Abstract

A preindustrial control run and an ensemble of twentieth-century integrations of the Community Climate System Model, version 4 (CCSM4), are evaluated for Antarctic sea ice climatology, modes of variability, trends, and covariance with related physical variables such as surface temperature and sea... Show moreA preindustrial control run and an ensemble of twentieth-century integrations of the Community Climate System Model, version 4 (CCSM4), are evaluated for Antarctic sea ice climatology, modes of variability, trends, and covariance with related physical variables such as surface temperature and sea level pressure. Compared to observations, the mean ice cover is too extensive in all months. This is in part related to excessively strong westerly winds over ~50°-60°S, which drive a large equatorward meridional ice transport and enhanced ice growth near the continent and also connected with a cold bias in the Southern Ocean. In spite of these biases in the climatology, the model’s sea ice variability compares well to observations. The leading mode of austral winter sea ice concentration exhibits a dipole structure with anomalies of opposite sign in the Atlantic and Pacific sectors. Both the El Niño-Southern Oscillation and the southern annular mode (SAM) project onto this mode. In twentieth-century integrations, Antarctic sea ice area exhibits significant decreasing annual trends in all six ensemble members from 1950 to 2005, in apparent contrast to observations that suggest a modest ice area increase since 1979. Two ensemble members show insignificant changes when restricted to 1979-2005. The ensemble mean shows a significant increase in the austral summer SAM index over 1960-2005 and 1979-2005 that compares well with the observed SAM trend. However, Antarctic warming and sea ice loss in the model are closely connected to each other and not to the trend in the SAM. Show less