Oceanogr.. 41(5), 1996,966-976 0 1996, by the American Society of Limnology and Oceanography, Inc. Limnol. Effects of glacial meltwater inflows and moat freezing on mixing in an ice-covered antarctic lake as interpreted from stable isotope and tritium distributions Laurence G. Miller U.S. Geological Survey, Water Resources Division, Menlo Park, California 94025 GeorgeR. Aiken U.S. Geological Survey, Water Resources Division, Boulder, Colorado 80303 Abstract Perennially ice-covered lakes in the McMurdo Dry Valleys have risen several meters over the past two decadesdue to climatic warming and increased glacial meltwater inflow. To elucidate the hydrologic responses to changing climate and the effects on lake mixing processes we measured the stable isotope (al80 and 6D) and tritium concentrations of water and ice samples collected in the Lake Fryxell watershed from 1987 through 1990. Stable isotope enrichment resulted from evaporation in stream and moat samples and from sublimation in surface lake-ice samples. Tritium enrichment resulted from exchange with the postnuclear atmosphere in stream and moat samples. Rapid injection of tritiated water into the upper water column of the lake and incorporation of this water into the ice cover resulted in uniformly elevated tritium contents (> 3.0 TU) in these reservoirs. Tritium was also present in deep water, suggesting that a component of bottom water was recently at the surface. During summer, melted lake ice and stream water forms the moat. Water excluded from ice formation during fall moat freezing (enriched in solutes and tritium, and depleted in IsO and 2H relative to water below 15-m depth) may sink as density currents-to the bottom of the lake. Seasonal lake circulation, in response to climate-driven surface inflow, is therefore responsible for the distribution of both water isotopes and dissolved solutes in Lake Fryxell. interactions between antarctic lakes and their surrounding watersheds and, specifically, how climate-driven changes in the inflow of water and solutes effect vertical mixing processes and the estimate of the age of basin refilling in Lake Fryxell. Lake levels in the McMurdo Dry Valleys, the largest of the polar desert oases on the antarctic coast, have risen several meters since measurements began in the 1970s (Chinn 1993). Lake Bonney, in the Taylor Valley, rose 13.5 m since it was fortuitously surveyed in 1903 during the Scott expedition. Other Dry Valley lakes, including Fryxell, continue to rise an average of 0.15 m yr- l, suggesting that the present climate is changing toward increased energy available for meltwater production (Wharton et al. 1992; Chinn 1993). Decreased ice-cover thickness in Lake Hoare, adjacent to Lake Fryxell, also suggests a response to increased summer temperatures and melting (McKay et al. 1985; Wharton et al. 1992). Water in closed-basin antarctic lakes is predominantly meteoric in origin (Ragotzkie and Friedman 1965; Matsubaya et al. 1979). Freezing, evaporation, sublimation, and mixing of water from different reservoirs alter the isotopic and chemical composition of meteoric water, resulting in a distribution of water isotopes and dissolved solutes that provides a distinctive signature of lake processes. The stable isotope ratios of water (P*O and 6D) are conservative, such that only mixing and physical processessuch as evaporation, sublimation, and freezing will alter their values. Tritium (3H as HTO) is a useful tracer of water that has been in recent contact with the atmosphere, both because of its 12.3-yr half-life and because Seasonal or perennial ice covers protect many polar lakes from freezing completely in winter and represent a balance between cold winter temperatures and warm summer temperatures (Doran et al. 1996). In closed-basin antarctic lakes, perennial ice covers restict vertical mixing in the water column and influence the biogeochemical processesthat occur there and in the sediments (e.g. Lawrence and Hendy 1985, 1989; Green et al. 1988). Vertical profiles of dissolved solutes in ice-covered lakes have been previously applied in estimating the lifetime of restricted mixing with one-dimensional diffusion models and interpreted as basin refilling ages(Wilson 1964; Hendy et al. 1977; Lawrence 1982; Lawrence and Hendy 1985; W. B. Lyons pers. comm.). The purpose of this study is to examine the effects of changing climate on the Acknowledgments We thank R. Smith, D. McKnight, M. Brooks, E. Furlong, R. Quadry, R. Van Etten, B. Howes, and the staff at Berg Field Center, McMurdo, for invaluable assistance in the field. We especially thank the crews of VXE-6 for safe conduct through a harsh environment. M. Huebner and D. White provided stable isotope analyses, and R. Michel analyzed tritium via scintillation counting. We thank S. Drenkard, R. Weppernig, and P. Schlosser for providing tritium analyses by 3He ingrowth. The manuscript was greatly improved by suggestions from T. Torgersen, W. Carrothers, R. Michel, B. Lyons, D. McKnight, and P. Mulholland. Special thanks to J. Garcia. The W. M. Keck Foundation provided generous support for establishment of the Lamont-Doherty Earth Observatory (LDEO) helium isotope facility. 