Accuracy of digital elevation models; under canopy vs. open field

Abstract

Digital Elevation Models have been known as the representations of terrain elevation data captured, processed and released by remote sensing capabilities. They could either be called Digital Surface Models, which include elevation data of the Earth along with the objects later placed by the humans, or Digital Terrain Models, which only reflect the real z values of the bare ground. This is a conundrum for the latter definition since a considerable amount of earth surface is covered with natural and artificial objects, forests being the most noteworthy. Furthermore, the usage of them on areas covered by forests is a popular phenomenon deserving an in depth questioning. For this reason, the measurement of how accurate they are, is necessary on terrains covered by forests vs. on terrains which barely involves vegetation or no vegetation at all. In this study, the elevation data of Shuttle Radar Topography Mission C-band SAR 30 m Global DEM, Shuttle Radar Topography Mission X- band SAR 25 m partial Global DEM, ALOS Phased Array type L-band SAR 12.5 m Global DEM, ALOS World 3D Precise 30 m Global Digital maps and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER v3) 30 m Global DEM were compared to a high density sample of z values generated by Real Time Kinetic Global Positioning System. Results showed that under the forest canopy, the margin of error increased across the elevation data.