Study on Variation of Forestland in Lugu Lake Scenery District Based on Technique of Spatial Data Mining

By using the remote sensing images of Lugu Lake scenery district in 1990 and 2005, the information about variation of forestland for 15 years has been extracted. Based on the technique of spatial data mining (SDM), both the universal geometrical characteristics of forestland variation patches and the knowledge of spatial association rules have been mined by adapting statistical analysis method and cluster relation analysis method. On the basis of calculating and statistic analyzing the minimum, maximum, mean, variance, and mode of the objective geometric characteristic eigen values, the spatial distributed discipline and the association rule of variation patches have been analyzed, as well as the inner disciplines in forestland variation process; thus to provide a foundation for the study of driving forces of land use variation and the management of forest ecosystem in Lugu Lake scenery district.