Groundwater residence times in Shenandoah National Park, Blue Ridge Mountains, Virginia, USA: a multi-tracer approach

Chemical Geology 179 Ž2001. 93–111 www.elsevier.comrlocaterchemgeo

Groundwater residence times in Shenandoah National Park, Blue Ridge Mountains, Virginia, USA: a multi-tracer approach a L.N. Plummer a,) , E. Busenberg a , J.K. Bohlke , D.L. Nelms b, R.L. Michel c , ¨ P. Schlosser d a

US Geological SurÕey, MS 432, Reston, VA 20192, USA US Geological SurÕey, 1730 East Parham Rd., Richmond, VA 23228, USA c US Geological SurÕey, MS 434, Menlo Park, CA 94025, USA Lamont-Doherty Earth ObserÕatory and Department of Earth and EnÕironmental Engineering, Columbia UniÕersity, New York, Palisades, NY 10964, USA b

d

Abstract Chemical and isotopic properties of water discharging from springs and wells in Shenandoah National Park ŽSNP., near the crest of the Blue Ridge Mountains, VA, USA were monitored to obtain information on groundwater residence times. Investigated time scales included seasonal Žwet season, April, 1996; dry season, August–September, 1997., monthly ŽMarch through September, 1999. and hourly Ž30-min interval recording of specific conductance and temperature, March, 1999 through February, 2000.. Multiple environmental tracers, including tritiumrhelium-3 Ž3 Hr3 He., chlorofluorocarbons ŽCFCs., sulfur hexafluoride ŽSF6 ., sulfur-35 Ž35S., and stable isotopes Ž d18 O and d2 H. of water, were used to estimate the residence times of shallow groundwater discharging from 34 springs and 15 wells. The most reliable ages of water from springs appear to be based on SF6 and 3 Hr3 He, with most ages in the range of 0–3 years. This range is consistent with apparent ages estimated from concentrations of CFCs; however, CFC-based ages have large uncertainties owing to the post-1995 leveling-off of the CFC atmospheric growth curves. Somewhat higher apparent ages are indicated by 35S Ž) 1.5 years. and seasonal variation of d18 O Žmean residence time of 5 years. for spring discharge. The higher ages indicated by the 35S and d18 O data reflect travel times through the unsaturated zone and, in the case of 35S, possible sorption and exchange of S with soils or biomass. In springs sampled in April, 1996, apparent ages derived from the 3 Hr3 He data Žmedian age of 0.2 years. are lower than those obtained from SF6 Žmedian age of 4.3 years., and in contrast to median ages from 3 Hr3 He Ž0.3 years. and SF6 Ž0.7 years. obtained during the late summer dry season of 1997. Monthly samples from 1999 at four springs in SNP had SF6 apparent ages of only 1.2 to 2.5 " 0.8 years, and were consistent with the 1997 SF6 data. Water from springs has low excess air Ž0–1 cm3 kgy1 . and N2 –Ar temperatures that vary seasonally. Concentrations of He and Ne in excess of solubility equilibrium indicate that the dissolved gases are not fractionated. The seasonal variations in N2 –Ar temperatures suggest shallow, seasonal recharge, and the excess He and Ne data suggest waters mostly confined to gas exchange in the shallow, mountain-slope, water-table spring systems. Water from wells in the fractured rock contains up to 8 cm3 kgy1 of excess air with ages in the range of 0–25 years. Transient responses in specific conductance and temperature were observed in spring discharge within several hours of large precipitation events in September, 1999;

)

Corresponding author. Tel.: q1-703-648-5841; fax: q1-703-648-5832. E-mail address: [email protected] ŽL.N. Plummer..

0009-2541r01r$ - see front matter. Published by Elsevier Science B.V. PII: S 0 0 0 9 - 2 5 4 1 Ž 0 1 . 0 0 3 1 7 - 5

94

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

both parameters increased initially, then decreased to values below pre-storm base-flow values. The groundwater residence times indicate that flushing rates of mobile atmospheric constituents through groundwater to streams draining the higher elevations in SNP average less than 3 years in base-flow conditions. Published by Elsevier Science B.V. Keywords: Springs; Groundwater age; Residence time; Shenandoah National Park; Environmental tracers

1. Introduction The concentrations of atmospheric environmental tracers in groundwater can provide information on groundwater residence times ŽCook and Solomon, 1997; Solomon and Cook, 1999; Plummer and Busenberg, 1999.. Although several detailed studies have compared age information obtained when a series of environmental tracers were measured in recharge areas of aquifers ŽEkwurzel et al., 1994; Cook et al., 1995; Szabo et al., 1996., few studies have measured suites of environmental tracers in groundwater discharge, such as at springs. Springs provide an important, and in some instances, the only source of hydrologic information for groundwater systems. However, springs represent points of discharge, and are often complex mixtures of waters of varying ages. Interpretation of environmental tracer data in spring discharge requires consideration of groundwater-mixing processes, and the temporal information that can be resolved is usually an estimate of the mean residence time of water in the groundwater reservoir, rather than the travel time of a parcel of water through the system ŽEriksson, 1958; Vogel, 1967; Maloszewski and Zuber, 1982; Maloszewski et al., 1983; Focazio et al., 1998; Cook and Bohlke, 1999; Manga, 1999.. ¨ In addition to mixing, a number of processes can affect the concentrations of environmental tracers to varying extents in spring discharge, requiring a multi-tracer approach to resolve conflicting results. For example, groundwater-dating methods that are based on concentrations of dissolved gases assume that shallow groundwater is closed to gas exchange following recharge ŽCook and Solomon, 1997., yet, little is known about the actual extent of gas confinement of water discharging from shallow, water-table springs. In this study, concentrations of chlorofluorocarbons ŽCFCs., tritium Ž3 H., tritiumrhelium-3 Ž3 Hr 3 He., sulfur hexafluoride ŽSF6 ., sulfur-35 Ž35 S., sta-

ble isotopes of water Ž d18 O and d2 H., and dissolved nitrogen ŽN2 ., argon ŽAr., helium ŽHe., and neon ŽNe. were measured in water from springs and wells in Shenandoah National Park ŽSNP., located in the Blue Ridge Mountains of Virginia, USA. The data were used to investigate gas exchange processes, groundwater mixing and residence times for shallow groundwater in SNP.

2. Hydrogeologic setting Shenandoah National Park ŽSNP. covers an area of approximately 800 km2 in the Blue Ridge Mountains of Virginia ŽFig. 1.. Land surface elevation ranges from 170 m in the northwest part of the park near Front Royal, VA, to 1230 m in the central part of the park. Most peaks range in elevation from 600 to 1200 m, with 60 peaks reaching elevations of more than 900 m. The maximum relief is about 900 m, with an average relief of about 600 m. Annual precipitation at Big Meadows in the central part of the park averaged about 114 cm over the period 1962 through 1971 ŽDeKay, 1972., and 138 cm during the period 1981 to 1997 ŽNational Park Service Air Resources Division data.. Annual precipitation for the years 1994, 1995, 1996 and 1997 was 138, 171, 162 and 124 cm, respectively. The mean annual air temperature was 7.88C at Big Meadows Želevation 1070 m. and 11.28C at Park Headquarters Želevation 340 m. between 1962 and 1971 ŽDeKay, 1972.. Most of SNP is underlain by fractured rocks of Precambrian to Cambrian age ŽGathright, 1976.. Limited groundwater supplies occur in relatively thin Ž0–24 m, and commonly about 9 m. residuum and colluvium that overlie metabasalts and granodiorites, and in fractures in the crystalline rocks ŽDeKay, 1972; Lynch, 1987.. Fracture density decreases with depth, and few wells yield water from depths greater

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

95

Fig. 1. Location of springs and wells sampled in Shenandoah National Park.

than 90 m below land surface. DeKay Ž1972. reports average densities of water-bearing fractures from 30 of the 38 test holes drilled in the 1960s of approximately 0.14 my1 for the interval 0–15 m, 0.09 my1 for the interval 15–30 m, 0.04 my1 for the interval 30–45 m and approximately 0.01 my1 for the interval 75–90 m. Twenty of the test holes were completed as water-supply wells and produce less than 6 l sy1 and typically less than 1 l sy1 . Most of the sampled springs ŽFig. 1. are used for public supply and were modified by sub-surface construction in the 1930s to enhance discharge and collection. In some cases, the sub-surface development was fairly extensive and permits contact of spring discharge with air. Construction records indicate that some springs combine discharge from shallow fractures in bedrock with water draining from shallow colluvium and residuum that overlies the fractured bedrock. Throughout most of the Piedmont

and Blue Ridge Provinces of the southeast US, the shallow fracture system in metamorphic and crystalline bedrock is hydraulically connected to water stored in the shallow overlying residuum ŽLeGrand, 1967; Lynch, 1987.. Recharge occurs locally and travels relatively short distances to discharge at springs and streams. Most of the springs sampled as a part of this study are at relatively high altitude within SNP and discharge shallow groundwater from nearby water-table sources. No perennial streams exist in the higher elevations of SNP. The two largest developed springs in the Park, Lewis Spring at Big Meadows, and Furnace Spring at Skyland ŽFig. 1., have maximum discharges of about 20 and 11 l sy1 , respectively, but most springs in the park have discharges of less than several liters per second. During periods of drought, spring discharges are typically only 10% of the maximum flows. Although flow is greatly diminished during

96

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

periods of drought, only 3 of 34 springs sampled during the peak wet season ŽApril, 1996. had insufficient flow to permit sampling Ž- 0.01 l sy1 . during a prolonged drought in late summer, 1997.

