Sea ice, as the result of freezing of sea water, can have various textures determined by the prevailing thermal and hydrodynamic conditions during ice formation and subsequent growth (Weeks and Ackley, 1982; Eicken and Lange, 1989). In the Antarctic, flooding of surface snow also often leads to snow-ice formation, giving rise to a meteoric component in the sea tee (Lange and others, 1990; Eicken and others, 1994), which has a granular texture with grains of several millimeters up to some centimeters in size. Snow-ice is believed to form up to 4% of the total sea-ice volume in the Weddell Sea (Eicken and others, 1994) and perhaps more than twice as much in the Ross. Amundsen and Bellingshausen Seas (Jeffries and others, 1994).
Here, we report a novel finding of glacial-ice fragments(up to several centimeters in diameter) in a sea-ice core obtained in the Bellingshausen Sea north of Alexander Island, Antarctica. To our knowledge, glacial ice has not previously been described as an important component of sea ice, although in this core it amounted to 20% of the core volume. The features of the core, including texture, salinity and δ18O measurements, are reported and the relevance of glacial ice fragments as a component of the sea ice in this region of the Antarctic is discussed.
General Ice-Core Properties
Ice coring was part of a larger sea-ice program during an expedition of RV Polarstern to the Bellingshausen and Amundsen Seas in the late austral summer of 1994 (Fig. 1; Miller and Grobe, 1995). The ice core was obtained on 29 January 1994 during an ice station at 68°40.2′S, 72°37.3′W, approximately 11 nautical miles (n.m.) south of the sea-ice edge and 20 n,m. from the coast The ice concentration was 10/10 and the ice floes were mostly 10-100 m in diameter with piled-up brash in between. There was hardly any evidence of ice deformation (see Haas and Viehoff (1994) for a detailed description of ice conditions). The floe sampled was approximately 30 m × 40 m in size and had a mean ice thickness of 3.1 ± 1.1 m covered with 0.51 ± 0.13 m of snow. A thickness profile derived from drilling (Fig. 2) revealed that, although the surface was rather flat the underside of the ice was highly irregular. During drilling, many gaps or voids several centimeters thick were penetrated. These were mainly in the uppermost ice and may be characteristic of summer sea ice in this region (Jeffries and others, 1994). Some larger voids were measured as indicated in Figure 2.
A 2.12 m long ice core (10 cm diameter.) was obtained at position 0 of the thickness profile (Fig. 2). The mean bulk salinity of the core was only 1.52 ± 0.83 psu, the lowest value of the 30 cores drilled during the expedition (Fig. 1) which had a mean salinity of 2.55 psu. The ice was of orbicular granular texture throughout, with a small section of columnar ice between 1.34 and 1.47 m (see Eicken and Lange (1989) for nomenclature). The glacial-ice pieces were distributed mainly within the lower two-thirds of the core (Fig. 3).
Properties of Glacial-Ice Fragments
Image analysis of Figure 3 showed that the fraction of glacial-ice pieces within the total area of the ice thick sections was 20.7%. This can be considered the volume fraction of glacial ice in the whole core. The fragments varied in size, ranging from sub-centimeter to a maximum of 15 cm in diameter, although the majority were less than 5 cm. The edges and corners of the pieces were generally rounded.
Glacier-ice pieces could mainly be distinguished from sea ice by their characteristic pore structure in the ice thick sections (Fig. 4a, c): they had a high density of equally distributed small pores (less than 1 mm in diameter) and in many pieces the pores were elongated and appeared to he aligned linearly along layers of some millimeters width. Generally, no pores surrounded the fragments to indicate the boundary between glacial and sea ice. Fine cracks, some centimeters in length, which crossed each other at various angles were often visible (Fig. 4a, c). Ice thin sections, viewed between crossed polarizers, yielded information about crystal structure (Fig. 4b, d). The texture of the pieces was typical of glacier ice (e.g. interlocking gains) and different from the polygonal granular structure of snow or sea ice (Eicken and Lange, 1989), However, because grain-sizes ranged from some millimeters to some centimeters (Fig. 4d), it was difficult to distinguish between sea ice and iceberg ice from thin-section analysis.
Some of the larger fragments were cut from the ice and the salinity and the ratio of oxygen isotopes 18O/16O measured. Salinities were zero or slightly above zero. δ18O, the difference between the ratio 18O/16O in the sample and in a standard (VSMOW: Vienna Standard Mean Ocean Water), ranged from −14.4 to −10.8‰ (mean -11‰). These values are probably small overestimates, because of a possible contamination with sea salt during sample preparation, indicated by the slightly elevated salinities.
