CO 2 outgassing in a combined fracture and conduit karst aquifer near Lititz Spring, Pennsylvania

Transcription

1 Geological Society of America Special Paper CO 2 outgassing in a combined fracture and conduit karst aquifer near Lititz Spring, Pennsylvania Laura Toran Department of Geology, Temple University, Philadelphia, Pennsylvania 19122, USA Eric Roman New Jersey Geological Survey, P.O. Box 427, Trenton, New Jersey 08625, USA ABSTRACT Lititz Spring in southeastern Pennsylvania and a nearby domestic well were sampled for 9 months. Although both locations are connected to conduits (as evidenced by a tracer test), most of the year they were saturated with respect to calcite, which is more typical of matrix flow. Geochemical modeling (PHREEQC) was used to explain this apparent paradox and to infer changes in matrix and conduit contribution to flow. The saturation index varied from 0.5 to 0 most of the year, with a few samples in springtime dropping below saturation. The log P CO 2 value varied from 2.5 to 1.7. Lower log P CO 2 values (closer to the atmospheric value of 3.5) were observed when the solutions were at or above saturation with respect to calcite. In contrast, samples collected in the springtime had high P CO 2, low saturation indices, and high water levels. Geochemical modeling showed that when outgassing occurs from a water with initially high P CO 2, the saturation index of calcite increases. In the Lititz Spring area, the recharge water travels through the soil zone, where it picks up CO 2 from soil gas, and excess CO 2 subsequently is outgassed when this recharge water reaches the conduit. At times of high water level (pipe full), recharge with excess CO 2 enters the system but the outgassing does not occur. Instead the recharge causes dilution, reducing the calcite saturation index. Understanding the temporal and spatial variation in matrix and conduit flow in karst aquifers was increases here by geochemical modeling and calculation of P CO 2 values. Keywords: karst, geochemical modeling, CO 2 outgassing, spring, well, coefficient of variation of hardness. INTRODUCTION The discharge of groundwater in karst aquifers can exhibit variations in water chemistry due to rapid changes in both volume of flow and residence time (Hess and White, 1988). Variations in residence time lead to different degrees of equilibrium between groundwater and the carbonate rock. By analyzing variations in concentrations of dissolved constituents at different locations, internal drainage parameters of karst groundwater systems may be better understood. This information can aid in predicting flow properties, contamination transport, and residence times. Toran, L., and Roman, E., 2006, CO 2 outgassing in a combined fracture and conduit karst aquifer near Lititz Spring, Pennsylvania, in Harmon, R.S., and Wicks, C., eds., Perspectives on karst geomorphology, hydrology, and geochemistry A tribute volume to Derek C. Ford and William B. White: Geological Society of America Special Paper 404, p , doi: / (23). For permission to copy, contact editing@geosociety.org Geological Society of America. All rights reserved. 275

2 276 L. Toran and E. Roman Two end-member flow types occur in karst aquifers, diffuse flow and conduit flow, more commonly, a mixture of the two (White, 1988). Conduit flow occurs along solution-formed channels, whereas diffuse flow occurs in small fractures and fissures through the bedrock matrix. Conduit flow is characterized by short residence times that are highly affected by precipitation events. In a diffuse-flow system, the conduits may be absent or poorly connected (Shuster and White, 1971; Atkinson, 1977). Diffuse flow is characterized by long residence times with less seasonal and temporal variation following precipitation events. In karst terranes, the diffuse and conduit terminology applies from the spring site all the way back to the recharge areas. Thus, different recharge pathways are also manifest in variations in discharge geochemistry. Some recharge falls on noncarbonate areas, enters the karst system through sinkholes, and has a rapid transit time to the spring. This recharge can travel long distances and is referred to as allogenic. Some recharge enters through the soil zone above the karst (epikarst), then finds its way to conduits and matrix below. Recharge through the local soil zone is referred to as autogenic. The autogenic