Sensitivity to soil moisture by applying a model-based polarimetric decomposition to a time-series of airborne radar L-band data over an agricultural area

In this paper, the sensitivity to soil moisture variations over an agricultural area characterized by different vegetation covers has been investigated. The objectives of the study were identifying the correlation of one or more polarimetric parameters with soil moisture changes over time. In this respect, we assessed the possibility to separate three individual scattering mechanisms (i.e., surface, double-bounce, volume) offered by radar polarimetry. We apply a model-based polarimetric decomposition to a time-series of high-resolution SAR data collected at L-band by the NASA/JPL UAVSAR airborne radar over the Yucatan Lake region in Louisiana, USA. Thirteen flights were considered and five regions of interest characterized by different surface properties and vegetation covers were selected. The temporal evolution of different polarimetric parameters, obtained by applying the Freeman-Durden decomposition, is reported and discussed. The polarimetric features were compared not only to the NDVI variations derived from Sentinel-2 satellite, but also to precipitation data recorded by a nearby precipitation station as well as to the Soil Water Index derived from the ASCAT sensor onboard Metop satellites. The improved sensitivity of the polarimetric features with respect to the single backscattering coefficients at different polarizations was also assessed.