Sir,

I have read with interest the letter by Reference Allison, Allison, Young and MidhurstAllison and others (1985) regarding my interpretation of tonal variations in Landsat multispectral scanner (MSS) imagery over the Antarctic ice sheet.

Concerning their comments on accumulation rates in Lambert Glacier’s interior drainage basin, I should like to make two observations which support my contention of a lower value for net mass input. First, Allison and others claim that the probable strong katabatic winds of the area in question “can redistribute snow but do not result in significant net ablation at the very low temperatures of these inland areas”. This statement contradicts measurements made at Mizuho Station (Fujii and Kusunoki, 1982), where surface lowering by sublimation of approximately 54 kg m⁻² a⁻¹ was recorded by two independent methods. These measurements were made across a variety of surface types including hard glaze as discussed by Allison and others. The mean annual surface temperature at Mizuho Station is approximately –33°C (Fujii, 1979) and compares with an interpolated value of –40°C in that part of the Lambert Glacier drainage basin under discussion (McIntyre, in press). I must therefore conclude that significant net ablation could well occur, even towards the ice-sheet interior, and hence result in a reduction in estimated mean accumulation rates from 55 kg m⁻² a⁻¹ (Reference AllisonAllison, 1979) to 36 kg m⁻² a⁻¹ (McIntyre, 1985).

Secondly, the complex relationship between emissivity and structure of snow and ice surfaces is such that brightness temperatures could, indeed, be affected by differences in surface character, as pointed out by Allison and others. The measurements made by Zwally and Gloersen (1977), however, demonstrated unexpectedly low values across the entire interior drainage basin of Lambert Glacier and hence extend far beyond the area of blue ice or glazed surface under discussion. Some other basin-wide parameter, such as anomalously low accumulation rates, may offer an alternative interpretation.

Allison and others rightly observe that incorrect interpretation of remotely sensed data can occur without sufficient ground truth. This may be particularly true for sources such as Land sat photographic products. The relative brightness values, or grey scales, of different images may not be comparable, because they are uncorrected for differences in maximum and minimum measurable radiances; these vary both between Land sat satellites and for different calibrations of the same satellite (Dowdeswell and McIntyre, in press).

I would therefore suggest that the present uncertainty as to the cause of tonal variations seen in Land sat data over the Antarctic ice sheet could be resolved by converting digital MSS data into physical values in the form of radiances (Robinove, 1982). This may enable definition of spectral response patterns for different snow and ice types and hence automatic identification and absolute intercom-parison within and between Land sat scenes. Although this procedure is not a necessary precursor to the valuable mapping programmes being carried out with Land sat data (for instance, Swithinbank and Lucchitta, in press), it would provide the basis for the absolute comparison of spectral signatures and the quantitative identification of ice-sheet surface characteristics. Of course. such investigations cannot operate entirely in isolation and I look forward to the results of the field work proposed by Reference Allison, Allison, Young and MidhurstAllison and others for 1988.