Antarctic sea ice thickness and snow-to-ice conversion from atmospheric reanalysis and passive microwave snow depth

Passive microwave snow depth, ice concentration, and ice motion estimates are combined with snowfall from the European Centre for Medium-Range Weather Forecasting (ECMWF) reanalysis (ERA-40) from 1979-2001 to estimate the prevalence of snow-to-ice conversion (snow-ice formation) on level sea ice in the Antarctic for April-October. Snow ice is ubiquitous in all regions throughout the growth season. Calculated snow-ice thicknesses fall within the range of estimates from ice core analysis for most regions. However, uncertainties in both this analysis and in situ data limit the usefulness of snow depth and snow-ice production to evaluate the accuracy of ERA-40 snowfall. The East Antarctic is an exception, where calculated snow-ice production exceeds observed ice thickness over wide areas, suggesting that ERA-40 precipitation is too high there. Snow-ice thickness variability is strongly controlled not just by snow accumulation rates, but also by ice divergence. Surprisingly, snow-ice production is largely independent of snow depth, indicating that the latter may be a poor indicator of total snow accumulation. Using the presence of snow-ice formation as a proxy indicator for near-zero freeboard, we examine the possibility of estimating level ice thickness from satellite snow depths. A best estimate for the mean level ice thickness in September is 53 cm, comparing well with 51 cm from ship-based observations. The error is estimated to be 10-20 cm, which is similar to the observed interannual and regional variability. Nevertheless, this is comparable to expected errors for ice thickness determined by satellite altimeters. Improvement in satellite snow depth retrievals would benefit both of these methods.