Forest Analysis Using SAR Tomography and Maximum Likelihood Inspired Spectral Estimation

Abstract

This paper treats the synthetic aperture radar (SAR) tomography (TomoSAR) non-linear inverse problem, within the framework of maximum likelihood (ML) estimation theory. In this context, a novel non-parametric spectral analysis (SA) technique, addressed in a closed fixed-point iterative fashion, is presented. The main goal of the proposed approach is to provide resolution-enhancement, with suppression of artifacts and ambiguity levels re-duction, to an initial estimate of the continuous power spectrum pattern (PSP), retrieved using the celebrated matched spatial filter (MSF) beamforming technique. The feature enhancing capabilities of the proposed method are corroborated via processing L-band airborne multi-baseline SAR data of the German Aerospace Center (DLR), acquired by the F-SAR system over the forested test site of Froschham, Germany, in 2017.