Evaluation of basal melting parameterisations using in situ ocean and melting observations from the Amery Ice Shelf, East Antarctica

Ocean-driven melting of Antarctic ice shelves is causing accelerating loss of grounded ice from the Antarctic continent. However, the ocean processes governing ice shelf melting are not well understood, contributing to uncertainty in projections of Antarctica's contribution to sea level. Here, we analyse oceanographic data and in situ measurements of ice shelf melt collected from an instrumented mooring beneath the centre of the Amery Ice Shelf, East Antarctica. This is the first direct measurement of basal melting from the Amery Ice Shelf and was made through the novel application of an upward-facing acoustic Doppler current profiler (ADCP). ADCP data were also used to map a region of the ice base, revealing a steep topographic feature or "scarp" in the ice with vertical and horizontal scales of similar to 20 and similar to 40 m, respectively. The annually averaged ADCP-derived melt rate of 0.51 +/- 0.18 m yr(-1) is consistent with previous modelling results and glaciological estimates. There is significant seasonal variation around the mean melt rate, with a 40 % increase in melting in May and a 60 % decrease in September. Melting is driven by temperatures similar to 0.2 degrees C above the local freezing point and background and tidal currents, which have typical speeds of 3.0 and 10.0 cm s(-1), respectively. We use the coincident measurements of ice shelf melt and oceanographic forcing to evaluate parameterisations of ice-ocean interactions and find that parameterisations in which there is an explicit dependence of the melt rate on current speed beneath the ice tend to overestimate the local melt rate at AMO6 by between 200 % and 400 %, depending on the choice of drag coefficient. A convective parameterisation in which melting is a function of the slope of the ice base is also evaluated and is shown to underpredict melting by 20 % at this site. By combining these new estimates with available observations from other ice shelves, we show that the commonly used current speed-dependent parameterisation overestimates melting at all but the coldest and most energetic cavity conditions.