Using K-means clustering algorithm and GIS components ArcGIS Engine, the spatial data mining system based on continuous time sequence was developed. The system is applied to the data mining on the dry and wet spatial characteristics from 1644 to 1911 years in Jiangsu province. Through cluster analysis, The weight ratio of flood and drought are obtained. The result indicates that the weight ratio of flood and drought is different in different decades. Consistent characteristics not only exist in two types of values, there are also the first large values and the second category of lesser value, or the opposite characteristics. The different weight ratios of flood and drought reflect the various spatial distribution of wet and dry. In Jiangsu Province, the study is very important for measures it would be taken of preventing drought and flood disasters and discovery of its variation.