基于最优TS评分的多模式集成降水预报  

Multi-model ensemble precipitation forecast based on optimal threat score

作　　者：肖丹 胡超 李力根 XIAO Dan;HU Chao;LI Ligen(Meishan Meteorological Bureau,Sichuan Meishan 620010,China;Heavy Rain and Drought-Flood Disasters of Plateau and Basin Key Laboratory of Sichuan Province,Chengdu 610072,China;Hongya Meteorological Bureau,Sichuan Hongya 620300,China)

机构地区：[1]眉山市气象局,四川眉山620010 [2]高原与盆地暴雨旱涝灾害四川省重点实验室,成都610072 [3]洪雅县气象局,四川洪雅620300

出　　处：《气象研究与应用》2024年第3期37-43,共7页Journal of Meteorological Research and Application

基　　金：眉山市科技指导计划项目(2023KJZD167);高原与盆地暴雨旱涝灾害四川省重点实验室科技发展基金项目(SCQXKJYJXMS202205);川西南(雅安)暴雨实验室研究型业务重点专项(CXNBYSYSYWZD202401);眉山“天府粮仓”气象为农服务创新团队项目。

摘　　要：利用2020年1月1日至2023年1月2日欧洲中心、美国、德国、中国和日本降水预报资料和眉山市的逐日降水观测资料,用最优TS评分法对模式降水进行订正,再使用概率匹配平均、消除偏差集成平均和加权集成平均三种方案进行多模式集成预报,通过检验评分和个例分析。结果显示,概率匹配平均方案提升晴雨准确率,但对大雨和暴雨的预报效果差。消除偏差集成平均方案对降水预报的提升效果较小。加权集成平均方案和分类站点集成预报均能大幅提升晴雨准确率,并减小暴雨漏报,使暴雨TS评分得到较大提升。Using precipitation forecast data from European Centre for Medium-Range Weather Forecasts,National Centers for Environmental Prediction,German meteorological service,China Meteorological Administration,Japan Meteorological Agency,and the daily observed precipitation in Meishan from January 1,2020 to January 2,2023,the model precipitation is first revised using the optimal TS score method,and then the multi-model ensemble forecasts are carried out using probability-matching mean,bias-removed ensemble mean and weighted ensemble mean schemes,whose effects are experimented and compared.The results show that the probability-matching mean scheme improves the accuracy of clear-rainy forecast,but poorly forecasts heavy rain and rainstorm.The bias-removed ensemble mean scheme has a smaller improvement for rain prediction.Both the weighted ensemble mean and classified station-ensemble schemes improve the accuracy of clear-rainy forecast greatly,and reduce the omission rate of rainstorm,resulting in a larger improvement in the TS score of rainstorm.

关 键 词：最优TS评分 多模式集成 概率匹配平均 加权集成平均 分类站点集成 

分 类 号：P457.6[天文地球—大气科学及气象学]