Influence of Vegetation, Slope, and Lidar Sampling Angle on DEM Accuracy

Abstract

Detailed GIS studies across spatially complex rangeland landscapes, including the Aspen Parkland of western Canada, require accurate digital elevation models (DEM). Following the interpolation of last return lidar (light detection and ranging) data into a DEM, a series of 256 reference plots, stratified by vegetation type, slope and lidar sensor sampling angle, were surveyed using a total laser station, differential GPS and 27 interconnected benchmarks to assess variation in DEM accuracy. Interpolation using Inverse Distance Weighting IDW resulted in lower mean error than other methods. Across the study area, overall signed error and RMSE were � 0.02 m and 0.59 m, respectively. Signed errors indicated elevations were over-estimated in forest but under-estimated within meadow habitats. Increasing slope gradient increased vertical absolute errors and RMSE. In contrast, lidar sampling angle had little impact on measured error. These results have implications for the development and use of high-resolution DEM models derived from lidar data.