This paper describes the activity performed in the framework of the 1990 AVIOSAR 580 SAR and scatterometer Italian campaign over the Matera test site. The experiment has been carried out to ascertain the reliability of surface scattering models in order to tackle with the data inversion problem. To achieve this goal, physical and geometrical features of 38 homogeneous bare soil fields are considered as a result of ground truth data collection. X band - VV polarization σ° values from the HUTSCAT scatterometer at 45° and from the CCRS airborne SAR sensor at the incidence angles between 50° and 62° are compared with those obtained from modeled data. The SAR data have been calibrated using both corner reflectors and extended natural targets. The backscatter variability due to surface roughness has been analyzed. Due to the characteristics of the analyzed data set, i.e., high incidence angle and surfaces with large root mean square slope profile m, the SPM and PO models were not suitable for many fields in the scene; on the other hand, a good agreement between theoretical predictions and experimental data is found when the Geometrical Optics (GO) model is used. A threshold value for m as function of the incidence angle is found and a physical interpretation of this behavior is also proposed. After the assessment of the GO model results, the problem of retrieving the real part of the soil relative dielectric constant Re(∈r) and m value is addressed as well, considering the extended Kalman filter approach. The comparison between the obtained values and the experimental data shows the reliability of the inversion approach, thus encouraging the use of more sophisticated backscattering models.