中文摘要：

本文运用GRAPES_Meso业务模式（3.1版）通过四个同化地面不同观测要素方案及没有同化任何地面要素的控制试验对2012年7月初发生在四川的一次降水过程做了数值模拟,结果表明：五个模拟方案对不同区域不同量级的降水皆存在不同程度的空报和漏报现象,空报和漏报相对较严重的是控制试验和同化水平风场的同化方案,随着预报时效的增加各方案模拟可信度减小;四个同化方案的同化作用随着预报时效的增加表现出不稳定,相对较稳定的是同化所有地面观测要素的方案;在模式初期和后期,同化作用表现相对明显的是同化所有地面观测要素的方案,只同化水平风场的方案同化作用较小,同化中期各同化方案同化作用皆不稳定;随着预报时效的增加,几个模拟方案之间的差异程度越来越小,但同实况间的差异越来越大。

英文摘要：

Numerical experiments about the precipitation occurred in Sichuan Province in early July 2012 were carried out by using operational mesoscale numerical model GRAPES - MESO （ Version 3.1 ） with four kinds of surface observation data as- similation and the control experiment. The results are as follows： different degree of false precipitation and missing precipitation was discovered among the five numerical simulations of different area and different magnitude of precipitation. Among all these simulations, the severity of false and missing precipitation in the control experiment and the assimilation of horizontal wind were more than the others; the reliebility of each simulation decreased with increasing simulation time; the assimilation impact of four assimilation methods were instable as the time of simulation increased, the method assimilating all of surface observation factors were relatively stable; in the beginning and the end of simulation, the scheme that assimilating all the observation data had the relatively obvious effect, assimilation of horizontal wind had the little effect ; during the middle period of simulation, all the simulation scheme became instable;with the increasing of simulation time, the difference among all of the simulations decreased, but increased between the simulation and the observation.