Abstract: This paper proposes an forecast method for heavy rainfall cloud clusters based on FY2C satellite. Forecast storm position can be divided into three steps. A clustering technique is used to identify rainfall cloud. A combinatorial optimization method is utilized to track identified cloud. A linear extrapolation method is used to provide 30 min, 60 min and 90 min forecast. Experimental results show the efficiency of the proposed approach.

Key words: satellite cloud imagery, heavy rainfall cloud clusters, identification, tracking

摘要： 提出一种基于卫星数据的暴雨云团追踪预警方法，采用聚类法在卫星红外云图上识别出有效的暴雨云团，利用组合最优化方法实现暴雨云团的追踪，通过线性拟合法给出30 min、60 min、90 min的短时预警。实验结果表明，该方法能有效应用于暴雨云团的临近预报。

关键词: 卫星云图, 暴雨云团, 识别, 追踪

CLC Number: 

• TP391.41