3. Sample collection and measurement procedures Water from springs and wells was sampled with varying temporal resolution: Ž1. seasonally Žwet season, April, 1996; dry season, August–September, 1997., Ž2. monthly ŽMarch through September, 1999., and Ž3. hourly Ž30-min interval recording of specific conductance and temperature using data loggers, March, 1999 through February, 2000.. The seasonal sampling was conducted at 34 springs and 15 wells ŽFig. 1.. The monthly and hourly Ž30-min observations. sampling was conducted at five springs ŽLewis Spring, Furnace Spring, Byrd’s Nest 3, Browntown Valley Overlook Spring, and Hudson Spring, see Fig. 1.. Four of the five springs sampled monthly and at 30-min intervals are near Skyline Drive in SNP ŽFig. 1. and one ŽHudson Spring. is a public-supply spring at the base of the Blue Ridge Mountains at Luray, VA. The analyses from the seasonal and monthly sampling include major- and minor-element chemistry, dissolved nitrogen ŽN2 . and argon ŽAr., tritium Ž3 H., chlorofluorocarbons ŽCFC-11, CFC-12 and CFC-113., sulfur hexafluoride ŽSF6 ., tritiumrhelium-3 Ž3 Hr3 He. and stable isotopes Ž d2 H and d18 O. of water. Many of the monthly samples also included measurement of the activity of sulfur-35 Ž35 S. of dissolved sulfate. Monthly composite samples of precipitation collected from August, 1997 through December, 1999 at the Big Meadows Air Monitoring Station were analyzed for 3 H, d2 H, d18 O, and 35 S. Weekly air samples collected from October, 1995 through December, 1999 at the Big Meadows Air Monitoring Station were analyzed for CFC-11, CFC-12, CFC-113 and SF6 . A synoptic sampling of 30 streams for 3 H, d18 O, and d2 H near the base of the east and west flanks of the Blue Ridge Mountains ŽFig. 1. was conducted during April, 1996, coinciding with the wet-season sampling of springs and wells, and was repeated for nearly the same set of streams in late August and early September, 1997 during the dry season.

Details of the methods used to collect and analyze the water and air samples are summarized in Plummer et al. Ž2000a. and at the URL http:rrwater. usgs.govrlabrcfcr. The chemical and isotopic data for springs, wells, and surface water are tabulated in Plummer et al. Ž2000a.. The chemical and isotopic data on precipitation from Big Meadows are from Bohlke et al. Žsubmitted for publication.. ¨

4. Results 4.1. CFC and SF6 atmospheric mixing ratios Fig. 2 compares CFC and SF6 mixing ratios Ž1975 to present. in rural North American ŽNA. air with individual flask measurements of CFCs and SF6 in air from the Big Meadows Air Monitoring Station, SNP Ž1995–2000.. Mixing ratios express the dry air volume fraction of the gas in air and are given in parts per trillion by volume Žpptv.. Most of the

Fig. 2. Plot showing CFC and SF6 mixing ratios for rural NA air Žcurves. and measurements made on single-flask air samples from the Big Meadows Air Monitoring Station, SNP.

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

97

CFC-11 and CFC-113 mixing ratios from SNP closely follow the NOAA Niwot Ridge, CO values ŽElkins et al., 1993; http:rrwww.cmdl.noaa.govr., whereas the CFC-12 and SF6 mixing ratios are locally higher by approximately 2% and 5%, respectively, than those at Niwot Ridge ŽFig. 2.. The differences between local air in SNP and Niwot Ridge air are within the analytical uncertainty, but can contribute to slight low biases in CFC-12 and SF6 ages when applying the Niwot Ridge air curve to SNP samples. Although the effects of these enrichments are minor, age calculations with SF6 were based on the NA air curve corrected for a 5% local enrichment, and those ages based on CFC-12 used NA air values corrected for an enrichment of 2%. The peak in atmospheric mixing ratios for CFC-11 and CFC-113 in the 1990s ŽFig. 2. leads to ambiguity in age, or apparently modern ages for waters collected over a period that spans the decade of the 1990s ŽBusenberg and Plummer, 1992; Plummer and Busenberg, 1999.. The slow rise in CFC-12 atmospheric mixing ratios during this study permits age interpretation, but requires measurement of recharge temperature for individual samples, and resulting ages usually have uncertainties larger than those associated with SF6 or 3 Hr3 He.

4.2. DissolÕed nitrogen, argon, helium, and neon Concentrations of dissolved N2 , Ar, He and Ne were used to estimate recharge temperatures and excess air ŽHerzberg and Mazor, 1979; Heaton, 1981; Heaton and Vogel, 1981; Heaton et al., 1983; Busenberg et al., 1993; Stute and Schlosser, 1999., and to assess potential for gas exchange in the SNP waters. Waters sampled from springs have N2 –Ar temperatures that range from approximately 58C to 188C with low contents Žtypically 0–1 cm3 kgy1 ŽSTP.. of excess air ŽFig. 3A.. N2 –Ar temperatures for springs from summer, 1997 Žaverage 10.6 " 2.48C. were about 2.68C warmer than those from April, 1996 Žaverage 8.0 " 1.58C., the latter being near the mean annual air temperature ŽFig. 3A.. Many of the samples from springs have N2 –Ar temperatures near the measured water temperature ŽFig. 4.. In contrast, most samples from wells and some from springs

Fig. 3. Comparison of concentrations of dissolved N2 and Ar in water from springs ŽA. and wells ŽB., in relation to solubility equilibrium and excess air. The calculations assume an average elevation for the samples within SNP of 914 m.

apparently had been heated in the ground relative to their N2 –Ar temperatures ŽFig. 4.. Three samples with N2 –Ar temperatures significantly higher than the water temperature ŽFig. 4. were collected shortly after a large summer precipitation event and have warm recharge temperatures. The N2 –Ar temperatures for water from wells in fractured rock averaged 9.0 " 0.88C for samples from

98

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

4.3. Tritium in precipitation and streams Tritium in streams at the base of SNP averaged 9.4 " 0.8 TU in April, 1996, and 9.9 " 1.9 TU in September, 1997, whereas tritium in spring discharge averaged 9.6 " 1.3 and 8.7 " 1.9 TU in April, 1996, and September, 1997, respectively. The precipitation-weighted tritium content of precipitation at Big Meadows between August, 1997 and November, 1998 was 9.1 " 3.6 TU. Water from springs at high altitude in SNP and surface water near the base of

Fig. 4. Comparison of N2 –Ar temperatures for waters from springs and wells in relation to the field water temperature, the mean annual temperature, and season ŽApril, 1996 vs. August– September, 1997..

summer, 1997, and 8.3 " 1.68C in samples from April, 1996, with relatively high excess air contents that, in some samples, exceeded 8 cm3 kgy1 ŽFig. 3B.. There is no consistent evidence of seasonality in N2 –Ar temperatures for waters from wells, as there is in water from springs ŽFig. 3A.. Most water that recharges the fracture system infiltrates through the overlying residuum, and likely has low excess air content, similar to that of spring discharge from the residuum. The relatively high quantities of excess air in water from wells ŽFig. 3B. are thought to form within the fracture system during periods of rising water levels. Quantities of excess air determined from the mass-spectrometric measurements of dissolved Ne and He are in close agreement ŽFig. 5A., and within the analytical uncertainties, similar to quantities of excess air calculated from gas-chromatographic measurements of dissolved N2 and Ar ŽFig. 5B.. The measured N2 –Ar temperature for each sample was used in the initial evaluation to interpret ages from all gas-based tracers. Additional calculations evaluated uncertainty in age based on uncertainty in recharge temperature.

