The identification of factor and indicational features, which are characterized by the high informativity and field of view in relation to the soil cover organization, plays a very important role in the soil mapping. Such characteristics are more common for Unmanned Aerial Vehicles (UAV), which include spectrazonal imagery and digital elevation model (DEM) with ultrahigh spatial resolution, necessary for obtaining fine and large scale images. However, the agrogenic micro- and nanotopography is considered as a noise during the studies of the soil cover topographic differentiation under the conditions of plowland, as the genetic soil properties correlate with natural micro- and mesotopography. A filtration algorithm for the land surface roughness, which is not related to the spatial organization of the objective soil properties, is suggested in the paper. The stages of linear dimension identification for self-similar structures of the glacial and agrogenic topography based on two-dimensional Fourier decomposition are demonstrated using the example of a field topography digital model for the area of 125 hectares. Filtering in the frequency domain allowed restoring the natural field topography and substantiating the effective resolution of the DEM and the size of the area to calculate local morphometric specificities of the topography for digital soil mapping.