Radar Polarimetry: Classical Versus Quad-Pol Methodologies

In the synthetic aperture radar (SAR) context, “fully polarized” has two conflicting meanings: measurement of the scattering matrix, or an output image product that is a complete polarimetric characterization of an observed scene. Modern quadrature-polarimetric (quad-pol) SARs focus on the scattering matrix, while the typical user’s objective is a fully polarimetric output product. Conventional quad-pol polarimetric retrieval processing relies on matrix decomposition, which leads to output products that are not fully polarimetric. The Stokes parameters from classical radar architectures are fully polarimetric, thus meeting users’ objectives. A quad-pol SAR will produce fully polarimetric output products if and only if it is polarization-conserving, via the Mueller matrix for example. On this path, decomposition algorithms are not needed. Approximations and models are not needed. Classical full-polarimetry—which has no use for the scattering matrix—achieves the same goal by transmitting circular polarization and receiving two orthogonal polarizations, sufficient to evaluate the Stokes vector. Circular polarization (either ${L}$ or ${R}$ ) coupled with a dual-polarized receiver are required. Both classical and user-oriented quad-pol approaches are founded on fundamental principles from classical optics, the Stokes parameters. These provide a full polarimetric portrait of the incoming electromagnetic field, including polarized and unpolarized constituents. Both the approaches share a requirement for circularly polarized transmissions, actually realized in the classical precedent, but virtually emulated in a quad-pol radar only when polarization-conserving methods are used.