长江中下游地区梅雨期降水的集合预报试验

ENSEMBLE FORECASTING PRECIPITATION EXPERIMENT IN MID-LOW AREA OF YANGTZE RIVER IN MEIYU PERIOD

本文利用MM5模式,通过改变模式的积云参数化方案、边界层过程和云微物理方案,构造了一个15个成员的集合预报系统,对2001年长江中下游地区梅雨期降水做了集合预报试验。试验表明,集合预报与决定预报相比,可以有效地提高梅雨降水的预报准确度;在不同分辨率条件下集合预报相对于决定预报的提高程度是不同的,分辨率越低,集合预报的价值越大。尽管如此,集合预报还具有的内在缺陷是对小雨高估、大到暴雨低估。试验也揭示了,MM5模式本身对于降水有过分预报的倾向,粗网格模式配置对细网格内降水产生重大影响。

An ensemble forecasting system with 15 members is constructed by changing MM5 model configuration of cumulus parameterization, PBL and explicit moisture schemes. An ensemble forecasting precipitation experiment is done during the 2001 Meiyu period in the mid-low area of Yangtze river by the EF system. Experiment result shows that ensemble forecasting may enhance forecasting precipitation accuracy compared to EF system＇s member deterministic forecasting on the whole . But, with different model resolution,the enhancement is different. EF shows higher enhancement when model resolution is low. EF has intrinsic weakness which will overestimate precipitation on low precipitation threshold and underestimate precipitation on high precipitation threshold. Experiment also shows that MM5 has the tendency of overforecasting precipitation and model configuration with coarse grid domain will have an important influence on precipitation with nest domain.