New Perspectives on Observed and Simulated Antarctic Sea Ice Extent Trends Using Optimal Fingerprinting Techniques

Using optimal fingerprinting techniques, we performed a detection analysis to determine whether observed trends in Southern Ocean sea ice extent since 1979 are outside the expected range of natural variability. Consistent with previous studies, we find that for the seasons of maximum sea ice cover (i.e. winter and early spring), the observed trends are not outside the range of natural variability, and in some west Antarctic sectors may be partially due to tropical variability. However, when information about the spatial pattern of trends is included in the analysis, the summer and autumn trends fall outside the range of internal variability. The detectable signal is dominated by strong and opposing trends in the Ross Sea and the Amundsen/Bellingshausen Sea regions. In contrast to the observed pattern, an ensemble of 20 CMIP5 coupled climate models shows that a decrease in Ross Sea ice cover would be expected in response to external forcings. The simulated decreases in the Ross, Bellingshausen and Amundsen Seas for the autumn season are significantly different from unforced internal variability at the 95% confidence level. Unlike earlier work, we formally show that the simulated sea-ice response to external forcing is different from both the observed trends and simulated internal variability, and conclude that in general the CMIP5 models do not adequately represent the forced response of the Antarctic climate system.