Introduction

The Ross Sea is one of two major embayments around the edge of Antarctica (Fig. 1). As with the Weddell Sea, it represents a major incursion of the sea into the lower-elevation West Antarctica, but is bounded on the west by the higher mountains of Victoria Land on the edge of the Polar Plateau. These are part of the Transantarctic Mountains which transect the continent.

Fig. 1. Physical features of the Ross Sea area.

The Ross Ice shelf provides an abrupt coastline in the form of an ice cliff to the south, while the eastern side is less confined than that to the west. As with the Weddell Sea, the Ross Sea consists of relatively cold water which experiences a cyclonic circulation. It is not surprising that these two areas also experience the coldest coastal temperatures of the Antarctic (Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983).

In spite of these similarities, there are distinct differences in the sea-ice regime of the two areas. The polynya which forms in the eastern Weddell Sea is quite distinctive, as is the major area of pack ice which persists through the summer along its western side (Reference AckleyAckley, 1981). The reasons for these general differences in sea-ice concentrations are not yet known, as the mechanisms of the seasonal expansion and decay of Antarctic sea ice are only poorly understood (Reference BarryBarry, 1983). The major aim of this paper is therefore to examine the sea-ice regime of the Ross Sea using ESMR (Electrically Scanning Microwave Radiometer) data, and by reviewing recent literature, to develop hypotheses about the principal mechanisms involved. Such hypotheses are developed on the basis of observed seasonal sea-ice variations and the small amount of supplementary data available, as well as the results of previous research. The supplementary data provide only a superficial view of atmospheric and oceanic conditions. Previous research on sea ice is also limited, with ideas of ice drift dependent largely on the track of Aurora which was trapped in the ice around the turn ofthecentury (Reference WordieWordie,1921; Reference AckleyAckley, 1981). It is hoped, however, that the hypotheses developed here can be tested over the next few years as the Ross Sea AWS (Automatic Weather Station) network expands (personal communication from C. Stearns) and field data collection on sea ice increases.

Sea-Ice Processes and Data

Antarctic sea ice contrasts markedly with that of the Arctic largely because it is not confined by continental land masses but is totally surrounded by ocean (Reference AckleyAckley, 1981; Reference Andreas and AckleyAndreas and Ackley, 1982). The Antarctic sea ice experiences a much greater annual variation in extent (Reference Ackley and KeliherAckley and Keliher, 1976) which has meant an annual turn-over of ice. As a result, almost all of the pack is “first-year” ice (Reference AckleyAckley, 1981). Although most of the research on ice processes has concentrated on the Weddell Sea, it seems that the same general ideas of ice type and development also apply to the Ross Sea. In particular, advection (due to winds and currents) plays an important part in the growth of sea ice, particularly from March to June (Reference AckleyAckley, 1981). However, Reference CarletonCarleton (1981) has suggested that movement and distribution of the Ross Sea pack ice is influenced more by cyclonic weather systems, while the Weddell Sea pack ice is dominated by the oceanic gyre. In a limited study, Reference DeRyckeDe Rycke (1973) suggested that the general movement off the Getz Ice Shelf (Fig. 1) is due to ocean currents while air flow causes short-term variability.

Expansion of the pack ice occurs through the combined effect of low temperatures and offshore movement of ice (whether by air flow or ocean currents). Ice formed near the coast is advected northward leaving more open water to be frozen (Reference AckleyAckley, 1981; Reference Hibler and AckleyHibler and Ackley, 1983). Further growth of the pack ice may also be due to the continual development of open water by Ekman divergence of surface water and sea ice due to wind stress (Reference Gordon and TaylorGordon and Taylor, 1975). This open water within the pack ice freezes and the total ice area expands. In spring and summer the open water acts in absorbing solar radiation which subsequently assists in melting the ice. The outer ice edge, however, is determined by the balance between supply of sea ice by advection and melting due to higher air or sea temperatures. Consequently, where the ice edge appears stationary over several weeks, the melt rate is equal to sea-ice advection.

