Abstract: Mountain torrents are the result of many factors such as geological conditions, meteorological conditions and human activities. The relationship between rainfall and mountain torrent has always been the focus of research. It is explored that how the multi-source rainfall data can be effectively applied in the prediction and early warning of mountain torrent disasters. Rainfall prediction methods in areas without data are made up, based on the technology of satellite and radar estimation of rainfall, and combined with station observation data. The ensemble numerical prediction method is introduced, and the foresight period of rainfall forecast for mountain torrents is prolonged. A forecasting and warning system for mountain torrents is established. And a preliminary early warning platform based on multi-source rainfall data is designed with four levels, namely risk identification, risk prediction, risk simulation and risk assessment.

Key words: torrential flood disasters, multisource precipitation data, ensemble forecast

摘要： 山洪灾害是地质条件、气象条件、人类活动等多种因素综合作用的结果，降雨与山洪灾害的关系一直是研究关注的重点。探究多源降雨数据如何有效地应用于山洪地质灾害预报预警中，依据卫星与雷达估算降雨技术，融合站点观测数据，有效弥补无资料地区的降雨预报方法。通过引入集合数值预报，提高了山洪灾害降雨预报的预见期。建立山洪灾害预报预警体系，以风险识别、风险预报、风险模拟和风险评估4个层次为主要框架，初步设计基于多源降雨数据的预警平台。

关键词: 山洪灾害, 多源数据, 集合预报