Empirical and theoretical backscattering behaviour as a function of roughness for arid land surfaces

A study was carried out to assess the potential of using satellite images to identify a parameter that could characterize land surface condition-i.e., soil roughness, a factor that strongly influences radar backscattering from semi-arid land surfaces. A simple setting in terms of roughness-the flat Atar desert, Mauritania-was used for data extraction from an ERS-1 image, and two methods were defined for measuring the statistical parameters of roughness (standard deviation of heights and correlation length) from photographs of the ground. Analysis of these parameters showed that a close relationship exists between correlation length and standard deviation of heights for regs and that a single parameter enables the authors to describe roughness. Theoretical models (small perturbation model, integral equation model and geometrical optics model) were then used to establish data sets simulating ERS-1 backscattering coefficients for the different roughness surfaces. Based on the statistical parameters extracted from photographs, good agreement is observed between the modeled backscattering and the recorded backscattering from ERS-1 data. The relationship between the two statistical parameters was then used to simulate several surfaces, with the backscattering behaviour being studied through theoretical models. The results show backscattering as a function of the standard deviation of heights and define the possibilities and limitations for extracting roughness parameters