Electromagnetic Models for the Retrieval of Surface Parameters through SAR Images

Abstract

Nowadays, several private or public institutions require the knowledge of many ground physical parameters, i.e. permittivity, ground roughness, soil moisture content, vegetation biomass-index, and so on. In particular, the soil moisture content is a key parameter in the hydrological cycle and, moreover, it represents a fundamental information in several applications as in predicting rivers floods, rainfalls and landslides. In this thesis, a Polarimetric Two-Scale Model (PTSM), able to describe the scattering from a bare soil surface and to account for de-polarization and cross-polarization effects observed in real radar data, is presented. In particular, the model assumes the surface to be composed by rough randomly-tilted facets, which give rise to a random drift of the local incidence angle and a random rotation of the local incidence plane. Unlike other existing approaches, the PTSM accounts for both these effects - which come straight from the surface model – and allows us to compute the Normalized Radar Cross Sections (NRCS) for the whole surface by averaging over the facets slopes the power density scattered from a single facet. The prediction on the diffuse field provided by the NRCS is then employed in a co-polar cross-polar ratio based retrieval algorithm, from which it is possible to get maps of parameters from measured polarimetric Synthetic Aperture Radar (SAR) data. Accordingly, the retrieval procedure is performed on real data acquired at the same time of “in situ measurements”, both to test its performances and to provide the PTSM validation. Moreover, an innovative interpretation of the Kirchhoff scattering integral is also provided: it regards the physical meaning of the Kirchhoff solution for the electromagnetic scattering from both classical and fractal surfaces.