定量降雨模式预报精度比较及评定指标 Study on Comparison and Evaluation Index of Quantitative Rainfall Forecast Accuracy

定量降雨预报精度评定在国际上仍是一项亟待深入的研究工作，至今没有一个客观的统计指标或检验方法能单独地全面描述降水预报的可靠性。为给水文气象耦合洪水预报提供更多的参考信息，在现有降水预报检验方法的基础上，借鉴洪水预报中的精度评定方法，建立了一套检验评估降水预报精度的完整指标体系。选取清江流域为应用示范，采用建立的指标体系比较分析四种气象模式的降雨预报水平，结果表明，人工预报和ECMWF模式的降雨预报效果最好，其次为日本模式，WRF模式则相对较差。 The accuracy evaluation of quantitative rainfall prediction is still an urgent research job in the world. So far, there is no single index or test method can objectively and independently describe the reliability of rainfall forecast results. In order to provide more reference information for the hydrological and meteorological coupling forecast, a complete set of evaluation indexes for precipitation forecast was established by studying with the existing rainfall forecast accuracy assessment methods and the evaluation indexes used in flood prediction. The Qingjiang river basin was chosen as a case study to compare four different meteorological forecast models’ performance in rainfall forecast by using the established indexes. The results showed that the manual forecasting and the ECMWF model have the better effect on rainfall forecasting, followed by the Japanese model, while the WRF model is relatively poor.