Abstract:By analyzing the cloud characteristics during Meiyu periods from 2005 to 2008, t he heavy rain clouds were classified into three types, i.e. north type, center type and south type. The relations between the TBB of cloud top and correspon ding rainfall extreme and coverage of rainfall rate above 10 mm/h were studied. The results indicate that the north type and south type clouds have lots of co mmon feature, and show reverse features compared to the center type. The positi on of heavy rainfall corresponding to the moving path of clouds revealed that du ring Meiyu periods the one hour variation and gradient of TBB were not obviously correlative. Then the relations between environmental factors and clouds were analyzed. And the resul ts show that the macroscopic features of cloud are obviously correlative with ve rtical velocity and water vapor flux. At last based on atmospheric circumstances the n eural network forecast equations of TBB of cloud top and strong rainfall in matu re phase of clouds were developed. 