Recent work suggests that natural variability has played a significant role in the increase of Antarctic sea ice extent during 1979-2013. The ice extent has responded strongly to atmospheric circulation changes, including a deepened Amundsen Sea Low (ASL), which in part has been driven by tropica... Show moreRecent work suggests that natural variability has played a significant role in the increase of Antarctic sea ice extent during 1979-2013. The ice extent has responded strongly to atmospheric circulation changes, including a deepened Amundsen Sea Low (ASL), which in part has been driven by tropical variability. Nonetheless, this increase has occurred in the context of externally forced climate change, and it has been difficult to reconcile observed and modeled Antarctic sea ice trends. To understand observed-model disparities, this work defines the internally driven and radiatively forced patterns of Antarctic sea ice change and exposes potential model biases using results from two sets of historical experiments of a coupled climate model compared with observations. One ensemble is constrained only by external factors such as greenhouse gases and stratospheric ozone, while the other explicitly accounts for the influence of tropical variability by specifying observed SST anomalies in the eastern tropical Pacific. The latter experiment reproduces the deepening of the ASL, which drives an increase in regional ice extent due to enhanced ice motion and sea surface cooling. However, the overall sea ice trend in every ensemble member of both experiments is characterized by ice loss and is dominated by the forced pattern, as given by the ensemble-mean of the first experiment. This pervasive ice loss is associated with a strong warming of the ocean mixed layer, suggesting that the ocean model does not locally store or export anomalous heat efficiently enough to maintain a surface environment conducive to sea ice expansion. The pervasive upper-ocean warming, not seen in observations, likely reflects ocean mean-state biases. Show less
