Introduction
In geology, a facies is a rock or sediment unit that exhibits lithological, structural or faunal characteristics which enable it to be distinguished from other rock or sediment units. Although the term refers to all of the characteristics of a particular material, in practice the recognition of facies is based on the visible composition and structure of materials that are detectable in the field. As pointed out by Reference WalkerWalker (1984), facies-based approaches have been widely used in geology for a long time but only recently have they been applied to the study of glaciers (e.g. Reference LawsonLawson, 1979).
The ultimate purpose of facies designations and descriptions is to end up with an environmental interpretation. However, many facies have ambiguous interpretations and it is common to analyze all facies communally in context because the sequence in which they occur can contribute as much information as the facies themselves (Reference WalkerWalker, 1984). At Matanuska Glacier, Alaska, U.S.A., Reference LawsonLawson (1979) used a combination of ice characteristics, debris concentration and the mode of debris dispersion to define englacial and basal ice facies. Since Lawson’s study, facies-based approaches have become widely used and increasingly sophisticated in the study of the composition and origin of basal ice sequences (Reference Souchez and LorrainSouchezand Lorrain, 1991; Reference Sharp, Jouzel, Hubbard and LawsonSharp and others, 1994; Reference Hubbard, Sharp and LawsonHubbard and others, 1996). However, in the study of Earth systems, there is recognition that different initial states can lead to a similar end, i.e. different modes of origin may produce the same or very similar characteristics. This concept, known as equifinality, has proven to be an important one in Earth science because it works against a tendency to monocausal (single explanation) thinking.
The purpose of this paper is to examine the spatial variability in basal ice facies in a cold-based Antarctic glacier in order to understand whether ice facies that have the same or similar appearance may have a different origin. We document and discuss a situation from Suess Glacier, in Taylor Valley, Antarctica, which has recently been studied by Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999) and Reference Fitzsimons, Mcmanus and LorrainFitzsimons and others (1999, Reference Fitzsimons, Lorrain, Vandergoes, Maltman, Hubbard and Hambrey2000, Reference Fitzsimons, Mcmanus, Sirota and Lorrain2001). These papers focused respectively on ice composition analyses of the marginal basal zone, and on study of the mechanical properties of the basal ice encountered in a subglacial tunnel. In the present study, we are interested in both marginal and subglacial basal zones. We describe the field site and methods used, we introduce a facies classification common for the two sites, and we present the key analytical results and their interpretation.
Field Site and Methods
Suess Glacier is a small dry-based glacier damming a lake in Taylor Valley, South Victoria Land, Antarctica. It descends from 1750ma.s.l. on the Asgard Range to about 50ma.s.l. where it rests on the unconsolidated frozen sediments that compose the floor of the valley (Fig. 1). The right margin of the glacier terminus forms an 18–20m high cliff above an ice-and-debris apron adjacent to the dry-based Lake Popplewell (following Reference Chinn, Green and FriedmannChinn, 1993). The temperature at the base of the glacier is –17°C.
Fig. 1. Top: Location map of the study site (dotted lines represent melt streams). Bottom: Plan of the margin of Suess Glacier, showing the location of the tunnel and of the apron section, and the flow direction.
Two series of contiguous ice blocks (about 40 cm high, 10 cm x 20 cm wide) were sampled along the vertical: one from an excavation made in the apron and another from a shaft 4.5 m high dug at the end of a 25 m long tunnel (Fig. 1). All these ice blocks were taken with the aid of a chain-saw equipped with tungsten carbide teeth. Additionally, several ice cores were obtained with a SIPRE-type ice auger: seven 1.6 m long ice cores were retrieved along a longitudinal profile from the upper surface of the glacier, and a 51 m long core was retrieved from above the tunnel. The ice samples were then wrapped in polyethylene bags, stored at –18°C and transferred to the cold laboratory in Brussels.