Clearly, both these dynamic and thermal influences on the sea-ice regime are closely related to synoptic scale atmospheric circulation. Local air flow is also influenced by the strong katabatic outflow from the Antarctic continent (Reference ParishParish, 1982) but this too is modified by synoptic disturbances. The impact of strong katabatic outflow from the Polar Plateau in developing the Terra Nova Bay polynya appears to be greater than originally thought possible (Reference Kurtz and BromwichKurtz and Bromwich, 1983). However, this may be due to synoptic reinforcement of the drainage wind. Oceanic circulation also has a two-way relationship with the atmosphere. It has been suggested that ice transport from the Weddell Sea strongly influences mid-latitude temperature and presumably atmospheric circulation in the South Atlantic region (Reference SchwerdtfegerSchwerdtfeger, 1979). The location of cyclogenetic zones also appears to be determined by the sea-ice edge (Reference CarletonCarleton, 1981, Reference Carleton1983). The complex relationship between synoptic circulation, oceanic circulation, katabatic winds, and sea ice is still largely unknown. However, it is clear that the strength and tracking of cyclonic disturbances have a major impact on sea-ice concentrations and movement (Reference Schwerdtfeger and KachelhofferSchwerdtfeger and Kachelhoffer, 1973; Reference CarletonCarleton, 1981, 1983). Consequently, the year-to-year variability of sea-ice cover is largely a function of such synoptic scale characteristics, and these in turn are influenced by global circulation variations. The current lack of knowledge of the relationships between Antarctic sea ice and atmospheric circulation at all scales is a major cause for concern (Reference BuddBudd, 1982). Global climate modelling, for example, relies on an accurate parameterization of these relationships (Reference BennettBennett, 1982).

Study of the sea-ice regime of the Ross Sea requires repetitive observations of the whole area at adequate resolution. These observations should cover the complete annual cycle including the polar night period. Passive microwave data are the most suitable because of their all-weather all-year capabilities. Such data are available for 1973–76 from the ESMR sensor on board the Nimbus 5 satellite, at a resolution of 30 km × 30 km. This radiometer recorded brightness temperatures at the 1.55 cm wavelength (Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983). Although the relationship between sea-ice concentrations and brightness temperatures is complicated by varying surface characteristics of sea ice, there is a roughly linear correspondence. However, the resulting inaccuracies and limited resolution restrict the use of these data to study of regional rather than local ice conditions. The more sophisticated SMMR (Scanning Multichannel Microwave Radiometer) has a five-channel, dual-polarizing sensor system but data sets for 1979 and 1980 are only now becoming available to the research community. The U.S. Navy–NOAA JIC (Joint Ice Centre) sea-ice charts for the Antarctic are based on visual and infra-red data from NOAA and DMSP (Defense Meteorological Satellite Program) satellites as well as the microwave systems. However, as they are largely concerned with delimiting large areas of roughly similar ice concentrations, their coarse resolution is limiting. They also provide winter-time estimates of the area of open water that are too low (Reference Comiso and ZwallyComiso and Zwally, 1982). This seems to be due to the fact that ice-surface temperatures are not taken into account in interpreting ESMR images. Also, open water in leads and polynyi smaller than about 1 km resolution in visual and infra-red images cannot be included accurately in estimates of ice concentration (Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983). However, it can also be said that the digital ESMR data may overestimate the area of open water because of uncertainty as to the microwave emissivity of very thin ice under about 5 cm thick Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and Gloersen(Zwally and others, 1983). Following consideration of these problems, it was decided to base this study on the 1973–76 ESMR data. However, future work will incorporate the SMMR data as they become available.

