Polarimetric SAR Data for Forest and Deforestation Mapping in Guizhou Province, Southwest of China

Forestry inventory in southwest of China often suffers from the cloudy cover, and Synthetic Aperture Radar (SAR) can play an important role because of its all-weather and all-day capabilities. In this paper, we studied the potential of polarimetric SAR data for forest mapping in heavy cloud prone and rainy areas with polarimetric TerraSAR-X, Radarsat 2 data and how to improve the forestry classification accuracy through integration of SPOT 5 image and polarimetric SAR data. Field works were carried out in late August of 2007, and parameters were collected using a ground based Lidar and field measurements. For the high relief condition at the test site, a geometric correction strategy using two side looking direction SAR images and high resolution digital elevation model was proposed to overcome the geometric distortion of SAR image such as foreshortening, layover and shadow in hilly areas. Neural net method was suggested for classification of the SAR and SPOT images.