In the cold room, vertical thin sections were cut along the two profiles, and samples were taken for measurements of stable isotopes (δD and δ18O), total gas content and gas composition. To avoid any contamination, a vertical 2 cm thick slice was cut from the surface of the samples and discarded prior to any further sampling. Because of the presence of a great amount of debris, most of the thin sections were cut using a diamond-wire saw as described by Reference TisonTison (1994). For debris-free ice, the vertical thin sections were cut with a microtome. All thin sections were photographed between crossed polaroids, allowing ice-texture observations.
Sampling for isotopic analyses was done with the saw cutting horizontal short sticks, 5 cm long with a section of 1cm2. After melting, the samples were analyzed at the Nuclear Research Center of Saclay in France. Results are expressed in d-units normalized to Standard Mean Ocean Water (SMOW). Precision of the measurements is ±0.5% in δD and ±0.1%in δ18O. The gas analyses were performed in Brussels. Thegascompositionwasmeasured on ice samples of 70–80g using the technique described by Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999). Total gas content was measured on samples of about 30 g by a melting–refreezing method described by Reference Raynaud, Chappellaz, Barnola, Korotkevichand and LoriusRaynaud and others (1988) and Reference BlunierBlunier and others (1993).
Ice Facies
Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999) and Reference Fitzsimons, Mcmanus and LorrainFitzsimons and others (1999) have respectively studied the apron basal zone and the tunnel basal zone. These papers alreadydescribed the basal ice units present at these two sites, and pointed to the similarity of their structures. In the present paper, we introduce a facies classification common for both sites. Several facies are present: an amber facies, a solid facies andabasal stratified facies. Because of its variability in debris and bubble content, the latter facies has been subdivided into three subfacies: laminated, dispersed bubbly and dispersed clear subfacies.
-
The amber ice facies (Fig. 2a) consists of slightly ambercolored ice containing very dispersed particles.Thisfacies has been described by Reference HoldsworthHoldsworth (1974) at Meserve Glacier which is located not far from Suess Glacier but in Wright Valley (see Fig. 1for location).
-
The basal solid facies (Fig. 2b) consists of frozen sand and fine gravel that contains well-preserved sedimentary structures. Ice is only present in the pores between mineral grains.
-
The basal stratified facies, which lies below the amber ice, consists in a succession of layers that range from clean ice bands to bands with very high debris concentration, the latter being often deformed. Each of these layers belongs to one of the subfacies described below. According to the debris distribution, a laminated and a dispersed subfacies have been defined. Because of the great variability in the bubble content within the dispersed facies, the latter has been subdivided into a bubbly and a clear subfacies.
-
3.1. The laminated subfacies (Fig. 2c) contains stacks of clear ice layers (a few mm to a few cm thick) interspersed with debris-rich laminations <1mm to a few mm thick. The latter sometimes are superposed, forming layers a few cm thick.
-
3.2. The dispersed bubbly subfacies (Fig. 2d) contains dispersed debris in variable concentrations, but the debris content is much lower than in the debris-rich laminations of the laminated subfacies. It contains numerous bubbles that are mostly elongated (from tenths of mm to a few mm long) parallel to the inferred local flow direction. It sometimes contains a few layers of clear ice. A few layers and/or clusters of basal solid facies were observed within this facies.
-
3.3. The dispersed clear subfacies (Fig. 2e) has the same characteristics as the dispersed bubbly subfacies, but has a very low bubble content. When visible, bubbles are concentrated in layers aligned parallel to the inferred ice-flow direction.
Fig. 2. Pictures of the facies and subfacies encountered at the apron and tunnel sites: (a) amber ice facies; (b) clusters within the dispersed bubbly subfacies; (c) laminated subfacies; (d) dispersed bubbly subfacies with solid facies; (e) dispersed subfacies.
Figure 3a and b show how the different facies are arranged in the apron and in the tunnel profiles. The thickness and the relative position of the facies differ between the two locations. The basal amber facies, for example, is only 20 cm thick at the apron, which is very thin in comparison with the 125 cm thickness in the tunnel. No solid facies was sampled in the apron sequence, but two layers with a very high debris concentration were observed within the dispersed clear and bubbly subfacies. These debris layers, whose debris content is close to that of the solid facies, are respectively located between –320 and –370cm and between –429 and –437 cm. In the tunnel profile, the solid facies appears only as layers and clusters within the stratified facies. However, elsewhere in the tunnel it can be observed as a thick layer (up to 0.6 m thick and averaging 0.5 m thick in the rear part of the tunnel) between the amber and the stratified facies, as presented in figure 3 in Reference Fitzsimons, Mcmanus and LorrainFitzsimons and others (1999), which shows a synthetic view of several local observations from Suess tunnel.