Fig. 5. Comparison of excess air ŽApril, 1996 and August–September, 1997 data. in spring and well discharge. ŽA. Calculated from dissolved Ne and He data, and ŽB. calculated from dissolved N2 –Ar data, and dissolved Ne data.

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

99

the Blue Ridge Mountains have tritium activities consistent with recent recharge and cannot be separated on the basis of tritium content. Tritium concentrations higher than those of modern precipitation are present in water from some wells in SNP, the highest being 18 TU from the Park Headquarters well No. 1 ŽFig. 1.. Although there are no long-term tritium records of precipitation from SNP, a long-term record of tritium in precipitation at Washington, DC ŽR.L. Michel, unpublished data., approximately 140 km east of SNP, can be scaled to SNP by comparing tritium concentrations measured in precipitation during August, 1997 through July, 1998 at the two locations. Accordingly, the Washington, DC record was multiplied by a scaling factor of 0.73 to estimate a long-term tritium record at SNP. The adjusted tritium record for SNP suggests the yearly weighted tritium content of precipitation has not changed significantly during the past 10 years, averaging about 9.4 " 1.2 TU. Because we have limited data on 3 H in precipitation in SNP, calculations were performed using both the unmodified Washington, DC 3 H record and the scaled Washington, DC record. 4.4. CFCs, SF6 and 3H in groundwater from springs and wells Figs. 6 and 7 compare measured concentrations of CFC-11, CFC-12, CFC-113 and SF6 for all springs and wells sampled in April, 1996 and August–September, 1997, with calculations from three hypothetical mixing models—piston flow, exponential mixing and binary mixing ŽCook and Bohlke, 1999.. The ¨ CFC and SF6 concentrations are expressed as the gas mixing ratios in pptv, calculated at equilibrium with the measured concentration in water using the N2 –Ar temperature and local spring elevation. The normalization of the data to pptv removes differences in aqueous concentrations in samples that are caused by varying N2 –Ar temperatures, and permits direct comparison of the data to long-term records of atmospheric mixing ratios. The envelopes bounded by the modeled tracer concentrations ŽFigs. 6 and 7. define the regions that should contain the measured tracer concentrations Žpptv., if no other processes affect the samples. The endpoints for the curves at zero age correspond to samples taken in January, 1997 Žlocal

Fig. 6. Comparison of the mixing ratios of CFC-11 and CFC-12 ŽA., and CFC-113 and CFC-12 ŽB. for water from all springs and wells sampled in April, 1996 and August–September, 1997. The data are compared to three hypothetical mixing models. The dashed lines correspond to binary mixing of modern Ž1997. water with old Žpre-CFC. water. The long–short dashed lines correspond to exponential mixing in discharge from groundwater reservoirs with mean residence times of 0 to more than 100 years. The solid lines correspond to recharge of water in equilibrium with air Žpiston flow. for ages Žin years from 1997. of 0 to more than 30 years. Numbers referenced to the model lines are travel times Žpiston flow. or mean residence times Žexponential mixing.. The models are insensitive to variations in excess air content of 0–2 cm3 kgy1 , and were calculated without excess air.

SF6 concentration in air s 4.15 pptv., which is approximately the mid-point of the two sampling periods Žbetween April, 1996 and August–September,

100

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

Fig. 7. Plots showing relationship of SF6 and CFC-12 mixing ratios ŽA., and 3 H activity ŽTU. and SF6 mixing ratios ŽB. for all springs and wells sampled in April, 1996 and August–September, 1997. The data are compared to models of piston flow, exponential mixing and binary mixing ŽCook and Bohlke, 1999., calcu¨ lated without excess air Žheavy lines. and with 2 cm3 kgy1 of excess air added Žlight lines.. The letters in ŽA. identify samples from Furnace Spring ŽF. and Lewis Spring ŽL.. Two input functions for 3 H were considered in ŽB.. The upper line shows results from the exponential model with a tritium record from Washington, DC. The other lines assume a local tritium input function scaled to the Washington, DC record Žsee text for further explanation..

1997.. Corresponding curves for the actual sample dates would have SF6 endpoints at 4.01 ŽApril, 1996. and 4.29 pptv ŽAugust, 1997.. The differences in SF6 mixing ratios in air between April, 1996 and August–September, 1997 are not significant. The model calculations with CFCs ŽFig. 6. are nearly insensitive to expected variations in amount of excess air, and local air enrichment in CFC-12. The calculations involving SF6 are also insensitive to local enrichment, but depend significantly on variations in excess air ŽFig. 7A.. The relative sensitivity of SF6 models to excess air is a consequence of the low Henry’s law solubility of SF6 compared to those of CFCs. Excess air was added to the NA air curves for SF6 by addition of excess air to water-saturated values and re-calculation of the internal gas partial pressure for a recharge elevation of 823 m Žthe mean elevation of all samples from SNP, including some lower elevation springs and wells. and a recharge temperature of 9.28C Žthe average recharge temperature between April, 1996 and August–September, 1997.. The composition of the excess air added was assumed to be that of air in the corresponding recharge year. Most of the samples have concentrations of CFC11, CFC-12, and CFC-113 corresponding to low apparent ages ŽFig. 6A,B.. Based on CFC-113 and CFC-12 concentrations, most samples from springs Žand some wells. consist of water recharged within the past 0–7 years ŽFig. 6B.. Because of the similar nature of the atmospheric input functions for CFC-11 and CFC-12, the piston flow and exponential models cannot be distinguished based on CFC-11 and CFC12 ŽFig. 6A.. The spring sample with the lowest CFC concentrations ŽFig. 6. is from Hudson Spring at the base of the Blue Ridge Mountains at Luray, VA. A few samples, mostly from wells, plot near or within the envelope between the exponential and piston-flow models ŽFig. 6B.. These samples lie along hypothetical mixing lines that can be drawn between the piston flow line and old ŽCFC-free. water, and may be interpreted as binary mixtures of relatively young water with old water. Most other samples from wells have CFC-12, and to a lesser extent, CFC-113 concentrations that are greater than that possible for water in solubility equilibrium with modern Niwot Ridge air, or with air locally enriched

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

in CFC-12, as in Fig. 2. Some of these AcontaminatedB samples have as much as two to three times normal air–water equilibrium concentrations, plot beyond the limits of the graphs, and apparently indicate waters that have been affected by local anthropogenic sources of CFCs in SNP, beyond that of exchange with the atmosphere. Dissolved nitrate concentrations average 0.93 " 0.7 and 0.95 " 0.9 mg ly1 Žas N. in water from springs and wells, respectively, and are apparently unaffected by anthropogenic sources in SNP. However, because of the low detection limit of CFCs, and their association with anthropogenic sources, it is believed that the CFCs provide a sensitive and early indicator of susceptibility to contamination from human sources. Because most of the springs are from rural, uninhabited areas with virtually no opportunity for anthropogenic contamination beyond that of locally enriched air, other processes must be considered to account for samples from springs that plot outside of the region defined by the mixing models ŽFigs. 6 and 7.. One possible explanation is that estimated recharge temperatures, especially from August–September, 1997, are too high, and lead to calculated CFC and SF6 partial pressures greater than those possible based on the model calculations. Samples showing the greatest deviation from the model calculations in CFC-12 and SF6 ŽFig. 7A. are from Lewis Spring Žlabeled L. and Furnace Spring Žlabeled F.. These two springs have the most extensive subsurface development of all springs in SNP, and greatest opportunity for re-equilibration of gases in contact with air. However, the He and Ne data indicate little or no evidence for gas fractionation in samples from Furnace and Lewis springs. Two groups of waters are evident in the SF6 – CFC-12 data that cluster along the line for piston-flow recharge ŽFig. 7A.. The separation of the two sets of seasonal samples is, in part, a function of the increase in the local SF6 content of air between the April, 1996 Ž4.0 pptv. and August–September, 1997 Ž4.3 pptv. sampling dates. Also, if the waters are considered mixtures of recent recharge that spans several seasons, the N2 –Ar temperatures for the August–September, 1997 samples may be too high by as much as 38C. However, as discussed below, other factors may be responsible for the separation in

101

ages of spring discharge between the April, 1996 and the August–September, 1997 sampling. The data of Fig. 7B indicate that most waters from springs sampled in August–September, 1997 have 3 H and SF6 concentrations similar to those of modern waters. Spring samples from April, 1996 have 3 H activities similar to those from August–September, 1997 and SF6 concentrations somewhat lower than those from August–September, 1997. The few samples from wells with elevated tritium were prob-

Fig. 8. Comparison of apparent Žpiston flow. ages based on SF6 and 3 Hr3 He for water from all springs and wells ŽA. April, 1996, and ŽB. August–September, 1997. Error bars correspond to analytical errors in the 3 Hr3 He data and to variations of "5% in SF6 air mixing ratios.

