Experimenting with radar backscattering models using E-SAR imagery

This paper presents a study on the problem of separation and extraction of interrelated parameters, namely soil moisture and surface roughness, from a series of radar images acquired with the DLR airborne experimental sensor E-SAR. Many models have been developed in the active microwave domain to explain backscatter as a function of soil moisture and/or surface roughness. The complexity of the theoretical models and the non reliability of the empirical models motivated the authors to choose semi-empirical models. The results produced up to now using the OH-model show a rough compliance with the soil moisture terrain observations, but it has to be ameliorated significantly for the roughness estimation. The results obtained using the DUBOIS-model show a better agreement with C-band than with L-band. In this context, the relation between the estimated soil surface parameters (RMS-height and dielectric constant) and the relative and/or absolute sensor calibration errors were analysed. A review of the DUBOIS-model will also be described. It has been found that model parameter adaptions were necessary in the case of the authors hilly agricultural catchment area.