recharge may have either gradual or rapid transit. Both types of rapid recharge allogenic from distant sources or autogenic that finds a short path to the conduits lead to large variation in water chemistry. The influence of storm water rapidly infiltrating the subsurface through karst conduits leads to large variations in hardness (Ca and Mg), dissolved ion content, groundwater temperature, and saturation state with respect to calcite. In contrast, when autogenic recharge is diffuse, it leads to less variation in water chemistry (Jacobson and Langmuir, 1974; Scanlon and Thrailkill, 1987; Worthington et al., 1992; White, 1999). The slow infiltration of recharge water results in little variation in groundwater temperature, near constant hardness (Ca and Mg) and dissolved ion content, and saturation with respect to calcite. Generally, a combination of recharge types occurs. Quinlan and Ray (1995) reported up to 30% allogenic and 70% autogenic recharge in the karst area around Mammoth Cave, Kentucky. Contrasts in groundwater-flow paths in karst terrains may be reflected in geochemical signatures, such as the conductivity, the partial pressure of CO 2 (P CO 2 ), and saturation indices (SI) of carbonate minerals. Scanlon and Thrailkill (1987) identified variation in recharge from diffuse to rapid flow through variations in hardness and SI. The springs they sampled in the Bluegrass region of Kentucky had conduit openings, but showed little variation in water chemistry. They concluded that diffuse recharge led to more uniform water chemistry. Desmarais and Rojstaczer (2002) and Dreiss (1989) used Ca concentration to try to determine portions of new and old water flushed to springs during storms. Desmarais and Rojstaczer (2002) estimated ~5% new water in springs in Tennessee. They also observed a decrease in SI and δ 13 C values after the storm peak, suggesting that the new water enters toward the end of the storm. Dreiss estimated 25% new water at Merimac Spring, Missouri, although the percentage varied during the storm. Jacobson and Langmuir (1974) observed that conductivity and P CO 2 changes during storms had different patterns in diffuse versus conduit springs in central Pennsylvania. In the initial storm response, conduit springs had low-conductivity, low- P CO 2 water from fast-flow paths. Diffuse springs first discharged old water with high conductivity and high P CO 2, then transitioned to low-conductivity storm water. Shuster and White (1971) found little variation in P CO 2 in their study of diffuse and conduit springs in central Pennsylvania. They noted that the maximum and the minimum P CO 2 values were both found in diffuse-flow springs and speculated that variations in recharge likely caused the differences in P CO 2. However, only diffuse-flow springs were at saturation with respect to carbonate minerals; conduit-flow springs were undersaturated for all or most of the year. Furthermore, there are seasonal variations in P CO 2 due to increases in P CO2 in the soil zone (and thus recharge) due to the growing season. Wicks and Englin (1997) used differences in SI and carbonate concentration along a cave to estimate dissolution rates. In summary, because of the multiple flow paths, it can be difficult to identify recharge water source areas, and there remains uncertainty about mapping recharge areas and classifying karst aquifers. Multiple methods must be used to enhance interpretation. Understanding the recharge pathways is important for evaluating potential contamination sources, basin areas, and monitoring strategies. SITE DESCRIPTION The study area is located in Lititz Borough, Lancaster County, Pennsylvania (Fig. 1). Land use throughout the region is predominantly agricultural, but large businesses exist, such as Warner-Lambert Pharmaceuticals and Wilbur Chocolate. The geologic units are deformed, lying in west-to-east trending bands that form a basin with ridges of shale and quartzite to the north and south. Bedrock in the basin consists of crystalline limestone, interbedded limestone and dolostone, and minor shale of Ordovician age (Poth, 1977). The depth to bedrock ranges from ~1.5 to 5 m. The regional groundwater-flow field in the area is directed toward Lititz Spring in the center of the valley (Fig. 2). Lititz and the surrounding area (Warwick Township) use six municipal wells for the residential and commercial water supply. The wells were drilled to intersect a fracture network close to the spring level and are believed to be connected to the aquifer feeding the spring. The municipal water supply personnel report periodic turbidity in the water, and have seen some water-level fluctuations when local farmers turn on irrigation pumps, which suggests that the wells are connected to the karst network. The aquifer exhibits several karst features. Lititz Spring in the center of town discharges from a series of small caves (Fig. 3) and is the largest surface-water feature and discharge

