A Layered Vegetation Model for GPR Full-Wave Inversion

Abstract

The vegetation cover strongly affects microwave remote sensing backscattered signals and makes difficulties in retrieving the surface soil water content (SWC). In this paper, we developed a vegetation cover model for full-wave inversion of ground-penetrating radar (GPR) data assuming multiple layers with effective electrical properties. Measurements were performed in laboratory conditions over spelt wheat grown in a sandy soil. A stepped-frequency continuous-wave (SFCW) GPR system was set up using a vector network analyzer and a horn antenna operating in 1–4 GHz range in far-field conditions with different incidence angles and both VV and HH polarizations. In order to control the sand surface water content, a water table was fixed at 20 cm depth and 6 time-domain reflectometry (TDR) probes were horizontally placed at 2 and 20 cm depth. Measurements were performed and recorded hourly for about 1 month and the canopy height was measured daily. The effective electromagnetic properties of the vegetation layer were modeled using a complex refractive index model (CRIM) combined with a single Debye relaxation model. A scattering loss parameter was included in the electromagnetic full-wave GPR model for the vegetation layer. The proposed method was also applied in a $\sim 7$ ha agricultural field with a vegetation canopy height of $\sim 27$ cm (spring wheat). The laboratory measurements resulted in a strong link between the vegetation scattering losses parameter and canopy height (generally, the biomass structure), which cannot be retrieved directly from the GPR measurements. Applying this vegetation model to the GPR full-wave inversion led to a significant correction on GPR-derived SWC values from the field measurements.