目前气象预报业务中，预报员主要借助卫星云图，定性判断云团的移动趋势，缺乏形式化的定量评判方法。本文基于FY-2C与FY-2D的高时间分辨率的近红外影像(10.3—11.3μm)，采用亮温和面积阈值方法进行云团识别，然后根据最大相关系数云团匹配技术进行追踪，系统地实现强对流云团的自动临近预测。实验结果表明，本文提出的最大相关系数追踪比传统的交叉相关系数法具有更高的匹配精度和运行效率,而且研究发现云团质心外推明显优于最低亮温外推，平均亮温、面积、圆形度对云团的分裂合并有较好的指示作用，经列联表法检验，本文提出的自动识别追踪技术具有较高预测精度和预测时效，并且为卫星云图业务化应用提供了定量科学依据。

The movement of clouds is qualitative analyzed by forecasters with satellite images currently, which is, however, lack of objectivity and quantitativity. In this paper, based on the stationary satellite infrared (IR) channel (10.3—11.3 μm) images of FY-2C and FY-2D with the time resolution of 15 minutes, brightness temperature (BT) and area threshold are selected to identify the severe convective cloud (SCC). We then use the SCC matching algorithm of maximum correlation to track the shorttime automatic prediction of SCC systematically. The experiment results show that the tracking method proposed in this work has higher matching accuracy and efficiency compared with the traditional cross-correlation approach. The cloud center of gravity (CG) extrapolation is markedly superior to the minimum temperature, and the mean temperature, area and roundness all have better indications to the cloud split and merge. Tested by contingency table, the automatic identification and tracking technology has high prediction accuracy and timeliness. In addition, the research of this paper provides a scientific basis for the objective and quantitative application of satellite images to SCC short-time prediction in operation.