Representation of Arctic Moist Intrusions in CMIP5 Models and Implications for Winter Climate Biases

Abstract

Abstract This paper examines the wintertime northward moisture flux at 70°N from 1981–2005 in 31 of the CMIP5 models compared with the ERA-Interim reanalysis product. The models’ total zonally integrated northward moisture flux is found to agree reasonably well with the reanalysis, but with large compensating regional biases. Specifically, the models systematically underpredict the moisture flux in the Atlantic sector and overpredict it in the Pacific sector. The biases are predominantly due to misrepresentation of extreme moisture flux events, which are known to exert a significant control on Arctic climate. Biases in these high-intensity fluxes are almost entirely contributed by biases in the meridional velocity, suggesting a link with biases in storm-track activity at lower latitudes. The extent to which the deficit of moisture intrusions in the Atlantic sector and excess in the Pacific sector may account for biases in the climate of the respective sectors is assessed. Biases in the frequency of moisture intrusions explain roughly 17% of surface temperature and 24% of surface downward longwave radiation biases in the Atlantic sector, and about 14% and 16% of the gradient in these respective biases between the two sectors. The predicted bias gradients, while small in amplitude, are very highly correlated with the true bias gradients in the models, suggesting that the temperature bias directly induced by misrepresented intrusion statistics may be strongly amplified by sea ice feedback.