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

102

ably recharged at times nearer the mid-1960s bomb peak, and subsequently mixed with more recently recharged water.

deep springs at the base of the Blue Ridge ŽHudson and Yeager Springs.. The average SF6 apparent age for the August– September, 1997 sample set is 1.3 " 2.3 years ŽTable 1., compared to 4.4 " 1.4 years from the April, 1996 group Žusing the N2 –Ar temperatures.. Even if we assume both sample sets were recharged at an average of 88C, the average apparent age of the 1997 sample set Ž2.1 " 2.0 years. is less than that of the 1996 set Ž4.8 " 1.6 years.. The differences in the SF6 apparent ages between the April, 1996 and August– September, 1997 sampling dates Ž3.1 years using the N2 –Ar temperatures, or 2.7 years assuming 88C. is not understood and appears to be inconsistent with SF6 data collected approximately monthly during 1999 from four springs in SNP. Of the five springs sampled monthly in 1999, four are located in the higher elevations of SNP ŽByrd’s Nest 3, Browntown Valley Overlook, Furnace, and Lewis spring.. All four springs have low SF6 ages during early spring ŽMarch–April, 1999, range 0.3– 1.8 years. and higher ages during the low-flow period of late summer ŽAugust–September,1999, range from 0.7 to 3.2 years.. The more confined springs, ŽByrd’s Nest 3 and Browntown Valley Overlook. have generally higher SF6 ages than those from the lesser-confined ŽFurnace and Lewis. springs, and indicate a generally steady increase in apparent age between early spring and late summer. The 1999 SF6 results from these four springs agree well with previous SF6 results from August–September, 1997 but not with the earlier April, 1996 samples. This sug-

4.5. Comparison of 3H r3He and SF6 ages Fig. 8 compares piston-flow ages based on SF6 ŽBusenberg and Plummer, 2000. with ages calculated from the 3 Hr3 He data ŽSchlosser et al., 1988, 1989, 1998; Solomon and Sudicky, 1991; Solomon et al., 1993; Solomon and Cook, 1999. for all wells and springs from April, 1996 and August–September, 1997. No 3 Hr3 He data were obtained in monthly sampling in 1999, but by comparison to the April, 1996 and August-September, 1997 3 Hr3 He data, it is reasonable to assume that 3 Hr3 He ages in the range of 0–2 years would be obtained from the 1999 data set. The SF6 ages were calculated from the analyzed SF6 concentrations in water samples and based on the adjusted local air curve for SNP Ž5% enrichment., the measured N2 –Ar temperature for each spring, the elevation of the spring, and adjusted for excess air as determined by N2 –Ar concentrations. The 3 Hr3 He ages were corrected for air sources of He, but, for most samples, no terrigenic helium correction was warranted, because the helium concentrations were within the range expected for solubility equilibrium and addition of excess air. Only 7 of the 107 3 Hr3 He samples required correction for terrigenic helium and most of these were from a deep well in SNP ŽDickey Ridge well. and

Table 1 Comparison of 3 Hr3 He and SF6 apparent ages for ground waters sampled from Shenandoah National Park Sample set

Mean 3 Hr3 He age Žyears.

Median 3 Hr3 He age Žyears.

Range 3 Hr3 He age Žyears.

Springs, April, 1996 Springs, August– September, 1997 Four springs, March– April, 1999 Four springs, August– September, 1999 Wells, April, 1996 Wells, August– September, 1997

0.4 " 0.7 0.8 " 1.1

0.2 0.3

y1.1 to 2.6 y0.3 to 4.6

n.d.

n.d.

n.d. 6.5 " 6.8 8.0 " 7.9

n.d.—not determined.

Mean SF6 age Žyears.

Median SF6 age Žyears.

Range SF6 age Žyears.

4.4 " 1.4 1.3 " 2.3

4.3 0.7

0.3 to 7.8 y1.8 to 9.2

n.d.

1.2 " 0.5

1.2

0.3 to 1.8

n.d.

n.d.

2.1 " 0.8

2.4

0.7 to 3.2

3.7 6.1

0.3 to 23.7 0.2 to 25.3

10.3 " 4.5 7.4 " 5.9

9.8 4.2

3.3 to 18.3 1.4 to 17.7

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

gests that there may be some artifact associated with collection of the April, 1996 SF6 samples that is responsible for the differences of ages observed between spring, 1996 and spring 1999. Table 1 compares the average, median and range of ages of waters from springs and wells in 1996 and 1997 based on the 3 Hr3 He and SF6 data, and for the four springs sampled for SF6 in 1999. For the April, 1996 samples, the 3 Hr3 He data indicate ages of most spring discharge that are younger than those based on SF6 . For the dry-season samples in August–September, 1997, both the 3 Hr3 He and SF6 mean apparent ages are approximately 1 year ŽTable 1., and similar to the range of SF6 ages found throughout 1999. In contrast, both 3 Hr3 He and SF6 indicate a wide range of ages from near 0 to approximately 25 years for water from many of the wells. The ages of well waters based on SF6 are not as reliable as those determined from 3 Hr3 He data, because several processes affect the SF6 Žand CFC. concentrations. Three groups of ages of waters from wells are recognized ŽFig. 8.. Those samples with SF6 ages greater than the 3 Hr3 He age are probably mixtures in which the SF6 concentration associated with the young fraction has been diluted with old water, and the 3 Hr3 He age applies to the young fraction in the mixture. Some of these same samples have low CFC concentrations consistent with mixing ŽFig. 6A,B.. Well samples with SF6 ages less than the 3 Hr3 He age may contain an additional terrigenic source of SF6 ŽBusenberg and Plummer, 2000.. Other samples from wells and a few springs show reasonably good correspondence in age between SF6 and 3 Hr3 He ŽFig. 8B., but this apparent agreement could be caused in part by compensating errors of combined binary mixing and addition of terrigenic SF6 . 4.6. Seasonal Õariation in

18

103

unsaturated zone ŽDeWalle et al., 1997.. Therefore, residence times based on seasonal variations in d18 O can be greater than ages based on SF6 , 3 Hr3 He and CFCs which represent time elapsed since gas confinement at the water table. The isotopic composition of precipitation at Big Meadows, SNP Ž2-year record from 1997 through 1998. and that of water from all wells and springs ŽApril, 1996 and August–September, 1997. plot parallel to the global meteoric water line with a deuterium excess Ž d2 H excess ' d2 H–8 d18 O. of 16.4‰ for waters from springs and wells and 16.1‰ for precipitation ŽFig. 9.. The relatively high deuterium excess in SNP waters is similar to that in meteoric groundwater from other parts of the mid-Atlantic region of the eastern US Žsee for example, Dunkle et al., 1993.. The seasonal variation in d18 O of precipitation at Big Meadows ranges over nearly 10‰, while the d18 O of water from all springs and wells varies over less than 2‰. The isotopic composition of monthly discharge from five selected springs ŽApril, 1996–September, 1999. was remarkably constant relative to that of precipitation in SNP ŽFig. 10.. The separation in d18 O values between the five springs is related, in part, to spring elevation that

O

If the seasonal variation in d18 O Žor d2 H. in local precipitation is known, observations of seasonal variations in the stable isotope composition of spring discharge can be used to estimate mean residence time ŽMaloszewski et al., 1983; Burgman et al., 1987; Vitvar and Balderer, 1997; Vitvar, 2000.. Groundwater residence times based on d18 O variations include the residence time of water in the unsaturated zone and effects of mixing within the

Fig. 9. Comparison of the monthly composited stable isotopic composition of precipitation from the Big Meadows Air Monitoring Station ŽAugust, 1997 through August, 1999. with the range in isotopic composition of water from springs and wells in SNP.

