average 0.72 g/dm³ and 0.16 g/dm³, respectively). This can be explained either by an increase of lateral inflow from the adjacent marginal recharge zone or by a breaking of backward circulation loops in the gypsum due to steepe vertical gradients towards the quarry bottom. In places immediately adjacent to the foredeep, Cl–Na methane-bearing waters with TDS up to 7.5 g/dm³ were identified, incoming along faults from the foredeep, an oil- and gas-bearing basin.

Groundwaters entering the gypsum from the lower aquifer are very aggressive with respect to gypsum, being able to dissolve it at rates ranging from 2.48 mm/year to 25.57 mm/year. Average SR in the Jazovsky area is 9.8 mm/year and average SI_g from the SR-related chemical data is −1.16. These data characterise the disturbed situation; it can be assumed that SR by the waters of the lower aquifer at the natural stage would be somewhat lower as the sulphate content and Si_g were higher, and circulation intensity was lower before quarrying.

When circulating through the gypsum bed, groundwaters gain SO₄–Ca composition. The SO₄ content and dissolution rates vary substantially depending upon borehole gauge locations in the cave/fissure system relative to the internal currents structure. Currents directly rising from feeding channels in a conduit system may keep rather low SO₄ content and high aggressiveness even in the middle part of the gypsum, while the bulk water under sluggish circulation is more saturated with SO₄ (Klimchouk 1997b). Under the disturbed conditions in the Jazovsky area SO₄ content in the “mid” gypsum (borehole samples) ranges from 0.98 g/dm³ to 1.54 g/dm³ (average 1.21 g/dm³) and SI_g ranges from −0.2 to 0.05 (average −0.03). Correspondingly, SRs in the “mid” gypsum vary in a range of 3.54–0.42 mm/year (1.07 mm/year in average). When the water comes out of the gypsum (samples from the quarry sump), it contains 1.38 g/dm³ of SO₄and is eventually saturated with sulphates (average SI_g 0.05). Solution rates measured in this situation are quite negligible (0.05 mm/year).

We do not have water chemistry and SR data for the gypsum bed in the Jazovsky area during the pre-quarrying period prior to 1970 (natural conditions) although it can be assumed that the circulation had been more sluggish, SO₄ content somewhat higher and solution rates lower. With certain reservations, the data of the post-quarrying stage in the Nikolaevsky area can be assumed as being representative of the natural conditions. The respective average SR (0.22 mm/year) is about an order of magnitude lower than the “mid” gypsum value in the disturbed Jazovsky area.

In the upper aquifer, the waters discharged from the gypsum mix with the lateral flow component. Conformably, SO₄ and TDS content drops and SI_g rises, as compared to the gypsum aquifer, to (average) 0.3 g/dm³, 1.32 g/dm³ and − 1.4 respectively. It can be seen from Fig. 4a that in the Jazovsky area average SO₄and TDS contents in the upper aquifer under disturbed conditions are substantially lower compared to the respective values for the natural stage (SO₄ of 1.2 g/dm³ ; TDS of 2.75 g/dm³). This can be explained by the increase of the lateral inflow from the marginal recharge area and by the involvement of the vertical downward recharge through the capping clays within the drawdown cone of the quarry, both components diluting the waters rising from the gypsum.

Fig. 4

Mean TDS, major ion contents, SI_g and solution rates in waters of the confined Miocene aquifers: a Jazovsky area, pre-quarrying and quarrying conditions; b Nikolaevsky area, quarrying and post-quarrying conditions

Unconfined settings

Hydrogeological conditions and the locations of tablet stations in the entrenched karst zone (Sites 5 and 6) are generalised in Fig. 5a. In the subjacent karst zone (Sites 3 and 4) the situation is largely similar to the above-gypsum part of the cross-section, but differs in that the gypsum is inundated for most of the thickness and there are no nearby entrenchments and free drainage out of the aquifers located below the gypsum.

Fig. 5

a The formation of groundwaters and sampling locations in unconfined settings, gypsum karst of the Western Ukraine. b Mean TDS, major jon contents, SI_g and solution rates in the unconfined groundwater system. c Scheme illustrating water stratification in cave pools and morphological effects (horizontal notching) of enhanced aggressiveness of the upper water layer

In both zones the karst systems in gypsum are recharged via two routes: (1) through swallow holes in sinkholes (surface runoff formed on the capping clays from precipitation and discharge from the perched Plio-Quaternary aquifer) and, (2) vertical percolation from the Plio-Quaternary aquifer along faults or collapse zones that disturb the clay overburden. In both cases waters have similar HCO₃–Ca composition with TDS of about 0.87–0.89 g/dm³ . However, they differ considerably in the flow dynamics. Surface runoff is focused in streams with highly variable discharge and percolation is represented by drip/trickle localities in caves. The swallowed streams run some tens of meters inside the caves on the clay fill and then dissipate joining the groundwater lens whose surface is located either within the gypsum or in the sub-gypsum aquifer.

