Comparative Analysis of Forecast Evaluation for Rain Band Position in an Extreme Meiyu Rainfall Event

From June 30 to July 4, 2016, an extreme precipitation event occurred in Yangtze-River valley, which is the strongest rainfall process since the flood season. Obvious errors appeared in predicting the location of the rain band in operational forecast. Based on the deterministic and ensemble model data obtained from the European Centre for Medium-Range Weather Forecasting (ECMWF) and the National Centers for Environmental Prediction Global Forecast System (NCEP-GFS), the difference of the model performance between them is compared, and the causes of forecast error are analyzed. The results show that the forecast of NCEP is more accurate and that there are obvious northward errors in the rain band location forecasted by the ECMWF deterministic model. Further analysis from ensemble data reveals that despite the smaller forecast error of NCEP, which is caused by the obvious instability of its errors, it appears alternately northerly and southerly error from day to day. Although the ECMWF model predicts the rain band with further north location, this northerly error is stable. In addition, by dividing the ensemble members into accurate and bias groups, composite analysis between them reveals that the member with further north rain band location will have stronger precipitation and vice versa. Finally, the connection between the westerly trough on 500-hPa level and the position of rain band is discussed in this paper. The results show that when the strength of the westerly trough is stronger, the rain belt is located further north, and the intensity of rainfall is more obvious and vice visa. The ECMWF model forecasted that the westerly trough continued to be strong. Owing to this reason, this model gives a stable northerly position of the rain belt forecast. Because the NCEP model predicts stronger and weaker westerly troughs alternately, its forecast of the rain belt location shows corresponding northerly and southerly errors. This conclusion can serve as an important reference for the forecast errors correction of the rain belt location in the forecast operation and can also help to improve the forecast accuracy of rain band location in Meiyu season.