Accuracy Analysis of ensemble precipitation forecast data based on TIGGE

The accuracy evaluation of precipitation forecast is the key to the application of numerical precipitation forecast in practice. Taking the Tianzhuang reservoir watershed as the research object, three forecasting models of NCEP, CMA and UKMO from TIGGE center are selected, and the accuracy of the ensemble precipitation forecast data from May to September in 2007 to 2019 is analyzed by using test methods of precipitation classification forecast, Brier score and Talagrand map. The results show that: (1) the ensemble average precipitation forecast effects of no rain, light rain and moderate rain for the three forecasting models are almost the same, while those of heavy rain and rainstorm and above are quite different, among which the prediction effect of CMA is the best; (2) the Brier scores decrease with the increase of the precipitation grade and the forecast accuracy of CMA is the highest; (3) the Talagrand maps of the three forecasting models all show a rough “U” shape, indicating that the members of each forecasting model are not scattered enough. Generally speaking, the ensemble precipitation forecast of CMA has the highest accuracy.