Analysis of agricultural scenes based on SAR interferometry

Abstract

The aim of this work is to study the temporal behavior of interferometric coherence of natural scenes and use it to discriminate different classes of targets. The scattering properties of targets within a SAR resolution cell depend on their spatial distribution and dielectric constant. We focus on agriculture scenes. In case of bare soils, the radar cross section depends on surface roughness and soil moisture. Both quantities are strongly related to agriculture practices. The interferometric coherence can be modelled as the factorization of correlation terms due to spatial and temporal baselines, terrain roughness, soil moisture and residual noise. We use multivariate analysis methodologies to discriminate scattering classes exhibiting different temporal behaviors of the interferometric coherence. For each class, the temporal evolution of the interferometric phase and radar cross-section are studied.