Evaluation And Error Correction Of Subseasonal Summer Precipitation Hindcast Over Eastern China In ECMWF S2S Database

In this study,the European Centre for Medium-Range Weather Forecasts(ECMWF)subseasonal-to-seasonal(S2S)model hindcast is systematically assessed by focusing on the hindcast quality for the summer subseasonal accumulated ten-day precipitation over eastern China,and the evaluation metrics such as the Pearson correlation coefficient,root mean square error and false alarm ratio are used.Additionally,the hindcast error is corrected by utilizing the preceding sea surface temperature(SST),and singular value decomposition(SVD)also uses to correct hindcast error.The main conclusion are as follows:(1)The performance of the ECMWF,Beijing Climate Center(BCC)and Japan Meteorological Agency(JMA)S2S models drops as the lead time increases,and the forecasting quality of ECMWF model is best.In addition,the quality of ECMWF exhibits strong interannual and regional differences,the precipitation forecast skill of the ECMWF hindcast is best at approximately 15 days in some areas of Southeast China(The useful forecast skill of the model is considered as the temporal correlation coefficient exceeds the 95%confidence level).(2)Based on the relationship between the previous SST and the root mean square error or average error of the precipitation,error correction methods were established.After correction,the forecasting performance of ECMWF is significantly improved in the Huaihe River Basin,and the forecasting skill of the model enhances 5 days approximately in some areas of Southeast China.Comparing the performance of the ECMWF model to predict the different thresholds' precipitation events with and without error correction,it can be found that the model has a preferable performance at forecasting accumulated ten-day precipitation rates of approximately 20?50mm/10 d and the performance of the model for forecasting different precipitation events has been improved after correction.The error correction method using the relationship between the root mean square error of precipitation and preceding SST is more effective.However,the performance of correction methods is regional and interannual differences,and the reasons are that the relationships between the precipitation error and preceding SST are different in different correction area and the preceding SST patterns are not well considered when the error correction methods establish.(3)Based on the SVD decomposition of the whole China or subregional forecast and observation precipitation field,the error correction methods are established.After correction,the forecasting skill of the ECWMF model is enhanced 5 days in some areas of Southeast China and improved 10 days in some areas of Huaihe River Basin and North China.Additionally,the false alarm ratio of the model for forecasting precipitation events with50mm/10 d threshold is decreased,the probability of detection of the model for forecasting precipitation events with 20mm/10 d threshold is increased.It is evident that the numbers of SVD model with best correction result are different regardless of which variable(the lead time or subregion)is changed.