3 CO 2 outgassing in a combined fracture and conduit karst aquifer 277 Figure 2. Basin map showing approximate locations of tracer test injection site (star) and sampling points (circles), and regional flow from west to east based on water-level contours within the basin (20 m interval). Figure 1. Location of the town of Lititz in Lancaster County, Pennsylvania, United States. Lititz is at 40.1 N latitude, 76.4 W longitude. point within the study area. Lititz Spring is estimated to discharge around 10,000 m 3 /d. Kochanov (1990) created a sinkhole map for the Lititz region and plotted ~250 sinkholes and depressions per square mile, although the connectivity of these depressions with any network is unknown. Dissolution of rock follows weaknesses in the rock, such as bedding planes and fractures. Only during large storms with visible overland flow is recharge through sinks observed, such as a sinking stream located ~300 m north of Lititz Spring. A qualitative dye tracer test was conducted in August 1998 to study the connection between area sinkholes and Lititz Spring (Roman, 2000). The sinkhole used for injection (450 m west of Lititz Spring) was dry, so the connection with the cavities below the water table was unknown. A pre- and postinjection flush of water was used to push the dye through the sinkhole. The dye was detected at the spring at low concentrations (<1 µg/l) within hours of the injection. The dye also was observed in a domestic well 60 m away on the opposite side of the sinkhole, but only after a heavy rain ~5 d later. To better understand the types of connections portrayed by the tracer test response in the spring and the well, monthly sampling was instituted to look for geochemical signatures of conduit and matrix flow. This paper discusses the sampling and geochemical modeling to interpret results. Figure 3. Photo of Lititz Spring showing spring openings A, B, C, and D. METHODS Water samples were collected monthly at Lititz Spring and the well from September 1998 to May Lititz Spring is located in the center of town and is the main discharge point in the watershed. The water at Lititz Spring emanates from several fractures at the base of a hill and forms a pool ~7 7 m in diameter (Fig. 3). Samples were collected from one of the bedrock openings. The residential well, known as the Buch well (Fig. 2), is 1 m in diameter and ~18 m deep. The well is upgradient (~60 m) from the sinkholes according to the regional groundwater-flow direction. Local variations in the water table have not been mapped (Fig. 2), so the specific flow vectors around the Buch well are not known. Samples were analyzed for dissolved constituents, including alkalinity and ph. Lititz Spring was sampled by hand, and the Buch well was sampled with a bailer. The average depth to water in the well during the sampling period was 10.6 m. Water samples were filtered with a 0.45 µm Millipore filter.

4 278 L. Toran and E. Roman Temperature and ph measurements were taken in the field. Alkalinity was analyzed in the laboratory within 24 h. Major-ion concentrations (Ca 2+, Mg 2+, Na +, K +, NO 3, F, Cl, NH 4+, SO 2 4, and PO 4 ) were measured using a DIONEX 500 liquid ion chromatograph. Water-chemistry data were input into the U.S. Geological Survey (USGS) geochemical modeling program PHREEQC (Parkhurst, 1995) to calculate partial pressure of CO 2 (P CO 2 ) and saturation indices. The P CO 2 is calculated from the activity of the dissolved carbonate (as H 2 CO 3 ) and the Henry s Law constant. Since the carbonate concentration is dependent on calcite equilibria and ph, P CO 2 in a closed system can be calculated as: P CO 2 = (ahco 3 ah)/(k HCO3 K CO2 ), (1) where ahco 3 is the activity of bicarbonate, ah is the activity of hydrogen ion (derived from ph), K HCO 3 is the dissociation constant for HCO 3, and K CO 2 is the Henry s Law constant for