中文摘要：

利用中尺度非静力WRF（Weather Research and Forecasting）模式及其三维变分同化系统,对2007年7月淮河流域的一次强降雨过程进行多普勒雷达径向速度资料的三维变分同化试验,重点考察雷达资料的不同稀疏化方式对同化结果以及对暴雨数值模拟的影响。结果表明：同化多普勒雷达径向速度资料使得模式初始风场包含了更丰富的中尺度特征信息,有效调整了初始场的环流结构,能够改善模式对暴雨过程的模拟效果;以不同的稀疏化处理方式同化多普勒雷达径向速度资料对分析场会产生不同的影响,进而影响模式的降水预报效果,本次试验中当极坐标网格径向分辨率取10 km的时候降水过程的预报效果最好。

英文摘要：

Owing to a lack of sufficient observational data to provide an accurate initial field in numerical modeling,the numerical prediction of severe convective weather systems is always difficult.With the rapid development of atmospheric detection technology,nonconventional observed data（e.g.,radar data） have been widely applied.Compared with conventional data,radar data-with their higher spatial and temporal resolution-can extract more useful information to improve the model＇s initial field,improve rainstorm mesoscale weather forecasting,and promote the simulation process capability.In this paper,the impact of Doppler radar radial velocity on the prediction of a heavy rainfall event is examined.Based on the mesoscale nonhydrostatic Weather Research and Forecasting（WRF） model and its three dimensional variational assimilation system（WRF-3DVAR）,a control experiment and three assimilation experiments were conducted for a heavy rainfall process that happened over the Huai-he River basin in July 2007.The model background fields were provided by NCEP data,and the Doppler radar radial velocity data were obtained from Fu-yang station,which houses a new generation of S-band Doppler Weather Radar（Model：CINRAD/SA）.Because the effect of the radar data is related to the thinning method,this paper focuses on the influence of different thinning methods of radar data on assimilation results and rainstorm numerical simulations.To this end,a new radar data thinning method is introduced.Firstly,a horizontal polar coordinate plane with certain resolution is constructed;secondly,radar data on the elevation plane are projected onto the polar coordinate plane;then,the distances between the projected points and the polar coordinate plane grid points are calculated;and lastly,the nearest points are regarded as good points and the corresponding radar data are retained.In this method,the degree of radar data thinning depends on the density of the polar coordinate horizontal structured grid.The numerical results showed：（1