Quantitative analysis approaches to the driving forces of cultivated land changes on a national scale

By using digital satellite remote sensing data of China acquired in late 1980s and 2000 respectively, this paper explores the analysis approaches in the study of the driving forces of land use changes in large regions. At first an integrated regionalization was made based on the cultivated land changes, the natural situation and socio-economic changes using GIS and RS techniques. Then using BP neural network, the leading driving forces causing the decrease of gross cultivated land area were found in one class of the above regionalization as a case. In the case study area, the primary contributors are the rapid urbanization, the external forces from around the region namely urban influence, and the rapid development of second and third industry under market economy condition of China. A good result is achieved under 771 samples through BP neural network training, which indicates that neural network technique has a big power in the study of the driving forces of land use change. Meanwhile, it is proved to be feasible and reasonable to use the solution of regionalization first and then filtering variables. The quantitative research methods provide references for the study of other types of land use change in large areas.