CO 2. In addition, the coefficient of variation of hardness has also been used to classify karst flow regimes (Ternan, 1972). The coefficient of variation (CV) in percent was calculated for these waters as:, (2) where Ca 2+ and Mg 2+ concentrations are in mg/l, the quantity in parentheses is the total hardness expressed as mg/l CaCO 3, x is the mean, and σ is the standard deviation of hardness. Precipitation data for the site were obtained from the advanced flood warning system rain gauge network ( Rain gauge #2943 is located ~5 miles from Lititz Spring. RESULTS Water Chemistry ( ) ( ) σ 25. Ca Mg CV = 100 x 25. Ca = 41. Mg The water-chemistry analyses indicated Ca-HCO 3 type waters similar to those present throughout the karst region of central Pennsylvania (Langmuir, 1971; Shuster and White, 1971; Jacobson and Langmuir, 1974; Poth, 1977). The Ca/Mg ratio was 3:1, which indicated that the water flows through both calcite and dolomite. Lititz Spring exhibited properties of a diffuse-flow type end member of a karst aquifer. Cation concentrations had variations of <5 mg/l throughout the year (Fig. 4), with the exception of Ca 2+, which varied from 70 to 115 mg/l. Anion concentrations, except for HCO 3 in Lititz Spring, also exhibited the same pattern of concentration variations <5 mg/l throughout the year, regardless of changes in precipitation amounts (Fig. 4). The HCO 3 concentrations were relatively constant, between 260 and 285 mg/l. The concentration of NO 3 was between 20 and 25 mg/l in Lititz Spring, and the presence of NO 3 in these waters was probably due to fertilizers introduced into the groundwater from agricultural sites within the watershed. The overall water-chemistry variation is represented by the range in total dissolved solids (TDS), which was only mg/l over the course of this study. The Buch well water chemistry exhibited properties of conduit flow, having high variations in concentrations of cations (Fig. 4), especially Ca 2+, throughout the year due to variations in precipitation amounts. The Ca 2+ concentration varied from 25 to 110 mg/l. The Na + concentration varied from 7 to 17 mg/l. The HCO 3 concentration declined from 285 to 140 mg/l as precipitation increased at the beginning the However, the Buch well results mimicked the Lititz Spring results during the low-rainfall period (September 1998 to December 1998, Fig. 4). During this period, concentrations stayed at about the same level. The Buch well had low concentrations of NO 3 (~5 mg/l) throughout the year. The range in TDS was higher in the Buch well than in the spring, ranging from a low of 150 mg/l to a high of 340 mg/l. The coefficient of variation of hardness (CV) was 11% for Lititz Spring and 26% for the Buch well, or more than twice the variation observed at the spring. Hess and White (1988) found that the CV in conduit-type springs varied by 10% 24%, whereas the diffuse-flow springs had a relatively constant hardness and CV on the order of 5%. According to Wicks (1997), a CV even up to 12% indicates diffuse-type flow. The CV for the Buch well exceeded the typical CV for conduit flow. The Lititz Spring CV was at the low end for conduit flow and may indicate that the conduit system is influenced strongly by fractures that feed it or by diffuse recharge. Geochemical Modeling Lititz Spring samples were oversaturated with calcite during the entire sampling period (Fig. 5), with the exception of one sample with SI = 0.25 collected in March Dolomite followed a parallel trend. The observed saturation with respect to calcite suggests that the water had a sufficiently long residence time to reach equilibrium, which is typical of a diffuse system. The Buch well samples gradually decreased in calcite saturation index from SI = 0.5 in December 1998 to SI = 1.25 in March 1999 as rainfall increased. The Buch well was undersaturated with respect to calcite during three months (February, March, and April). The oversaturation with respect to calcite and dolomite requires an explanation, because the system is clearly drained by conduits. Fast-moving conduit water would be expected to be undersaturated, and slow-moving matrix water to be at saturation. These data tell a story about the contributions (timing and amount) of matrix and conduit water. One explanation for these data is that the recharge water picked up excess CO 2 over the atmospheric equilibrium value as it traveled through the soil zone. Subsequently, CO 2 out-