Fig. 3. Main characteristics of the basal ice: δD (▴), δ18O (o) and deuterium excess profiles with the arrangement ice facies (a) for the apron basal ice, (b) for the tunnel basal ice.
Results
Apron
In the amber ice, crystals are equigranular and very small (mean diameter 0.7 mm). In the stratified facies, crystal size varies between ∼1 and ∼1 0 mm in diameter, the biggest ice crystals being observed in the dispersed clear subfacies, and the smallest ones in the laminated subfacies.
The isotopic profiles of the apron site are presented in Figure the 3a. The δ18O and δD values in the amber ice have a narrow range (–33.51 to –32.67% and –264 to –260%, respectively) while they vary widely in the stratified facies (–38.64 to –28.1%for δ18O and– 286.8 to –228.5% for δD). The laminated subfacies presents generally low δ values (–36.93 to –32.38% in δ18O and –278.5 to –253.8% in δD). Besides, the dispersed bubbly subfacies shows a large range of δ values (–33.92 to – 28.31 % in δ18O and –267.6 to –228.5% in δD). This subfacies presents a very narrow range of isotopic values in the middle part of the profile, but it shows high δ values in the lower part. The clear ice subfacies contains the lowest and the highest δ values of the whole profile (–38.64 to –28.1% for δ18O and – 286.8 to –228.5% for δD). The deuterium excess (d = δD–8δ18O) varies from –4.5 to 22.3, as a mirror image of the δ profiles.
Notes: Values with a * are from Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999). Heights are from the limit between glacier ice and basal ice.
Results of the gas analyses from different ice facies (Table 1) show that CO2 concentrations range from 6674 to 229000 ppmv and that O2/N2 ratios vary between 0.20 and 0.52. The total gas content is generally lower than in meteoric ice.
Table 1. Results ofgas analyses ofa meteoric ice sample, and of basal ice samples from the tunnel and from the apron sites
Tunnel
Except in the amber ice, ice crystals are smaller in the tunnel than at the apron. In the stratified facies, the biggest ice crystals (∼5 mm) are observed in the dispersed clear subfacies, whereas the smallest ice crystals (∼0.7 mm) are encountered in the debris-rich layers. The dispersed bubbly ice is made up of crystals 1.5–2.5 mm in mean diameter.
The isotopic profiles of the tunnel site are presented in Figure 3b. δD and δ18O values for the amber ice are similar at the glacier margin and in the tunnel (values for the tunnel amber ice are –33.43 to –33.03% for δ18O and –266.7 to –264.1% for δD). There are, however, strong contrasts between the δ values for the stratified facies in the two locations. In the tunnel, the highest isotopic values (up to –28.17% for δ18O and –226.7% for δD) were found in the dispersed clear subfacies, and in a sample at –142 cm that corresponds to a clear layer in the dispersed bubbly sub-facies. This situation of higher isotopic values in the clear ice is similar to what Reference Iizuka, Satake, Shiraiwa and NaruseIizuka and others (2001) have observed for samples from Hamna Glacier, Soya Coast, East Antarctica.
The deuterium excess (d) varies in a narrow range (–0.4 to 2.3) from the top ofthe sequence all the way to –180 cm. d is higher (2.9–5.3) in the dispersed bubbly subfacies between –253 and –318 cm. d is lower (–4.8 to –0.6) in the dispersed clear and in the laminated subfacies, and reaches a much lower value in the solid facies near the base of the profile (–8).
