The purpose of the thesis is an analysis of selected existing algorithms for DEM generalization through case application in selected fields of geodesy and geoinformatics like cartography, spatial analyses in GIS, and quality control. In its practical part, I used the test area of Semič, consisting of a plain and the Semič hill rising above. Via ArcGIS program, I conducted several simple surface analyses: calculation of slope, angle, curvature, and analytical hillshading using DEM 5, DEM 12.5
and LIDAR data. By bilinear interpolation method, I resampled all three layers of spatial data from 12.5 resolution to 25m and 50m resolutions, and conducted analyses. I interpreted the analyses' results in the sense of three selected fields in geodesy and geoinformatics. LIDAR is the highest quality display regarding the level of detail. Especially in the flat part, DEM 12.5 and DEM 5 are more consistent, whereas DEM 5 and LIDAR tally better in the hilly tract. Analytical hillshading at 45° azimuth yields undefined surface displays, a reverse surface display at 135° azimuth, hills being valleys and valleys hills, while hillshading at a vertical 10° angle forms long shadows. The best contours are generated from LIDAR, where contours are generalized and reduced mainly in the flat tract. Resampling to better resolution from 12.5m to 1m, I searched for errors in the layers. Through indirect sampling, I obtained even more generalized displays compared to direct sampling. Exposition is least sensitive to generalization, while angle and curvature are much more so. In curvature, differences are well enough seen at 25m resolution, where we can also visually perceive the differences in concavity and convexity. The least differences are seen in plan curvature. When resampling to smaller resolution of 25m and 50m, the quality of display and DEM data decreases, but I obtained more equivalent inter-layer displays at 50m resolution.