966 Hydrology of Lake Fryxell of the peak in atmospheric tritium concentrations resulting from aboveground nuclear testing in 1950s and early 1960s. Here, we use measurements of 6180, 6D, and 3H in water samples collected from the possible sources and reservoirs in the Lake Fryxell watershed to examine the mixing processes that control lake circulation in response to climate-driven variation in surface inflow. Common features of many antarctic closed-basin lakes (including Fryxell, Bonney, and Vanda) include the formation of perimeter moats in summer and increasing dissolved solute concentrations with depth. Attempts to explain the origin of saline bottom water in closed-basin, antarctic lakes have focused on relict seawater (Hendy et al. 1977; Masuda et al. 1988) evaporative concentration of a pre-existing water body (Lawrence and Hendy 1985; Torii et al. 1989), and input of high-salinity groundwaters or modified surface waters (Green and Canfield 1984; Green et al. 1989). One or more of these processes may account for the distribution of dissolved solutes in icecovered coastal ponds (Matsubaya et al. 1979; Bird et al. 199 1) and in several dry valley lakes in the McMurdo Sound region of southern Victoria Land. Distinguishing between these processes is critical to understanding regional paleoclimate because the evaporative concentration hypothesis has been used to estimate the timing of desiccation in several McMurdo Dry Valley lakes (Wilson 1964; Hendy et al. 1977; Lawrence 1982; Lawrence and Hendy 1985; W. B. Lyons pers. comm.) These studies all assume that following desiccation and basin refilling, a solute concentration gradient existed between underlying brine and overlying dilute water, and that molecular diffusion was responsible for the total solute mass now present in the upper water. However, if solute mass were added to the surface, then the ages calculated would be overestimates of the true age of the basin refilling. Conversely, if solute mass were added to the deep water of the lake, then calculated ageswould be underestimates of the true age of basin refilling. This paper presents evidence for the addition of dissolved solutes to the deep water of the lake via advection of modified surface water. Site description Lakes of the McMurdo Dry Valleys region of western Antarctica are in a relatively ice-free region sustained by the precipitation shadow of the Transantarctic mountains and by the flow of persistently low-humidity air masses originating on the polar plateau (Green et al. 1989). Lake Fryxell lies near the eastern end of Taylor Valley at 77”37’S, 163’8’E and is the lowest (elevation, 18 m) in a series of ice-covered lakes extending to Lake Bonney (elevation, 57 m) at the foot of Taylor Glacier. Lake Fryxell occupies a closed basin between Canada Glacier to the west and New Harbor on McMurdo Sound to the east (Fig. 1); it drains a watershed area of 230 km2 (Lawrence and Hendy 1985). As many as 14 ephemeral meltwater streams drain into the lake from the surrounding alpine glaciers during summer, and the terminus of Canada Glacier is in direct contact with the water column of the lake at its western end. 967 %.--.-T.&p: 1.000 CONKIIJR ,NTuwAL M’LES Z,MM METERS 50 METERS WITH SUPFl EMFNTAW COPiTOURS DATUM IS hIEM4 5EA LEVEL AT 25 METER INTtRVALS Fig. 1. Map of the Lake Fryxell basin showingdrainagefrom Commonwealth and other alpinc glaciersand the proximity of Canada Glacier. Water-column sampling station occupied in 1987-l 989- x . Moat samples were collected between Huey Creek and CanadaStream. During summer, an ice-free moat commonly forms as a result of surface input and local melting of the ice cover. The combined sources of water maintain a positive water balance in the lake, as seen by a 2-m rise in lake level since 197 1 (Chinn 198 1, 1993). Surface-ice ablation is the main water loss (30-40 cm yr-l; Henderson et al. 1965). The ice-cover thickness of Lake Fryxell has varied little over time, ranging from 4.0 kO.5 m in 1963 and 1987-l 99 1 (Hoare et al. 1965; this study) to a minimum of 3.5a0.5 m in 1986 (Clow et al. 1988), suggesting that at present, freezing and ablation are in balance, but the water budget is not in hydrologic steady state because inputs exceed ablation plus moat evaporation. The water column of Lake Fryxell was vertically stable over the sampling period 1987-1990. Conductivity increased regularly with depth and gradients were consistent from year to year (Smith et al. 1993). Temperature was similarly consistent, ranging from 0.4”C below the ice surface to 2.6”C at 18 m (max depth, 19 m) and reaching a middepth maximum of 3.4”C between 9- and 10.5-m depth with little seasonality. Aiken et al. (199 1) calculated in situ density of the lake water from temperature and conductivity measured in 1987, which, corrected for the partial molar volume of dissolved solutes and gases(O,, Miller and Aiken 968 Table 1. Morphometry and hypsography of Lake Fryxell (estimated from bathymetry given by Lawrence (1982)). Zone Lake Moat Surface water Deep water Contour interval* (ml o-3 3-6 6-9 9-12 12-15 15-18 >18 o-1.5 5-8 >15 vol.+ Area? (x 1O-6 m2) (X 1O-6 m3) 17.1 7.14 4.27 11.3 8.39 3.26 5.36 2.33 2.99 1.24 1.47 0.75 0.35 0.23 2.15 1.44 9.80 3.60 1.82 0.75 * Relative to piezometric water surface. t Surface area of top of contour interval, except Moat surface calculated as area between shoreline and 1.5-m depth contour. $ Volume = (lower area)(interval) + [(upper area - lower area)(interval)/2]. Ar, N,), ranged from 1.00026 g cm-3 at 5 m to 1.006 12 g cm-3 at 18 m. They compared the average vertical density gradient with density gradients in the thermocline region of temperate stratified lakes. Using a relationship between density gradient and vertical eddy diffusion coefficient (Quay 1977), Aiken et al. estimated an average vertical eddy diffusivity of K, = 5.9 + 1.2 x 1Om5cm -2 s- l for the water column (5-l 8 m) of Lake Fryxell. This value was 6 + 1 times the tracer diffusion coefficient of water at the in situ temperature (Li and Gregory 1974) and was the basis for assuming that transport of dissolved solutes in the water column was diffusion controlled. The chemical and stable isotope mass balances we calculated are based on the bathymetric map provided for 1979 by Lawrence (Lawrence 1982; Lawrence and Hendy 1985). Integrated surface area and volume