Introduction
Sea-water infiltration into porous ice shelves has been observed at a number of Antarctic locations (Reference DubrovinDubrovin, 1962; Reference YevteyevYevteyev, 1962; Reference Stuart and BullStuart and Bull, 1963; Reference HeineHeine, 1968; Reference ThomasThomas, 1975; Reference Kovacs and GowKovacs and Gow, 1975, Reference Kovacs and Gow1977; Reference Kovacs, Kovacs, Gow and CraginKovacs and others, 1982[a]). Much of the past work has focused on either determining the geographical extent of the brine penetration inland using radio echo-sounding techniques (Reference CloughClough, 1973; Reference Kovacs and GowKovacs and Gow, 1975) or on studying ice-shelf brine thermodynamics (Reference ThomasThomas, 1975), Initially, brinelayer formation was believed to be due to vertical infiltration of sea-water into porous firn (Reference Stuart and BullStuart and Bull, 1963) but Reference Risk and HochsteinRisk and Hochstein (1967), and Reference Kovacs and GowKovacs and Gow (1975), have concluded from their observations on the McMurdo Ice Shelf that lateral infiltration of sea-water from the ice-shelf edge is the dominant mechanism of brine infiltration. More recent observations by Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others (1982(b)) show that lateral infiltration is itself dominated by wave-like intrusions of sea-water triggered by periodic break-out (calving) of the ice front in McMurdo Sound.
In order for brine to remain liquid at the temperature encountered in ice shelves, excess water must be eliminated by freezing within the pores of the permeable firn. Freeze-concentration could also lead to significant changes in brine chemistry. The only reported chemical measurements of the composition of the infiltrated brine are those of Reference Wilson and HeineWilson and Heine (1964) and Reference HeineHeine (1968), who observed a SO4 Footnote * depletion relative to Cl in the brine layer of the McMurdo Ice Shelf. They attributed this to preferential precipitation of mirabilite. Here we report results of more extensive chemical analyses of a series of ice cores taken from the top of the brine layer in the McMurdo Ice Shelf.
Experimental Procedures
Core drilling and sampling
An electro-mechanical drill designed by Reference Rand and SplettstoesserRand (1976) was used to obtain 10 cm diameter cores from the snow surface down to about 3 m into the brine layer. In this particular drill the electrical section leading to the motor was not designed to operate in liquids, so it was not possible to drill deeper than about 3 m into the brine layer. We obtained additional cores with a modified SIPRE coring auger. Cores were drilled at the locations shown in Figure 1. The cores labeled 77 High Step and 77 Low Step were drilled during January 1977; cores from locations A, B, C, D, and E were drilled during November 1978.
Sections of brine-soaked firn and ice were cut from cores at about 10 cm intervals, melted in polyethylene containers and transferred to glass vials for shipment to CRREL for chemical analysis.
Cation analyses
We performed cation analyses for Na, K, Ca, and Mg on the melted samples, which had been diluted by factors of 104 to 106, depending upon the element and the sample salinity. Concentrations were determined on 20μl aliquots by graphite–furnace atomic absorption using a Perkin—Elmer Model 403 AA spectrophotometer in conjunction with a Model 2200 heated graphite atomizer. Deuterium background correction was unnecessary at the low concentrations (μg/1) of the diluted samples. Analysis for Na, however, did require a maximum temperature (approximately 2700°C) heating step after each atomization to remove residual salts and to prevent interference in the signal of subsequent aliquots. Standards, prepared fresh daily from Fisher 1000 mg/l stock solutions, and samples were analyzed in triplicate.
Sulfate and chloride analyses
Sulfate concentrations were determined turbidimetrically using BaCl2 with a stablizing agent to form a BaSO4 suspension. Samples were diluted by factors of 2–100 to bring their SO4 concentrations into the analytical working range of 5–50 mg/l. Absorbances were measured in 2.5 cm cells at 450 nm with a Hach DR/2 spectrometer.
As a check on the above turbidimetric method, we also analyzed selected samples for SO4 using the more involved barium chloranilate technique (Reference Bertolacini and BarneyBertolacini and Barney, 1958; Reference ShaferShafer, 1967) modified as previously described (Reference Cragin, Cragin, Herron and LangwayCragin and others, 1975) without sample pre-concentration. A comparison (Table I) of results of these two methods shows excellent agreement.
