Classification of forest vegetation species based on parameters of tomography

In this paper, we propose a method to extract multi-parameters regarding to tomography from PolInSAR data and investigate the potential of applying them to classify the forest vegetation species. We first propose a general volume scattering model that includes three basic scattering profile structures according to the change of extinction. Then three basic geometries of trees are introduced and their corresponding scattering profiles in HH-VV and HV channels are developed by the combination of three presumptive profiles. Supposing the contribution values of the three geometries to the final complex coherence is a, β and γ respectively, the total vertical structure of a general tree can be reconstructed. By resolving the equations between coherence and each contribution component the three contribution values can be calculated and then are used to identify the tree species. Finally this method is validated using simulated data.