were determined for 3-m contour intervals, assuming that the 1979 and 1989 surface elevations were similar (Table 1). This ignores the approximately 0.6-m rise in lake level since 1979 (Chinn 1993). The estimated volume error is > 20% and is a result of both the uncorrected effect of surface elevation on water depths and the sparse areal coverage of the original bathymetric survey. Methods Sampling- Water-column samples were collected during summer 1987- 1989 by peristaltic pumping at < 1 liter min-l through Tygon tubing. Water-column depths reported were relative to the piezometric water level (0.2 m below ice surface), which introduced an uncorrected spatial error due to seasonal and yearly changes in lake surface elevation. Ice-cover sample depths reported were relative to the ice surface. Samples for stable isotopes (al80 and 6D) were collected unfiltered and stored in 20- ml glass vials with polyseal screwcap closures. Stream and moat samples were collected by directly filling glass vials after several rinses with the sample. Glacier, snow, and lake-ice samples were melted in air before filling the vials and stored unfrozen. Samples for tritium (3H) analyses were collected unfiltered in 2-liter glass bottles by overflowing several bottle volumes with a tube placed in the bottom to reduce contact with the atmosphere and stored unfrozen. Additional water-column tritium samples were collected by slowly pumping (qO.2 liter min-l) upward through vertically held sections of 0.95-cm-o.d. copper tubing and sealing the tubing with pinch clamps. Lake-ice tritium samples were collected in butyrate tubes, thawed in an argon atmosphere, and syphoned through vertically held sections of copper tubing as above. A moat sample was collected in a 2-liter glass bottle and, by keeping the outlet tubing away from the atmosphere, siphoned with minimal air contact through the copper tubing and stored unfrozen. Sediments were cored in 1990 by driving a lined corer (2.5-cm i.d.) into the bottom at locations in the lake corresponding to water depths of 8, 10, and 18 m. Upon retrieval, the liner was separated from the corer and sectioned on shore by extruding intervals of the core into gas-powered squeezers(Reeburgh 1967). Filtered (0.4 pm) pore waters were collected in plastic syringes and transferred to scintillation vials for transport, unfrozen, to Menlo Park for analysis. Analytical techniques -Stable isotopes of water were determined by CO2 equilibration (Epstein and Mayeda 1953) for 180, and zinc reduction (Kendall and Coplen 1985) for 2H and are reported vs. VSMOW. Stable isotope compositions are reported in standard delta (6) notation using units of per mil (7~) where &180, D) VSMOW = (R,,rnp~RVSMOW - ljx 1,000. (1) R is the ratio of 180 *. 160 or 2H : ‘H. The 2-o analytical precision was &O. lY& for al80 and + 1.57~ for 6D. Tritium was analyzed both by liquid scintillation counting after electrolytic enrichment (Thatcher et al. 1977) and by 3He ingrowth (Clarke et al. 1976). Tritium concentrations are reported in tritium units (TU), equal to 1 tritium atom per 1018hydrogen atoms, and are corrected for decay to the date of sampling. Precision (2 a) of tritium analyses was + 0.3 TU at the lowest levels of detection (0.3 TU) with counting and +O. 15 TU with 3He ingrowth. Chloride was analyzed on pore-water and water-column samples by ion chromatography (Oremland et al. 1987) after 1OO-fold dilution with deionized water. Results Stable isotope compositions of all water samples from the Lake Fryxell watershed are presented in Table 2 and Fig. 2, along with the global meteoric water line (6D = 8 k al80 + 10; Craig and Gordon 1965). Values of al80 ranged from - 24 to - 33o/oo and values of 6D ranged from - 200 to - 25 8a/oo. The lightest samples were found in deep 969 Hydrology of Lake Fryxell Table 2. Isotopic composition of water (vs. VSMOW) in the Lake Fryxell watershed. Sample Water column Water Moat and stream Moat Moat at creek Huey Creek Fryxell stream Pore water Sed. core, 8 m Date collected Depth* W-0 6’80 @d 18Dec87 18 Dee 87 18 Dee 87 18 Dee 87 4 Dee 88 4 Dee 88 4 Dee 88 4 Dee 88 4 Dee 88 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 26 Nov 89 5 7.5 11.5 18 5 8.5 10 15 18 5 6.5 8 9.5 11 12.5 14 15.5 17 18 -30.2 -31.0 -31.2 -32.0 -30.5 -31.3 -31.7 -32.1 -32.3 -31.1 -31.1 -31.3 -31.6 -31.6 -32.0 -32.1 -32.3 -32.3 -32.2 -232 -241 -240 -249 -233 -242 -243 -246 -244 -242 -243 -244 -246 -245 -249 -251 -253 -251 -250 2 Jan 88 4 Dee 88 15 Dec88 11 Dee 89 14 Dee 89 22 Dee 88 25 Dee 88 24 Nov 89 14 Dee 89 0.5 0.5 0.5 0.5 0.5 0.1 0.1 0.1 0.1 -25.8 -27.1 -26.8 -26.8 -28.4 -29.9 - 30.0 -30.5 -29.5 -211 -218 -216 -217 -228 -238 -240 -240 -234 - 30.4 -30.5 - 30.4 -30.3 -29.8 -29.4 -31.0 -31.1 - 30.9 -31.2 -30.7 -28.7 -30.0 -30.5 -29.5 -31.4 -30.4 -30.6 -31.1 -28.9 -31.4 -240 -241 -241 -239 -239 -235 -245 -245 -243 -244 -244 -236 -239 -241 -240 -247 -244 -244 -246 -238 -249 19 Dee 90 water§ 9 Dee 90 0.05 11 9 Dee 90 0.15 19 Dee 90 0.36 19 Dee 90 0.48 Sed. core, 10 m 19 Dee 90 water 19 Dee 90 0.05 19 Dee 90 0.15 19 Dee 90 0.25 19 Dee 90 0.35 19 Dee 90 0.46 19 Dee 90 0.57 Sed. core, 18 m 18 Dee 90 water 18 Dee 90 oozell 18 Dee 90 0.35 18 Dee 90 0.12 18 Dee 90 0.22 18 Dee 90 0.32 18 Dee 90 0.42 18 Dee 90 0.52 18 Dee 90 0.62 * Depth below piezometric water level. t Depth below ice surface. # Water underlying ice cover. 