Chloride concentrations were determined potentiometrically using an Orion solid-state Cl electrode and an Orion double-junction reference electrode with l M KNO3 filling solution in the outer chamber. Standards containing 5–50 mg Cl/l were prepared from a 1000 mg Cl/l stock solution of “Specpure” NaCl (Johnson, Mathey and Co., Ltd). Samples were diluted 10–1000 times to bring their Cl concentrations into the working range of the standards.
Precision of all the above chemical analysis methods was ±10% or better.
Sea-water analysis
As an additional verification of the accuracy of the analytical methods used, a sample of Copenhagen Standard Seawater (CSS) was diluted appropriately and analyzed along with the brine samples. Measured concentrations, included in Tables II and III, conform very closely to the expected values.
Results and Discussion
Core locations
Figure 2 presents a cross-section of the McMurdo Ice Shelf in the area of interest. It shows the location of drill holes that penetrated brine, together with the brine layer and ice-shelf bottom as determined by drilling and radar profilometry (Reference Kovacs and GowKovacs and Gow, 1977; Reference Kovacs, Kovacs, Gow and CraginKovacs and others, 1982[a] Reference Kovacs, Kovacs, Gow, Cragin and Morey[b]). Two cores were drilled during the 1976-77 austral summer on either side of a prominent 4.4 m high brine step (Fig. 3): the 77 High Step and the 77 Low Step cores. During the 1978-79 austral summer, five more cores were drilled at locations A, B, C, D, and E. Figure 2 is drawn for the brine layer as it existed in January 1977. Observations made during the second field season (November 1978) showed that the brine step had further infiltrated the ice shelf and that the it was now located at position C, a net inland movement of 700 m in a little less than 2 years (Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others, 1982 [b]).
Preliminary observations
This discovery of a migrating brine step determined for the first time the dynamic wave-like nature of brine infiltration in the McMurdo Ice Shelf. According to our perception of events, the 77 High Step core penetrated the leading edge or nose of a brine wave riding on top of an older brine layer from which the 77 Low Step cores were obtained just a few meters beyond the 77 High Step drill core. Preliminary chemical analyses conducted before the brine step was re-surveyed in November 1978 showed that brines from the 77 Low Step core had appreciably lower SO4/ Na ratios than brines from the 77 High Step core. At that time we interpreted this as indicating the presence of two chemically dist inct brine layers in the immediate vicinity of the brine step.
Brine chemistry
((Table II)) and III present the complete data of chemical concentrations and ionic weight ratios for samples from all seven drill holes. It should be noted that sample salinities and elemental concentrations vary because the brine has been diluted with melt water from the firn of the cores. This dilution results in all samples having salinities and concentrations very much less than those of sea-water. The amount of dilution is different for each sample so chemical concentrations or salinities per se cannot be used directly to determine whether any salts are selectively precipitated. However, because the dilution water is known to be derived from firn, which is chemically much purer than sea-water, ratios of elemental concentrations are not affected by dilution and should reveal any chemical fractionation that has occurred. The purity of McMurdo Ice Shelf firn is shown by chemical analysis of four samples from above the brine zone in the 77 High Step core where Na concentrations averaged 4.3 mg/l, which is approximately three orders of magnitude less than Na concentrations in the brine-soaked firn.
A plot of Na concentration versus salinity shows a linear relationship with a least-squares slope of 0.299 (r = 0.972). This agrees well with the expected slope of 0.319 for unfractionated sea-water. There is no discernible difference between the Na/salinity ratio of fractionated and unfractionated core samples. Most of the scatter at the high-salinity end of the curve is caused by samples from the 77 Low Step and 77 High Step cores where, as will be discussed later, chemical fractionation of the brine was actively taking place.
In studies of chemical fractionation of sea-water, NaCl is generally considered to be chemically conservative, and elemental ratios are usually calculated relative to Cl or Na (Reference Duce, Duce, Stumm and ProsperoDuce and others, 1972). We used Na as the reference element simply because the analytical measurement was less time-consuming than that for Cl.
Sulfate is the primary ion of interest. When sea-water freezes, one of the first major salts to separate is sodium sulfate decahydrate (mirabilite), Na2SO4·10H2O, which begins to precipitate at –8.2°C after sufficient water has been removed as ice (Reference Lewis and ThompsonLewis and Thompson, 1950; Reference Nelson and ThompsonNelson and Thompson, 1954). Sodium chloride dihydrate (NaCl·2H2O) then follows at -22.9°C (Reference Nelson and ThompsonNelson and Thompson, 1954) but the in-situ McMurdo Ice Shelf temperatures, typically –12° to –16°C in the deeper parts of the brine zone (Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others, 1982[b]), are not low enough to cause its precipitation.
