Evaluating Southern Ocean sea-ice for the Last Glacial Maximum and pre-industrial climates: PMIP-2 models and data evidence

Sea ice is recognized as an important factor controlling climate conditions at high latitudes. It has a major impact in controlling the ocean to atmosphere exchanges (e.g. for CO2) as well as surface oceanic conditions and deep water formation in the high latitude oceans. Being at the interface of two major components of the climate system, sea ice is useful to evaluate and intercompare different climate models. We here analyze the results of eight coupled climate models (from which seven are from the PMIP-2 database) against available reconstructions for winter and summer sea ice extent around Antarctica. To evaluate the ability of models in representing different climate states, we compare their results for both the pre-industrial and the Last Glacial Maximum (LGM). We find that models generally have identical seasonal ranges for the two climate states analyzed, which is not supported by data inferences. Indeed, data indicate that the LGM seasonality was enhanced compared to the pre-industrial. The shape of the sea ice distribution is also different in models (circular around the Antarctic continent) and in reconstructions (oval-shaped distribution; similar to modern sea ice distribution). We show that models underestimating the sea ice extent for the pre-industrial are also underestimating the Last Glacial Maximum extent. Our results thus point to difficulties for climate models in representing the sea-ice dynamics in the Southern Ocean.

► PMIP-2 simulations compared to Southern Ocean sea-ice data for present-day and LGM.
► LGM winter spatial shape is not reproduced by climate models in general.
► Simulated sea-ice seasonality is similar for present-day and Last Glacial Maximum.
► Current climate models have difficulties representing Southern Ocean sea-ice dynamics.