Arctic sea ice area changes in CMIP3 and CMIP5 climate models’ ensembles

Abstract

The shrinking Arctic sea ice cover observed during the last decades is probably the clearest manifestation of ongoing climate change. While climate models in general reproduce the sea ice retreat in the Arctic during the 20th century and simulate further sea ice area loss during the 21st century in response to anthropogenic forcing, the models suffer from large biases and the results exhibit considerable spread. Here, we compare results from the two last generations of climate models, CMIP3 and CMIP5, with respect to total and regional Arctic sea ice change. Different characteristics of sea ice area (SIA) in March and September have been analysed for the Entire Arctic, Central Arctic and Barents Sea. Further, the sensitivity of SIA to changes in Northern Hemisphere (NH) temperature is investigated and dynamical links between SIA and some atmospheric variability modes are assessed. CMIP3 (SRES A1B) and CMIP5 (RCP8.5) models not only simulate a coherent decline of the Arctic SIA but also depict consistent changes in the SIA seasonal cycle. The spatial patterns of SIC variability improve in CMIP5 ensemble, most noticeably in summer when compared to HadISST1 data. A better simulation of summer SIA in the Entire Arctic by CMIP5 models is accompanied by a slightly increased bias for winter season in comparison to CMIP3 ensemble. SIA in the Barents Sea is strongly overestimated by the majority of CMIP3 and CMIP5 models, and projected SIA changes are characterized by a high uncertainty. Both CMIP ensembles depict a significant link between the SIA and NH temperature changes indicating that a part of inter-ensemble SIA spread comes from different temperature sensitivity to anthropogenic forcing. The results suggest that, in general, a sensitivity of SIA to external forcing is enhanced in CMIP5 models. Arctic SIA interannual variability in the end of the 20th century is on average well simulated by both ensembles. To the end of the 21st century, September variability is strongly reduced in CMIP5 models under RCP8.5 scenario, whereas variability changes in CMIP3 and in both ensembles in March are relatively small. The majority of models in both CMIP ensembles demonstrate an ability to capture a negative correlation of interannual SIA variations in the Barents Sea with North Atlantic Oscillation and sea level pressure gradient in the western Barents Sea opening serving as an index of oceanic inflow to the Sea.