The last column in Table II shows the reduction in SO4/Na ratio for the brine from drill cores D and E; SO4/Na ratios in the brine layer of this part of the ice shelf are an order of magnitude lower than the normal seawater SO4/Na ratio of 0.25. This lower ratio could be due either to a depletion of SO4 or to an enrichment of Na relative to Cl. To distinguish between these two mechanisms, we measured Cl concentrations in 14 samples of high and low salinities, and from fractionated (low SO4/Na ratio) brines (see Table III). The average Cl/Na ratio of 1.8 ± 0.3 is the same as the seawater Cl/Na ratio, indicating that Na is not enriched. Thus, the low SO4/Na ratios observed in the brine from cores D and E must be due to depletion of SO4.
The SO4 depletion observed in the brine of drill cores D and E could result from precipitation of any of a number of alkali or alkaline-earth salts of sulfate. The question is, which one? The least soluble of the SO4 salts is CaSO4 and Reference RichardsonRichardson (1976) reported the presence of CaSO4·2H2O in sea ice. But the average Na/Ca ratio (Table III) of three SO4-depleted brine samples is 26, which is not appreciably higher than the sea-water Na/Ca ratio of 25. Neither do changes in Ca concentrations parallel changes in SO4 concentrations. Furthermore, there is not enough Ca in sea-water to precipitate the amount of SO4 that was removed. If all the Ca in sea-water, 0.43 g Ca/l (21 meq Ca/l) is removed as CaSO4, then 1.73g SO4/l (36 meq SO4/l) would remain; the SO4/Na ratio would then be 0.16, which is still six times greater than that observed for the fractionated samples. This means that some other cation is removing SO4.
Even though sea-water contains enough Mg conceivably to remove all of the SO4, the good agreement between the sea-water Na/Mg ratio and that of the samples (Table III) indicates that Mg is not being removed. This is probably because the lowest temperature in the ice shelf (–16°C) is warmer than the observed temperatures of deposition of MgCl2·8H2O from sea-water (–18°C) (Reference RichardsonRichardson, 1976). Neither does K appear to be affected, as shown by the Na/K ratios.
It is more likely that Na2 SO4 would precipitate, since it is less soluble than K2 SO4 and since sea-water contains a much higher concentration of Na than of K. In fact, sea-water contains such a high concentration of Na that precipitation of all SO4 as Na2 SO4 would change the Na/K ratio by only 6% (from a normal 27.8 to 26.1). Thus, the high sea-water concentration of Na permits the SO4 to be removed as Na2 SO4 without causing an appreciable change in the ratios of Na/K, Na/Mg, or Na/Ca. Also, because of the small relative change in Na concentration when Na2 SO4 precipitates, changes in the SO4/Na ratio essentially reflect changes in the SO4 concentration.
Sulfate is depleted as much as ten-fold in samples from cores E, D, and 77 Low Step (Table II). Cores E and D are the farthest inland from the ice front and correspond to the deepest and oldest brine layers (Fig. 2). Samples from the 77 Low Step cores are also from an older brine zone currently being over-ridden by a new wave of brine that was sampled first at 77 High Step and then approximately 22 months later at location C (see Fig. 2). Samples from these latter two cores together with brine samples from cores A and B show no significant SO4 depletion and have SO4/Na ratios that are normal or greater than sea-water; this is reasonable because cores A and B are located closest to the ice front and contain brine from the most recent intrusion of sea-water.
Brine-wave dynamics
The brine step at location C, as of 28 November 1978, represents the leading edge of a brine wave migrating through the ice shelf in a direction opposite to the movement of the ice shelf itself. Because of glacial flow, the McMurdo Ice Shelf is moving toward McMurdo Sound at rates of 47–118 m/year (Reference HeineHeine, 1968), depending upon location. At Heine’s station 307, located very close to drill site E, the ice shelf is moving at a measured velocity of 84 m/year. Drill site E is very close to the inland boundary of brine infiltration where the rate of migration of the deepest brine layer is only about 12 m/year (Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others, 1982[b]). This means that, in terms of its geographical location, the brine terminus is currently moving seaward with the ice shelf at a velocity of about 72 m/year.
