An analysis of the late twentieth century and the twenty-first century predictions of Antarctic circulation patterns in a ten model ensemble of global climate system models, using the method of self-organizing maps (SOMs), is presented. The model simulations, and this analysis, were conducted in support of the Intergovernmental Panel on Climate Change (IPCC) fourth assessment report. An application of the methodology to twentieth century reanalyses resulted in conclusions congruent with earlier synoptic climatologies that used different techniques. It was found that the SOM method is a useful tool for diagnosing differences among a large cohort of often quite divergent simulations, allowing the rapid identification of outliers.
Some models do rather poorly at simulating the present day Antarctic circulation, but taking these outliers into account and considering the models as an ensemble, the simulations of twentieth century circulation are reasonable. The most notable outlier was the Centre National de Recherches Météorologiques model (also known as ARPEGE), which was unable to simulate Southern Ocean cyclones. Also of interest was the excessive cyclogenesis displayed by the National Center for Atmospheric Research CCSM3 model.
The trend towards increasing cyclonicity and stronger zonal winds is quite consistent among models, and is also reflected in an increase in positive Antarctic oscillation (AAO) index. The coherence of temperature and precipitation anomaly patterns and their trends reflects the extent to which these are related to circulation. It is clear from this analysis that several of the models in this ensemble are capable of predicting the Antarctic Peninsula warming as the rest of the continent cools—this is an important advance in our simulation capacity. Copyright © 2006 Royal Meteorological Society.