104

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

Ridge mountains ŽBrowntown Valley Overlook Spring, 884 m.. The smoothing of the isotopic signal from precipitation in spring discharge is characteristic of output from a mixed reservoir that could have a variety of age distributions. Fig. 11 shows a schematic example in which the precipitation record from Big Meadows is repeated as input to a groundwater reservoir with mean residence time of 2 years. In this example, the input function Žprecipitation composited approximately monthly. has an amplitude and standard deviation of 9.6‰ and 2.8‰, respectively, in d18 O. If the mean residence time of transit were 2 years, and flow lines were exponentially mixed in discharge, we would observe an amplitude of 1.1‰ and standard deviation of 0.3‰ in d18 O of spring discharge. CalculaFig. 10. Comparison of the stable isotopic composition of water from five springs from 1996 to 1999 with the stable isotopic composition of precipitation from the Big Meadows Air Monitoring Station.

ranges from 233 ŽHudson Spring. to 1045 m ŽFurnace Spring., and whether the spring is recharged from the east side of the Blue Ridge mountains ŽFurnace Spring, Lewis Spring Ž1012 m. and Byrd’s Nest 3 Spring Ž914 m.. or the west side of the Blue

Fig. 11. Schematic showing the smoothing of the measured isotopic composition of atmospheric precipitation in exponential discharge from a groundwater reservoir with mean residence time of 2 years. The input function is the stable isotope record for precipitation at Big Meadows and has an average d18 O composition of y8.2‰, with amplitude of 9.6‰ and standard deviation of 2.8‰. If discharge from a spring represented an exponential mixture of flow lines from the groundwater reservoir with mean residence time of 2 years, the spring discharge would have an average d18 O composition of y8.2‰, with amplitude of 1.1‰ and standard deviation of 0.3‰. The modeled amplitude and standard deviation of the isotopic composition of spring discharge decrease with increasing mean groundwater residence time.

Fig. 12. Lines showing calculated seasonal variation in d18 O of spring discharge as one standard deviation and the maximum amplitude of isotopic variation calculated using the isotopic composition of precipitation in SNP and assumed exponential mixing of discharge from springs and wells with groundwater residence times of 0.5 to about 20 years. The points represent apparent ages based on CFC-12, SF6 and 3 Hr3 He as a function of their observed seasonal variation in d18 O ŽApril, 1996 minus August– September, 1997.. Most tracer-based ages of spring discharge are biased younger than those mean residence times implied by the stable isotope data. The average seasonal variation in d18 O for water from all springs and wells in SNP is 0.12‰, corresponding to a mean residence time for groundwater in SNP of about 5 years.

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

105

isotope-based age estimate includes the groundwater age, mixing and transit time in the unsaturated zone. 4.7.

35

S

35

Fig. 13. Activity of 35 S of sulfate in precipitation at the Big Meadows Air Monitoring Station, SNP, 1997–1999, and in discharge from approximately monthly samples from five springs in SNP Ž41 measurements in all. during 1999.

tions similar to that of Fig. 11 were repeated for mean groundwater residence times of 0.1–20 years Žsimilar to that of Burgman et al., 1987., leading to the relation of Fig. 12 showing calculated seasonal variation in d18 O as a function of mean residence time for transit through the system. The average seasonal variation in d18 O for all springs and wells is 0.12‰, which is close to the 1 s analytical precision of the d18 O measurements of "0.1‰. The observed range in seasonal variation of stable isotope data from springs implies a range of mean groundwater residence times of about 0–10 years in SNP. Although seasonal variations of - 0.1‰ are not statistically significant for a single pair of measurements from a single site, as a population, the mean seasonal variation in d18 O for all springs and wells of 0.12‰ corresponds to a mean exponential groundwater residence time in SNP of about 5 years ŽFig. 12.. This age estimate based on seasonal variations in d18 O is at least several years greater than the apparent ages of spring discharge based on CFC-12, SF6 , and 3 Hr3 He. The difference in age estimates is in part a function of the fact the gas-based tracer ages more closely represent time since recharge and gas confinement, while the stable

S is produced in the atmosphere by cosmic-ray spallation of Ar ŽTanaka and Turekian, 1995., and decays with a half-life of 87 days. 35 S has been used to estimate mean ages of water discharging from springs and runoff in small mountainous watersheds in Colorado ŽMichel and Turk, 1995; Sueker et al., 1999; Michel et al., 2000., Virginia ŽBohlke et al., ¨ 1996, submitted for publication. and New York ŽBurns et al., 1998.. The activity of 35 S of sulfate in precipitation at the Big Meadows Air Monitoring Station ranges from about 10 to 35 milli-Becquerels per liter of sample ŽmBq ly1 . ŽFig. 13., and averages 18.5 mBq ly1 . Fig. 13 indicates that 35 S activity of sulfate in discharge from five springs sampled approximately monthly over 0.5 years in 1999 was zero or nearly so

Fig. 14. Comparison of SF6 partial pressures with 35 S activities measured in water from five springs in 1999. The lines correspond to model calculations assuming piston flow, exponential mixing and binary mixing of modern and old water. The model curves assume addition of 0.5 cm3 kgy1 of excess air, near the average excess air content of water from the four springs in SNP. Three samples from Byrd’s Nest 3 had excess air of about 2.0 cm3 kgy1 and plot above the other SNP samples. The bar on the right side shows the range of initial SF6 pptv values if the initial water contained between 0 and 2 cm3 kgy1 of 1999.5 excess air.

106

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

within the uncertainties of the measurements. Of the 41 measurements of 35 S activity at these five springs ŽFig. 13., only eight values were higher than the 1 s uncertainty above zero, the largest being 1.2 " 0.3 mBq ly1 in a sample from Browntown Valley Overlook Spring collected 2 days after a major precipitation event in September, 1999. The low 35 S activities imply ages greater than 1.5 years, which is equal to or greater than the SF6 ages that averaged 1.2–2.5 years at the four springs sampled monthly in SNP in 1999 ŽFig. 14.. Samples from two deep springs at the base of the Blue Ridge Mountains near Luray, VA have very low 35 S activities and small excesses of Žterrigenic. SF6 ŽFig. 14.. Because most of the 35 S activities are beyond the dating range of 35 S, the 35 S data indicate groundwater residence times greater than 1.5 years. The maximum detected 35 S activity in spring discharge Ž1.2 mBq ly1 . could correspond to a mean exponential residence time of more than 15 years if transported conservatively, but the data of Fig. 14 show little evidence of exponential mixing on time scales of 15 years. If the spring waters are interpreted as mixtures, then the low 35 S activities could indicate that dissolved SO42y does not move conservatively with groundwater in SNP. Sulfate concentrations decreased 32% in spring discharge between April, 1996 and August–September, 1997, while dissolved chlo-

ride concentrations increased 17%, indicating that the 35 S activity of recharge may have decreased somewhat by exchange with the biomass, and possibly by sorption onto soils. The maximum observed 35 S activity in spring discharge is consistent with a mean exponential mixing residence time of about 5–7 years, if approximately 50% of the 35 S is removed by the biomass andror sorption during recharge. 4.8. Transient response in spring discharge Measurements of water temperature and specific conductance at 30-min intervals in discharge from five springs for periods of 6–12 months were compared to the timing of precipitation events and changes in air temperature recorded hourly at the Big Meadows Air Monitoring Station Žsee Fig. 15 for results from Furnace Spring, March, 1999 through February, 2000.. Seasonal warming of spring discharge lags several months behind warming of air temperature ŽFig. 15.. Following significant precipitation from remnants of Hurricane Floyd in early September, 1999, water temperature and specific conductance increased abruptly within several hours of the event at Furnace Spring. This was followed within several hours by falling water temperature and decreasing specific conductance ŽFig. 15.. The

Fig. 15. Continuous Ž30-min. measurements of specific conductance and water temperature for discharge from Furnace Spring from March 1999 through February, 2000. Also shown are daily precipitation and average daily air temperature from the Big Meadows Air Monitoring Station.

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

initial rise in specific conductance and water temperature probably represents rapid recharge of soil pore water in the vicinity of the spring. Water temperature then fell to early summer values even though the air temperature was approximately 108C warmer than the groundwater temperature. This suggests that significant recharge of water to the catchment from the event in early September, 1999 caused increased discharge of water from somewhat deeper sources ŽEvans and Davies, 1998.. Although spring discharge increases significantly in SNP following major precipitation events, these results suggest that the increased flow may include a significant fraction of pre-event groundwater ŽKennedy et al., 1986; Rice and Hornberger, 1998., or at least, discharge of water from somewhat greater depths in the groundwater reservoir.