0 Overlying water collected with core. I( Depth below sediment-water interface. ll Organic surface collected with core. Sample Ice Glacier ice Snow Surface ice Moat ice Ice core 6 Ice core 9 Ice core 10 Ice core 11 Ice core 12 Ice core 15 Ice core 16 Date collected Depth? Cm) 6’80 (~4 24 Dee 87 24 Nov 89 24 Nov 89 18 Nov 89 16 Dee 88 16 Dee 88 16 Dee 88 16 Dee 88 16 Dee 88 11 Dee 89 30 Nov 89 30 Nov 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 2 Dee 89 6 Dee 89 6 Dee 89 6 Dee 89 6 Dee 89 8 Dee 89 8 Dee 89 8 Dee 89 8 Dee 89 8 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 9 Dee 89 0 0 0 0 0 0 0 1.0 0 0 4.2 water* 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.6 water 3.6 0.25 0.75 1.25 1.75 2.25 2.75 2.5 3.25 3.6 water 0.75 1.75 2.25 3.25 3.75 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 4.25 4.75 water -32.3 -31.0 -32.8 -32.1 -26.8 -26.1 -24.2 - 26.4 -25.7 -26.3 -27.8 - 30.6 -24.2 -25.7 -25.5 - 26.0 -25.8 -26.2 -26.5 -27.6 - 30.4 -27.5 -25.5 -25.6 -25.8 -25.5 -27.1 -26.3 -26.8 -26.5 -27.6 - 30.4 -26.6 -26.6 -26.2 -27.0 -26.7 -26.4 -26.4 -26.8 -26.6 -26.7 -27.5 -26.7 -27.3 -27.4 -26.9 -29.5 -249 -247 -258 -255 -219 -214 -202 -214 -210 -217 -226 -242 -208 -213 -215 -217 -215 -215 -219 -222 -242 -226 -212 -213 -214 -213 -222 -215 -218 -219 -222 -242 -219 -218 -218 -218 -217 -217 -216 -218 -217 -218 -223 -219 -222 -222 -221 -237 970 Miller and Aiken Water column -Stable isotope composition of water decreased with depth (Fig. 3A,B). Data from 1988 were fit to a logarithmic equation, and 1989 data were best fit to a linear equation. The 1989 6D data were systematically 4o/oo lighter than the 1988 and 1987 data, except that 18-m water (1987-l 989) agreed within the analytical error. No systematic difference existed in the multiyear al80 profiles, except that 1989 samples collected just below the ice cover were lighter. Water-column tritium data from 1989 and 199 1 were combined to produce one profile of 3HH0 vs. depth (Fig. 3C). Both L-DE0 and USGS data were used in this plot and the observed agreement between methods is excellent. Tritium concentrations ranged from 4.0 TU at 5 m to below the detection limits (co.3 TU) at 12 and 18 m with elevated concentrations in four samples at 14-, 15.5-, and 17-m depths. -210 -220 g-230 s -240 -250 -260 -34 -32 -30 -28 -26 -24 6’“O (%o) Fig. 2. P*O vs. 6D of all waters sampled in the study: Cllake water column; A- moat; 0 -streams; e -pore water; + Canada Glacier; O-lake ice; v-surface lake ice; + -snow. Shown also is the global meteoric water line (GMWL) and the regional evaporation trend line (slope = 5.6). Error bars show +l SD. levels of the water column, fresh snow at lake level, and ice from Canada Glacier. Heavier samples were associated with sublimation of lake ice and evaporation of the moat. -32 Ice cover-Stable isotope composition of water decreased with depth in the ice cover (Fig. 5) with the lightest values at the bottom of the ice cover corresponding to 3HH0 (TU) 6D (%o) 8’80 (%o) -33 Pore water-Mean pore-water values of stable isotopes and the isotope compositions of overlying water collected with the cores were heavier than those of the water-column samples collected in the center of the lake at depths corresponding to the sediment-water interface (Table 3, Fig. 4A,B). Pore-water chloride concentrations, however, were similar to corresponding lake-water values and showed no trend with depth (Fig. 4C), suggesting that the upper sediments were open to exchange with lake water. -31 -30 -255 -250 -245 -240 -235 -230 0 1.0 2.0 3.0 4.0 5.0 12 lb..16” 18 - A Fig. 3. Vertical profiles of 6’*0, 6D, and 3HH0 vs. depth in the water column: O- 1987; A- 1988; Cl- 1989. Solid lines are log (1988) and linear (1989) fits to the stable isotope data and polynomial (1987-1989) fit to 3HH0 data. Tritium samples from 1989 measured by 3He ingrowth (Cl-L-DEO), all others by scintillation counting (O-USGS). Error bars show Ifl 1 SD. Hydrology of Lake Fryxell 971 Table 3. Stable isotope composition (%) and chloride concentration (mM) of bottom water and sediment pore water in Lake Fryxell. (Not determined--d.) 89 10 18 Pore water? Bottom water* Depth Cm) 6’80 6D -31.2f0.2 -243&l (n = 3) -31.4kO.3 (n = 3) -32.3kO.3 (n = 3) (n = 3) -244f2 (n = 3) -248+3 (n = 3) Cl-$ 6D 6180 -240+ 1 (n = 5) -242+4 (n = 5) -243+4 (n = 9) -30.320.3 (n = 5) -30.7kO.8 (n = 5) -30.4LO.9 (n = 9) 28.4 43.0 115 Clnd 66.4f4.0 (n = 5) 1lOk4.8 (n = 6) * Mean f 1 SD of samples collected 1987-l 989. t Mean + 1 SD of all samples collected 1990. I#I1987 Water column Cl- from Aiken et al. 1996. 0 Depth of core. Discussion equilibrium freezing of surface waters (Craig and Gordon 1965). Values of both al80 (Fig. 5A) and 6D (Fig. 5B) were heaviest at the ice surface, presumably due to sublimation of ice. Generally constant values were observed between the surface and bottom of the ice cover. Tritium values in the ice were greater than in corresponding samples collected in the water column just below the ice cover (Table 4). Tritium concentrations in the deepest ice samples increased from 4.0 TU at the center of the lake to 6.5 TU at a site near Canada Glacier and the outlet of Canada Stream. 6’80 (%o) -34 -32 6D (%o) -30 $8 -260 B 20 - +s Oki’0 A v I 1 , overlying water ,y’ 7’ l-w IJiP / %I+ 50 -230 0 1 I C I ‘, 150 100 ,A, ‘,V’ I overlying water l-o-l \ ‘0 El I $30 E. - I 0 loo o"40 - / O\ q \ 50 - , I, Chloride (mM) -240 -250 I 10 - Water budget -Because Fryxell is a terminal lake, its surface level readily responds to changes in the water budget, particularly meltwater input. Since 197 1 the surface level of the lake has increased an average 10.5 cm yr-l, although some yearly decreases in elevation were also measured (Chinn 1993). Surface-ice ablation determined in Lake Fryxell (30-40 cm yr-I, Henderson et al. 1965) is similar in nearby Lake Vanda and Lake Hoare (30-35 cm yr- l, Chinn 1993; Clow et al 1988) and is s, 60 - Fig. 4. Vertical profiles of 6l8O, 6D, and Cl vs. depth in sediment pore waters: O-core at 8 m; Cl-core at 10 m; 0 -core at 18 m. Open symbols above sediment-water interface represent water collected during coring process; solid symbols are watercolumn samples collected