Interdecadal changes in the Asian winter monsoon variability and its relationship with ENSO and AO

Interdecadal changes in the Asian winter monsoon (AWM) variability are investigated using three surface air temperature datasets for the 55-year period of 1958-2012 from (1) the National Centers for Environmental Prediction-National Center for Atmospheric Research reanalysis 1 (NCEP), (2) combined datasets from the European Centre for Medium-range Weather Forecasts (ECMWF) 40-yr reanalysis and interim data (ERA), and (3) Japanese 55-year reanalysis (JRA). Particular attention has been paid to the first four empirical orthogonal function (EOF) modes of the AWM temperature variability that together account for 64% of the total variance and have been previously identified as predictable modes. The four modes are characterized as follows: the first mode by a southern warming over the Indo-western Pacific Ocean associated with a gradually increasing basin-wide warming trend; the second mode by northern warming with the interdecadal change after the late 1980s; the third and fourth modes by north-south triple pattern, which reveal a phase shift after the late 1970s. The three reanalyses agree well with each other when producing the first three modes, but show large discrepancy in capturing both spatial and temporal characteristics of the fourth mode. It is therefore considered that the first three leading modes are more reliable than the rest higher modes. Considerable interdecadal changes are found mainly in the first two modes. While the first mode shows gradually decreasing variance, the second mode exhibits larger interannual variance during the recent decade. In addition, after the late 1970s, the first mode has a weakening relationship with the El Nio-Southern Oscillation (ENSO) whereas the second mode has strengthening association with the Artic Oscillation (AO). This indicates an increasing role of AO but decreasing role of ENSO on the AWM variability. A better understanding of the interdecadal change in the dominant modes would contribute toward advancing in seasonal prediction and the predictability of the AWM variability.