ECMWF集合预报产品在重庆暴雨预报中的检验与应用

利用重庆地区34个国家气象站降水资料和ECMWF集合预报降水资料,系统检验和评估了集合预报统计量产品及后处理技术产品对2014—2016年5—9月重庆暴雨的预报性能。结果表明:集合统计量产品中最大值、90%分位数、融合产品、概率匹配平均、75%分位数对暴雨预报有一定参考性,其中90%分位数和融合产品对暴雨落区预报较好,最大值对暴雨强度预报有一定指示意义,但表现为明显的湿偏差。集合预报后处理技术产品的暴雨TS评分较控制预报和集合平均有明显提高,其中概率预报、最优百分位、融合—概率匹配、频率匹配法的暴雨TS评分超过最大值,对暴雨强度预报具有较好的指导意义,其预报偏差均表现为湿偏差,融合—概率匹配和频率匹配法对暴雨落区预报较好,概率匹配—融合对降低暴雨空报率较好。

Verification and application on forecasting heavy rainfall over Chongqing region by using ECMWF ensemble forecast products

The precipitation data of 34 Chongqing national weather stations and ECMWF ensemble forecast precipitation data were used to test and evaluate the forecast performance of ensemble forecast statistic products and post-process technique products on forecasting heavy rainfall over Chongqing region from May to September of 2014-2016.Resultsshow that the maximum product, 90% quantile product, fusion product, probability matching average product and 75% quantile product have certain reference for heavy rainfall forecast. The 90% quantile product and fusion product are better for rainstorm area forecast. The maximum product is better for rainstorm intensity forecast, but it shows obvious wet deviation. The threat score of ensemble forecast post-process technology products are significantly higher than control forecast and ensemble average product. The threat score of probabilistic forecast, optimal percentile, fusion-probability matching and frequency matching method exceed it of the maximum product. They are better for rainstorm intensity forecast. The fusion-probability matching and frequency matching method are better for rainstorm area forecast. The probability matching-fusion is better to reduce the false alarm ratio.