“集合统计量集成法”在湖北短期暴雨预报中的应用研究  

Applications of ensemble statistic-integrating method to short-range torrential rain forecast in Hubei province

作　　者：李武阶[1] 车钦[2] 姜杰 祝赢[2] 陈赛男[2] LI Wujie;CHE Qin;JIANG Jie;ZHU Ying;CHEN Sainan(Institute of Heavy Rain,CMA,Wuhan 430205;Wuhan Central Meteorological Observatory,Wuhan 430074)

机构地区：[1]中国气象局武汉暴雨研究所,武汉430205 [2]武汉中心气象台,武汉430074

出　　处：《暴雨灾害》2018年第5期455-461,共7页Torrential Rain and Disasters

基　　金：湖北省气象局科技基金重点项目(2014Z02;2018Z01)

摘　　要：利用2013年11月—2014年11月ECMWF全球集合预报的51个成员降水预报资料和湖北省78个国家气象站逐日降水实况,对集合平均值、分位值、概率匹配平均值、众数(Mode值)等10种集合统计量在湖北省不同预报分区内的降水预报效果分别进行检验评估。在此基础上,采用在不同降水量级上选取TS评分最优的集合统计量的原则,设计出适用湖北降水预报的最佳集合统计量集成方案,并检验了该方法在2015年和2016年6—8月湖北降水预报的应用效果。结果表明:将集合统计量集成法应用于湖北降水预报时,集合统计量集成方案应随着预报分区的改变而改变;改进后的ECMWF集合统计量集成方案对湖北72 h内大雨及以上降水预报的TS评分均有不同程度的提高,且空报率和漏报率总体上有所降低;与ECMWF确定性预报相比,ECMWF集合统计量集成预报产品对湖北24 h内各降水量级的预报均优于ECMWF确定性预报,且对湖北72 h内的暴雨预报准确率均高于ECMWF确定性预报。Based on the precipitation forecast data from 51 members of the ECMWF Global Ensemble Prediction from November 2013 to November 2014 and the daily precipitation observation data at 78 national meteorological stations in Hubei province, the skill of precipitation forecast for 10 kinds of ensemble statistic products including ensemble mean, quantiles, probability-matching mean and mode value was evaluated respectively in different prediction regions in Hubei province. Moreover, according to the principle of selecting ensemble statistic which TS score was the highest at different precipitation levels, the best ensemble statistic-integrating scheme suitable for precipitation prediction in Hubei province was designed,and then the application performance of this method was tested in the precipitation forecast of Hubei province from June to August in 2015 and 2016. The results were showed as follows. When the ensemble statistic-integrating method was applied to the precipitation forecast in Hubei province,its scheme should be changed with the change of prediction region.The improved ECMWF ensemble statistic-integrating scheme has improved the TS scores of precipitation forecast for heavy rain and above within 72 hours in Hubei province,and its false rate and fail rate have been reduced overall.Compared to ECMWF deterministic forecast, the precipitation forecast results of ensemble statistic-integrating products are all better than those of deterministic forecast products in Hubei province within 24 hours, and the accuracy of torrential rain forecast within 72 hours in Hubei province is higher.

关 键 词：集合统计量集成法 概率匹配法 暴雨预报 

分 类 号：P458.121.1[天文地球—大气科学及气象学]