Vertical positional accuracy assessment of a topographic lidar point cloud using conventional surveying as reference

Abstract

Lidar is an active remote sensing technology that reduces the cost and time required to collect topographic data. The vertical accuracy of lidar data is normally set by commercial data providers at a maximum root mean square error of 0.150 m. However, the results of accuracy assessments where field data at least three times more accurate than lidar data have been used suggest that that accuracy is only met when the density of the lidar data is higher than 1 ground points per square meter. Unfortunately, the number of these studies is limited and more accuracy assessments are needed in order to establish the vertical accuracy of lidar data. In this paper, the vertical accuracy of a topographic lidar cloud point with a density of 1.02 ground points per square meter is assessed for different types of terrain. Reference data was collected on the field using a total station with millimeter accuracy in order to meet the ASPRS requirements for the accuracy assessment of lidar data. The results suggest that the vertical accuracy of topographic lidar data exceeds the accuracy specifications of commercial data providers and, also, are consistent with previous studies where the vertical accuracy of lidar was assessed under similar conditions. Therefore, it is suggested to set the vertical positional accuracy (EPV, after its Spanish acronym) of the lidar point cloud in 0.200 m for mixed terrain with constant changes and in 0.150 m for areas with smooth topographic changes. However, under the presence of systematic errors, the EVP does not convey the magnitude of 95% of the vertical errors, as intended; therefore, it is recommended to include the 95th error percentile in the documentation of lidar data.