Oceanographic and Meteorological Processes in the Ross Sea

Knowledge of oceanographic and meteorological processes in the Ross Sea area is limited. This is largely due to the inhospitable conditions for both personnel and equipment. However, in spite of the continued annual movement of people and supplies though the area, the data base is much less than that for the Weddell Sea. This lack of data is clearly evident in the writings of other researchers. The prime example is the use of the track of Aurora, which was trapped in the ice between 1915 and 1916, to indicate ice drift (e.g. Reference AckleyAckley, 1981). Only Reference DeRyckeDe Rycke (1973) has used satellite imagery to determine ice movement and that study was both temporally and spatially limited. Reference PerchalPerchal’s (1975) ice-forecasting guide for the western Ross Sea and McMurdo Sound is similarly tentative in describing surface air flow and currents in the Ross Sea area. Data bases, such as that of Reference Jenne, Jenne, Crutcher, van Loon and TaljaardJenne and others (1974), are based more on educated speculation than concrete observations as far as the Ross Sea area is concerned. Several oceanographie studies have been made but they have been largely concerned with describing and classifying sub-surface water masses using limited period survey data (e.g. Reference Jacobs, Jacobs, Amos and BruchhausenJacobs and others, 1970; Reference Jacobs and HainesJacobs and Haines, 1982). Because of severe environmental conditions, it has not been possible to obtain the appropriate data, particularly in winter. However, the deployment of buoys would provide a useful data source, as would further extension of the AWS network to cover the coastline and offshore islands of the Ross Sea. The network of AWS centred on McMurdo Station since 1979/80 has become an increasingly important data source, as it has extended to other parts of the Ross Sea area (including the Ross Ice Shelf and Franklin Island) (see Fig. 1). The period of observation of available data is still very short. It is hoped that, as both the SMMR and AWS data become available over the next few years, the conclusions of the present study can be further examined.

On a very local scale, study of the Terra Nova Bay polynya has provided some idea of sea-ice processes around the edge of the Ross Sea Reference DeRycke(Kurtz and Bromwich, 1983; Reference Bromwich and KurtzBromwich and Kurtz, 1984) (Fig. 1). Several of their conclusions were based largely on conjecture and require further analysis before they can be confirmed. It appears that katabatic flow is frequently concentrated down glaciated valleys and may be a controlling factor in the development of coastal polynyi.

Seasonal Variations of Sea Ice

As already mentioned, the extent and distribution of sea ice in the Ross Sea is the result of a complex energy balance involving a range of factors including radiation fluxes, air temperature, sea temperature, wind stress, and ocean salinity. These are controlled by broader atmospheric and oceanic circulation features which cause upwelling, ice break-up, and advection. It is interesting, given their broad similarity in physical character and the dominant role of advection in both sea-ice regimes (Reference AckleyAckley, 1981), that the Ross Sea ice distribution varies distinctly from that of the Weddell Sea. The differences are said to be due to the greater role of oceanic circulation in the Weddell Sea and the deflecting effect of the Antarctic Peninsula (Fig. 1), while the Ross Sea is said to be influenced more by cyclonicweathersystems(Reference CarletonCarleton,1981).

