【摘要】 利用ECMWF集合预报降水资料和重庆市自动站降水资料,运用晴雨、TS评分、预报偏差等检验方法对重庆地区2014—2016年ECMWF集合预报降水产品在短期时效的预报性能进行检验分析。结果表明:最小值的晴雨预报准确率最高。对于TS评分检验,小雨量级可优先参考最小值、10%分位数和融合产品,中雨量级参考平均数和概率匹配平均,大雨和暴雨量级分别参考75%分位数和90%分位数。对于预报偏差检验,小雨量级可优先参考最小值、Mode,中雨量级参考融合产品、中位数,大雨量级参考控制预报、融合产品,暴雨量级参考90%分位数。对于百分位值预报产品和概率预报产品,小雨量级可参考5%～10%分位数和80%～90%概率预报产品,中雨量级可参考45%～55%分位数和40%概率预报产品,大雨量级可参考70%～80%分位数和20%概率预报产品,暴雨量级可参考90%～95%分位数和10%概率预报产品。更多还原

【Abstract】 Based on ECMWF ensemble forecast precipitation data and Chongqing precipitation observational data, and with clear-rain prediction(PC), Threat Score(TS) and Bias Score(BS)methods, the precipitation forecasting performance in short-time prediction of the ECMWF ensemble precipitation products from 2014 to 2016 in Chongqing area were analyzed. The results showed that the highest PC accuracy appeared in the minimum product. For the TS value, the minimum product,the 10% quantile product and the statistic fusing product could be referred to with priority for light rain, while the mean product and probability matching product for moderate rain, and 75% quantile product for heavy rain, 90% quantile product for rainstorm. For the BS value, the minimum product,and the Mode product could be used as prior reference for light rain, the statistic fusing product and the medium product for moderate rain, while the control forecast product and the statistic fusing product for heavy rain, and the 90% quantile product for rainstorm. For quantile products and probability forecast products, the 5%-10% quantile product, the 80%-90% probability forecast product could all be the reference for light rain, while the 45%-55% quantile products and the 40%probability forecast product for moderate rain, the 70%-80% quantile products and the 20%probability forecast product for heavy rain, and the 90%-95% quantile products and the 10%probability forecast product for rainstorm.更多还原