From the properties described above, it is evident that the fragments contained within the sea ice were of glacial origin: salinity and δ18O values, together with shape and pore structure, made it possible to differentiate between the sea ice and the glacial ice. Texture, as determined from thin-section analysis, was the main aid in distinguishing between fragments of snow-ice and glacial ice. This underlines the importance of careful inspection of sea-ice cores to avoid misinterpretation of glacial fragments, since δ18O of snow-ice can be very close to the values for glacial ice measured here (Jeffries and others (1994) measured δ18O as low as −18‰ in the western Bellingshausen Sea). Since glacial-ice fragments could also have sizes in the range of some meters, they could also contribute to lager ice-floe features such as surface roughness being misinterepreted a pressure ridge.
It is most likely that the glacial fragments were incorporated in autumn or winter during sea-ice formation in an area with glacial brash or in the vicinity of icebergs. The glacial fragments were distributed mainly over the lower two-thirds of the core (Fig. 3) down to 2.12 m depth. Since the glacial fragments were bouyant, than must have been turbulent conditions during initial ice formation or later rafting and deformation. Probably bath processes were involved. Ice growth, as in “the pancake-cycle” (Lange and others, 1989), influenced by Strong wave action would explain the orbicular granular texture. Although the surface of the floe was rather flat, the undulations and gaps along the thickness profile indicate that deformational processes acted during the history of the floe [Fig. 2). Additionally, the section of columnar ice at 1.34 m depth may demarcate the former underside of the floe, where congelation of ice commenced after a closed ice cover had formed. Later rafting may have contributed to Further thickening of the floe, including iceberg pieces that were originally incorporated further above.
Crocker (1993) described an inverse exponential relationship in the size distributions of particles calved per calving event from Arctic icebergs. Due to his methodology, the distribution ends at a lower limit of 0.5 m, but he stated that below this limit numbers of pieces-decrease towards zero for very stall fragments;. For the pieces described here, this seems not to be the case, since no larger fragments were found and no growlers were observed along the thickness profile (Fig. 2). We speculate that this iceberg brash had already been in the water for a considerable time, and was close to complete melting. Therefore, the bits found in the core are the last remnants of that brash but, although smaller, they may still resemble the size distribution after calving.
The fact that this is the first recording of glacial-ice fragments within sea ice raises the question of how widespread this phenomenon is and whether or not it is peculiar to the Bellingshausen Sea? During the expedition, a large number of icebergs (mostly less than 1 km in size) were observed, especially around Alexander Island and in Ronne Entrance where more than 100 icebergs were counted in many of the regular 2 h ice observations (Haas and Viehoff, 1994). Within the pack ice they were mainly tabular, while in the open sea they were smaller and of no particular shape. In the open sea, the icebergs were often surrounded by areas (up to several square kilometers) of iceberg brash, including growlers and very small ice pieces less than a few meters in diameter. Thus, the Bellingshausen Sea is evidently a region in which a high degree of iceberg disintegration will occur. This would suggest that glacial ice is likely to contribute to sea-ice thickness over a wider area.
The δ18O data do not allow the determination of the region or ice shelf from which the ice fragments originated, Clearly, it is most likely that they came from the nearby Wordie, George VI, Wilkins or Bach Ice Shelves: This is supported by δ18O values from the Alexander Island region, given by Peel and Clausen (1982), who measured a range with a mean of 15‰, similar to our values, especially considering the probable contamination of our samples with sea water.
Speculating about the consequences of atmospheric warming, It is possible that glacier fragments may become a more important component to overall sea-ice thickness. A scenario similar to the recent rapid disintegration of the Wordie Ice Shelf (Doake and Vaughan, 1991) could also occur within the ice shelves west of Alexander Island. This would result in even higher numbers of icebergs, although at least far the Wilkins ice Shelf, Vaughan and others (1993) have predicted a malting of the ice shelf rather than it breaking off Considering the huge volume of ice contained within the ice shelves and icebergs, glacially derived brash, either floating or incorporated into the sea ice, could even partially compensate for the sea-ice. retreat in: this region observed by Jakobs and Comiso (1993). This could also apply to the Weddell Sea, if the Larsen Ice Shelf, located near the climatic limit of viability, starts to disintegrate. However, general ice-drift patterns may result in a significant difference between the Bellingshausen Sea and the Weddell Sea: calving from, and deterioration of icebergs, mainly results from wave action, and is therefore mostly limited to open water (Wadhams, 1988). For iceberg fragments to occur within sea ice, it is therefore necessary that the sea around the iceberg brash freezes shortly after the event. Little is known about drift patterns in the Bellingshausen Sea but, if they are oriented toward the continent (Talbot, 1988), this ice could be preserved within, the pack-ice region. In the western Weddell Sea, the Weddell Gyre is known to transport icebergs away from the sea-ice zone, so making the incorporation of glacial ice into sea ice a less probable event.
We are grateful to the crew of RV Polarstern for their assistance in the field and to H. Eicken for his help during the preparation of this manuscript. This is publication No. 891 of the Alfred-Wegener-Institut für Polar- und Meeresforschung.