Polarimetric SAR Tomography of Forested Areas: a Covariance Matching Approach

Abstract:
A covariance matching based technique is presented to perform tomographic analyses of forested areas through multibaseline SAR acquisitions. We study the dependance of the second order statistics of the data with respect to the spatial distribution and backscattered power of each target within the resolution cell. This approach may be extended in order to perform tomographic analyses on the basis of a fully polarimetric data, when available. In this case, we force the hypothesis that the scatterers which determine the HH, HV, and VV polarimetric channels share the same spatial distribution, whereas their intensity only is allowed to change from one channel to another. The forest structure is modeled as being constituted by two elements, representing the ground and the canopy layers, so that the set of the unknowns is given by the spatial distribution and the three backscattered powers - one for polarimetric channel - for both the ground and the canopy layer. The results of a P-Band tomographic analysis of the forest site of Remningstorp, Sweden, are presented.