Soil roughness characterization by photogrammetric techniques Towards SAR modeling

Abstract

Soil roughness and volumetric moisture are the two main soil-related variables influencing the SAR backscattering coefficient. Since it is usually challenging to decouple the effect of each, in this work, we propose a methodology to estimate soil roughness, mainly in agricultural environments. The traditional methods, using graduated tables and profilometers, are laborious and spatially limited. The use of lasers is feasible, but they are normally inaccessible. This work proposes a methodology based on photogrammetric techniques using primarily educational software. We tested the method over two bare soil 1m2-areas, one with a random surface pattern and another where we simulated typical crop rows (52 cm apart). As a result, highly accurate surface three-dimensional digital models were obtained. We then extracted the standard deviation of the surface height and the correlation length, the main roughness parameters required in SAR modeling. Additionally, we were able to extract other relevant information, such as the predominant spatial structure directions and height profiles from the autocorrelation function and the multiscale roughness through Fourier analysis. Given the excellent results of this fast and low-cost methodology, we estimate it could provide precise and systematic information on soil  roughness for operational applications in the SAR context in view of Argentina's SAOCOM missions.