separately. Dashed lines connect 18-m data. Error bars show + 1 SD. 972 -32 Miller and Aiken -30 6’80 -28 (9&I) -26 -24 -22 '50 -240 6D (YAo) -220 -230 -210 -200 Table 5. Stable isotope composition and tritium concentration of the well-mixed reservoirs and water masses in Lake Fryxell (mean + 1 SD of samples collected in 1987-l 989). (Not determined - nd.) Glacier Stream Moat Surface (5-8 m) Deep (15-18 m) Fig. 5. Vertical profiles of 6’*0 and 6D vs. depth in selected cores from the ice cover: 0 -ice core 6; O-ice core 9; O-ice core 15; A-ice core 16; + -moat ice. Solid symbols are underlying water. Error bars show + 1 SD. likely constant with time. Using 30 cm yr-l ablation and Lawrence’s ( 1982) estimates of surface area (7.1 X 1O6m2) and volume (47 x lo6 m3), we calculate a steady-state (ignoring lake level increases) residence time of water as r= volume ablation x area’ (2) yielding a whole-lake residence time of 22 yr. This is a maximum residence time because evaporation of water from the moat is not included and because we chose the lower estimate for ablation. It is not a true whole-lake residence time, however, because there is much less than complete mixing in the water column. Rather, this is considered an estimate of how quickly water entering the lake at the surface is mixed in the upper water column, frozen in the ice cover, and advected through the ice to the surface. For an ice-cover thickness of 4.5 m and ab- Table 4. Tritium concentration of the ice cover and surface water below the ice (samples collected in 1989). Location IC-9 IC-9, WC? IC-10 IC-15 IC-15, WC/ Depth* b-d 3.5-3.7 3.8 3.5-3.7 0.5-l .o 1.5-2.0 2.0-2.5 3.0-3.5 3.5-4.0 5.0, 5.5 6.12kO.15 4.00+0.15 6.53kO.15 4.61 f0.18 4.79f0.17 3.63kO.15 2.72f0.15 3.95kO.15 3.75kO.15 (12= 2) ~ * Depth below ice surface, except WC depth below piezometric water level. WC depth is -0.2 m lower. t Mean k 1 SD of concurrent below-ice water samples. a’*0 W) -32.0+0.9 (n = 3) -3O.OkO.5 (n = 3) -26.6f0.6 (n = 4) -31.2kO.l (n = 3) -32.3kO.l (n = 6) ;i -251+6 (n = 3) -238+3 (n = 3) -216+3 (n = 4) -243+2 (n = 3) -249f3 (n = 6) (;“u) nd 3.lkO.3 (n4=22) ’ 3.7&O. 1 (n = 5) 0.4kO.3 (n = 3) lation of 30 cm yr - l, a transit time of water through the ice cover is calculated as thickness r= (3) ablation ’ yielding an ice-cover residence time of 15 yr. These calculations, coupled with the observations of elevated tritium concentrations in surface lake water and in the ice cover relative to incoming streams (Tables 4, 5), suggest that much of the new water entering the lake is rapidly cycled through the upper water column and the overlying ice. Sources of water- Liquid water has continuously occupied the Lake Fryxell basin for at least 1,000 yr (Lawrence and Hendy 1985; Green et al. 1988, 1989; W. B. Lyons pers. comm.). The possible sources of water entering the lake are direct input of surface meltwater from Canada Glacier to the ice cover, the proglacial streams and their hyporheic zones (McKnight and Andrews 1993), the moat, permafrost melting (Stuiver et al. 198 l), and groundwater (S. Tyler pers. comm.). The measured water reservoirs are distinguished by their isotope compositions (Table 5), although information is lacking for permafrost 6D and groundwater in general. Canada Glacier ice is a source of dilute water with the lightest stable isotope composition in the watershed (Fig. 2, Table 5). Canada Glacier is 2-3?&~lighter in 180 than the other glaciers draining into Lake Fryxell (Matsubaya et al. 1979; Stuiver et al. 198 1). Tritium was not measured; however, glacier ice away from accumulation zones should contain tritium only in surface samples. Stream water is likewise chemically dilute compared to lake water (Green et al. 1989) but is enriched in both 180 and 2H due to evaporation (Table 5). The trend of the evaporation line (Fig. 2) is described by a slope of 5.6, consistent with low humidity control of the process (Gat and Tzur 1967). Stream-water tritium concentrations are similar to expected ambient concentrations resulting from exchange with the atmosphere. Hydrology of Lake Fryxell Moat water is chemically similar to stream water. Concentrations of S042- and Cl- in moat water collected in 1987 (35.0 and 269 PM; L. Miller unpubl. data) were 14& 1% greater than the volume weighted S042- and Clconcentrations of all streams measured in 1982 (3 1.O and 235 IAM, Green et al. 1988). However, moat water is greatly enriched in 180 and 2H compared to stream water (Table 5) and lies further along the evaporation line (Fig. 2). Moat water is also enriched in tritium over stream water (Table 5), but it is unclear whether enrichment is a result of greater contact with the atmosphere, evaporation, or selective storage of water with higher tritium content (i.e. memory effect of higher tritium concentrations in the past decade). The three remaining reservoirs - permafrost, hyporheic zone water, and groundwater- were not sampled in this study. However, surface permafrost al80 measured during the Dry Valley drilling project ( 1975) was similar to glacier al80 in the vicinities of Commonwealth and Canada Glaciers (cores 11 and 12, Stuiver et al. 198 1). For example, the 6180 of permafrost from 1O-m depth in both cores was -32%~. Permafrost melting is considered an unkown but possible source of subsurface water to nearby Lake Hoare (S. Tyler pers. comm.). The hyporheic and saturated zones are environments where solutes (including aerosols and weathering products) are likely to be dissolved into water, and both zones offer restricted contact with the atmosphere. Unfortunately, the extent and chemical and isotopic compositon of these subterranean waters are unknown. Lake chemistry and isotopes-Surface lake water is chemically similar to moat and stream water (Green et