Short-range ensemble forecasting of quantitative precipitation

In order to study the performance differences of different ensemble forecast products for temperature and precipitation forecasts in Chongqing, based on the Chongqing Convective-scale Ensemble Prediction System (CQCEPS) which has been operationally implemented, a comprehensive comparison and analysis are carried out for the differences as well as temporal and spatial distribution characteristics in the forecast performance of ensemble forecast application products of surface elements such as the control forecast of 24 h cumulative precipitation and 2 m temperature, ensemble mean forecast and ensemble quantile forecast for the whole year of 2020, as well as the probability matched mean forecast of 24 h cumulative precipitation. The results show that (1) the precipitation prediction performance of each ensemble forecast product decreases with the increase of predicting lead time. When predicting lead time is consistent, ensemble mean forecast, probability matched mean forecast, and 60% and higher quantile forecasts are better than control forecast. The 90% quantile forecast is the best among quantile forecast products. (2) In summer and autumn, the forecast results of each product are quite different, and the 90% quantile forecast, probability matched mean forecast and ensemble mean forecast are better than others. (3) The SEEPS scores of control forecast, ensemble mean forecast, probability matched mean forecast and the 90% quantile forecast show higher accuracy in eastern Sichuan and poorer accuracy in the southeast and northeast of Chongqing, indicating that the prediction performance may have a certain relationship with the terrain. (4) For temperature forecasts with different lead times, overall, the ensemble mean forecast performs the best among all of the prediction products, and the 70% quantile forecast performs the best among all quantile products.