Introduction
Distinct morphological features exist on the surface of a perennial ice layer that caps Patterned Lake at the northern end of the Central Masson Range in Mac.Robertson Land, East Antarctica (Fig. 1). The remarkable ice patterning on the surface of this lake was found in 1979 (Reference PickardPickard and Adamson, 1983) and a detailed study was carried out in February 1985. This paper describes these surface features and relates them to the fabric observed in the ice. Patterns of crystal orientation developed in frozen bodies of water vary considerably within both lake ice (Reference MichelMichel and Ramseier, 1971) and sea ice (Reference WeeksWeeks and Ackley, 1982). Observed c-axis variations can be related both to ice morphology and to the processes of ice growth.
Horizontal c-axes are a well-documented feature of both sea ice and lake ice, and alignment of horizontal c-axes in sea ice has been studied extensively (e.g. Reference WeeksWeeks and Ackley, 1982).
Perennially frozen pro-glacial lakes are a feature of the Antarctic continent and are common in the Framnes Mountains (Reference PickardPickard and Adamson, 1983). Patterned Lake is about 20 ha in area and has a permanent ice cover at least 4 m thick. It is typical of perennially frozen lakes found at glacier/rock margins in that it occupies a topographic low and represents decay of stationary glacial ice that has been trapped during a general lowering of the East Antarctic ice sheet. It is further isolated from the effects of glacial movement by the topography of the enclosing mountains. Characteristically, perennially frozen lakes in the Framnes Mountains have a moat isolating central perennial ice which is elevated a few tens of centimetres above the level of the moat (Fig. 2). Water accumulation occurs, particularly in the moat, during summer through the melting of winter snow banks of adjacent rocky bluffs and activation of glacial melt streams which drain into the lake. Since there is no apparent surface or sub-surface drainage leaving Patterned Lake, evaporation is the only known method of water loss. Thus, ice that forms at the bottom of the perennial ice block will eventually be exposed and ablated from the lake surface. This process may take around 20 years to complete if ablation rates measured in the vicinity of the Masson Range (Reference CrohnCrohn, 1959; Reference BuddBudd, 1967; Illingsworth, unpublished) of between 18 and 25 cm of ice per year are typical. However, the occurrence of inclusions of melt water, which form during summer below about 0.7–1.0 mm from the surface, complicate this simple model. In February 985, the perennial ice contained numerous pockets of melt water which prevented the recovery of ice cores. We infer that the ice block had become porous due to heat absorption, without substantial thinning. Lakes in the Framnes Mountains are fresh water with the depths of some exceeding 100 m (Reference PickardPickard and Adamson, 1983). The depth of Patterned Lake has not been determined but it is likely to be at least 50 m by comparison with other lakes of similar size in the embayments of the mountains.
Description of Ice Types of Patterned Lake
Three distinct ice types exist within Patterned Lake: glacier ice, annual moat ice, and perennial lake ice. The distribution of these different ice types is shown in Figure 2.
Glacier ice is whiter than other types. It shows typical glacier-ice features such as flow lines and cryoconite holes. A study of glacier-ice fabrics in the vicinity of the Framnes Mountains has been made by Reference KizakiKizaki (1969[a], [b]) and the micro-structure of glacier ice is not further considered here. It is, however, interesting to note that the large rafts of glacier ice visible in the 1960 air photographs had been reduced in 1985 to about one-half of their size. The reduction in size of these glacier remnants over 25 years can be explained by melting as a result of thermoerosion by lake water and eventual incorporation of glacier ice into the lake. Hence, perennially frozen lakes are now expanding at the expense of the adjacent glaciers, a process well illustrated at Bicuspid Lake (Reference PickardPickard and Adamson, 1983).
The moat varies depending on the season. At the height of summer, the moat is a shallow trough of melt water and granular ice underlain by perennial ice. The moat surrounds the central perennially frozen ice block. As the temperature drops towards the end of summer, freezing forms a smooth ice surface which is bluish in colour. It may have a lace-like appearance (Fig. 3c) caused by small air bubbles between crystals that are generally greater than 2 cm in diameter.
Perennial ice has a coarser texture than the ice types previously described (Fig. 3a and b). In all perennially frozen lakes at glacier/rock margins, ablational etching has, by late summer, given rise to an uneven ice surface with a variation in relief of around 10 cm. Only at Patterned Lake are large-scale individual patterns of several metres clearly developed. On other lakes, small-scale patterned areas may occur, for example, on Rumdoodle Lake. An individual pattern is formed of sets of parallel elongated ridges and troughs approximately 10 cm relief and 3 cm spacing. Patterns reach sizes of up to 3 m – 7 m, and penetrate to depths of greater than 1 m. The maximum depth of individual patterns could not be determined because of incomplete core recovery. To the west of the major fracture that cross-cuts the perennial block (xx in Figure 2), patterns are not greater than 50 cm – 50 cm. East of this major fracture, individual patterns are usually greater than 100 cm – 50 cm. Patterns are variably orientated with inter-secting and cross-cutting relationships (Fig. 5) and, since they are not uniform, it is unlikely that ablational factors such as sun or wind, or any unidirectional force, are responsible for their formation.
