Studies of the Earth's land surface involving scatterometers are becoming an increasingly important application field of microwave remote sensing. Similarly to scatterometer observations of ocean waves, the backscattering coefficient (sigma0) response of land surfaces depends on both the incidence and azimuth angle under which the observations are made. In order to retrieve geophysical parameters from scatterometer data, it is necessary to account for azimuthal-modulation effects of the backscattered signal. In the present study, this paper localizes the regions affected by a strong azimuthal signal dependence when observed with the European Remote Sensing Satellite Scatterometer and the SeaWinds Scatterometer on QuikSCAT (QSCAT). The possible physical reasons for the azimuthal effects, relating the very detailed QSCAT azimuthal response to the spatial orientation of special topographic features and land cover within the sensor footprint, were then discussed. Different methods for normalizing the backscattering coefficient with respect of observation azimuth angle were also proposed and evaluated. First, the mean local incidence angle of the sensor footprint using the shuttle radar topography mission digital elevation model (DEM) were modeled and concluded that the resolution of the DEM is too coarse to characterize most of the observed azimuthal effects. A more effective way of normalizing the backscatter with respect to azimuth is then found to be by using historical backscatter observations to statistically determine the expected backscatter at each observation azimuth and incidence angle as well as time of the year. The efficiency of this method is limited to the availability of past measurements for each location on the Earth