Abstract

Local Elevation Surface Modeling using GPS-Derived Point Clouds
Track: Lidar and 3D Mapping
Authors: John Whitman

Public DEM datasets don't allow high fidelity representation of small extents. GPS measurements are easily made at fine spatial resolution, but have large positional uncertainties.

This study developed an elevation surface from dense, arbitrarily spaced GPS positions exported to ArcGIS as point shapefiles. It maintained horizontal resolution equal to or better than the 1/3-arcsecond National Elevation Dataset. A raster model was selected. The Point to Raster tool set a desired horizontal resolution and reduced elevation uncertainty by averaging elevation values within cells. Low pass filtering using the Filter tool provided additional smoothing. Visualization employed the Contour tool.

Local surface fidelity was validated by replication using contours derived from GPS positions collected on multiple different days. The surface was judged locally superior to the NED dataset.

The methods described required extensive GPS data collection. In areas of heavy forest canopy and terrain blockage, contour spacing of 3 meters was found justifiable.