90. On the Accuracy of LiDAR Derived Digital Surface Models

The use of LiDAR data for developing high resolution topographic models of urban and rural areas has been widespread for more than a decade. In particular, the derived high resolution digital surface models (DSMs) are commonly used as input (among other models and parameters) for predicting the coverage of flooded regions. The accuracy of such digital surface models varies depending on a number of factors, from system components (including positioning technology, e.g., GPS/INS, laser scanner and boresighting parameters), data processing (point clouds, interpolation) and signal-target/surface interaction (i.e., backscatter signal strength, laser incidence angle/geometry). The main objective of this paper is to quantify the absolute and relative accuracy of the derived digital surface models. LiDAR data collected over an urban area in Canada from an airborne platform is used for the analysis and compared to precise DGPS survey data. In addition to inter-comparative first order statistical measures, the iterative closest point matching of point clouds method is employed for a consistent analysis. Results provide an indication of the contributing factors to the total error budget for vertical heights in this region which is found to be at the 3 cm level (1 sigma).