Research of the vertical accuracy assessments according different methods are in the paper presented. The following data were compared: GPS RTK (as a reference), air- borne stereo model from ADS40 camera, cloud of points from the ALS, and a GRID model created from the ALS data. Raw ALS data were initially preprocessed in TerraScan for classifying of ground points (using the active TIN model algorithm). Triangulations in a radius of 20 m around the GPS control points were performed. Then the height corresponding to GPS position (x, y) was from the triangle plane calculated. In the same way height for GPS position was obtained from ADS 40 stereo model. NMT in GRID model of 1 m grid size was generated in TerraScan basing on the points early classified as a ground. The lowest of the all data set was GPS surveying (average, more than 0.2 m below the ALS data). ALS points cloud and GRID model were the closest to each other but the NMT was an average of 0.1 m above the raw data.

13-24