Predictability of the two temperature modes of the East Asian winter monsoon in the NCEP-CFSv2 and MRI-CPSv2 models.

Using hindcast and forecast data from advanced prediction systems of NCEP CFSv2 and JMA/MRI CPSv2 for the winter 1982/1983–2017/2018, this study investigates the predictability of the climatology and dominant modes of winter-mean surface air temperature (SAT) over East Asia. Although the simulated climatological mean SAT has a large bias over most of the mainland China, both models show high forecasting skills for the principal components (PC1 and PC2) of the two major modes (northern and southern temperature modes) of the winter-mean SAT over East Asia 1 month in advance. In comparison, the MRI-CPSv2 performs better than the CFSv2 in simulating the spatial distribution of the northern temperature mode and the associated circulation anomalies aloft, but the CFSv2 has better performance than the MRI-CPSv2 in simulating the spatial distribution of southern temperature mode and its linkage to the Arctic Oscillation and Eurasian pattern. Furthermore, both models well simulate the impacts of the dipole sea surface temperature (SST) anomalies over the mid-latitude North Pacific and El Niño–Southern Oscillation on the northern and southern temperature modes, respectively. Accordingly, the forecasting skill and signal-to-noise ratio of the PC1 (PC2) are significantly improved in the years with strong SST anomalies over the mid-latitude North Pacific (tropical central-eastern Pacific), particularly for the CFSv2 (MRI-CPSv2) predictions. These results are beneficial for understanding the interannual predictability of the East Asian winter climate. [ABSTRACT FROM AUTHOR]