Forest AGB Estimation using L-Band Polarimetric SAR Features

Forest biomass plays an essential role in forest carbon reservoir studies, biodiversity protection, forest management, and climate change mitigation actions. Currently, polarisation information shows great potential for reducing saturation problems and improving estimation accuracy. 137 SAR features including backscatter coefficients, texture characteristics and features extracted from H/A/a decomposition and so on 9 decomposition methods were extracted for L-band airborne PolSAR data at two test sites, respectively for forest L-band scattering mechanisms analysis and AGB estimation. A multiple linear stepwise regression (MSLR) model and a fast iterative feature selection for K-nearest neighbor (KNN-FIFS) method are used to estimate the forest AGB at the two test sites. In the present study, there was evident site dependence of the L-band forest scattering mechanisms, while KNN-FIFS performed better in the estimation of forest AGB. The best AGB estimation was acquired at the Hainan test site with RMSE = 28.88 t/ha and rRMSE = 18.46%.