Introduction

That part of the Antarctic which drains into the Amery Ice Shelf has long been of interest to glaciologists because of its size. It forms a prominent and distinctive feature of the Antarctic ice sheet and has been identified as the continent’s fourth largest drainage system (Reference Giovinetto and MellorGiovinetto, 1964). Its main tributary is Lambert Glacier, which flows through the rift be-tween the Prince Charles Mountains and the Mawson Escarpment. There, it is over 50 km wide and reaches a thickness of 2500 m (Reference Morgan and BuddMorgan and Budd, 1975). Several attempts have been made to assess the mass balance of the Amery-Lambert drainage system (Reference MellorMellor, 1959, Reference Mellor1964; Reference Giovinetto and MellorGiovinetto, 1964, Reference Giovinetto[1970]; Reference Budd, Budd, Landon–Smith, Wishart and ŌuraBudd and others, 1967; Reference AllisonAllison, 1979) and all have calculated positive mass budgets. Although based on conservative estimates of accumulation rates, a mean surface elevation rise of +0.04 m a⁻¹ in the Amery-Lambert system (Reference AllisonAllison, 1979) has supported the results of modelling the East Antarctic ice sheet (Reference Budd, McInnes, Pittock, Pittock, Frakes, Jenssen, Peterson and ZillmanBudd and McInnes, [^c1978]) and has been taken to indicate that the system may, at present, be undergoing a post-surge build-up.

These mass-balance studies have all been hindered by the scarcity of data for this remote part of the Antarctic, particularly of accumulation rates in the interior of the drainage basin. The purpose of this paper is to re-assess the budget of this part of the drainage basin of Lambert Glacier in the light of new data concerning its area and the accumulation rates within it. The interior drainage basin is defined here as the accumulation area which drains into the Lambert Glacier system delimited by the eleven ice-movement stations reported by Reference AllisonAllison (1979).

Area of the Interior Drainage Basin

Previous studies have been inhibited by the sparseness of surface elevation data with which to delimit the interior drainage basin of Lambert Glacier. Errors have typically been estimated to be 20 to 30% (Reference MellorMellor, 1959, Reference Mellor1964; Reference Giovinetto and MellorGiovinetto, 1964, Reference Giovinetto[1970]; Reference Budd, Budd, Landon–Smith, Wishart and ŌuraBudd and others, 1967). The most recent estimate of 1.09 × 10⁶ km² ± 20% (Reference AllisonAllison, 1979) was based on the American Geographical Society 1 : 5 000 000 map of Antarctica published in 1970. This was modified with data from airborne radio echo-sounding (Reference Morgan and BuddMorgan and Budd, 1975) and oversnow survey (Reference AllisonAllison, 1979) by the Australian National Antarctic Research Expeditions. Surface elevations were also available from traverses by the Japanese Antarctic Research Expeditions which intersected the south-western limits of the drainage basin (Reference Fujiwara, Fujiwara, Kakinuma, Yoshida and MurayamaFujiwara and others, 1971; Reference Shimizu and WatanabeShimizu, 1977). However, there was a complete lack of data in the central parts of the drainage basin.

The present study used the most recent compilation of ice surface elevations in the Antarctic (Reference Drewry and DrewryDrewry, [^c1983]) to re-define the boundaries of the interior drainage basin. This includes altimetry from two sources (radio echo-sounding and constant-density balloons) not available to previous mass-balance studies. The distribution of these data in the interior drainage basin is shown in Figure 1 with the general ice-sheet surface, contoured at an interval of 500 m.

Fig. 1. Delimitation of the interior drainage basin of the Lambert Glacier system (redrawn from Reference Drewry and DrewryDrewry, [^c1983]). The distribution of ice surface elevations from TWERLE (dots) and radio echo-sounding (dashed lines) is also shown.

Surface elevations were available along 5100 km of flight track completed during the joint radio echo-sounding programme of the National Science Foundation, the Scott Polar Research Institute, and the Technical University of Denmark (Reference Kotlyakov, Kotlyakov, Barkov, Loseva and PetrovDrewry and others, 1982). These cover the southern and south-eastern parts of the drainage basin. The method used to calculate elevations, based on barometric altimetry, has been described by Reference Drewry, Drewry, Jordan and JankowskiDrewry and others (1982). Errors in areas away from control points, such as in the interior drainage basin, are estimated to be better than 50 m and, at best 30 m (Reference Drewry and DrewryDrewry, [^c1983]). Some 280 elevations, widely distributed across the drainage basin, are also available from constant-density, super-pressure balloons which opportunely drifted over the Antarctic during the Tropical wind Energy and Reference Level Experiment (TWERLE) in 1975–76. The method and results have been described by Reference LevanonLevanon (1982). Worst-case errors are thought to be of the order of ±60m.

