Performance of the CMIP5 models in simulating the Arctic Oscillation during boreal spring

This study evaluates the performance of thirty coupled models from the Coupled Model Intercomparison Project Phase 5 (CMIP5) in capturing the Arctic Oscillation (AO) during boreal spring. We focus on model's ability in simulating spring AO's spatial structure, dominant temporal frequencies and climatic impacts. The anomalous anticyclone over the North Pacific and the North Atlantic in the positive spring AO phase is much stronger in most of the models than in the observations. This may be due to that most models simulate a stronger interannual variability of atmospheric anomalies over the North Pacific and North Atlantic. In addition, the center of the anomalous anticyclone over the North Pacific in the MME shifts obviously southeastward and that over the North Atlantic shift northeastward compared to the observations. Inter-model longitudinal spread is larger than the latitudinal spread in the anomalous anticyclone center over both the North Pacific and North Atlantic. Most of the models produce periods longer than the observed 4-year spectral peak in spring AO index. In addition, the model's ability in reproducing the vertical structure of zonal wind tends to have a close relation with the model's performance in capturing the vertical structure of air temperature related to the spring AO. Furthermore, there exist larger spreads among the models in simulating the spring AO-related surface air temperature over the Russian Far East and the west coast of North America. These spreads were partly related to the biases of the models in capturing the atmospheric circulation anomalies related to the Pacific center of the spring AO.