The Establishment and Verification of the Operational Ensemble Forecast System for Shanghai Regional Precipitation

The operational ensemble forecast system for Shanghai regional precipitation is established. The model is based on MM5 in the system, with a one way two nesting procedure. The coarse grid domain of 101×101 grid points has a horizontal grid spacing of 45 km, while the inner grid domain of 70×70 grid points has a horizontal grid spacing of 15 km. Both domains have 21 vertical sigma levels. Eight different configurations of MM5 are used to produce an 8 member ensemble forecast. The approach to generating eight members is to use different convective parameterization schemes and different PBL parameterization schemes of the inner grid domain. In this system, all steps from data collection, data processing and model running to output automatically are carried out by computer. The system starts operating on August 1, 2005. The system operates twice everyday and each time it runs to 48 h forecast. The forecasts are then displayed on Website. Since the start of the operation, the system has shown stability and reliability. The forecasts from August to October are verified in the fields of ensemble mean, probabilistic forecast and spread. The results show the ensemble means of precipitations of low thresholds are more skillful than the ensemble means of precipitations of high thresholds. For precipitations of low thresholds, the larger the forecasting probability is, the larger the usefulness of probabilistic forecast is. For precipitations of high thresholds, the smaller the forecasting probability is, the larger the usefulness of probabilistic forecast is. The results also show the system has the problem of too small spread. In conclusion, the performance of the system is alright in forecasting precipitation. The precipitation forecasts of low thresholds seem better than the forecasts of high thresholds. The products of the system, especially the products of probability forecasts can be used as a reference. However, there are still some problems in the system, which should be improved in the future.