Results of the gas analyses from different ice facies are presented in Table 1. CO2 concentrations range from near 909 ppmv in the amber ice to a concentration of 14 179 ppmv in the dispersed bubbly subfacies at –317 cm. CO2 concentrations are constant in the amber ice and have a large range in the two other subfacies. O2/N2 ratios vary between 0.21 and 0.26. Total gas content is very high in the amber facies (about 0.1 cm3 g–1) and in the dispersed bubbly facies (about 0.8 cm3 g–1). These values are in the range reported by Reference Martinerie, Raynaud, Etheridge, Barnola and MazaudierMartinerie and others (1992) for glacier ice from polar ice cores. In contrast, the values from the dispersed clear facies are much lower: as low as 0.007 cm3 g–1. These gas data differ from those of the apron basal ice, which has generally a higher CO2 concentration and a lower total gas content (Table1).
Discussion
Isotopic and gas composition
The results presented above demonstrate that although the basal ice facies exposed at the apron and inside the tunnel look the same, they clearly present differences in ice composition.
Figure 4 shows the δD–δ18O diagram for the different ice samples taken in the Suess Glacier environment. The local meteoric water-line (LMWL) has been obtained from glacier ice samples collected along the longitudinal profile and glacier ice samples from the 51 m core cited above. The slope obtained has the following equation: δD = 8.21δ18O + 11.5 (r2 = 0.98), which is nearly equivalent to the LMWL defined in Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999) from the samples of the glacier surface only. The amber ice samples do not plot significantly away from this line, indicating that amber ice cannot be isotopically distinguished from glacier ice, at least at the scale of our sampling, which can be coarser than the scale of potential individual freezing events.
Fig. 4. δD–δ18O diagram showing samples from: the surface ice cores and the 51m long ice core (9); the amber ice from apron and tunnel sites (□); the tunnel stratified ice (◊); and the apron stratified ice (A). The solid line corresponds to the LMWL (regression line from the glacier ice samples); the dashed line corresponds to the regression line from the apron basal stratified facies. S is the value of the slope of the nearby regression line.
In contrast, ice samples from the stratified facies at the apron site are aligned on a much lower slope, the value of which is 5.6 (Fig. 4). It can be considered as a freezing slope since its value is nearly identical to that calculated by Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999) for this environment. The fact that this slope is lower than 8.21 (slope of the LMWL) explains the mirror image of the d profile vs the δ profiles (Fig. 3a and b). Another striking feature of the isotopic composition of the apron ice samples is that some samples, from different ice facies, are considerably depleted in heavy isotopes of oxygen and hydrogen along the freezing slope. Such characteristics imply the more-or-less complete freezing of liquid-water volumes. This has been theoretically and experimentally developed by Reference Jouzel and SouchezJouzel and Souchez (1982) and Reference Souchez and JouzelSouchez and Jouzel (1984). As the freezing front progresses into water, ice is enriched in heavy isotopes with regard to water. When the residual liquidwater becomes very restricted, it impoverishes in deuterium and oxygen-18, and the forming ice is therefore impoverished in heavy isotopes as well. Such isotopic results are consistent with the hypothesis that basal ice at Suess apron has been built up by ice and debris accretion to the sole of the glacier during its advance into a wet-based lake (Reference FitzsimonsFitzsimons, 1996; Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others, 1999, Reference Lorrain, Sleewaegen, Fitzsimons and Stievenard2002).
In the tunnel, the basal ice samples from the stratified facies are generally enriched in heavy isotopes compared to the glacier ice samples and are clearly offset from the LMWL. They are also not aligned on the freezing slope obtained from the apron ice samples. Such a situation nevertheless implies phase changes, probably from minor liquid-water volumes. This indicates that liquid water makes a less significant contribution to the formation of basal ice in the tunnel (minor contribution) than at the apron (major contribution).
The gas data also point to this variable contribution at the apron and in the tunnel. Thebasal ice samples from the apron show some very high CO2 concentrations, high O2/N2 ratios and low total gas contents, in striking contrast to meteoric ice (Table 1). Lorrain and others (1999) explained these characteristics by considering freezing of liquid water. Indeed, this process causes gas rejection and preferential incorporation of CO2 because of its higher solubility. Furthermore, as mentioned by Reference Lorrain, Fitzsimons, Vandergoes and StievenardLorrain and others (1999), some organic processes associated with the presence of liquid water may have produced the unusual measured gas composition.
