A Statistical Prediction for East Asian Winter Monsoon Based on Sea-ice-air System

Abstract

East Asian winter monsoon (EAWM) is one of the most crucial circulation systems in the Northern Hemisphere during winter, significantly influencing the weather and climate of East Asia. Therefore, predicting EAWM variations is considered as a key issue in winter climate prediction. The EAWM intensity index, as defined by Liu Shi (ISA) has shown a strong and consistent correlation with the interannual and interdecadal variations of winter temperature in Northeast China. However, the precursors influencing the EAWM (ISA) changed significantly with the decadal shift of the EAWM in the late 1990s. Predictions of EAWM have become less effective, and it is necessary to identify new predictors. Therefore, correlation analysis is conducted to identify the key factors influencing ISA based on the sea-ice-air system using reanalysis data produced by the National Centers for Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR), as well as optimum interpolation SST V2 data from the National Oceanic and Atmospheric Administration (NOAA). EAWM precursor factors are established and their possible interactions are discussed. Factors are used to construct a statistical prediction model using multiple linear regression method, which is evaluated through cross-validation. Results reveal a significant positive correlation between ISA and the horseshoe-shaped sea surface temperature (SST) pattern over the tropical Pacific autumn, as well as SST over the Gulf Stream and the Eurasian mid-high latitude circulation pattern in stratosphere. ISA shows a stronger and more consistent negative correlation with the sea ice concentration of the Barents Sea than that of the Kara Sea and Laptev Sea. These precursors influence ISA through land/sea thermal differences, winter atmospheric circulation patterns such as the East Asian trough, Ural blocking, and the East Asian subtropical westerly jet. The aforementioned prediction model demostrates a good fit and can be utilized to predict EAWM intensity under the current interdecadal background, with a consistency in the anomaly sign rate of 81.8% (9/11) during 11-year hindcast from 2012 to 2022. An analysis of two years of prediction failures reveals that the winter Arctic Oscillation (AO) forecasts, as well as the abrupt transition of the AO from autumn to winter, should be considered in the EAWM prediction process.

Keywords

• statistical prediction
• external forcing factors
• stratosphere
• interdecadal background