al. 1989; Aiken et al. 199 1; McKnight et al. 1993). The stable isoltope composition of surface lake water is likely a result of equilibrium freezing of moat or stream water, resulting in heavier ice and lighter residual water (Craig and Gordon 1965). The equilibrium fractionation factors for freezing in Lake Fryxell are 1.004 for al80 and 1.020 for 6D (CU factors derived from Table 2). Yearly variations in the stable isotope composition of surface water (Fig. 3) can then be explained by fluctuations in the amount of “new” input (heavier 6180and 6D) if the annual amount of freezing is constant. The resulting fluctuation in stable isotope composition of ice (lym for 6180 and loo/o0for 6D) is less than the variability observed in the ice cover values (Fig. 5). Enhanced vertical mixing in the upper water column is evident from nearly uniform concentrations of dissolved solutes between the bottom of the ice cover and 6-7-m depth (Aiken et al. 199 1, 1996; Smith et al. 1993; McKnight et al. 1993). This effect is also seen in the tritium profile (Fig. 3C). Stream and moat water entering the lake has an estimated density of 1.00004 g cm-3 at 1.5”C (derived from stream chemistry of Green et al. 1989), which would allow it to sink only to 5 m. During the melting period, surface water is transported from the edges of the lake toward the center by horizontal eddy diffusion (KX = 10-3-10-1 cm2 s-l, Quay 1977; Coleman and Armstrong 1983) and by horizontal advection of wa- 973 ter down gradient (1 .Ocm s-l, Ragotzkie and Likens 1967; McKnight and Andrews 1993). Enhanced vertical mixing therefore may result from both local convection and shear below the ice cover. During the freezing period, surface water may move laterally under the ice in the opposite direction, toward the shoreline (Gade et al. 1974; Welch and Bergmann 1985). Freezing of surface water to the bottom of the ice cover in winter causes salt exclusion and local convection of the upper water column (Canfield et al. 1983) and increased concentration of salts toward shore (Ferris et al. 199 1) in seasonally ice-covered lakes. This cryogenic concentration of solutes in the nearshore surface zone may contribute to the formation of bottom water in Lake Fryxell during fall. The vertical structure between 8 and 15 m is described by regularly increasing density with depth resulting from nearly uniform temperature (3.2+0.2”C, n = 20) and increasing dissolved solute concentrations (Aiken et al. 199 1; Smith et al. 1993). Biogeochemical processesare complex within this region of the water column, including calcium carbonate precipitation between 8 and 9 m (Lawrence and Hendy 1989). Overall, this region is characterized by nearly linear profiles of dissolved conservative constituents, including stable isotopes of water (Fig. 3A,B), leading to a uniformly high density gradient and low K,. This middle region is the most stable vertical zone in the water column. Below 15-m depth, the water column is less stable. Deep water has elevated concentrations of major and minor elements (Lawrence and Hendy 1985; Green et al. 1989; Aiken et al. 1991; McKnight et al. 1993; Smith et al. 1993); however, the density gradient between 15 and 18 m is less steep, reflecting a smaller vertical gradient in dissolved solute concentrations (Aiken et al. 199 1). The deep lake water is isotopically lighter than water higher in the column (Fig. 3A,B). In addition, tritium is present in water collected at 14-, 15.5-, and 17-m depths (Fig. 3C), suggesting that some component of bottom water was in contact with the postnuclear atmosphere and that this signal has not decayed to below the detection limit for 3H. This tritiated deep water cannot be explained by diffusion from either above or below the zone. Deep water is isotopically lighter than pore water (Table 3; Fig. 4A,B). To further distinguish sources of bottom water, the 1989 water-column stable isotope values are plotted against water-column chloride data (Aiken et al. 1996) in Fig. 6, along with the 1990 pore-water samples collected at 10 and 18 m. The linear trends in watercolumn values suggest end-member mixing between surface and deep water. Pore waters plot off the trend, suggesting two possibilities: if pore water is the source of solutes and water isotopes to the bottom water, then some fractionation in stable isotope ratios should occur during transport across the sediment-water interface, or the pore water is not a mixing end-member for water isotopes. Advection of bottom water-To explain the bottomwater distribution of stable isotopes and tritium, we propose that modified surface water, enriched in solutes, 974 Miller and Aiken tagged with tritium, and depleted in 180 and 2H, is injected at 15-l 8-m depth. Bottom water (> 15 m) presently has a density > 1.00523 g cm-3, hence surface water such as stream or moat water would have to be significantly concentrated to sink as density currents to 15-m depth. Two mechanisms for solute enrichment are considered feasible: dissolution of soluble salts during subsurface transport and salt exclusion during fall freezing of the moat. Both mechanisms rely on density currents to transport modified surface water along the basin boundary to the deepest part of the lake. Stream water entering the moat has a well-defined chemical and isotopic signature, whereas water transported in the hypohreic zone (4 times the cross-sectional stream area; McKnight and Andrews 1993) may be more enriched in solutes. Seepagemeters in Lake Hoare sampled groundwater transported through the shallow sediments of the lake; this water was enriched in chloride and 0.5%0 lighter in I80 than the ambient lake water at the same site (S. Tyler pers. comm.). If we assume that moat water is essentially stream water and begin with