The main application of airborne laser scanning (ALS) technology is the data collection for creating high quality digital elevation models (DEM). In Poland, almost entire area of the country has been scanned for the implementation phase of extraordinary hazards (mostly water hazards) protection system (ISOK). The quality of acquired data was a subject of inspection at the acquisition time. In this study, an alternative methodology was applied to evaluate the height component accuracy of the ALS data. For the inner accuracy evaluation (data consistency), subsets of the point cloud representing flat surfaces (e.g. roofs) were used. This data was approximated by a plane using least squares method. Based on residuals between approximated plane and the ALS data a mean square error was calculated. Numerical tests were executed for 36 planes representing roofs, roads, meadows and arable fields. For the anthropogenic areas the estimated mean square error is similar to the accuracy of distance measurement by a scanning system and ranges from two to four cm. In the case of natural surfaces that are characterized by high roughness, the error increases to a value of three to twenty cm. In order to assess the external (absolute) accuracy of the ALS data, heights of the reference points measured by geodetic techniques were compared with the heights of corresponding (neighboring) ALS points. The accuracy assessment was carried out for three classes of the land use (arable fields and meadows, roads, forests) and four test areas, located in the area of Widawa River Valley. The absolute accuracy varies for different types of land use from slightly more than ten cm for roads to more than double the value for forests.

27-38