Automated Generation of Digital Terrain Model using Point Clouds of Digital Surface Model in Forest Area

At present, most of the digital data acquisition methods generate Digital Surface Model (DSM) and not a Digital Elevation Model (DEM). Conversion from DSM to DEM still has some drawbacks, especially the removing of off terrain point clouds and subsequently the generation of DEM within these spaces even though the methods are automated. In this paper it was intended to overcome this issue by attempting to project off terrain point clouds to the terrain in forest areas using Artificial Neural Networks (ANN) instead of removing them and then filling gaps by interpolation. Five sites were tested and accuracies assessed. They all give almost the same results. In conclusion, the ANN has ability to obtain the DEM by projecting the DSM point clouds and greater accuracies of DEMs were obtained. If the size of the hollow areas resulting from the removal of DSM point clouds are larger the accuracies are reduced.