A METHOD OF FOREST TYPE CLASSIFICATION USING POLINSAR DATA

摘要

Forest type mapping is of great significance forrnregional forest carbon estimation as forest typesrndistribution information is always the critical priorrninput information to forest carbon stock mappingrnmodel using remote sensing. Polarimetricrninterferometric synthetic aperture radar (Pol-InSAR)rndata acquired by DLR airborne SAR system (ESAR) inrnthe Traunstein test site in Germany was used to studyrnforest type classification method in this paper. A newrnunsupervised PolInSAR classification method based onrncoherent optimization (R-) matrix was proposed torndistinguish coniferous forest, deciduous forest and otherrnland cover types. It not only considers the fullrnpolarimetric information of single Polarimetric SARrn(PolSAR) data set but also the coherent information of arnpair of PolSAR data. The results show that thernclassification algorithm proposed in this paper is thernbest method with higher accuracy comparing with thernclassical method based on T_6 matrix.