The Good, the Bad and the Ugly: Modeled Versus Observed Regional Trends in Late 20th Century Antarctic Sea Ice

Abstract:

The total Antarctic sea ice extent has increased over the satellite record. This slight net increase results from the sum of much larger opposing regional trends. Although many atmospheric and oceanic processes have been proposed as drivers for these differing regional trends (e.g. atmospheric temperature or wind stress changes due to stratospheric ozone loss or increases in the Southern Annular mode; ocean circulation responses to land-based ice sheet melting; or ocean-atmosphere interactions and feedbacks), explanations remain uncertain.

Global climate models have been unable to explain observed increases in Antarctic Sea Ice extent. In contrast to the observations, nearly all of the Coupled Model Intercomparison Project phase 5 (CMIP5) simulations show decreasing trends in Antarctic sea ice extent over the late 20^th century. Analysis of regional ice area trends reveals that discrepancies between observed and modeled trends in ice extent are largest and most consistent (between models) in the Ross Sea area during austral fall – a region and season of large observed increases in sea ice. A consideration of pre-industrial control simulations and large ensembles of 20^th century runs suggests that this discrepancy cannot be explained by internal variability. We use output from the CMIP5 models as well as several variations of the Community Earth System Model (CESM) to explore possible reasons for this consistently large discrepancy between observed and modeled Antarctic Sea Ice extent trends.