During the MAESTRO 1 Campaign, mulli-frequency polarimetric SAR data were acquired over the Dutch site ‘Flevoland’. At the same time, an extensive ground data collection took place in order to measure several different bio- and geophysical parameters for bare soils. The SAR data are used to assess the performance of several theoretical surface scattering models for rough surfaces. These models are extended in order to yield all the elements of the covariance matrix, particularly the phase and the correlation factor between the HHand W channels. Using the ground data, input elements such as the soil permittivity and the surface roughness are fed into the models to compute the elements of the covar ance matrix. A comparison for Land C band between the theoretical predictions and the measured backscattering coefficients is then presented and clearly shows that the models predict in a satisfactory manner the elements of the covariance matrix.