三种数值模式对山西降水预报性能的对比检验

Comparing and testing of three kinds of numerical model for the precipitation in Shanxi

利用MM5、T213和Grapes3种数值模式的降水预报产品和山西省108个标准测站的降水实况资料，采用客观统计检验方法，对2008年7月各模式在山西省的累加降水预报进行了对比检验。结果表明：24h中雨以下预报1、213优于MM5，中雨以上MM5则略优于T213，48h预报各级降水MM5都优于T213，T213和MM5对暴雨都有一定的预报能力。无论哪个预报时效和降水量级，Grapes均无明显优势。Grapes预报降水量级和降水范围都偏小，空报较少，漏报严重，尤其48h和72h10mm以上降水基本都漏报。MM5预报降水量级和预报范围都偏大，10mm以上降水TS评分较其它模式高，但同时空报也比较严重。3种模式TS评分均随降水量级的增大而减小，T213和Grapes的TS评分随预报时效的增加而减小，MM5的TS评分随预报时效的增加变化不大。

Using the precipitation forecast products of MM5, T213 and Grapes numerical model and the actual precipitation data from 108 standard weather stations in Shanxi, by the objective statistical verification methods, the cumulative precipitation of three models in July 2008 in Shanxi Province were compared and verified. The results showed that： 24h T213 is superior to moderate rain forecast following MM5; moderate rain for more than a little better than MM5 is T213; d8h precipitation forecast MM5 at all levels are better than the T213; T213 and MM5 have certain heavy rain forecasting capability. Grapes has no obvious advantage in forecasting time efficiency and precipitation forecast level. It is lower than actual value that Grapes forecast precipitation level and scope. Its false forecasting is less but its dismissal forecasting is serious and it is almost dismissal especially for more than 48h and 72h in more than 10mm precipitation forecasting. It is upper than actual value that MM5 forecast precipitation level and scope. MM5＇ s TS score of precipitation more than 10mm is better than the others but at same time its false forecasting is more serious than the others. The TS score of three models decrease with increasing precipitation level. TS score of T213 and Grapes decrease with increasing the forecasting time efficiency. TS score of MM5 has no change with increasing the forecasting time efficiency.