Airborne SAR polarimetric decompositions for soil moisture retrieval

Soil moisture is one of the key parameters needed to study the hydrological properties of the ground, playing an important role in many land applications such as agricultural management and flooding risk assessment. In this context, polarimetric SAR decompositions can separate individual scattering mechanisms (surface, double-bounce, volume), giving the possibility to study how soil moisture changes not only over bare soils but also over agricultural areas covered by different crops during the year. In this work, the sensitivity to soil moisture of different scattering mechanisms discriminated by an airborne polarimetric radar operating at L-band has been investigated. The main objective was assessing the capability of a fully polarimetric system to distinguish the change in soil moisture under different vegetation covers. We used data collected by the NASA UAVSAR airborne radar flying over the Yucatan Lake site in Louisiana. Six overflights were analysed, and six regions of interest characterized by different vegetation covers were selected. The NDVI derived from the Sentinel-2 satellite and data from a nearby precipitation gauge were used as an indication of the vegetation growing stage and moisture changes. The temporal trends of the magnitude of different scattering mechanisms, according to the Freeman-Durden and Nonnegative Eigenvalue decompositions, of NDVI and rain rate are analysed and discussed with reference to the theoretical expectations. The Freeman-Durden physical decomposition exhibited quite a reasonable behaviour, with the co-polarization ratio of the surface and double bounce components being particular effective for the forest plots to identify change in moisture.