5 CO 2 outgassing in a combined fracture and conduit karst aquifer 279 Figure 4. Water sample analyses of major anions and cations for Lititz Spring and Buch well: (A) Lititz Spring anions; (B) Buch well anions; (C) Lititz Spring cations; and (D) Buch well cations. gassing occurred when the matrix water entered the conduit. Outgassing of CO 2 (represented as H 2 CO 3 in solution) drives the carbonate equilibrium toward more calcite precipitation: 2+ ppt CA + 2HCO CaCO + H CO outgas. (3) Because bicarbonate is lost through conversion to CO 2 and subsequent outgassing, the ph must rise to maintain the charge balance. (Another way to view this balance is that a decrease in P CO 2 results in a decrease in H+ [Eq. 1].) This ph increase further moves the reaction toward precipitation of calcite. However, outgassing occurs faster than CaCO 3 precipitation, which leads to oversaturation (e.g., Freeze and Cherry, 1979). Specifically, precipitation of calcite doesn t tend to occur until the SI is over 0.5, based on field data collected by Dreybrodt et al. (1992) and travertine deposits in a stream flowing out from a cave in Virginia (Lorah and Herman, 1988). The log P CO 2 values of the springs in Virginia were relatively high, varying from 2.02 to Oversaturation was also observed in drip waters of caves, as reported by White (1997). Vesper and White (2004) discussed an example of high-p CO 2 water entering a karst system in Kentucky during diffuse recharge. They saw variations in P CO 2

6 280 L. Toran and E. Roman Figure 5. Saturation indices (SI) with monthly rainfall values for Lititz Spring and Buch well. Monthly rainfall is presented since samples were not collected to follow a particular recharge event. during recharge that also demonstrate potential pathways through diffuse and conduit recharge. The trend of P CO 2 versus SI calcite supports the hypothesis that variations in outgassing are related to pathways and dissolution trends. As the log P CO 2 decreases from 1.9 to 2.4 (more outgassing), the system becomes more oversaturated with respect to calcite (Fig. 6). The outgassing of CO 2 can be modeled using PHREEQC. Assuming that the matrix water is close to saturation, the February 1999 samples were used as the initial solution and were then stepped through progressively decreasing values of P CO 2. The modeled outgassing exhibited a decrease in P CO 2 and increase in SI, which follows the observed P CO 2 versus SI trend at both Lititz Spring and the Buch well. The March 1999 sample, which was undersaturated, could represent a dilution trend (Fig. 6). This monthly sample followed a period of higher rainfall (Fig. 5). During periods of high rainfall, overland flow increases and has been observed to recharge sinkholes (John Diehl, Pennsylvania Department of Environmental Protection, August 2002, personal commun.). Sinkhole recharge would have a dilution effect in the conduits. Alternately, diffuse recharge could fill the conduits and inhibit outgassing. The log P CO 2 value remained high in March ( 1.71 at the Buch well and 1.91 at Lititz Spring), so there was still a significant portion of water from soil recharge. However, the higher water levels in conduits prevent outgassing. Physical evidence for this model comes from a comparison of water levels and P CO 2 in the Buch well. For the portion of the year monitored, log P CO 2 values increased from 2 to 1.65 as the water level in the well rose (Fig. 7). Data were not collected over the entire year to observe whether the growing season affected P CO 2 levels. Thus, the Lititz karst system seems to be fed by a fractured matrix that discharges to a conduit network then to the main spring. It exhibits features of both matrix and conduit flow along its flow path, with variation in the proportions in time and space. Figure 6. Measured and modeled P CO2 versus saturation index (SI) of calcite showing the outgassing of supersaturated samples for (A) Lititz Spring and (B) Buch well. Alternate dilution path is shown for low SI sample. Note that P CO2 enhancement for log 1.7 is 63 times more than atmospheric; log 2.5 is 10 times more than atmospheric. Figure 7. Increase in P CO2 related to increasing water levels in the Buch well, which prevents outgassing. Note that P CO2 enhancement for log 1.7 is 63 times more than atmospheric; log 2.0 is 28 times more than atmospheric.