volume-weighted average major element concentrations (Green et al. 19SS), we calculate a moat density of 1.000042 g cm-3 at 1.5”C. If freezing were 90% efficient at excluding dissolved solutes (Canfield et al. 1983), then an enrichment factor of [0.00523/ (0.000042 x 0.9)] or 138 is required to increase moatwater density to 15-m water density. This could occur only when the moat is nearly completely frozen and < 1% of its volume remains as a saline brine, presumably just below the moat ice. The resulting moat ice would be isotopically heavier than the previous moat and the residual brine would be depleted in the stable isotopes of t -I -230 10 m pore water - B P 18 m pore water- -240 2 s 1989 water column -250 r = 0.9574 -260 0 20 40 60 80 100 120 Chloride (mM) water. Beginning with moat isotope composition (Table 5), water remaining after freezing would follow a Rayleigh curve, depending on the isotopic fractionation factor cy and the fraction of water remaining (f): 6 = 1,000(~+1) - I), (4) Fig. 6. Dissolved chloride vs. 6180 and 6D in water-column and pore-water samples. Solid lines are linear regressions of 1989 water-column data (Cl) where Cl- (Aiken et al. 1996) increasesregularly with depth. Pore waters are from cores collected in 1990 at 10-m (0) and 18-m (A) water depths. resulting in brine compositions of 6180 = - 43.4?& and 6D = - 299o/oofor a = 1.004 and 1.020, repectively (from Table 2), and forf = 0.0 1. Tritium may fractionate twice as much as 2H. This implies a tritium fractionation factor cy= 1.040 and results in a calculated decrease in tritium of 16.8%. Advection of 1% of the moat volume as brine with this isotopic composition would add 2 x lo4 m3 of water, enriched in tritium and depleted in stable isotopes, to the bottom water (15-18-m depth) annually. This is calculated as an isotope mass balance: = c3v3. CJ, -I- c,v, (5) could supply the observed tritium in the bottom water over the four decades of elevated surface tritium deposition. It is likely that this process is not at steady state (i.e. there may be years when little or no brine is formed or the brine formed is not dense enough to sink to the bottom). In years when a larger moat develops due to increased glacial meltwater inflows, greater production of high density brines may occur during moat freezing. Over time, this input could result in significant and measurable depletion of bottom-water stable isotopes and contribute to the present distribution of water isotopes and dissolved solutes in the water column. The seasonal input of glacial meltwater to the lake surface and of modified surface water to the deep-water column promotes stability in the water column by increasing Cl, C2, and C3, are the isotope ratios (180 : 160, 2H : ‘H) and tritium concentration of the initial water at 15 m, the moat, and the final composition at 15 m, and V1, V2, and V3, are the volumes of 15-18-m water, the volume of added brine, and the final volume. This annual addition (N 1% of the deep-water volume) would result in decreases in 6l 8O and 6D of 0.12 and 0.6%0 and an increase in tritium of 0.04 TU. Although the input function of tritium in Antarctica is not well known, this mechanism the solute concentration gradient. This, in turn, affects any regional climatic inferences based on estimates of the basin refilling age. If the concentration gradient reflects both diffusive and advective processes, then conceptually advection would tend to steepen the gradient over time, and diffusion would tend to flatten the gradient. Calcu- Hydrology of Lake Fryxell lated diffusion-cell ages, therefore, may be underestimates of the age of basin refilling if the concentration gradient has a strong advective component. It is unclear what fraction of dissolved solutes in the bottom water originates by the above mechanism vs. transport from the sediments. To evaluate the relative contribution of surface brines to the deep-water solute flux we must construct mixing models using time-dependent geochemical tracers with known input functions. For example, measurement of the anthropogenic gases CFC11 and CFC-12 in the water column may help to identify . the extent and timing of bottom-water advection. Determining lateral gradients in below-ice concentrations of solutes, 3H, and stable isotopes of water (al80 and 6D) during fall freezing of the moat and throughout winter would demonstrate the significance of our proposed hypothesis and link surface phenomena (e.g. changes in meltwater input as a function of climate change) to bottom-water formation processes. References AIICEN,G., D. M&NIGHT, R. HARNISH, AND R. WERSHAW. 1996. Geochemistry of aquatic humic substances in the Lake Fryxell basin, Antarctica. Biogeochemistry. In press. R. L. WERSHAW, AND L. MILLER. 1991. Evidence for ihe diffusion of fulvic acid from the sediments of Lake Fryxell, Antarctica, p. 75-88. In R. Baker [ed.], Organic substances and sediments. Lewis. BIRD, M I., A. R. CHIVAS, C. J. RADNELL, AND H. R. BURTON. 199 1. Sedimentological and stable-isotope evolution of lakes in the Vestfold Hills, Antarctica. Palaeogeogr. Palaeoclim. Palaeoecol. 84: 109-l 30. CANFIELD, D. E., R. W. BACHMAN, AND M. V. HOYER. 1983. Freeze-out of salts in hard-water lakes. Limnol. Oceanogr. 28: 970-977. CHINN, T. J. H. 198 1. Hydrology and climate in the Ross Sea area. J. R. Sot. N.Z. 11: 373-386. . 1993. Physical hydrology of the Dry Valley lakes, p. l-5 1. In W. J. Green and E. I. Friedman [eds.], Physical and biogeochemical processes in Antarctic lakes. Antarct. Res. Ser. 59. AGU. CLARKE, W. B., W. J. JENKINS, AND Z. TOP. 1976. Detcrmination of tritium by mass spectrometric measurements of 3He. Int. J. Appl. Radiat. Isot. 27: 5 15-522. CLOW, G. D., C. P. McI(AY, G. M. SIMMONS, JR., AND R. A. WHARTON, JR. 1988. Climatological observations and predicted sublimation rates at Lake Hoare, Antarctica. J. Climatol. 