Evaluation of sub-seasonal prediction skill for an extreme precipitation event in Henan province, China

Abstract

A severe torrential rain attacked Henan province from July 19 to 21, 2021, resulting in extensive social and economic damages. The models’ sub-seasonal prediction skill for this extreme event remains to be evaluated. Based on the real-time data of 5 models (CMA, ECMWF, NCEP, KMA, and UKMO) from the sub-seasonal to seasonal (S2S) prediction project, our study compared the models’ predictability and explored the possible reasons. Results indicate that most models can predict the spatial distribution of accumulated precipitation for this event 1 week in advance. Two models (NCEP and CMA) still have specific reference values in predicting precipitation intensity 2–3 weeks ahead. However, the predicted maximum rainfall is only about 20% of the observation, and all models cannot catch the extremes of this event. While large-scale atmospheric circulation can be predicted with some accuracy, there are still significant deviations in predicting the location and intensity of the western North Pacific subtropical high (WNPSH) and Typhoon In-Fa. The models predict weaker intensity of the southeast airflow transporting water vapor into the rainstorm area, resulting in significantly weaker precipitation. This is mainly attributed to unsatisfactory predicted typhoon circulation in most models. The model ECMWF and KMA predict a better moisture flux at 925hPa, about 60% of the observations. The characteristics of local high SST centers in the Sea of Japan cannot be caught, resulting in the position of the predicted WNPSH eastward and weak. Therefore, to improve the prediction skill for extreme precipitation events, it is imperative to enhance the interaction mechanisms among atmospheric circulation systems within the model.