Knowledge Based Grasshopper Habitat Classification Approach Supported by GIS in Qinghai Lake Region

Grasshopper is a kind of primary pests damaged to a wide rangeland region in the west of China. According to its ecological features, classifying habitat types of grasshopper using remotely sensed is an important research project. The result is very important for rangeland management. For Qinghai Lake region, we applied the Landsat TM images, auxiliary geographical data including DEM, the specific map of rangeland types ect, and GPS Located field data. From the monolithic view point of image processing, geographical data and specialist knowledge, the grasshopper habitat was classitied by applying the methods based on knowledge. The classification accuracy is 84.23%. And three kinds of rengelard: Stipa krylovii, Achnatherum splendens, Stipa purpurea belong to frequently occurred site of grasshopper which take up 29.09% of whole research area. This result provides a basis for scientific management of rangeland pests.