中文摘要：

为了了解初值场对华南短时临近降水预报的影响，文中利用GRAPES区域中尺度模式，针对华南一次暖区暴雨过程分别进行控制试验、同化地面探空资料、nudging雨水资料和同化雷达径向风等四个模拟试验。分析结果表明：（1）同化地面探空资料有助于改善24小时的降水落区及其量级；（2）nudging雨水资料对临近降水预报有积极影响；（3）同化雷达径向风能使24小时的降水落区、量级得到明显的提升。这些结论为下一步的华南地区短时临近降水预报研究提供了重要的技术参考。

英文摘要：

In order to understand the impact of initial conditions upon prediction accuracy of nowcasting and short-term precipitation over south of China, four experiments, which include a control experiment, the assimilation of routine surface and sounding observational data, and the assimilation of radar radial wind information and nudging rainwater, are respectively conducted to simulate a heavy rainfall case occurred in warm region over south of China, by using GRAPES_MESO. Results show that （1） assimilating routine surface and sounding observational data helps to improve both the predicted precipitation area and magnitude of 24 hours accumulated precipitation close to those of observation; （2） nudging rainwater contributes to an obvious improvement in nowcasting, while （3） assimilation of radar radial wind information distinctly improves the predicted precipitation area and magnitude of 24 hours accumulated precipitation close to those of observation. These results are significant technical references for the study of predicting short-term and nowcasting of precipitation over south of China in the future.