Predictability and applicability evaluation of winter temperatures in China based on Eurasian Arctic sea ice concentrations in autumn

Abstract

The current trends of a warming Arctic and a reduction in Arctic sea ice lead to remote effects on climate anomalies in mid-latitude regions. We investigate the predictability of winter temperatures in China using an empirical regression model to forecast those temperatures based on the Eurasian Arctic sea ice concentration (SIC) in autumn and explore the applicability of this prediction method. Result shows the September Eurasian SIC achieves a highly skilled seasonal prediction of winter temperature anomalies in China. A cross-validated hindcast for the leading principal component of winter temperatures in China using the September SIC within the region (40–150° E, 65–85° N) yields a correlation skill of 0.47 from 1979 to 2018. This suggests that 22% of winter temperature variance in China can be predicted by the Eurasian SIC two months in advance. Winter temperature hindcast/forecast results indicate that September SIC demonstrates a positive temporal anomaly correlation coefficient at most stations in China, with a spatial average reaching 0.32/0.29. However, its forecasting ability for the magnitude of temperature anomalies is relatively weak. Lower tropical Pacific Ocean temperatures, a weak polar vortex, and a strong Ural blocking (UB) in autumn could improve Eurasian SIC's predictive performance.