中文摘要：

以中尺度非静力WRF模式的格点预报结果作为云模式的初值集合，经云模式的多初值雷暴预报及预报结果的集合分析，建立了雷暴云的集合预报方法。将该方法应用于南京周边地区未来一天雷暴天气的特征预报，并利用南京夏季9个雷暴天气的多普勒雷达资料（SCIT，storm cell identification and tracking）进行预报效果的检验。结果表明，雷暴云的集合预报对研究区域内未来一天雷暴强度、分布预报效果较好，尤其对强雷暴的分布有较强的预警预测能力。此外，雷暴持续时间概率密度分布的集合预报产品，在雷暴影响范围概率预报上的应用，提高了雷达对雷暴的预警监测能力。

英文摘要：

A thunderstorm ensemble forecasting approach is perti~rmed by the ensemble analysis of the cloud model＇ s forecast results according to the initial condition aggregation provided by the grid fore-cast results of the mesoscale non-hydrostatic WRF model. The approach is applied to forecast the char-acteristics of thunderstorm around Nanjing one day in advance and is verified by the Doppler radar data of the thunderstorms in summer in Nanjing. Results show that the intensity and distribution of thunder-storms over the research area is reasonably predicted one day in advance. It is particularly good to fore-cast the distribution of the strong thunderstorms by using the ensemble forecasting technique. Moreover, the application of the PDF（probability density function）of thunderstorms＇ duration into the probability forecasting of the influenced area of thunderstorms improves the radar＇ s warning and monitoring capac-ity of local thunderstorms.