A hybrid dynamical-statistical scheme to enhance summer precipitation prediction skill over eastern China one year ahead

Abstract

Accurate prediction of summer precipitation is vital for disaster prevention in eastern China. However, skillful forecasts one year ahead remain a significant challenge. This study evaluates the predictability of summer precipitation over eastern China one year ahead in the Japanese SINTEX-F2 model. Evaluations indicate that the SINTEX-F2 model lacks skills in directly predicting the dominant modes and interannual variations of summer precipitation. Nevertheless, the model demonstrates skills in predicting some key drivers of summer precipitation over eastern China: sea surface temperature anomalies over the Maritime Continent and the tropical Atlantic, and an anomalous cyclone over southern Maritime Continent. We therefore establish a statistical relationship between these drivers and summer precipitation in eastern China in observations. By applying the SINTEX-F2 predicted drivers to the observed statistical relationship and subsequently using a cumulative distribution function matching correction, a hybrid dynamical-statistical prediction scheme is developed. This scheme overall improves forecast skills for precipitation interannual increment over eastern China, notably increasing the anomaly correlation coefficient from approximately 0.15 to 0.45 over Central-North China. It also effectively predicts the amplitude of the observed interannual variability. Additionally, benefiting from the improved prediction of interannual increment, the summer precipitation prediction skill also increases over South China, the northern part of North China, and the northwestern part of Northeast China. These findings highlight a scalable hybrid prediction scheme that improves the long-term predictability of eastern China summer precipitation. This methodology holds significant potential for enhancing early warning systems and informing climate adaptation planning.