The seasonal variation of sea ice is illustrated in Figure 2. The major features are, first, a south-east to north-west axis of higher concentrations, and secondly a north to south zone of lower concentrations extending from the Ross Ice Shelf into and across the axis of higher concentrations. Along this zone, ice decay extends from the south in spring and lower concentrations persist throughout the year. Bearing in mind that sea-ice concentrations in any area are a function of formation, decay, and advection of ice, and that the prevailing winds are south-easterlies, it appears that sea ice moves generally along the south-east to north-west axis. This is supported by earlier ice-drift evidence discussed by Reference AckleyAckley (1981) and satellite observations of Reference DeRyckeDe Rycke (1973). It appears that differing rates of ice formation, advection, and decay produce convergence and divergence zones along this axis. For example, the eastern side of the Ross Sea consistently experiences higher concentrations, presumably receiving new supplies of ice from the east faster than it is removed or melted (i.e. ice convergence). The central and southern Ross Sea has lower ice concentrations at all times, suggesting that this is a zone of ice melting or divergence (Fig. 2). Potential causes are either stronger southerly winds moving the sea ice out of the area or upwelling of warmer or more saline water melting the ice (or a combination of these two) (Reference Jacobs, Jacobs, Amos and BruchhausenJacobs and others, 1970). By examining the succession of maps covering the melt period (October–January), it is apparent that this zone develops northward from the Ross Ice Shelf, suggesting the dominant role of southerly winds (Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983). An area of ice accumulation appears to occur north of Victoria Land near the Balleny Islands (see Figs 1 and 2), which is thought to be an area of convergence (Reference StretenStreten, 1973; Reference AckleyAckley, 1981). The mean air flow, based on Reference Jenne, Jenne, Crutcher, van Loon and TaljaardJenne and others’ (1974) data, indicates strong southerly and south-westerly winds over the western Ross Sea with a marked reduction in velocity between the Ross Sea and the Balleny Islands area.

Fig. 2. Average monthly percentage concentrations of sea ice in the Ross Sea based on ESMR digital data.

More minor features of the sea-ice regime include the small-scale variability of concentrations around the coastline. In interpreting these features, the coarse resolution of the data and geographical complexity of the coast should be considered. The brightness temperatures used to derive concentrations in some cases will be derived from surfaces other than sea ice, including land, ice shelf, and multi-year ice. However, it is apparent from the trend of sea-ice concentrations within a few hundred kilometres of the coast that a lot of these zones of lower concentration along the coast are real and represent leads created by offshore air flow. It is also apparent that some sections of coast experience higher concentrations than others (Fig. 2). These include the coastline at the eastern end of the Ross Ice Shelf front and to the west of Cape Adare. Another area of interest is along the coast in the south-east corner of the study area illustrated in Figure 1. Here, there is a clear seasonal variation in the sea-ice concentrations, with very high concentrations in winter(April–October) but with development of two distinct polynyi in summer. These latter are probably due to localized katabatic outflow. A similar change occurs off the Ross Ice Shelf north of Roosevelt Island(see Fig. 1), where a polynya develops from November to December before merging with the more extensive open-sea area in January and February (Fig. 2). This feature is also probably due to the channelling of strong katabatic flow from the Ross Ice Shelf. Ice concentrations along the western coast of the Ross Sea show much local variation, including the McMurdo Sound and Terra Nova Bay polynyi (Reference StonehouseStonehouse, 1967; Reference Bromwich and KurtzKurtz and Bromwich, 1983; Reference Kurtz and BromwichBromwich and Kurtz,1984).The Terra Nova Bay polynya shows up quite clearly through out the year, although the coast immediately to the north experiences persistently higher ice concentrations (Fig. 2). A band of higher concentrations also exists 100–200 km offshore and parallel to the coast during the winter months May–October (Fig. 2). It is likely that these features are strongly influenced by the interaction of regional katabatic and synoptic air flow. The convergence of south-easterly synoptic flow with the westerly katabatic winds along the coast could well contribute to a sea-ice accumulation zone offshore. As suggested by Reference Kurtz and BromwichKurtz and Bromwich (1983), cyclonic weather systems tracking through the Ross Sea area appear on occasions to enhance the strong katabatic flow so that offshore movement of sea ice is greater than that expected from katabatic flow alone. However, the modelling experiments by Reference ParishParish (1982) suggest that katabatic outflow would be concentrated locally due to channelling by the topography and particularly down Reeves Glacier in Terra Nova Bay. This explains the variability observed along the coastline.

Another minor feature which provides some insight into regional sea-ice processes is the irregular advance of the ice edge between summer and winter. The monthly average sequence of maps from February to June shows the development of a bulge in the ice edge which grows north eastward over this period (Fig. 2). It would seem that this development relates first to the temperature gradient but, secondly, to the cyclonic movement of ice in the Ross Sea. Figure 3 provides a generalized view of sea-ice movement in the Ross Sea on the basis of the information and ideas discussed in this paper. The monthly development of the bulge in the sea-ice front is indicated by the open circles. This generalized view would, of course, vary seasonally in response to dynamic and thermodynamic factors.

