Peatland Extraction Based on ENVISAT ASAR, Landsat TM and DEM Data

Peatlands are one of the most important types of wetlands and significant to the balance of global change and ecosystems. On the basis of field investigation and backscatter coefficient comparison of different land cover types in different polarization radar images, ENVISAT ASAR, Landsat TM, and DEM data were taken as basic data for a classification method combining object-oriented and decision tree approaches that were to extract peatlands and adjacent land cover types in a typical peatland zone located in West Xiao-Xing'an Mountains. The overall classification accuracy was 93.54% with a Kappa coefficient of 0.92, which indicates that the proposed method used is effective for peatland extraction. Compared to previous work, there is some improvement in classification accuracy.