The previous retrieval models of soil moisture based on active microwave remote sensing are applicable for non-periodic random surfaces, not available to periodic surfaces with row structure yet. This study was based on fully polarimetric data and field data. It was found that the like-polarized (hh and vv) backscattering coefficients react obviously to the azimuth angles, up to 4 dB. while the cross-polarized (vh) backscattering reacts insensitively to row-direction changes, then the response functions of hh and vv polarizations were given. A distance parameter that denotes the distance between a sampling point and response function curve was constructed to remove the influence of azimuth angles. The correlation coefficients between the distance parameter and soil moisture were 0.7095 and 0.8723, in hh and vv polarization modes, respectively. Meanwhile the influence of the error of roughness parameters in field measurement was removed by three different polarization combination schemes. The effectiveness of this model was demonstrated by the estimated soil moisture of checkpoints. This study selected 18 sampling points as checkpoints.The correlation coefficient between estimated soil moisture and measured soil moisture was up to 0.9461, the standard deviation was 0.0366 cm3/cm3. The model was reliable and applicable for periodic surfaces.