Rates of solution by the swallowed streams can be quite high due to the high aggressiveness of water (average SI_g of −1.26) and high flow velocity. We did not measured SR in streams because of their highly irregular flow regimes. Such data would be difficult to generalise due to influence of flow velocity that is difficult to control. SR values for surface streams from estimates made by different methods in various regions are in orders of a few tens to a few hundreds mm/yr (Klimchouk et al 1996). However, the streams dissolve the gypsum only along selected short routes in the immediate vicinity of swallow holes and do not contribute much to the development of the relict maze caves. Only a few short linear caves attribute their origin solely to such streams.

Focused vertical percolation is responsible for the development of characteristic dissolution pipes (“comins”) that are 0.5–3 m in diameter and are commonly superimposed on the relict labyrinthic passages. Solution rates generated by such focused vertical percolation (averaged at 0.52 mm/year) vary greatly between stations and seasons, reflecting the highly irregular percolation regime and local conditions of the water-rock interaction. Hence, these data are difficult to generalise. We roughly classified drip/trickle localities into three categories according to the degree of contact between water and rock: (A) poor contact (almost free-falling dripping or trickles with short contact with gypsum), (B) medium contact (1–2 m), and (C) close contact (few metres). Average SI_g values vary between C and A from −0.02 to −1.29.

From the viewpoint of contemporary karst development in unconfined settings, the most important situation is the saturated zone of the gypsum aquifer. The gypsum strata can be drained in full or be inundated to levels varying between the areas, tectonic blocks, seasons and years. The aquifer in the gypsum is represented by hydraulically connected pools (“flow pools” in Table 1) whose numbers and sizes depend on the water level position relative to the cave system configuration. Flow is quite sluggish. In the Zoloushka cave, where the water table has been progressively lowered during the last 40 years due to pumping from the nearby quarry, there are also pools perched on the clay fill, degrading with time.

Our studies suggest that there is a distinct density stratification in the pools due to chemical differences. In the flow pools the average SO₄ contents change from 0.8 g/dm³ in the uppermost layer (<5 cm) through 1.0 g/dm³ in the horizon between 5 and 15 cm to 1.3 g/dm³ at depths below 20 cm. Correspondingly, average SI_g decreases between these levels from −0.26 through −0.11 to 0.03 and SRs decrease from 10.8 through 6.59 to 1.32 mm/year. The fact that SRs in the uppermost layer are somewhat ten times higher than in the bulk water at depth is well illustrated by morphological evidences in the caves: horizontal notching in the passage walls (Fig. 5c), mushroom-like shape of the rocks projecting from the water in pools, etc. In the perched pools SO₄ contents are a little higher but SI_g and SR values are lower as compared to the flow pools: SR averages in the respective datasets are 0.02 mm/year and 5.29 mm/year. Chemical stratification is less distinct.

Considering that most measurements were made in the active circulation conditions in the entrenched karst zone (Site 5) and that circulation in the gypsum in the subjacent karst zone is probably somewhat more sluggish, the values measured in the flow pools should be regarded as maximums for the unconfined gypsum aquifer. For the purposes of speleogenetic modelling we would generalise solution rates for the bulk aquifer as being of the order of 0.1 mm/year and for the upper water layer as being of the order of 1.0 mm/year. It is remarkable that the former figure roughly corresponds to the above-derived characteristic solution rate for the undisturbed confined circulation in the gypsum; both situations represent quite sluggish circulation in the gypsum. The latter figure is close to the average SR for the “mid” gypsum in the disturbed confined conditions; both situations represent somewhat more live circulation. Also, the average SR for the upper water layer under active unconfined circulation (10.8 mm/year) is close to the SR in the waters entering the gypsum from below in the disturbed confined circulation (9.93 mm/year). This makes sense considering natural convection effects and similar chemistry of recharge sources. Hence, this order of solution rates can be taken as a characteristic in both the confined and unconfined settings for “fresh” waters at the initial contact with gypsum.