7 CO 2 outgassing in a combined fracture and conduit karst aquifer 281 CONCLUSIONS The karst network near Lititz Pennsylvania presents a system that is neither diffuse-flow dominated (such as many central Pennsylvania springs; Shuster and White, 1971) nor matrixflow dominated (such as the springs described in the Kentucky Bluegrass region; Scanlon and Thrailkill, 1987). Instead, at Lititz, both matrix and conduit water mix along the flow path from recharge areas to discharge areas. Although mixtures of these end members have been recognized in previous work, this study points out that the mixing can occur along different pathways (open versus closed) at different times and locations within the same basin. Evidence that mixing in different proportions occurs at different locations is provided by the contrasts between Lititz Spring and the Buch well. The well had a higher CV for chemical constituents, and a tracer from a nearby sinkhole showed up only after a heavy rain, suggesting that a conduit became saturated, resulting in an alternate pathway. Lititz Spring had a lower CV for chemical constituents, suggestive of greater contributions from diffuse flow, and the tracer showed up relatively fast, but with a large dilution factor. Both the conduit discharge at Lititz Spring and the Buch well water were oversaturated with respect to calcite, indicating matrix water contributes to the conduits. Modeling showed that CO 2 outgassing likely occurs when matrix water reaches the open-conduit system then moves rapidly to the discharge point before calcite can precipitate. Temporal variation was evidenced by P CO 2 values that varied by a factor of six over the year (10 60 times greater than atmospheric P CO 2 ). In the rainy season, pipe-full conditions were suggested by a high-p CO 2 value and the low SI of calcite. At other times of year, the P CO 2 value was lower, indicating that outgassing in conduits was occurring. Unfortunately, this mixture of water makes it more difficult to identify recharge areas in the system. Some recharge water is autogenic and enters through the soil zone. Some recharge water is allogenic, during the rainy season when sinkhole recharge occurs. Evidence shows that the temporal variation in geochemistry may be caused in part by switching from an open to a closed system as the conduits fill with water. The mixing of the recharge sources in the conduits makes it difficult to find the geochemical signature of the allogenic water, and thus to map the recharge area. The variations discussed here follow the concept of storage in karst aquifers and threshold events that change the geochemistry, as discussed by Loop and White (2001), and illustrate the importance of long-term monitoring to characterize karst aquifers. ACKNOWLEDGMENTS We wish to thank M. Field of the Environmental Protection Agency (EPA) and J. Diehl of the Pennsylvania Department of Environmental Protection (PADEP) for initiating this project, providing guidance, and conducting the tracer injections. REFERENCES CITED Atkinson, T.C., 1977, Diffuse flow and conduit flow in limestone terrain in the Mendip Hills, Somerset, Great Britain: Journal of Hydrology, v. 35, p , doi: / (77) Desmarais, K., and Rojstaczer, S., 2002, Inferring source waters from measurements of carbonate spring response to storms: Journal of Hydrology, v. 260, p , doi: /S (01) Dreiss, S.J., 1989, Regional scale transport in a karst aquifer. 