5. Discussion 5.1. Gas confinement and groundwater age From the perspective of dating groundwater with environmental tracers, information on dissolved N2 , Ar, Ne and He provides insights into processes that may be affecting the concentrations of other gases, such as CFCs, SF6 , and 3 HeŽtrit. , on which age estimates are based. Groundwater age based on CFCs, SF6 and 3 Hr3 He is defined as the time elapsed since the water sample was isolated from the unsaturatedzone air, and can be sensitive to recharge temperature, excess air and gas-exchange processes. Three processes can be considered to explain the observed changes in dissolved gas concentrations: Ž1. seasonal recharge with subsequent gas confinement, Ž2. diffusional gas exchange between trapped bubbles and unsaturated-zone air in the recharge area, and Ž3. maintenance of solubility equilibrium in response to temperature change subsequent to recharge. Gas bubbles can form near the water table in samples with excess air. Gas diffusion into and out of gas bubbles can cause dissolved gases to be fractionated and lead to uncertainty in calculated excess air and recharge temperatures ŽStute and Schlosser, 1999.. The aqueous diffusion coefficients

107

of N2 and Ar ŽNg and Walkley, 1969; O’Brien and Hyslop, 1977. are nearly identical, and thus, gas exchange, if it occurs, would not be evident in differences in excess quantities of dissolved N2 and Ar. However, because the aqueous diffusion coefficient of He is nearly twice that of Ne ŽJahne et al., ¨ 1987., dissolved He and Ne data can be sensitive indicators of gas-exchange processes. If gas exchange were driven by diffusion ŽStute and Schlosser, 1999., the excess air determined from He, Ne and N2 –Ar would be in the order He - Ne ŽN2 –Ar.. However, the observed order of excess air quantities in spring discharge is He f Ne f ŽN2 –Ar. ŽFig. 5.. Thus, the He and Ne data are not fractionated, and excess air quantities calculated from He, Ne and N2 –Ar are nearly identical. This suggests gas confinement following recharge. Recharge probably occurs in SNP throughout the water year, and it is thought that spring discharge represents mixtures containing fractions of seasonal recharge. Although it is likely that less recharge occurs during the summer months than during late winter–spring, spring discharge likely contains fractions of seasonal water throughout the year. In the simplest case then, the dissolved gases may represent mixtures of mostly confined seasonal recharge. Gas confinement following recharge is likely because the excess air quantities calculated from He, Ne and N2 –Ar concentrations ŽFig. 5. are similar, indicating that diffusional gas loss is not significant. Apparent gas confinement is even observed in discharge from Furnace and Lewis springs that have large subsurface pipes and collections conduits that permit more contact with air than most other springs in SNP. Solomon et al. Ž1997. reported that waters with apparent CFC ages of 10–20 years exchanged little with the atmosphere after discharge to a stream and flow of 1 km in contact with the air. They concluded that because of slow gas exchange, CFC ages of groundwater are at least partially preserved in streams in base flow. There is less opportunity for gas exchange in spring discharge in SNP than in the surface water study of Solomon et al. Ž1997.. The possibility of some gas-exchange with the unsaturated zone air cannot be entirely excluded because the springs are shallow and contact air along the water table and in some discharge areas. Gas exchange, if it occurs, would cause a low bias in

108

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

groundwater ages based on CFCs, SF6 and 3 HeŽtrit. . Because the He diffusion coefficient is approximately seven times that of CFCs or SF6 , there is greater potential for loss of 3 HeŽtrit. to the unsaturated zone air than diffusional exchange of CFCs and SF6 with the soil air. If we exclude the April, 1996 SF6 results, the 3 Hr3 He and SF6 ages agree well ŽTable 1, Fig. 8B. within the uncertainties of the dating methods, thus, there is little evidence for differential loss of 3 He relative to SF6 from the water samples. Although these relations appear to be qualitatively consistent with the age and dissolved gas observations, further study is needed. Some samples from springs, and most samples from wells, show no seasonality in recharge temperature and indicate confined warming in the ground following recharge ŽFig. 5.. The ages of water from these samples more closely reflect time since recharge.

5.2. Mixing of groundwater discharge from wells Most wells are completed in fractured metamorphic rocks and intercept only a few water-bearing fractures. While water from springs appears to be mixtures from only young sources, water from wells can be mixtures of water from different fracture sources and age. The apparent piston-flow CFC and SF6 ages of well water, if not corrected for mixing, can cause apparent ages that are older than 3 Hr3 He ages. 3 Hr3 He ages are based on an isotope ratio that is little affected by mixing with old Žpre-nuclear detonation. water ŽPlummer et al., 1998a,b, 2000b.. Plummer and Busenberg Ž1999. describe CFC results from one well in SNP ŽDickey Ridge well. that had an apparent recharge date of 1974 based on CFC-11 and CFC-12, and 1980 based on CFC-113. Such a separation in apparent age based on CFCs is characteristic of binary mixtures of young and old water ŽPlummer and Busenberg, 1999; Plummer et al., 2000b.. Assuming a binary mixing model, the calculated discharge from the well consists of 46% water recharged in mid-1986 with 54% pre-CFC water. The dilution-corrected CFC-based age of 11.3 years for the young fraction compared favorably with 9.7 years determined by the 3 Hr3 He method.

5.3. Summary of groundwater residences times of spring discharge Apparent groundwater residence times obtained for spring discharge based on CFCs, SF6 , 3 Hr3 He, 35 S and seasonal variation in stable isotope data all indicate that young water discharges from springs in SNP, yet there are some apparent differences between the individual tracers. Several processes contribute to the observed age differences. Ages of spring discharge based on CFCs were consistent with water recharged in the 1990s but, except for some of the CFC-12 data, could not be resolved further due to the nearly constant atmospheric mixing ratios of CFCs during the past 10 years. Ages based on 35 S are complicated by the very low activities of 35 S in SNP and possibly by the retardation of SO4 by exchange with plants, travel time through the unsaturated zone, and soil sorption. Nevertheless, a minimum age of 1.5 years is indicated by the observed maximum activity of 35 S. The SNP 35 S results differ from the 35 S data for spring discharge and runoff from mountainous catchments in the Colorado front range that contained 35 S and a significant fraction of snowmelt sulfate from the previous winter ŽMichel and Turk, 1995; Sueker et al., 1999; Michel et al., 2000.. However, comparison of sulfate residence time with water residence time is probably complicated by SO4 uptake by plants andror sorption by soils. The average residence time of spring discharge based on d18 O variations is on the order of 5 years. The relatively older residence time based on d18 O appears consistent with the mean age of spring discharge in April, 1996, based on SF6 which averaged 4.4 years. However, the SF6 results from the April, 1996 samples were anomalous in comparison to SF6 results from the August–September, 1997 samples, or in comparison with samples from any season during 1999 at four springs sampled monthly in SNP. We have tentatively excluded the April, 1996 results, though further study of spring discharge in SNP under high-flow conditions is warranted. The SF6 ages from August–September, 1997, and those from monthly sampling in 1999 are in agreement with low ages obtained from 3 Hr3 He dating in April, 1996 and August–September, 1997, and in the range of 0–3 years for spring discharge throughout

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

SNP. Groundwater residence times and ages based on 18 O and 35 S data are higher than those based on gases, because the estimates based on 18 O and 35 S include the transit time through the unsaturated zone and mixing processes within the unsaturated zone, while the gas-based ages correspond to only the time since recharge and gas confinement at the water table. The 3 Hr3 He ages are younger by less than 1 year relative to those from SF6 ŽTable 1.. This difference may be due to incomplete confinement of 3 He in spring discharge, however, the small difference is within the uncertainties of the two dating methods. Sulfur hexafluoride appears to give reliable age estimates for water from springs in SNP, because of relatively rapid throughput of water through shallow springs, which minimizes the accumulation of terrigenic SF6 in spring discharge. The results from SF6 agree well with those from 3 Hr3 He and are consistent with the CFC data.

