Polar climate simulation in FGOALS_g1.1. Polar cli mate simulation in FGOALS-g1.1.

The polar climate simulations (with emphasis on sea ice) in the latest version o f LASG climate system model (FGOALS-g1.1) were evaluated using a variety of observational data. The results show that FGOALS-g1.1 does a reasonable job i n simulating primary characteristics of the Arctic and Antarctic sea ice, includ ing spatial distribution, seasonal variation, and decadal trend. However, some d iscrepancies are noteworthy. (1) FGOALS-g1.1 produces more (less) total seaice area in the Arctic (Antarctic) as compared to the observations. In the Arctic, there is excessive (insufficient) ice cover in north Atlantic all year long, an d in east Siberia Sea (Beaufort Sea) in summer. In the Antarctic, there is insuf ficient ice cover in the Weddell and Ross Seas in winter. Extremely large area w ith small ice concentration is found far beyond the sea ice edge zone in both th e Arctic and Antarctic, particularly in summer. (2) The simulated sea ice velo city is systematically too large as compared to the observations, mainly near th e sea ice edge in both the Arctic and Antarctic as well as near Antarctic coasta l regions. In the Arctic, the model can not capture the Beaufort Gyre, and produ ce unrealistic ice motion and thickness distribution around the North Pole due t o the impropriate way to handle the North Pole. Biases in sea ice simulations id entified here are closely associated with how well the simulations are in atmosp here and ocean components of FGOALS-g1.1. Further analyses show that weak Icel andic Low and Antarctic Circumpolar Westerly simulated in FGOALS-g1.1 partly c ontribute to the aforementioned biases in sea ice simulations through influencin g atmospheric and oceanic poleward heat transport. Additionally, the amplificati on of biases in the subcomponent of FGOALS-g1.1 due to atmospheresea iceoc ean interaction is briefly discussed.