Soil Moisture Estimation from Multi-Frequency Synthetic Aperture Radar (SAR) data. Towards an Integrated Agricultural Drought Monitoring Framework

Over the last fifty years, microwave remote sensing has established itself as a key tool for soil moisture retrieval. In particular, Synthetic Aperture Radar (SAR) is capable of obtaining high-resolution measurements, which are required by several applications, such as precision agriculture, hydrological and flood monitoring, and drought forecasting. However, the SAR backscattered signal is influenced not only by variations in the soil’s dielectric properties, but also by a range of bio-geophysical factors, including soil roughness, canopy structure, vegetation biomass, and water content. Presently, SAR-based soil moisture retrievals exhibit higher uncertainty under dense canopies, steep terrain, and no sensitivity in forest ecosystems. The use of longer wavelengths is expected to improve soil moisture retrieval on vegetated areas, due to their penetration capability and thus greater sensitivity to soil properties, yet their limited availability has so far constrained their application. A new generation of L-band missions, including NISAR (2025) and ROSE-L (2028), has been designed with soil moisture monitoring as one of the primary objectives. These will accomplish a global C-band and L-band coverage of the Earth, posing the challenge of combining asynchronous acquisition in a System-of-Systems approach, dealing with different configurations and geometries. The main goal of this thesis is to advance soil moisture retrieval from SAR time series by exploiting the synergy between C- and L-band observations. The resulting soil moisture estimates are integrated with eco-hydrological indices to support applications related to drought monitoring and vegetation productivity. The research is structured around the following topics: (i) assessing the sensitivity of C- and L-band data for soil moisture monitoring through regressions and change detection methods; (ii) developing a roughness mitigation approach based on polarimetric entropy; (iii) quantifying the impact of incidence and azimuth angles on multi-frequency data; (iv) evaluating the effects of radiometric terrain flattening, masking, and downsampling on anisotropy effects; (v) developing a framework to integrate C- and L-band soil moisture estimates, together with Solar-Induced Chlorophyll Fluorescence (SIF) for empirical vegetation correction; and (vi) assessing eco-hydrological responses to drought with retrieved SM, meteorological and vegetation indices. The long-term change detection method provided the basis for these analyses, which were conducted employing SAOCOM and Sentinel-1 SAR systems. The test areas comprised a region in Spain, characterized by the availability of in-situ soil moisture data, and the Po Basin, for the variety of land cover and topography conditions, as well as for its sensitivity to drought events. Overall, the thesis demonstrates that integrated multi-frequency approaches can advance operational soil moisture retrieval and support applications in water management, early warning systems, and food security.