1. Component separation of spring flow hydrographs: Water Resources Research, v. 25, p Dreybrodt, W., Buhmann, D., Michaelis, J., and Usdowski, E., 1992, Geochemically controlled calcite precipitation by CO 2 outgassing: Field measurements of precipitation rates in comparison to theoretical predictions: Chemical Geology, v. 97, p , doi: / (92)90082-G. Freeze, R.A., and Cherry, J.A., 1979, Groundwater: Englewood Cliffs, New Jersey, Prentice Hall, p Hess, J.W., and White, W.B., 1988, Storm response to the karstic carbonate aquifer of south-central Kentucky: Journal of Hydrology, v. 99, p , doi: / (88) Jacobson, R.L., and Langmuir, D., 1974, Controls on the quality variations of some carbonate spring waters: Journal of Hydrology, v. 23, p , doi: / (74) Kochanov, W.E., 1990, Sinkholes and karst related features of Lancaster County, Pennsylvania: Pennsylvania Department of Environmental Resources, Bureau of Topographic and Geologic Survey, Open-File Report 90-01, 6 p., 18 maps. Langmuir, D., 1971, The geochemistry of some carbonate groundwaters in central Pennsylvania: Geochimica et Cosmochimica Acta, v. 35, p , doi: / (71) Loop, C.M., and White, W.B., 2001, A conceptual model for DNAPL transport in karst ground water basins: Ground Water, v. 39, p , doi: /j tb00357.x. Lorah, M., and Herman, J., 1988, The chemical evolution of a travertinedepositing stream: Geochemical process and mass transfer reactions: Water Resources Research, v. 24, p Parkhurst, D.L., 1995, User s Guide to PHREEQC A computer program for speciation, reaction-path, advective-transport, and inverse geochemical calculations: Lakewood, Colorado, U.S. Geological Survey, WRIR , 143 p. Poth, C.W., 1977, Summary groundwater resources report of Lancaster County, Pennsylvania: Pennsylvania Geological Survey, 4th series, Water Resources Report 43, 80 p., 1 map. Quinlan, J.F., and Ray, J.A., 1995, Normalized base-flow discharge of ground water basin: A useful parameter for estimating recharge area of springs and recognizing drainage anomalies in karst terranes, in Beck, B.F., ed., Karst geohazards: Rotterdam, Balkema, p Roman, E., 2000, Hydrogeologic characterization of a karst groundwater system in Lititz, Lancaster, County, Pennsylvania [Master s thesis]: Philadelphia, Pennsylvania, Temple University, 98 p. Scanlon, B.R., and Thrailkill, J., 1987, Chemical similarities among physically distinct spring types in a karst terrain: Journal of Hydrology, v. 89, p , doi: / (87)90182-X. Shuster, E.T., and White, W.B., 1971, Seasonal fluctuations in the chemistry of limestone springs: A possible means for characterizing carbonate aquifers: Journal of Hydrology, v. 14, p , doi: / (71) Ternan, J.L., 1972, Comments on the use of calcium hardness variability index in the study of carbonate aquifers; with reference to the Central Pennines, England: Journal of Hydrology, v. 16, p , doi: / (72) Vesper, D.J., and White, W.B., 2004, Storm pulse chemographs of saturation index and carbon dioxide pressure: Implications for shifting recharge

8 282 L. Toran and E. Roman sources during storm events in the karst aquifer at Fort Campbell, Kentucky/Tennessee, USA: Hydrogeology Journal, v. 12, p , doi: /s White, W.B., 1988, Geomorphology and hydrology of karst terrains: New York, Oxford University Press, 464 p. White, W.B., 1997, Thermodynamic equilibrium, kinetics, activation barriers, and reactions mechanisms for chemical reactions in karst terrains: Environmental Geology, v. 30, p , doi: /s White, W.B., 1999, Karst hydrology: Recent developments and open questions, in Beck, B.F., Pettit, A.J., and Herring, J.G., eds., Hydrogeology and engineering geology of sinkholes and karst: Rotterdam, Balkema, p Wicks, C.M., 1997, Origins of groundwater in a fluviokarst basin: Bonne Femme Basin in central Missouri, USA: Hydrogeology Journal, v. 5, no. 3, p , doi: /s Wicks, C.M., and Englin, J.F., 1997, Geochemical evolution of a karst stream in Devils Icebox Cave, Missouri, USA: Journal of Hydrology, v. 198, p , doi: /S (96) Worthington, S.H., Davies, G.J., and Quinlan, J.F., 1992, Geochemistry of springs in temperate carbonate aquifers: Recharge type explains most of the variation, in Proceedings of the 5th Conference on Limestone Hydrology and Fissured Media, Neuchatel, Switzerland: Besancon, France, Université de Besancon, p MANUSCRIPT ACCEPTED BY THE SOCIETY 22 SEPTEMBER 2005 Printed in the USA