EVALUATING ERRORS IN LEAST-SQUARES BASED SURFACE ROUGHNESS MODELS FOR ANALYZING TERRESTRIAL LIDAR DATA

In recent years, terrestrial light detection and ranging (LiDAR) instruments have become widely utilized for high resolution terrain surface mapping at the outcrop scale. The rapid growth of this technology has led researchers to look beyond the commonly invoked digital terrain model for extracting higher order information from the high resolution spatial data. One such data processing technique is surface roughness, which is a measure of point variability over a segmented point cloud dataset. Although several LiDAR-based surface roughness models have been proposed, the most frequently invoked model measures surface roughness over a segmented dataset as the standard deviation of point distances from a reference datum defined by local (grid cell scale) least-squares regression planes. In the research presented here, we evaluate the accuracy of least-squares based surface roughness models experimentally by constructing a surface of known roughness, acquiring LiDAR scans of the surface at 25 angular orientations, and comparing the results of three different least- squares based surface roughness models to the known value. Surface roughness estimation errors are shown to result from angular and occlusion (shaddowing) sampling error, and a recently proposed surface roughness model based on orthogonal distance regression is suggested as a generally robust technique for minimizing the effects of sampling error on surface roughness estimates.