LAPS云分析中卫星资料的应用及对GRAPES模式短时预报的影响

THE APPLICATION OF SATELLITE DATA IN LAPS CLOUD ANALYSIS AND ITS INFLUENCE ON THE NOWCASTING IN GRAPES MODEL

通过2013年5月16日广东暴雨个例,利用LAPS分析系统在GRAPES背景场的基础上融合了地面、探空、雷达和卫星资料,比较了LAPS分析系统是否融合卫星资料的云微物理变量(云水、雨水、云冰、云雪)差异;进一步对比讨论了模式提前6 h滚动预报与LAPS反演的云微物理变量的异同,在此基础上探讨了两者对模式降水预报的影响。结果发现,LAPS分析系统融合了卫星资料后,能较好地再现中尺度云团的形态和强度,且能减少一些云微物理变量(云水、云冰、云雪)的虚假中心,大值中心与强降水中心吻合较好;模式滚动预报与LAPS反演的云微物理变量水平和垂直分布差异较大,后者的分布形态基本与观测降水、雷达回波相似,大值中心与强降水中心吻合较好;从两者对模式降水预报的效果来看,LAPS分析的云微物理变量作为模式初始场中的云参数,能改善模式1 h的降水预报,而模式滚动预报的云微物理变量对模式降水预报基本没改善。

Based on the background field of the GRAPES model, the surface, sounding, radar and satellite data are integrated into the LAPS analysis system. Through the rainstorm case on May 26, 2013, the differences of cloud microphysical parameters（cloud water, rain water, cloud ice, cloud snow） are first compared between the ingested satellite data and uningested satellite data in the LAPS system, and their differences are then analyzed between the forecast of the GRAPES model and the analysis of LAPS at 00：00 on May 26, 2013. The influence of both cloud parameters on the model precipitation forecast is further discussed. The results show that the formation and intensity of mesoscale cloud clusters are reproduced, the false center of some cloud microphysical parameters is decreased, and a large-value center fits well with the center of strong precipitation after the satellite data is ingested into the LAPS system. There exist significant differences of vertical and horizontal distribution in cloud microphysical parameters between model rolling forecast and LAPS analysis, especially in the junction between Guangxi and Guangdong and the east of Guangxi, with the latter much closer to the observation than the former. From the model simulation of the precipitation, we learn that at the first hour it is improved only when the cloud microphysical variables of LAPS are used as cloud parameters in the initial field of the model, but not improved when the cloud microphyciscal parameters of model rolling forecast are used in the initial field of the model.