Wetland is a land with seasonal or yearly water, on which lives with specific vegetations and animals. Wetland specifies with both ecologic and scientific values. With the study area of international importance, supervised classification supported by exports was approached successfully to Landsat TM based on the scale of community. Our research indicated that efficient preprocess of image enhancement and precise selection of signatures is the key to classify succesfully. Spatial structure of virgin riparian wetlands of North China was presented according to the classifiation result of community types. Our research also demonstrated that very high proportion of Ass.Calamagrostis angustifolia-sundry herbs within the study area made it clear the occurent loss of the wettability of study area as typical mires in North China, so that urgent measurements were requested in case of further degeneration.