论文部分内容阅读
This paper is focused on the method for extracting glacier area based on ENVISAT ASAR Wide Swath Modes (WSM) data and digital elevation model (DEM) data,using support vector machines (SVM) classification method.The digitized result of the glacier coverage area in the western Qilian Mountains was extracted based on Enhanced LandSat Thematic Mapper (ETM+) imagery,which was used to validate the precision of glacier extraction result.Because of similar backscattering of glacier,shadow and water,precision of the glacier coverage area extracted from single-polarization WSM data using SVM was only 35.4%.Then,texturefeatures were extracted by the grey level co-occurrence matrix (GLCM),with extracted glacier coverage area based on WSM dataand texture feature information.Compared with the result extracted from WSM data,the precision improved 13.2%.However,theglacier was still seriously confused with shadow and water.Finally,DEM data was introduced to extract the glacier coverage area.Water and glacier can be differentiated because their distribution area has different elevations; shadow can be removed from the classification result based on simulated shadow imagery created by DEM data and SAR imaging parameters; finally,the glacier coverage area was extracted and the precision reached to 90.2%.Thus,it can be demonstrated that the glacier can be accurately semi-automatically extracted from SAR with this method.The method is suitable not only for ENVISAT ASAR WSM imagery,but also for other satellite SAR imagery,especially for SAR imagery covering mountainous areas.