Characterizing Bidimensional Roughness of Agricultural Soil Surfaces for SAR Modeling

In the description of agricultural soil roughness, the hypothesis of surface isotropy is currently admitted, and linear measurements are often used to characterize the soil roughness considered as a single-scale process. However, multiscale roughness is frequently observed, and tillage practices created oriented roughness. This paper presents a new technique to measure precisely the bidimensional soil roughness. Digital elevation model derived using photogrammetric technique reproduces the millimeter-scale height variations of three different soil surfaces (ploughed, smoothed, and row structured field) over about 8 m² . A single surface measurement is sufficient to accurately measure the soil roughness parameters. Geostatistic parameterization allows the measurement of the roughness anisotropy. For smooth surface, a two-scale roughness is observed. Anisotropy is observed in the larger scale roughness. The proposed method allows the computation of the bidimensional correlation function, which is required by the integral equation method model for the simulation of the SAR signal over anisotropic soil surfaces.