1: 715-728. COLEMAN, J. A., AND D. E. ARMSTRONG. 1983. Horizontal diffusivity in a small ice-covered lake. Limnol. Oceanogr. 28: 1020-1026. CRAIG, H., AND L. I. GORDON. 1965. Deuterium and oxygen18 variations in the ocean and marine atmosphere, p. 277379. In Paleotemperatures and isotopic oceanography. Proc. 3rd Spoleto Conf. DORAN, P. T., AND OTHERS. 1996. Climate forcing and thermal feedback of residual lake-ice covers in the high Arctic. Limnol. Oceanogr. 41: 839-848. EPSTEIN, S., AND T. MAYEDA. 1953. Variation of 0’8 content of waters from natural sources. Geochim. Cosmochim. Acta. 4: 2 13-224. FERRIS, J. M., J. A. E. GIBSON, AND H. R. BURTON. 975 199 1. Evidence of density currents with the potential to promote meromixis in ice-covered saline lakes. Palaeogeogr. Palaeoclim. Palaeoecol. 84: 99-l 07. GADE, H. G., R. A. LAKE, E. L. LEWIS, AND E. R. WALKER. 1974. Oceanography of an Arctic bay. Deep-Sea Res. 21: 547-57 1. GAT, J. R., AND Y. TZUR. 1967. Modification of the isotopic composition of rainwater by processes which occur before groundwater recharge, p. 49-60. In Isotopes in hydrology. Proc. Synp. AEA. GREEN, W. J., M. P. ANGLE, AND K. E. CHAVE. 1989. The geochemistry of Antarctic streams and their role in the evolution of four lakes in the McMurdo Dry Valleys. Geochim. Cosmochim. Acta 52: 1265-1274. AND D. E. CANFIELD. 1984. Geochemistry of the Onyx R:ver (Wright Valley, Antarctica) and its role in the chemical evolution of Lake Vanda. Geochim. Cosmochim. Acta 48: 2457-2467. AND OTHERS. 1988. Geochemical processesin the Lake F&xell Basin (Victoria Land, Antarctica). Hydrobiologia 172: 129-148. HENDERSON, R. A., AND OTHERS. 1965. An ablation rate for Lake Fryxell, Victoria Land, Antarctica. J. Glacial. 6: 129133. HENDY, C. H., A. T. WILSON, IS. B. POPPLEWELL, AND D. A. HOUSE. 1977. Dating of geochemical events in Lake Bonney, Antarctica, and their relation to glacial and climatic changes. N.Z. J. Geol. Geophys. 20: 1003-1022. HOARE, R. A., AND OTHERS. 1965. Solar heating of Lake Fryxell, a permanently ice-covered antarctic lake. J. Geophys. Res. 70: 1555-l 558. KENDALL, C., AND T. B. COPLEN. 1985. Multisample conversion of water to hydrogen by zinc for stable isotope determination. Anal. Chem. 57: 1437-1440. LA=-, M. J. F. 1982. Origin and occurrence of antarctic lacustrine carbonates, with special reference to Lake Fryxell, Taylor Valley, Antarctica. M.S. thesis, Univ. Waikato. 246 p. AND C. H. HENDY. 1985. Water column and sediment characterictics of Lake Fryxell, Taylor Valley, Antarctica. N.Z. J. Geol. Geophys. 28: 543-552. -, and -. 1989. Carbonate deposition and Ross Sea ice advance, Fryxell basin, Taylor Valley, Antarctica. N.Z. J. Geol. Geophys. 32: 267-277. LI, Y.-H., AND S. GREGORY. 1974. Diffusion of ions in sea water and in deep-sea sediments. Geochim. Cosmochim. Acta 38: 703-714. MCKAY, C. P., G. D. CLOW, R. A. WHARTON, AND S. W. SQUYRES. 1985. Thickness of ice on perennially frozen lakes. Nature 313: 56 l-562. M&NIGHT, D. M., G. R. AIKEN, E. D. ANDREWS, E. C. BOWLES, AND R. A. HARNISH. 1993. Dissolved organic material in dry valley lakes: A comparison oflake Fryxell, Lake Hoare, and Lake Vanda, p. 119-l 33. In W. J. Green and E. I. Friedman [eds.], Physical and biogeochemical processesin Antarctic lakes. Antarct. Res. Ser. 59. AGU. AND E. D. ANDREWS. 1993. Hydrologic and geochemical processesat the stream-lake interface in a permanently ice-covered lake in the McMurdo Dry Valleys, Antarctica. Int. Ver. Theor. Angew. Limnol. Verh. 25: 957-959. MASUDA, N., S. NAKAYA, H. R. BURTON, AND T. TORII. 1988. Trace element distributions in some saline lakes of the Vestfold Hills, Antarctica. Hydrobiologia 165: 103-l 14. MATSUBAYA, O., H. SAKAI, T. TORII, H. BURTON, AND K. KERRY. 1979. Antarctic saline lakes-stable isotonic ratios. chem- 976 Miller and Aiken ical compositions and evolution. Geochim. Cosmochim. Acta 43: 7-25. OREMLAND, R. S., L. G. MILLER, AND M. J. WHITICAR. 1987. Sources and flux of natural gases from Mono Lake, California. Geochim. Cosmochim. Acta 51: 2915-2929. QUAY, P. D. 1977. An experimental study of turbulent diffusion in lakes. Ph.D. thesis, Columbia Univ. 194 p. RAGOTZKIE, R. A., AND I. FRIEDMAN. 1965. Low deuterium content of Lake Vanda, Antarctica. Science 148: 1226-l 227. -, AND G. E. LIKENS. 1967. The heat balance of two antarctic lakes. Limnol. Oceanogr. 9: 412-425. REEBURGH, W. S. 1967. An improved interstitial water sampler. Limnol. Oceanogr. 12: 163-165. SMITH, R. L., L. G. MILLER, AND B. L. HOWES. 1993. The geochemistry of methane in Lake Fryxell, an amictic, permanently ice-covered, antarctic lake. Biogeochemistry 21: 95-l 15. STUIVER, M., I. C. YANG, G. H. DENTON, AND T. B. KELLOG. 198 1. Oxygen isotope ratios of antarctic permafrost and glacier ice, p. 13 l-l 39. In L. D. McGinnis [ed.], Dry Valley drilling project. Antarct. Res. Ser. 33. AGU. THATCHER, L. L., V. J. JANZER, AND K. W. EDWARDS. 1977. Methods for determination of radioative substances in water and fluvial sediments, p. 79-8 1. In Techniques of waterresource investigations. U.S. Geol. Surv. TORII, T., AND OTHERS. 1989. Chemical characteristics of pond waters in the labyrinth of southern Victoria Land, Antarctica. Hydrobiologia 172: 255-264. WELCH, H. E., AND M. A. BERGMAN. 1985. Water circulation in small arctic lakes in winter. Can. J. Fish. Aquat. Sci. 42: 506-520. WHARTON, R. A., AND OTHERS. 1992. Changes in ice cover thickness and lake level of Lake Hoarc, Antarctica: Implication for local climatic change. J. Geophys. Res. 97: 35033513. WILSON, A. T. 1964. Evidence from chemical diffusion of a climatic change in the McMurdo Dry Valleys 1,200 years ago. Nature 210: 176-177.
© Copyright 2026 Paperzz