Fig. 3. Generalized pattern of sea-ice movement in the Ross Sea. The open circles represent the most northerly point on the ice edge in the months given.

The sea-ice circulation pattern of the Ross Sea differs from that of the Weddell Sea (Reference AckleyAckley, 1981) largely due to the Antarctic Peninsula, which helps to trap sea ice on its eastern side, as well as causing it to move more strongly to the north-east. In contrast, at about lat. 70°S. there is a region of divergence at the entrance to the Ross Sea (Fig. 3). The polynya of the eastern Weddell Sea is not duplicated in the Ross Sea, probably due to oceanographic differences between the two. This polynya is thought to be due to upwelling and subsequent basal melting (Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983, p. 121). It is also apparent from the persistent ice cover in the eastern Ross Sea that sea ice continues to be produced further to the east during summer. This contrasts markedly with the Weddell Sea as there is no significant source of sea ice for almost 80° of longitude to the east in summer (see Reference Zwally, Zwally, Cosimo, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983, p. 75). This presumably relates to different thermal conditions of both air and sea in these source areas.

Summary and Conclusion

This is the first study to deal specifically and in detail with the sea-ice regime of the Ross Sea. Until recently, there has been too little information to allow such a paper. Even now, it is only possible to provide a general overview to be used as the basis for future research. It is clear that much more analysis is required both of satellite-derived data and information collected in the field.

The information so far available provides a good indication of the nature of the sea-ice regime and the influence of dynamic and thermodynamic processes. After the Weddell Sea, this area experiences the greatest seasonal change in sea-ice concentrations in the Antarctic and is therefore of major climatic significance. In spite of some physical similarities, there are major differences in the distribution of sea ice in the two major embayments. Regional air flow seems to be the major control of the Ross Sea regime. Offshore winds play an important role in the formation and movement of ice through the Ross Sea, except north of about lat. 68° to 70°S., where the strong mid-latitude westerly winds become increasingly important. Along the coast, marked variability of sea-ice concentrations is due to topographic channelling of katabatic winds from the interior of the continent. Glaciated valleys are particularly important in generating strong outflow at the coast which helps to move sea ice offshore. This offshore flow also appears to have a more general effect in providing a zone of lower ice concentrations around the coastline. In particular, offshore flow along the west coast and southerlies off the Ross Ice Shelf appear to produce a convergence zone further offshore in the western Ross Sea parallel to the coast.

A zone of lower ice concentrations trending north to south in the central Ross Sea is due to the strong southerly air flow, causing greater ice advection here. The eastern Ross Sea and near the Balleny Islands experience higher ice concentrations probably due to slower ice movement, convergence, and accumulation in these areas.

It is apparent that both dynamic and thermodynamic processes are important in the sea-ice regime. It seems that oceanographie factors are of lesser importance than atmospheric ones in determining ice distribution and concentration in the Ross Sea. However, Reference Jacobs, Jacobs, Amos and BruchhausenJacobs and others (1970) suggested that upwelling of circum-polar deep water in the southern Ross Sea assists in the first opening up of pack ice in spring.

It is clear that, although suitable sea-ice data are becoming increasingly available, further understanding of sea-ice processes in the Ross Sea requires much more atmospheric and oceanic data. The forcing fields to be measured include ocean currents and temperatures, wind, and air temperatures, as well as sea-ice characteristics (both structural and radiational). These measurements would require a co-ordinated programme involving ocean buoys, automatic weather stations, and field-data collection supported by satellite systems. This would provide information on sea-ice formation, movement, and decay, as well as the forcing functions responsible. It is hoped that such a programme could be executed in the next few years.