Evaluation and correction of sub-seasonal dynamic model forecast of precipitation in eastern China

Assessing the capability of sub-seasonal rainfall forecast of dynamic model and proposing correction method is quite an important topic in current climate research field. From the perspective of rainfall amount, rainy days and rainfall-belt evolution, the sub-seasonal forecast ability of the European Centre for Medium-Range Weather Forecasts (ECMWF) model for the main rainy-season rainfall in eastern China is evaluated. Evaluation results show that the forecast biases increase gradually with the increase of forecast lead time, characterized by the predicted rainfall amount being obviously higher and the rainy days being much longer than observation. In order to reduce the forecast biases of sub-seasonal rainfall forecast of ECMWF model, the rainy day-based correction (RDC) method is proposed in this study. Cross validation results indicate that RDC method can modify the number of rainy days forecast of ECMWF model with the SCC of rainy days increasing by 12.96% ~ 18.62%, and the RMSE decreasing by 56.49% ~ 63.78%. The problem of maximum continuous rainy days being too long in the model forecast can be also improved. Meanwhile, the spatial correlation coefficient (SCC) of rainfall amount forecast of the ECMWF model with the observation weakly increases by 0.61% ~ 1.56% and the root mean square error (RMSE) decreases by 3.50% ~ 7.60% after the RDC treatment. Therefore, RDC method presents a good performance on improving the sub-seasonal forecast of rainy days, and maximum continuous rainy days, which can be further applied in other models’ sub-seasonal forecast error correction.