6. Conclusions Concentrations of CFCs, 3 H, 3 Hr3 He, SF6 , 35 S, O, 2 H, N2 , Ar, He and Ne were used to investigate groundwater residence times for discharge from springs and wells in Shenandoah National Park ŽSNP. in the Blue Ridge Mountains of Virginia, USA. Some of the conclusions are as follows: Ž1. Mean groundwater residence times of 0–3 years were determined for discharge from springs in SNP based on SF6 and 3 Hr3 He data. Ž2. Groundwater from wells and deep springs in the vicinity of SNP ranges in age from 0 to 25 years. The CFC and 3 Hr3 He data indicate that some water samples from wells are mixtures of relatively young Ž0–10 year old. water with older, pre-CFC water. Ž3. Concentrations of CFC-11, CFC-12, CFC-113 and SF6 in spring discharge in SNP are all consistent with young ages. The CFC input functions cannot be used to distinguish ages using piston flow or exponential mixing models for waters that recharged in the past 10 years. Ž4. The 35 S data indicate minimum residence times of 1.5 years Žpiston flow. and mean ages as large as 15 years Žexponential mixing. for atmospheric sulfate in water from five selected springs. 18

109

However, transport of dissolved sulfate is probably retarded by uptake and exchange of sulfate in plants, andror soil sorption. If approximately 50% of the 35 S is lost to the biomass, exponential residence times of 5–7 years for water are possible. A similar transit time of about 5 years for transport through the soils to spring discharge is indicated by the 18 O data. Ž5. The recharge temperatures of spring water calculated from N2 –Ar data are approximately 2.68C warmer in the August–September, 1997 sample set relative to the waters from April, 1996. Amounts of excess air in spring discharge range from about 0–1 cm3 kgy1 , with nearly identical quantities of excess air calculated from He, Ne and N2 –Ar. The absence of gas fractionation between He and Ne indicates rapid confinement, and the seasonal variation in N2 –Ar temperatures indicates that shallow recharge may occur throughout the year. Excess air in water from wells ranges from 0–8 cm3 kgy1 , and is probably added to water in the fracture system during rapid rises in water levels. Ž6. Water temperature and specific conductance measurements at 30-min intervals at five springs over a period of as long as 12 months indicate that the increase in spring discharge following major precipitation events contains an increased proportion of pre-event base flow. Acknowledgements This study was conducted in cooperation with the National Park Service. The work was supported by funds from the US Geological Survey National Research Program, and the Office of Ground Water, US Geological Survey. Precipitation and air samples were collected by Shane Spitzer, Shenandoah National Park at the Big Meadows Air Monitoring Station. Meteorological data were provided by the National Park Service Air Resources Division. We thank Tyler Coplen, US Geological Survey, Reston, VA for stable isotope data. Field assistance from Julian Wayland, Gerolamo Casile, Michael Doughten, Peggy Widman, Wandee Kirkland, and Anne Burton ŽUSGS, Reston, VA., Steve Richards ŽNPS, SNP., Vanessa Trompetter ŽInstitute of Geological and Nuclear Sciences, Lower Hutt, New Zealand., and John Weaver ŽCSIR, Stellenbosch, South Africa. is gratefully acknowledged. We thank

110

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111

Gerolamo Casile and Julian Wayland for chlorofluorocarbon analyses, Peggy Widman for dissolved N2 and Ar analyses, Wandee Kirkland for SF6 analyses, and Michael Doughten for water chemistry analyses. Janet Hannon assisted with the precipitation isotope study. Computer and graphical assistance from Sarah Parnes and Brian Norton is gratefully acknowledged. The manuscript was improved by reviews from Karen C. Rice ŽUSGS, Charlottesville, VA., Stephanie D. Shapiro ŽUSGS, Reston, VA ., Martin Stute ŽLamont-Doherty Earth Observatory, Palisades, NY., Peter Cook ŽCSIRO, Australia., D. Kip Solomon ŽUniversity of Utah, Salt Lake City, UT., and an anonymous reviewer. Lamont-Doherty Earth Observatory of Columbia University contribution No. 6208.

References Bohlke, J.K., O’Brien, A.K., Michel, R.L., 1996. Isotopic evi¨ dence for discharge of old atmospheric sulfate from a forested mountain catchment in Virginia. EOS, Trans. Am. Geophys. Union 77, S102. Bohlke, J.K., O’Brien, A.K., Michel, R.L., submitted for publica¨ tion. Isotopic evidence for delayed discharge of atmospheric sulfate from a forested watershed: results of 3 H, 35 S, 34 S, 18 O, 2 H, and mass balance monitoring of two streams and a spring at Massanutten Mountain, Virginia. Hydrol. Processes. Burgman, J.O., Calles, B., Westman, F., 1987. Conclusions from a ten year study of oxygen-18 in precipitation and runoff in Sweden. Isotope Techniques in Water Resources Development. IAEA, Vienna, Austria, pp. 579–590. Burns, D.A., Murdoch, P.S., Lawrence, G.B., Michel, R.L., 1998. Effect of groundwater springs on NOy 3 concentrations during summer in Catskill Mountain streams. Water Resour. Res. 34, 1987–1996. Busenberg, E., Plummer, L.N., 1992. Use of chlorofluorocarbons ŽCCl 3 F and CCl 2 F2 . as hydrologic tracers and age-dating tools: example—the alluvium and terrace system of central Oklahoma. Water Resour. Res. 28, 2257–2284. Busenberg, E., Plummer, L.N., 2000. Dating young groundwater with sulfur hexafluoride: natural and anthropogenic sources of sulfur hexafluoride. Water Resour. Res. 36, 3011–3030. Busenberg, E., Weeks, E.P., Plummer, L.N., Bartholemay, R.C., 1993. Age dating ground water by use of chlorofluorocarbons ŽCCl 3 F and CCl 2 F2 ., and distribution of chlorofluorocarbons in the unsaturated zone, Snake River Plain aquifer, Idaho National Engineering Laboratory, Idaho. U.S. Geol. Surv. Water-Resour. Invest. Rept. 93-4054, 47 pp. Cook, P.G., Bohlke, J.K., 1999. Determining time scales for ¨ groundwater flow and solute transport. In: Cook, P., Herczeg, A. ŽEds.., Environmental Tracers in Subsurface Hydrology. Kluwer Academic Publishing, Amsterdam, pp. 1–30.

Cook, P.G., Solomon, D.K., 1997. Recent advances in dating young groundwater: chlorofluorocarbons, 3 Hr3 He, and 85 Kr. J. Hydrol. 191, 245–265. Cook, P.G., Solomon, D.K., Plummer, L.N., Busenberg, E., Schiff, S.L., 1995. Chlorofluorocarbons as tracers of groundwater transport processes in a shallow, silty sand aquifer. Water Resour. Res. 31, 425–434. DeKay, R.H., 1972. Development of ground-water supplies in Shenandoah National Park, Virginia. Virginia Division of Mineral Resources Report 10, Charlottesville, Virginia, 158 pp. DeWalle, D.R., Edwards, P.J., Swistock, B.R., Aravena, R., Drimmie, R.J., 1997. Seasonal isotope hydrology of three Appalachian forest catchments. Hydrol. Processes 11, 1895–1906. Dunkle, S.A., Plummer, L.N., Busenberg, E., Phillips, P.J., Denver, J.M., Hamilton, P.A., Michel, R.L., Coplen, T.B., 1993. Chlorofluorocarbons ŽCCl 3 F and CCl 2 F2 . as dating tools and hydrologic tracers in shallow ground water of the Delmarva Peninsula, Atlantic Coastal Plain, United States. Water Resour. Res. 29, 3837–3860. Ekwurzel, B., Schlosser, P., Smethie Jr., W., Plummer, L.N., Busenberg, E., Michel, R.L., Weppernig, R., Stute, M., 1994. Dating of shallow groundwater: comparison of the transient tracers 3 Hr3 He, chlorofluorocarbons and 85 Kr. Water Resour. Res. 30, 1693–1708. Elkins, J.W., Thompson, T.M., Swanson, T.H., Butler, J.H., Hall, B.D., Cummings, S.O., Fisher, D.A., Raffo, A.G., 1993. Decrease in the growth rates of atmospheric chlorofluorocarbons 11 and 12. Nature 364, 780–783. Eriksson, E., 1958. The possible use of tritium for estimating groundwater storage. Tellus 10, 472–478. Evans, C., Davies, T.D., 1998. Causes of concentrationrdischarge hysteresis and its potential as a tool for analysis of episode hydrochemistry. Water Resour. Res. 34, 129–137. Focazio, M.J., Plummer, L.N., Bohlke, J.K., Busenberg, E., Bach¨ man, L.J., Powers, D.S., 1998. Preliminary estimates of residence times and apparent ages of ground water in the Chesapeake Bay watershed and water-quality data from a survey of springs. U.S. Geol. Surv. Water-Resour. Invest. Rept. 97-4225, 75 pp. Gathright II, T.M., 1976. Geology of the Shenandoah National Park, Virginia. Va. Div. Miner. Resour., Bull. 86, 1–93 Charlottesville, Virginia. Heaton, T.H.E., 1981. Dissolved gases: some applications to groundwater research. Trans. Geol. Soc. S. Afr. 84, 1–97. Heaton, T.H.E., Vogel, J.C., 1981. AExcess airB in groundwater. J. Hydrol. 50, 201–216. Heaton, T.H.E., Talma, A.S., Vogel, J.C., 1983. Origin and history of nitrate in confined groundwater in the Western Kalahari. J. Hydrol. 62, 243–262. Herzberg, O., Mazor, E., 1979. Hydrological applications of noble gases and temperature measurements in underground water systems: examples from Israel. J. Hydrol. 41, 217–231. Jahne, B., Heinz, G., Dietrich, W., 1987. Measurement of the ¨ diffusion coefficients of sparingly soluble gases in water. J. Geophys. Res. 92 ŽC10., 10767–10776. Kennedy, V.C., Kendall, C., Zellweger, G.W., Wyerman, T.A.,