The sub-gypsum (lower) aquifer in the unconfined settings receives its recharge in two ways: (1) through the karst systems in the gypsum, where waters gain SO₄–Ca composition and, (2) through direct infiltration in areas where both the capping clays and the gypsum were removed by denudation (see Fig. 5a). In catchments where the gypsum is entirely removed, the springs have HCO₃–Ca composition and low TDS (0.6 g/dm³).

In the entrenched karst zone almost all the waters from the gypsum discharges via springs draining the sub-gypsum aquifer on the valley slopes. Their chemistry depends on the proportion between the two above recharge sources, the saturated thickness in the gypsum and the general circulation intensity in a particular basin. Table 2 characterises three typical basins (tectonic blocks) studied in the Seret-Nichlava interfluve (Site 5). Basin 1 (Glybochek) contains a little gypsum and the sub-gypsum aquifer discharges waters of SO₄–HCO₃–Ca composition. In Basin 2 (Ozernaya cave) the gypsum and the caprocks occupy most of the catchment and recharge though the karst system is dominating. In Basin 3 (Optimistychna cave) the sub-gypsum rocks are exposed through a notable proportion of the catchment, so that direct infiltration recharge considerably dilutes the waters that passed through the gypsum. Calculations using mixing equations suggest that in Basin 2 about 87 % of discharges form at the expense of the waters coming through the karst system (the remaining 12 % being formed at the expense of direct infiltration from the non-karstic areas) while in Basin 3 this proportion is 58% versus 42 %. This is in agreement with geomorphological characteristics of the basins.

Table 2 Generalization of gypsum solution rate data from the Western Ukraine for the purposes of calibration, adjustment and verification of speleogenetic and karst development models

The Upper Cretaceous aquifer in the entrenched karst zone discharges through numerous springs on the valley slopes. The dominating anions are HCO₃ and/or SO₄ and dominating cations are Ca and Mg. Variations in the chemical facies and TDS are dictated mainly by the varying SO₄ content that changes through the area from 0.05 g/dm³ to 1.4 g/dm³. This reflects varying degrees of hydraulic communication with the above Miocene aquifer and varying proportions between the recharge from the above aquifer and direct infiltration recharge.

The springs from the Devonian aquifer discharge mainly waters of the HCO₃–Ca or HCO₃—Ca–Mg composition with a mean TDS of 0.91 g/dm³ . Sulphate content is normally low (0.1–0.2 g/dm³). In places, waters of HCO₃–SO₄ composition occur with SO₄ content as high as 0.8–1.0 g/dm³ . This indicates local leakage from the Miocene aquifer via major faults.

Conclusions

The waters in the Miocene aquifers adjacent to the gypsum bed, as well as the main recharge sources, commonly have HCO₃–Ca composition and low TDS before they come into contact with gypsum. Circulating in the gypsum bed, they gain SO₄–Ca composition. Discharged to the adjacent aquifers in the downgradient direction of the circulation system (the upper aquifer in confined settings and the lower aquifer in unconfined settings), sulphate waters mix in different proportions with HCO₃-Ca waters, resulting in composite chemical types.

Within the gypsum aquifer, the contents of SO₄ and solution rates vary dramatically depending upon the internal circulation structure within a cave/fissure system. Under generally sluggish flow conditions, this structure results in a complex interaction of forced and natural convection circulation. In the confined settings currents that directly rise from the feeding channels at the gypsum bottom may keep rather low SO₄ content and high aggressiveness even in the middle part of the bed while the bulk water in a conduit system is much more saturated with SO₄. In the unconfined settings, the effect of density (chemical) stratification is pronounced, with the upper water layer being of lower SO₄ content and higher aggressiveness than the bulk water. Considering such a great variability, any generalisation of the water chemistry and dissolution processes based on limited occasional sampling can be misleading; comprehensive regime studies are therefore needed encompassing different flow components, situations and regimes.

Solution rates depend on the water chemistry but also on the dynamic regime of water (flow rate and velocity). Fig. 6 is a summary graph of gypsum solution rates measured in different aquifers and situations. For the purposes of conceptual and mathematical modeling of speleogenesis it is important to derive a valid generalization of solution rates for the most common situations encountered in the gypsum aquifer. Such an attempt, based on the reported data and the above discussion, is presented in Table 2.

Fig. 6

Gypsum solution rates and SI_g in different situations of water–rock interaction