Brine migration within the ice shelf is controlled by the permeability and temperature of the firn and ice, and is maintained or renewed by sea-water entering at the ice front. As demonstrated by Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others (1982[b]], the overall process of infiltration is dominated by wave-like intrusions of sea-water triggered by periodic break-out of the ice front in McMurdo Sound. (See Figure 4 for a schematic of break-out.) The 4.4 m brine step, originally detected at the 77 High Step and 77 Low Step drill sites and which subsequently migrated to location C, represents the most recent example of a wave generated by a major break-out of the McMurdo Ice Shelf. This break-out is postulated to have taken place about 1970. Several other steps (waves) have been identified in radio-echo profiling records by Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others (1982[b]), including several very prominent steps near the terminal boundary of brine migration.
Infiltration of sea-water into the McMurdo Ice Shelf is accompanied by freeze-concentration of the brine and freeze-fractionation of component salts, particularly mirabilite. The younger brine waves penetrate the ice shelf by simple permeation through porous firn. On 10 January 1977 the leading edge of the 4.4 m brine wave was located 2.85 km from the ice-shelf edge and by 24 November 1978 it had migrated to a position 3.55 km from the ice-shelf edge, an average rate of migration of 1.04 m/d.
When sufficient water (approximately 80%) is removed by freezing and the brine temperature reaches -8.2°C, mirabilite begins to precipitate. Precipitation occurs at the leading edge of the brine wave where the salt concentration is the highest and the temperature is lowest. The chemical fractionation is clearly demonstrated in samples from the 77 High Step and 77 Low Step cores. In samples from the 77 High Step core, SO4/Na ratios lie between 0.5 and 0.68 (Table II), which is about twice the normal sea-water ratio of 0.25. This excess SO4 in the 77 High Step samples can be attributed to inclusion of freshly precipitated mirabilite at the forward edge or nose of the brine wave. By contrast, SO4/Na ratios in samples from the 77 Low Step core located directly inland of the brine wave are much lower (2.3-4.5 times less) than that of normal sea-water. The 77 Low Step cores are depleted in SO4 because they were obtained from an older fractionated brine layer over which the new brine wave is migrating. The 77 High Step and 77 Low Step cores were taken within 60 m of each other, so that the dramatic ten-fold change in SO4/Na ratios occurs over a very short horizontal distance.
The deepest and oldest brine layer, sampled at drill site E near the inland boundary of brine penetration, contains highly concentrated brine in which the SO4/Na ratio is an order of magnitude less than that of normal sea-water (Table II). Despite the fact that this brine is now enclosed within impermeable ice (Reference Kovacs, Kovacs, Gow and CraginKovacs and others, 1982[a]), migration of the brine has not ceased. Movement on the order of 0.04 m/d has been measured (Reference Kovacs, Kovacs, Gow, Cragin and MoreyKovacs and others,1982[b]) and one possible explanation of this continued migration of brine is in the eutectic dissolution of ice by the concentrated brine as it is transported into deeper and warmer parts of the ice shelf by the combined process of surface snow accumulation and bottom melting.
Conclusions
Chemical analyses of cores from a 4.4 m high step in the brine-soaked firn of the McMurdo Ice Shelf confirm that this step is the leading edge of a brine wave that originated from sea-water entering the ice front after a major break-out of the ice shelf in McMurdo Sound around 1970. As the sea-water begins to permeate the porous firn in the ice shelf, water is removed by freezing and concentrated brine begins to form. When the brine reaches a concentration of about five times that of the original sea-water and a eutectic temperature of –8.2°C, mirabilite (Na2SO4·10H2O) begins precipitating at the wave front. As the brine wave advances further inland, mirabilite continues to separate until the residual brine contains less than 10% of the original sea-water SO4. Once the brine has traveled far enough into the ice shelf to reach the firn-ice transition, further movement of the brine is slowed dramatically by impermeable ice.
Thus, the overall brine zone consists of superimposed waves that originate at the ice-shelf front during ice breakouts and migrate laterally at different velocities, depending on their position within the ice shelf. This wave-like migration through firn, in which brines become increasingly concentrated and dramatically depleted in sulfate, represents the principal mechanism by which sea-water infiltrates the McMurdo Ice Shelf.
Acknowledgement
Field work was supported by a grant from the Division of Polar Programs of the National Science Foundation, grant NSF-DPP 77-19565. Laboratory studies were funded under DA Project 4A161102AT24, Cold Regions Surface Boundary Layer Physics and Chemistry. The authors thank Dr S. Colbeck and D. Leggett, both of CRREL, for reviewing this report and the Polar Ice Coring Office (PICO) of the University of Nebraska for its assistance with core drilling during November 1978.