基于T639细网格模式的陕西省秋淋天气预报效果评估  

Verification of precipitation forecasting for autumn continuous rainfall by T639 refined grid model in Shaanxi province

作　　者：潘留杰[1,2] 张宏芳 袁媛 王建鹏 薛荣 陈法敬[5] 

机构地区：[1]中国气象局成都高原气象研究所开放实验室,四川成都610072 [2]陕西省气象台,陕西西安710044 [3]陕西省气象服务中心,陕西西安710014 [4]西安市气象台,陕西西安710016 [5]中国气象局数值预报中心,北京100081

出　　处：《气象与环境学报》2015年第6期9-17,共9页Journal of Meteorology and Environment

基　　金：中国气象局预报员专项(CMAYBY2014-070);中国气象局成都高原气象研究所开放基金(LPM2014006);陕西省预报员专项(2015Y-10)共同资助

摘　　要：利用CMORPH卫星与自动气象观测站的逐时降水量融合资料,采用经验正交函数EOF、技巧评分及针对细网格模式的面向对象诊断检验方法 MODE,评估T639细网格模式对2014年陕西省秋淋天气降水预报的效果。结果表明:T639模式整体较好的表现了2014年陕西省秋淋天气过程降水的空间分布,但预报的降水量级显著偏小;逐12 h降水EOF分析表明,此次降水过程可分为系统性和对流性降水两种模态,T639模式对系统性降水的预报性能较好,空间模态与观测一致,但对局地性和对流性降水的预报性能相对较弱;模式整体预报能力较好,晴雨预报TS评分和ACC相关系数平均达0.6以上;空间和时间上预报降水频次偏多,是影响TS评分和ACC等晴雨评分技巧下降的主要原因。MODE检验表明,T639模式对大面积降水对象预报偏多,对局地性降水对象预报偏少;对中等强度降水对象偏多,对强降水及弱降水对象预报偏少;预报对象与观测对象的整体匹配性较好,但两者的面积存在一定差异。Using fusion data of hourly rainfall from automatic stations and CMORPH （ NOAA Climate Prediction Center Morphing Method） satellite data, the EOF analysis, TS and ACC skill scores, and MODE （Method for Object-Based Diagnostic Evaluation） technology were used to verify the forecasting performance of T639 refined grid model for autumn continual rainfall of 2014 in Shaanxi province. The results show that spatial distribution of pre- cipitation could be described well by the T639 model, while the intensity of precipitation forecast is obviously weak. According to EOF analysis of 12 hours precipitation, the process is divided into two modes, i. e. large scale precipitation and convective precipitation. The T639 refined grid model has excellent performance on large scale precipitation and its spatial mode is consistent with the observation, while its forecasting score for convective and local precipitation is lower. In general, the model has a very good forecasting ability, and averages of TS score and ACC correlation coefficient both reach above 0. 6. Too many frequency precipitation predictions are the main rea- sons that affect the skill score. The MODE test shows that the T639 model forecasts more great area precipitation objects, while it doesn＇t forecast enough local precipitation enough. Similarly, the model forecasts more moderate intensity threshold rainfall objects, while it doesn＇t forecast enough high and low threshold precipitation enough. The objects of forecast and observation have good matching ability, while there still exists differences in forecasting areas.

关 键 词：细网格模式 秋淋 EOF MODE 

分 类 号：P456.7[天文地球—大气科学及气象学]