Abstract:  Based on observed data from meteorological stations, gridded data and forecast data of intelligent grid, SWCWARMS (southwest center WRF ADAS realtime modeling system) and ECMWF, inspection evaluation was performed for the area rainfall forecast effect from June to October in 2019 in the Dadu River basin by using mean absolute deviation, fuzzy grading, accuracy and threat score. The inspection results of mean absolute deviation, accuracy and fuzzy grading indicated that the forecast effect of intelligent grid forecast was overall superior to the others. With the rising level of precipitation, the threat score and false alarm rate decreased, the miss alarm rate increased, and the forecast ability reduced. The ECMWF model was better in forecast of light and middle rainfall, and the intelligent grid model was better in forecast of heavy rainfall and rainstorm. The rainfall levels of forecast results from three models were bigger than observed value for light and middle rainfall, and were smaller than observed value for heavy rainfall and rainstorm. The area rainfall forecast results for typical rainfall process from each model showed that the result of SWCWARMS model was bigger than observation, and the results of intelligent grid model and ECMWF model were bigger than observation for light and middle rainfall but smaller than observation for heavy rainfall.

Key words: Key words: the Dadu River basin, area rainfall forecast, numerical model, verification of forecast

摘要： 基于站点观测资料、格点实况资料和智能网格、西南区域中心业务运行的中尺度模式系统（southwest center WRF ADAS realtime modeling system, SWCWARMS）及欧洲中期天气预报中心（ECMWF）模式资料，以面雨量为研究对象，采用平均绝对误差、模糊评分、正确率、TS评分、偏差分析等，对2019年6—10月大渡河流域面雨量预报效果进行检验评估。结果表明：预报平均绝对误差、预报正确率及模糊评分检验显示，智能网格的预报效果总体上优于其他模式。随着面雨量等级的增大，TS评分逐渐降低，空报率逐渐减小，漏报率逐渐增大，模式的预报能力逐渐降低。ECMWF模式在小雨和中雨面雨量预报中优势明显，智能网格在大雨和暴雨等级面雨量预报中表现较优。3个模式在小雨和中雨等级面雨量预报中预报的等级偏大，在大雨和暴雨等级面雨量预报中预报的等级偏小。各模式对典型降水过程面雨量预报结果表明，SWCWARMS模式对面雨量的预报等级均偏大，而智能网格和ECMWF模式对小雨和中雨的预报等级偏大，对大雨预报等级偏小。

关键词: 大渡河流域, 面雨量预报, 数值模式, 预报检验