Abstract:Based on the data from the CMA-CPSv3 and ECMWF-S2S models, a comprehensive assessment of their predictive skills regarding the climatological state of summer subseasonal precipitation, pentad-by-pentad precipitation, precipitation during the key period of the Meiyu season, and precipitation processes in Anhui Province was conducted using multiple scoring methods. Additionally, their predictive capabilities for key circulation systems were analyzed. The results revealed that the climatological state of the summer daily precipitation rate was underestimated by both models, while the frequency of light rain was significantly overestimated, and the frequency of no-rain conditions was significantly underestimated. The predictive skills of the models for precipitation in the 1st to 4th pentads were found to decrease with an increase in the lead time, after which little change was observed. When compared, it was found that the predictive skills of the CMA model for precipitation in the 4th to 5th pentads were generally lower than those of the EC model, with little difference in predictive skills between the two models thereafter. The assessment results for precipitation prediction during the key period of the Meiyu season were similar, and it was found that multi-time ensemble averaging could improve predictive skills to a certain extent. Fifteen to twenty-five days in advance, the Threat Score (TS) for predicting rain/no-rain conditions and significant precipitation events was found to be lower for the CMA model than for the EC model, with little difference observed thereafter. The overall lower predictive skills of the CMA model for precipitation in the 4th to 5th pentads may be attributed to deficiencies in its prediction of key circulation systems such as the East Asian-Pacific (EAP) pattern and the subtropical high.