How predictable is the winter extremely cold days over temperate East Asia?

Skillful seasonal prediction of the number of extremely cold day (NECD) has considerable benefits for climate risk management and economic planning. Yet, predictability of NECD associated with East Asia winter monsoon remains largely unexplored. The present work estimates the NECD predictability in temperate East Asia (TEA, 30°–50°N, 110°–140°E) where the current dynamical models exhibit limited prediction skill. We show that about 50 % of the total variance of the NECD in TEA region is likely predictable, which is estimated by using a physics-based empirical (P-E) model with three consequential autumn predictors, i.e., developing El Niño/La Niña, Eurasian Arctic Ocean temperature anomalies, and geopotential height anomalies over northern and eastern Asia. We find that the barotropic geopotential height anomaly over Asia can persist from autumn to winter, thereby serving as a predictor for winter NECD. Further analysis reveals that the sources of the NECD predictability and the physical basis for prediction of NECD are essentially the same as those for prediction of winter mean temperature over the same region. This finding implies that forecasting seasonal mean temperature can provide useful information for prediction of extreme cold events. Interpretation of the lead–lag linkages between the three predictors and the predictand is provided for stimulating further studies.