Qualification de modeles numeriques de terrain lidar pour l'etude de l'erosion : Application aux badlands de draix

This study deals with altimetry quality assessment of digital elevation models (DTM)using data from Aerial LiDAR surveys on a highly dissected terrain devoid of vegetation with a badland morphology. The study aims to quantify the deviation of the LiDAR measurement and the accuracy which can be expected for the derived DTM. The paper proposes anassessment of the LiDAR point cloud density required to describe correctly the terrain. It compares the elevation accuracy ofthree derived DTM. The results validate the DTM mostsuitable for hydrological and erosion process studies The three DTM are calculated from two point clouds obtained by two different recalibration systems. They use the same filtering method as Iterative Triangulation but with different sets of parameters. The qualificationshows that the filtering method, as driven by the lowest points, underestimates theterrain elevation. When the point cloud density is increased, for thethird DTM, this bias can be neglected and the quality of the resultsis quite satisfactory: the bias is 1.8 cm andthe standard deviation is 15.1 cm. The spatial distribution of the elevation deviations is analysed according to the landforms and slope angles. The altimetricerrors increaseforslope angles over 30° the DTM accuracy is better than 10 cm on flat areas howeverthe error can exceed 20 cm ondivides or steep slopes. The final DTM quality fulfils the needs of the users for their research purposes. However, this DTM accuracy isnot sufficient to quantify and monitor the erosion rate on an annualscale. A possible improvement wouldbe to adapt the filtering techniques to take into account the vegetation and the relief type.