This paper concerns three dimensional imaging of volumetric natural environments, such as snow packs and sea ice, using a ground based SAR system and X and Ku bands. The ground-based acquisition system consists of a vector network analyzer and an multi-static antenna system, moved along two orthogonal directions, so as to obtain a two dimensional synthetic array. Tomographic processing techniques, based on the coherent combination of the acquired SAR signals and adapted to the three-dimensional focusing of very high resolution data, are proposed. Specific automatic correction approaches, aiming to compensate for acquisition errors and uncertainties, are developed in order to improve the quality and descriptiveness of the tomograms. It is then shown that three dimensional imaging permits to retrieve the vertical structure of shallow volumetric media at a resolution of few centimeters and to emphasize physical scattering phenomena that cannot be observed from classical two-dimensional SAR data sets

The radar response of vegetated environments, and forested areas in particular, are usually modeled using a very simple structure made of a random volume, representing a cloud of vegetation particles, lying over a semi-infinite medium with a rough interface, associated with the underlying ground. This Random Volume over Ground model can efficiently handle double-bounce scattering mechanisms, or arbitrary volume reflectivity profiles. This paper proposes to analyze a specific component of the Random Volume over Ground simplified scattering model, which concerns the double-bounce interaction between the ground and the volume. This specific contribution is not considered by classical characterization techniques and is studied in this work using a controlled experiment involving a Synthetic Aperture Radar operated in a Polarimetric and Tomographic configuration in order to image in 3D a controlled miniaturized scene composed of volume lying over a ground. It is shown that ground/volume double-bounce scattering, which remains focused at the ground level even in 3D imaging mode, and has polarimetric patterns that differ largely from those usually expected from double-bounce reflections, with volume-like features, such as a strong cross-polarized reflectivity or decorrelation between co-polarized channels. Moreover, it is shown that the full rank polarimetric patterns of the ground-volume mechanism are tightly linked to the reflectivity of the volume and may mask the ground response. As a consequence, isolating the ground response using 3D imaging does not permit to avoid a generally very strong distortion of the soil response by the double-bounce reflection, and the estimation of different geophysical parameters of the ground, such as its humidity or roughness are significantly altered.

No abstract

No abstract