Antarctic surface temperature and sea ice biases in coupled climate models linked with cloud and land surface properties

Abstract:

This study investigates the magnitude, structure and potential causes of biases in summertime surface air temperature (SAT) and related variables across 29 coupled models over Antarctica and the Southern Ocean. These models, included in the Coupled Model Intercomparison Project phase 5 (CMIP5), exhibit local climatological biases of up to 10°C during December and January. In the 50°S-90°S domain, the spread of SAT across models is highly correlated with absorbed solar radiation (ASR). Over the ice-free ocean domain of 40°S-60°S, models have a spread in ASR of about 50 W/m² that arises from cloud properties. On the Antarctic Ice Sheet (AIS), ASR is strongly correlated with surface albedo. Inter-model biases in the open ocean, sea ice and land domains are largely uncorrelated, but similar biases within the open ocean and land domains occur among coupled models that share atmospheric and land component models, respectively. ASR is proposed as a diagnostic that captures model skill in simulating summertime surface climate across all three domains. ASR differences dominate the spread in net surface radiation and net surface energy flux (including sensible and latent heat). CMIP5 models’ net radiation and upward heat fluxes are predominantly too large over the ocean. This work will also illustrate how the radiation budget explains the large spread in sea ice extent among CMIP5 models.