Skip to Main Content
In the 1990s forestland in the panda's habitat, Southwest China Mountains, underwent rapid degradation since the natural forest was converted into agricultural land. Remote sensing technology has not only provided a vivid representation of the forestland's surface but also become an efficient source of thematic maps such as the deforestation in this area. Landsat-5 TM data in 1994 and Landsat-7 TM data in 2002 are available for detecting the forestland degradation in the study area. The foggy, cloudy and snowy weather and mountainous landscape make it difficult to acquire remotely sensed data with high quality in the panda's habitat. Supervised classification is performed in the image process and a maximum-likelihood classification (MLC) is applied using the spectral signatures from the training sites. According to the topographical and meteorological conditions, different training sites are created such as forest-forest, river valley, forest, crop, town, water, snow, cloud, shadow and non-forest. As the result, forestland degradation map provides much information for forest degradation. Classification accuracy assessment is carried out by ERDAS software and the overall classification accuracy is up to 82.81%.