In comparison with what is shown at the apron, the basal ice samples from the tunnel generally present lower CO2 concentrations and lower O2/N2 ratios, but higher total gas content (Table 1). In comparison with the glacier ice above, the basal ice from the tunnel presents much higher CO2 concentrations and slightly lower O2/N2 ratios. Concerning its total gas content, however, the basal ice shows the same values as for the above glacier ice (0.070 cm3 g–1), except for the clear ice samples. These results indicate that the tunnel basal ice has undergone an increase in CO2 without significant loss of gas.
The two different signatures for tunnel basal ice and apron basal ice can also be illustrated by a CO2–δ18O diagram (Fig. 5). In this figure, the very high CO2 concentrations from the apron correspond to depleted δ18O values, while CO2 concentrations in the tunnel are independent of δ18O values. These data reinforce the interpretation given about the occurrence of complete freezing of liquid-water volumes at the apron.
Fig. 5. CO2–δ18O diagram showing samples from apron basal ice (▵) and tunnel basal ice (♦).
Thus, while the microbiological activity is not excluded as a mechanism to change gas composition in basal ice, lower CO2 concentrations in the tunnel compared to the apron, combined with lower O2/N2 ratios and higher total gas contents, suggest that freezing of liquid water has played a lesser role in the formation of basal ice in the tunnel than in the apron.
Build-up and spatial continuity
At the base of Meserve Glacier cited above, amber ice is in direct contact with the bed except in a 20 m wide zone toward the marginal ice cliff. In this zone the amber layer rises gradually and over-rides an apron made of marginal debris containing ice blocks fallen from the cliff, windblown snow and refrozen meltwater (Reference HoldsworthHoldsworth, 1974; Reference CuffeyCuffey and others, 2000). In contrast, at the base of Suess Glacier, even at 25 m inside the tunnel, the amber ice lies on top of 2.15 m of stratified ice, well above the effective bed (defined as the place where displacement is zero). Although the formation of the stratified layer has been attributed to zones of weakness such as cavities or ice lenses within the glacier substrate (Reference Fitzsimons, Mcmanus, Sirota and LorrainFitzsimons and others, 2001), the entrainment processes have yet to be fully understood. However, it seems clear that the stratified facies has been formed by entrainment of fragments of the substrate at the base of the glacier. Such a mechanism implies that the amber ice that lies above the stratified facies has been formed upstream of and before the stratified facies.
The flow direction (see Fig. 1) and the similarity of facies exposed in the tunnel and at the apron indicate that a spatial continuity exists between these two locations. However, the strong differences in ice composition between the two sequences suggest that important changes occur towards the margin. Strong arguments have been developed in favour of water freezing and debris accretion to explain the apron ice, but it is clear that, despite the similarities of facies, the basal ice in the tunnel has not been formed by the same mechanism.
Conclusion
Isotopic analyses and gas data indicate that there is a strong spatial variability in the composition of basal ice in Suess Glacier. The strong contrast in isotopic composition suggests that liquid water plays an increasingly important role in basal ice formation toward the glacier margin.
A comparison of the ice composition data presented in this paper, together with descriptions of facies from the tunnel and ice-marginal sampling site, demonstrate that ices with similar visible characteristics can have distinctly different chemical composition. This observation suggests that facies descriptions can be unreliable indicators of origin and that careful composition alanalysis is required to decipher the mechanisms involved.
Acknowledgements
We would like to thank Antarctica New Zealand for providing the logistical support for this study. The Marsden Fund of the Royal Society of New Zealand and the University of Otago provided financial support. Thanks are also due to M. Stievenard from the Nuclear Research Center in Saclay, France, for the isotopic analyses and to R. Souchez for his constructive comments on an early draft of this paper. This paper is a contribution to the Belgian Scientific Program on Antarctica (OSTC–SPSDII). D. Samyn acknowledges grant support from Fonds pour la formation a la Recherche dans l’Industrie et dans l’Agriculture (FRIA), Belgium. We are also indebted to B. Hubbard for critical reviews of the paper, which resulted in substantial improvements to the text.
