The Grey Level Co-occurrence Matrix(GLCM) approach was firstly used in this paper to extract the textural parameters,the intrinsic attribute of different types of clouds,from 100 satellite images of cloud cluster samples,and then,four texture parameters of cirrus,cumulus,cumulonimbus,stratus and cloudless areas were clustered by using the Mahalanobis distance as the comparability measure standard.In the experiment,textural features of the clouds,of the strong rainstorm in the period of July 4-5,2003 in the middle and lower reaches of the Changjiang River,and their relations with precipitation were analyzed.Results of the quadratic polynomial regression analysis show that the developing tendency of texture parameters is parallel with one of precipitation.These results are of practical significance for the diagnoses and numerical prediction of weather processes,especially of short-medium range rainstorm processes.