A maximum likelihood analysis of the RVoG model for forestry studies in polarimetric SAR interferometry

Abstract

This work analyses the validity of the RVoG model as- sumption in case of PolInSAR data in forestry analy- sis. In order to assure that the interferometric coherence describes a line, as imposed by the RVoG model, the normalized PolInSAR coherency matrix must be affine equivalent to an Hermitian matrix. On the basis of this observation, two tests for the RVoG model hypothesis are proposed. A test based on the analysis of the coherence region shape, and a test based on a Maximum Likelihood approach, on the basis of the zero mean, complex, Gaus- sian distribution. Preliminary results would indicate a larger validity of the model at P-band over L-band and, in case of Tropical forests.