A semi-empirical approach for surface soil water content estimation from radar data without a-priori information on surface roughness

In this study, the spatial distribution of soil water content in an agricultural area of 30 km2 in Southern Italy has been estimated by using high-resolution space-borne Synthetic Aperture Radar data. Multi-polarised SAR images acquired during the SIR-C mission in April 1994 have been analysed by using the semi-empirical surface backscattering model derived by Oh et al. (1992). A site-specific calibration procedure of the cited model has been proposed to derive soil dielectric constant values without a-priori information on the surface roughness by using ground measurements on a regular grid in two bare-soil fields. The calibrated model applied to L-band data reproduced quite satisfactorily the spatial variability of the soil dielectric constant in the two fields. Diversely, C-band data gave poor results. Successively, the calibrated Oh’s model was applied to estimate the soil dielectric constant in bare soil and low vegetation fields of the entire irrigation district, where the output of a distributed simulation model of soil water balance were available. From the comparison between the Oh’s backscattering model and the soil water balance model, it was confirmed that, under bare soil conditions, the values of soil water content near the soil surface estimated from SIR-C L-band data differ by ± 20% from the simulated ones. Furthermore, as expected, the presence of a fractional vegetation cover, even if small, reduced the sensitivity of radar backscattering to soil moisture. The results of this study confirmed that L-band SAR data represent a minimum requirement for possible assimilation schemes in regional hydrological modelling.