L.N. Plummer et al.r Chemical Geology 179 (2001) 93–111 Avanzino, R.J., 1986. Determination of the components of stormflow using water chemistry and environmental isotopes, Mattole River basin, California. J. Hydrol. 84, 107–140. LeGrand, H.E., 1967. Ground water of the Piedmont and Blue Ridge Provinces in the southeastern states. U.S. Geol. Surv. Circ. 538, 1–11. Lynch, D.D., 1987. Hydrologic conditions and trends in Shenandoah National Park, Virginia, 1983–84. U.S. Geol. Surv. Water-Resour. Invest. Rept. 87–4131, 115 pp. Maloszewski, P., Zuber, A., 1982. Determining the turnover time of groundwater systems with the aid of environmental tracers: 1. Models and their applicability. J. Hydrol. 57, 207–231. Maloszewski, P., Rauert, W., Stichler, W., Herrmann, A., 1983. Application of flow models to an Alpine catchment area using tritium and deuterium data. J. Hydrol. 66, 319–330. Manga, M., 1999. On the time scales characterizing groundwater discharge at springs. J. Hydrol. 219, 56–69. Michel, R.L., Turk, J.T., 1995. Use of sulfur-35 and tritium to study rates of sulfur migration in the Flat Tops Wilderness Area, Colorado. Isotopes in Water Resources Management. IAEA, Vienna, pp. 293–301. Michel, R.L., Campbell, D., Clow, D., Turk, J.T., 2000. Timescales for migration of atmospherically derived sulphate through an alpine watershed, Loch Vale, Colorado. Water Resour. Res. 36, 27–36. Ng, W.Y., Walkley, K., 1969. Diffusion of gases in liquids: the constant size bubble method. Can. J. Chem. 47, 1075–1077. O’Brien, R.N., Hyslop, F., 1977. A laser interferometric study of the diffusion of O 2 , N2 , H 2 , and Ar into water. Can. J. Chem. 55, 1415–1421. Plummer, L.N., Busenberg, E., 1999. Chlorofluorocarbons. In: Cook, P., Herczeg, A. ŽEds.., Environmental Tracers in Subsurface Hydrology. Kluwer Academic Publishing, Amsterdam, pp. 441–478. Plummer, L.N., McConnell, J.B., Busenberg, E., Drenkard, S., Schlosser, P., Michel, R.L., 1998a. Flow of river water into a karstic limestone aquifer—1. Tracing the young fraction in groundwater mixtures in the Upper Floridan aquifer near Valdosta, Georgia. Appl. Geochem. 13, 995–1015. Plummer, L.N., Busenberg, E., Drenkard, S., Schlosser, P., McConnell, J.B., Michel, R.L., Ekwurzel, B., Weppernig, R., 1998b. Flow of river water into a karstic limestone aquifer—2. Dating the young fraction in groundwater mixtures in the Upper Floridan aquifer near Valdosta, Georgia. Appl. Geochem. 13, 1017–1043. Plummer, L.N., Busenberg, E., Bohlke, J.K., Carmody, R.W., ¨ Casile, G.C., Coplen, T.B., Doughten, M.W., Hannon, J.E., Kirkland, W., Michel, R.L., Nelms, D.L., Norton, B.C., Plummer, K.E., Qi, H., Revesz, K., Schlosser, P., Spitzer, S., Wayland, J.E., Widman, P.K., 2000a. Chemical and isotopic composition of water from springs, wells, and streams in parts of Shenandoah National Park, Virginia, and vicinity, 1995– 1999. U.S. Geol. Surv. Water Resour. Invest. Rept. 00-373, 70 pp. Plummer, L.N., Rupert, M.G., Busenberg, E., Schlosser, P., 2000b. Age of irrigation water in ground water from the Eastern

111

Snake River Plain aquifer, south-central Idaho. Ground Water 38, 264–283. Rice, K.C., Hornberger, G.M., 1998. Comparison of hydrochemical tracers to estimate source contributions to peak flow in a small, forested, headwater catchment. Water Resour. Res. 34, 1755–1766. Schlosser, P., Stute, M., Dorr, H., Sonntag, C., Munnich, K.O., 1988. Tritiumr3 He dating of shallow groundwater. Earth Planet. Sci. Lett. 89, 353–362. Schlosser, P., Stute, M., Sonntag, C., Munnich, K.O., 1989. Tritiogenic 3 He in shallow groundwater. Earth Planet. Sci. Lett. 94, 245–256. Schlosser, P., Dunkle-Shapiro, S., Stute, M., Aeschbach-Hertig, W., Plummer, L.N., Busenberg, E., 1998. TritiumrHe dating of young groundwater: chronologies for environmental records. Isotope Techniques in the Study of Environmental Change. IAEA, Vienna, pp. 165–189. Solomon, D.K., Cook, P.G., 1999. 3 H and 3 He. In: Cook, P., Herczeg, A. ŽEds.., Environmental Tracers in Subsurface Hydrology. Kluwer Academic Publishing, Amsterdam, pp. 397– 424. Solomon, D.K., Sudicky, E.A., 1991. Tritium and helium 3 isotope ratios for direct estimation of spatial variations in groundwater recharge. Water Resour. Res. 27, 2309–2319. Solomon, D.K., Schiff, S.L., Poreda, R.J., Clark, W.B., 1993. A validation of the 3 Hr3 He method for determining groundwater recharge. Water Resour. Res. 29, 2951–2962. Solomon, D.K., Parry, W.T., Forster, C.B., 1997. Groundwater dating in mountainous terrain. Geol. Soc. Am. Abstr. Programs 29, A-77. Stute, M., Schlosser, P., 1999. Atmospheric noble gases. In: Cook, P., Herczeg, A. ŽEds.., Environmental Tracers in Subsurface Hydrology. Kluwer Academic Publishing, Amsterdam, pp. 349–377. Sueker, J.K., Turk, J.T., Michel, R.L., 1999. Use of cosmogenic 35 S for comparing ages of water from three alpine-subalpine basins in the Colorado Front Range. Geomorphology 27, 61–74. Szabo, Z., Rice, D.E., Plummer, L.N., Busenberg, E., Drenkard, S., Schlosser, P., 1996. Age-dating of shallow groundwater with chlorofluorocarbons, tritiumrhelium 3, and flow path analysis, southern New Jersey coastal plain. Water Resour. Res. 32, 1023–1038. Tanaka, N., Turekian, K.K., 1995. Determination of the dry deposition flux of SO 2 using cosmogenic 35 S and 7 Be measurements. J. Geophys. Res. 100 ŽD2., 2841–2848. Vitvar, T., 2000. Comparison of groundwater residence times obtained by O-18 and basin topography in the Neversink watershed, Catskill Mts, NY. EOS, Trans. Am. Geophys. Union 81 Ž19., S249. Vitvar, T., Balderer, W., 1997. Estimation of mean water residence time and runoff generation by 18 O measurements in a Pre-Alpine catchment ŽRietholzbach Eastern Switzerland.. Appl. Geochem. 12, 787–796. Vogel, J.C., 1967. Investigation of groundwater flow with radiocarbon. Isotopes in Hydrology. IAEA, Vienna, pp. 355–369.