In this study,the time sequential images of the FY-2E VISSR(Visible and Infrared Spin Scan Radiometer) IR(Infrared) and WV(Water Vapor) bands are used to recognize and short-time forecast deep convective clouds.In many researches regarding the distinguishing of deep convective clouds from cirrus clouds,the BT (Brightness Temperature) threshold technique is a frequently applied method,of which the defect lies in its variance with time and space,rendering it difficult to find a proper threshold to all weather conditions.The fact that water vapor has strong absorption in the location of the WV band along with the vertical distribution of water vapor in the atmosphere makes it difficult for satellites to receive the radiation emitted at the WV band by clouds under the height of around 400 hPa and at ground,while the satellite detected energy of the IR band mainly originates from the middle-low level of atmosphere.With the aid of disparity in the radiation source,the increase in optical depth of high-level clouds leads to a gradual change in the distribution pattern of pixels of satellite images in IR-WV spectral space,which is invulnerable to time and space,in contrast to the BT threshold technique.In the present study,pixels are identified as deep convective clouds if the fitted slope of their IR and WV BT is greater than a given threshold.The backward trajectory method is used to predict the location and shape of the detected cloud in future hours.The motion vector field of the target area is retrieved using the cross-correlation method from two neighbouring images,with a time resolution of 30 minutes.The detection and forecast methods are applied to an MCC(Mesoscale Convective Complex) which occurred in southeastern China,and the evolution of the MCC during its entire lifecycle is obtained by the analysis of its cloud top TBB(Temperature of Black Body) distribution.The results show that the detection algorithm in this article,compared to other methods using IR data only,functions more effectively in discriminating thin cirrus from intense convective clouds,as well as in detecting convective clouds with lower height.The forecast approach performs well in a short time range,and the results are more accurate for clouds with large spatial dimensions than small ones.