Imaging radar polarization signatures: Theory and observation

TLDR

By adding two orthogonally polarized antennas and recording amplitude and absolute phase, the authors converted a conventional SAR into an imaging radar polarimeter that measures the full complex scattering matrix per pixel, from which the Stokes matrix is derived to synthesize scattering coefficients for any transmit/receive polarization combination, enabling analysis of polarization‑dependent scattering. The authors demonstrate that the derived polarization signatures agree with theoretical models—ocean and smooth lava flows match a slightly rough dielectric surface model, urban areas resemble dihedral corner reflector geometries—and that the coefficient of variation indicates ocean returns are highly polarized and can be selectively filtered, whereas vegetated areas show high variation and are largely insensitive to polarization changes.

Abstract

We have converted a conventional synthetic aperture radar system into an imaging radar polarimeter by employing two orthogonally polarized antennas and recording both the amplitude and absolute phase measurements of the received electric fields. This enables us to measure the complete complex scattering matrix for each individual resolution element in the radar image. The Stokes matrix, derived from the scattering matrix, is then used to synthesize the scattering coefficient for any desired combination of transmit and receive antenna polarizations, allowing us to measure the variation of the scattering coefficient with polarization. This polarization information can be displayed in a new and useful way, producing the polarization signature for each point in an image. Comparison of the theoretical and observed polarization signatures permits identification of the dominant scattering mechanisms contributing to the measured radar backscatter. For example, the ocean and smoother lava flows have polarization signatures which are similar to those predicted by a slightly rough dielectric surface model; thus the model likely approximates the physical mechanisms responsible for the reflected energy from these surfaces. Urban areas, on the other hand, exhibit polarization signatures which resemble those predicted by dihedral corner reflector geometries. The coefficient of variation, a parameter derived from the polarization signature, is a measure of resolution scale variation of a surface and indicates the degree to which polarization filtering can be used to enhance or suppress the return from a collection of scatterers. Our observations show that the ocean has a low coefficient of variation; thus the return from the ocean is highly polarized, and polarization effects can be used successfully to suppress or enhance the return from the ocean. Parks and other vegetated areas exhibit a large coefficient of variation, suggesting that the return from these areas contains a considerable component that is randomly polarized; varying the transmit and receive polarizations will not change the return from the vegetated areas appreciably.