The combination of these two new sources and those (Fig. 1) available to previous workers has enabled a more precise delimitation of the area contributing to the discharge of ice through the network of ice-movement stations reported by Reference AllisonAllison (1979). Flow lines were constructed from a 1 : 6 000 000 map contoured every 100 m (Reference Drewry and DrewryDrewry, [^c1983]) on the principle that, over distances several times the ice thickness, ice flow is normal to the trend of the contours. The resulting drainage basin has an area of 902 000 km². It is difficult to place an error on this figure but, given that the margin of the drainage basin has been located to within ±25 km, the actual area is likely to be between 813 000 and 994 000 km². This represents an error of ±10%. The significance of this re-calculation of the area is that it is 17% less than that defined by Reference AllisonAllison (1979) which was used to estimate the mass balance.

Pattern of Accumulation

Accumulation data within the drainage basin are also sparse, being largely confined to traverse routes along the southern periphery. In some areas, where interpolation between adjacent areas is necessary, interpretations have differed significantly (Reference Bull and QuamBull, 1971; Reference Kotlyakov, Kotlyakov, Barkov, Loseva and PetrovKotlyakov and others, 1974). Reference AllisonAllison (1979) modified the available data to account for the low values recorded at most of the ice-movement stations; this resulted in the 100 kg m⁻²a⁻¹ isopleth passing north of the head of Lambert Glacier. Despite these conservative values for the accumulation rate, there are several sources which suggest that the interior drainage basin is an area of exceptionally low accumulation. This may in part be the cause of the imbalance between accumulation within the basin and the discharge of ice from it (Reference Robin and RobinRobin, 1983).

Figure 2 shows a band 7 multispectral scanner (MSS) Landsat image centred at lat. 76°15’S., long. 68°18’E., 200 km south-east of the outlying nunataks of Mawson Escarpment. As well as substantial surface undulations and crevasse trains due to ice flow over the irregular subglacial Gamburtsev Mountains, there are considerable areas of what is interpreted to be bare ice (darker tones) interspersed with variable amounts of firn. Visual confirmation of this interpretation comes from comparison with Landsat imagery (Fig. 2) of blue Ice in the Queen Fabiola Mountains area (Reference Williams, Williams, Meunier and FerrignoWilliams and others, 1983) and near Allan Hills, Victoria Land, Similar patterns of tonal variations across and between undulation crests occur in all three areas. Further confirmation could be achieved with an analysis of spectral response patterns derived from digital Landsat data.

Fig. 2. Band 7 Landsat MSS image centred at lat. 76°15’S., long. 68°18’E. in the region of the subglacial Gamburtsev Mountains. Patches of bare ice (darker tones) can be seen as well as surface undulations and crevassed areas. Sun angle is 21°. Path 126, row 115. 1479–02223.

Compilation of a mosaic of 16 MSS images, covering most of the drainage basin up to 500 km inland of Lambert Glacier, enabled the area of bare ice to be mapped (Fig, 3). Since the bare ice occurs in isolated patches which form a complex pattern, this was taken as the region in which more than half of the terrain was composed of exposed ice. It was found to cover 56 000 km², that is, over 6% of the area of the interior drainage basin.

Fig. 3. Revised distribution of accumulation rates in the interior drainage basin of the Lambert Glacier system. Area of bare ice is shaded. Contours have been extrapolated between the bare ice and measured values on the basis of brightness temperatures. The 50 kg m^–2a^–1 isopleth of Reference AllisonAllison (1979) is shown (dashed line). Measured accumulation rates are given for the 11 ice-movement stations (Reference AllisonAllison, 1979) and previously published values near the periphery. Outcrops near Lambert Glacier are shown as black areas.

Blue-ice areas have been attributed to ablation by sublimation, wind scouring, and surface polishing by wind-driven snow (Reference Williams, Williams, Meunier and FerrignoWilliams and others, 1983). They are, therefore, most likely to occur in areas of strong katabatic winds. Where mean surface slopes are greater than 0.002 radians, katabatic forces tend to exceed considerably those arising from synoptic pressure gradients (Reference BallBall, 1960). Within the drainage basin of Lambert Glacier, these gradients, calculated over 50 km segments, are exceeded below elevations of 3500 to 3800 m. In the area reported to contain blue ice, which is between 2000 and 3100 m, slopes range from 0.003 to 0.008 radians, with a mean of 0.0045 radians. This is the most steeply sloping part of the interior drainage basin and hence that which is most likely to be subject to the strongest katabatic winds. Although katabatic winds are funnelled down-slope by topographic embayments, Reference ParishParish (1981) reports that they are deflected by 30 to 50° left of the fall line by the Coriolis force. This can be seen in the present case where the main axis of the zone of bare ice lies approximately 30° to the left of the maximum regional slope. That these winds are capable of ablating the ice-sheet surface at a sufficient rate to produce blue ice is suggested by the observation at Mizuho Station (lat. 70°41’53“S., long. 44°19’54”E., 2230 m above sea-level). Here, sublimation (54 kg m⁻²a⁻¹) is significant in reducing the annual accumulation (Reference Fujii and KusunokiFujii and Kusonoki, 1982). Gradients on the slope of the Mizuho Plateau with (Fig. 3) strong katabatic wind are directly comparable with those in the Lambert Glacier drainage basin. It is therefore possible that these very high rates of sublimation also occur in the catchment area of Lambert Glacier and are in part responsible for the exposed blue ice indicated by Landsat imagery.

