Analysis of Land Use/Cover Dynamic Change in XiŽan, using Modified BP Neural Network Classification for Remote Sensing Images

Abstract

Land use/cover change (LUCC) is one of the most important fields of global environment change research. It reveals the evolution process of natural ecological environment influenced by human activity. In this paper, land use and cover in XiŽan region in 2000 and 2003 were achieved by modified back propagation neutral network (BPNN) using remotely sensed data Landsat TM/ETM and DEM. Meanwhile the result of the BPNN classification was compared with the Maximum Likelihood Classification (MLC). The result showed that the BPNN had a higher accuracy. By analysing the results, we drew conclusions as follows: (1) the classification accuracy of BP neutral network was improved obviously; (2) the area of urban built-up land increased quickly, and its annual increasing rate was 12% from 2000 to 2003; (3) the area of woodland also rose fast whose annual increasing rate was 2.48%; (4) the increasing rate of orchard was astonishing which reached to 15.30% per year. These figures showed that since the great development of western China, the economy in XiŽan region had grown very rapidly and the ecological environment here had been also improved a lot.