Modelling the sensitivity of ice loss to calving front retreat rates in the Amundsen Sea Embayment, West Antarctica

Abstract. Ice sheet modelling studies of the Amundsen Sea Embayment (ASE) in West Antarctica have provided estimates of its future impacts on sea level rise. However, many of these studies have not considered the impacts of calving, a key process in the dynamics of marine-terminating glaciers. Sensitivity to calving front retreat is not well understood, so we set out to investigate it in systematic manner. In this study, we quantify the sensitivity of modelled future mass loss to ice front retreat in the ASE, including Pine Island and Thwaites Glaciers. We find that prescribing constant frontal retreat rates from 0.1 to 1 km a−1 progressively increases the contribution to sea level rise when compared to experiments with a fixed ice front. The result with our highest rate of retreat is up to 21.4 mm additional sea level contribution by 2100, and 239 mm by 2300. We identify specific buttressing thresholds where loss of contact with bedrock features causes changes in the ice dynamics. These are reached at different times depending on the retreat rate, and are the main cause of sensitivity to movement of the ice front. We compare the variability in our results using different retreat rates to that when using ocean melt and surface mass balance (SMB) forcing derived from different earth system models for ISMIP6, as these climate forcings are major factors in determining the future evolution of the Antarctic ice sheet. We find that the variability due to these two factors is similar. We also find that the additional loss of ice due to a prescribed retreat rate is not heavily dependent on mass balance forcing, so can be quantified independently of the ocean-induced melt and SMB. Our results demonstrate the importance of accurately representing calving processes in models, showing that they can be as important as climate forcing and therefore deserve a similar amount of attention in future model development work.