The notion that the entire interior drainage basin of Lambert Glacier is an area of exceptionally low accumulation is also supported by an analysis of passive microwave radiometry (Reference Zwally and GloersenZwally and Gloersen, 1977). Brightness temperatures of polar firn are closely although complexly related to mean annual temperatures and to accumulation. The anomalously low values found throughout the drainage basin were attributed by Reference Zwally and GloersenZwally and Gloersen (1977) to particularly low accumulation rates, thereby supporting the suggestion that sublimation by katabatic winds across the regionally steep slopes may significantly reduce the mass input to the basin.

On the basis of the above observations, the distribution of accumulation rates was re-contoured as shown in Figure 3. Even though ablation rates of up to 60 kg m⁻²a⁻¹ have been recorded for blue ice areas in the Queen Fabiola Mountains (Reference Yokoyama and IshidaYokoyama, 1978), the present delimitation of bare ice includes inter-vening areas of firn. A mean net accumulation of zero has been used to account for this local variability. Accumulation rates in the rest of the drainage basin have been contoured at an interval of 25 kg m⁻²a⁻¹ using brightness temperatures to provide an empirical extrapolation between the bare ice and the few known values at the periphery of the catchment area. The contouring attempts to show the regional patterns in accumulation and individual values have not been honoured in all cases. Local variability may, in part, be associated with the presence of areas of outcrop which significantly affect the heat balance and wind pattern at the ice-sheet surface (Reference AllisonAllison, 1979).

Re-Calculation of the Mass Budget

Using the distribution of accumulation rates shown in Figure 3, the annual mass input to the interior drainage basin was found to be 32 Gt a⁻¹. This represents a 47% reduction from the figure of 60 Gt a⁻¹ calculated by Reference AllisonAllison (1979). There is only a 6% imbalance between input to the basin and discharge from it as opposed to the previous estimate of 100%, thereby allowing for a balanced mass budget. The mean rate of mass input across the drainage basin is 36 kg m⁻²a⁻¹. Table I gives details of the mass budget which, for comparative purposes, has been extended to include the Lambert Glacier system and the (Table I) Amery-Lambert system. The limits shown are for the worst possible combination of magnitude and sign of the estimated errors, as discussed by Reference AllisonAllison (1979). Other than the much-reduced overall budget for the interior drainage basin (+2 rather than +30 Gt a⁻¹), it should be noted that the lower budget limit falls below zero by a significant amount. The errors for the change in surface elevation (−0.014 to +0.016 m a⁻¹ water) allow for the ice sheet to be in balance or actually thinning; this was not the case in any previous budget calculations.

Table I. Re-Calculation of the Mass Balance of Lambert Glacier on the Basis of the Revised Area and Accumulation Rates Shown in Figures 1 and 3 (Bracketed values are those of Reference AllisonAllison (1979).)

The mass budget for the whole Amery-Lambert drainage system is reduced by 67% to +14 Gt a⁻¹. This represents a mean surface increase of 0.015 m a⁻¹ water. The lower limit indicates that the entire drainage basin may be in balance while the upper limit is significantly less than previous estimates.

Discussion

The immediate conclusion of the downward revision of Lambert Glacier’s mass balance is to question the possibility of the basin being in a post-surge buildup (Reference AllisonAllison, 1979). This contention is supported both by the fact that the overall budget is significantly smaller than previous estimates and that the likely limits for the errors actually fall below zero. Although there is geomorphological evidence that ice levels in the Prince Charles Mountains may have fluctuated by several hundred metres in the recent past (Mellon, 1959; Reference TingeyTingey, 1974; Reference WellmanWellman, 1982) and al-though the results of modelling (Reference Budd, McInnes, Pittock, Pittock, Frakes, Jenssen, Peterson and ZillmanBudd and McInnes, [^c1978]) show that Lambert Glacier is likely to be subject to surging behaviour, the present data do not confirm these suggestions.

Identification of substantial areas of bare ice in the drainage basin of Lambert Glacier must also serve as a warning for the calculation of mass budgets, especially in regions of the Antarctic where data on elevations and accumulation rates are scarce. The possibility of significant local variations in mass balances, as have been found on Mizuho Plateau (Reference NaruseNaruse, 1979), makes the extrapolation of accumulation rates over large distances difficult and uncertain. This is particularly true towards coastal regions where slopes are steeper and katabatic winds stronger. Errors in the estimation of discharge from these zones with high accumulation rates will be proportionately more important to calculations of the output of ice from the ice sheet.

Finally, it is possible that the areas of blue ice identified in the Lambert Glacier basin represent a zone of meteorite concentration. Reference Whillans and CassidyWhillans and Cassidy (1983) and Meier (quoted by Reference Williams, Williams, Meunier and FerrignoWilliams and others, 1983) identified regions with high vertical strain-rates and high ablation rates as the most likely areas for them to collect. It is possible that the presence of the very irregular subglacial Gamburtsev Mountains and the steep surface slopes resulting in strong katabatic winds may offer suitable conditions to warrant a meteorite collection expedition.