Artificial neural network classification of Karst rocky desertification degree using SPOT satellite imagery and DEM data

Karst rocky desertification is a significant environmental and ecological problem in Southwest China. In this paper, the spectral information, spatial context and topography information were utilized to synthetically discriminate the Karst rocky desertification degree, which are derived from The SPOT satellite imagery and DEM. By the back-propagation neural network, we proposed the classification model structure and classified the rocky desertification levels in Du'an County of Guangxi province, China. The results verified the classification model of Karst rocky desertification degree is efficient and accurate.