Soil water content estimation using ground-based active and passive microwave remote sensing : ground-penetrating radar and radiometer

(eng) Soil water content is widely recognized as a key component of the water, energy and carbon cycles and knowledge of its spatiotemporal distribution is in particular needed for developing optimal and sustainable environmental and agricultural management strategies. In that context, we analyzed and further developed advanced ground-penetrating radar (GPR) and microwave radiometry techniques for high-resolution mapping and monitoring of shallow soil water content at the field scale. First, far-field ultra-wideband GPR and L-band radiometer were used for mapping soil water content over two test sites with bare soils and the results were compared to reference ground truths. For GPR, soil water content was derived from full-wave inversion focusing on the surface reflection while for radiometer a radiative transfer model was used. Both techniques provided relatively good results, especially for reconstructing spatial moisture patterns in relation to topography and forced conditions (differential irrigation and soil tilth). Nevertheless, absolute estimates were subject to inherent discrepancies that were attributed to the different characterization scales and local variability. Second, we addressed the roughness modeling problem. For GPR, we combined the full-wave GPR model with a roughness model derived from the Kirchhoff scattering theory. Laboratory experiments showed that this approach performs well for roughness amplitudes reaching up to one fourth the wavelength. For the radiometer, we used an empirical equation which requires calibrating ground truths. This approach was successfully validated in field conditions. Finally, GPR and radiometer measurements were performed over a sand box subject to hydrostatic equilibrium with a range of water table depths. For each technique, all measurements were aggregated in an inversion scheme to reconstruct the vertical water content profiles, which were constrained using the van Genuchten water retention equation. The results were in close agreement with reference time-domain reflectometry measurements. Our results open promising research and application perspectives for the joint use of active and passive microwave remote sensing for soil moisture retrieval. In that respect, we addressed new avenues for characterizing crop canopies and water-stress related phenomena.