Technique
An area of 10 m – 15 m with large patterns was selected for study (Fig. 4). Additional ice samples from the moat and from the zone of small patterns west of the major fracture (xx in Figure 2) were obtained. Thin sections of ice were prepared and examined on a Rigsby stage in the field as a guide for sampling. The results presented here are from blocks returned frozen for crystallographic analysis in the School of Earth Sciences, University of Melbourne.
Ice Microfabric
Table I summarizes the relationships between c-axes and the samples analysed. The microscopic observations (Fig. 5a and b) indicate that individual patterns are composed of large grains that have horizontal dimensional orientations which coincide with that of the surface pattern. Individual grains are tabular in shape and are elongated perpendicular to the ice surface. The same picture can be discerned for other patterned areas. These large grains (long dimension ≥ 2cm) have c-axes that lie perpendicular to the direction of grain elongation. Intergranular and smaller grains (≤ 1 cm diameter) located between the larger grains (Fig. 5) are variably orientated with c-axes lying within a sub-horizontal plane. The c-axes of these small grains cannot be related to the surface patterns.
Ice surfaces devoid of patterns and lying between patterned areas have either (i) c-axes variably orientated in a horizontal plane that parallels the ice surface (Fig. 6, block 6) or (ii) vertically orientated c-axes (Fig. 6, block 7).
Grain-sizes in regions of intersecting patterns are in general smaller than in areas of large single patterns. The plot of c-axes in Figure 6 is inconclusive due to the abundance of randomly orientated but horizontal small grains generally less than 1 cm in diameter. However, the distribution diagram also presented in Figure 5b suggests that two specific orientation groups exist. Each dominant orientation is perpendicular to the elongation of one of the intersecting patterns. The relationship suggests competitive growth of patterns has occurred.
The areas of ice with small patterning are generally composed of between one and four grains (Fig. 5c). As a consequence, the overall stereoplot in Figure 6 shows no pattern of preferred orientation. In fact, grains are aligned in small groups as shown in Figure 5c with c-axes perpen-dicular to the elongation of the patterns.
Moat ice forms more rapidly than the perennial ice and consequently grain-sizes are smaller (Fig. 5d). The c-axes of most grains are horizontal but no marked pattern of preferred orientation has developed. This is clearly shown in the distribution diagram in Figure 5d and in the stereoplot in Figure 6.
Discussion and Conclusions
In many perennially frozen lakes (Reference MichelMichel and Ramseier, 1971) and in most sea ice formed by unidirectional freezing, c-axis horizontal orientations develop rapidly after an initial ice skin forms, and they dominate the rest of the ice growth (Reference WeeksWeeks and Ackley, 1982). The fast-growth direction of ice crystals, parallel to the basal plane, may be controlled by the temperature gradient (Reference WeeksWeeks and Ackley, 1982). In a unidirectional temperature gradient, grains with the basal plane parallel to the temperature gradient can then grow more rapidly than their neighbours. In Patterned Lake, the temperature gradient will be vertical between the underlying stored liquid water and the external Antarctic air. Hence, a c-axis horizontal position is favoured. In the perennial ice, individual patterns are composed of elongate tabular grains orientated perpendicular to the surface with c-axes showing a preferred orientation in the horizontal plane. This is consistent with the anisotropic growth of individual ice grains parallel to their basal planes (Reference HobbsHobbs, 1974), and with grain boundaries meeting the ice–air or ice–water surfaces at a high angle in order to minimize the grain-boundary interfacial energy (Reference SmithSmith, 1964). In fact, the growth would be very similar to the geometric selection process described in sea ice by Weeks and Ackley (1982). Grains that are not favourably orientated with respect to adjacent grains do not grow as rapidly, or may be eliminated through a process of thermal annealing and migration of boundaries associated with an annealing effect during the following summer. Each summer, partial internal melting and ablation of the perennial ice block occurs preferentially along the aligned grain boundaries, leaving a framework of ice below the rigid surface. The framework of partially melted ice probably contains most of the larger crystals from the previous years. Towards the end of summer, crystal growth begins and is controlled by this pre-existing fabric through a process of epitaxial growth. Given that such seasonal processes are operating, it is not hard to see how a large-scale crystallographic orientation develops and how larger grains are able to grow over a long period of time.
Small patterning, as found to the east of the major fracture, probably represents an intermediate stage between that of the moat ice and the perennial ice. This part of Patterned Lake is adjacent to the north-west corner where most water runs into the lake during summer, and it is likely that more extensive internal and surface melting of the ice block occurs. As a result, large patterns have not developed. There is a greater abundance of small grains of uniform distribution of c-axes in a horizontal plane in ice with small patterns.
Acknowledgements
The Australian Antarctic Division is thanked for logistic support and the supply of cold–room facilities used for analysis of the ice samples. J.L.C.C. acknowledges a University of Melbourne summer studentship awarded by the Science Faculty, which enabled him to undertake this work. I. Allison is thanked for his discussion of our field observations. D.A.A. acknowledges financial support from Australian Research Grants Scheme and Macquarie University. J. Pickard’s contributions to the study of perennially frozen lakes near Mawson and his help with drilling on Patterned Lake in 1979 were invaluable.