Comparing Remote Topographic Measurements of a Coastal Beach/Dune/Marsh Complex

Accurate coastal maps are necessary for many applications including topographic and landcover change analysis, habitat monitoring, restoration, engineering, and emergency response. Lidar, in particular the topographic/bathymetric system operated by the Joint Airborne Lidar Bathymetry Technical Center of Expertise (JALBTCX), is the operational standard for large-scale coastal mapping. Structure from motion (SfM) provides an alternative technique for generating topography using aerial photos and ground control points (GCPs) with known locations. In this work, we compare SfM photogrammetry with JALBTCX lidar data acquired on the same day in the dunes and wetlands of the Sandy Neck / Great Marshes region of Cape Cod, Massachusetts. The accuracy and precision of both the SfM and JALBTCX datasets were evaluated by comparing them with 247 transect points and 27 GCPs that were surveyed in the study area. Vertical elevations of the SfM digital elevation model (DEM) were biased by an average of 0.025 m with root mean squared error (RMSE) of 0.096 m compared to the transect points, whereas the lidar DEM elevations were biased by 0.080 m with the same RMSE. The accuracy and precision of lidar elevations depended on the landcover: the most common landcover (dune sands) had an RMSE of 0.079 m, and the greatest errors (RMSE=0.157 m) occurred at locations with marsh vegetation. The horizontal difference of target locations (compared with survey measurements) in the JALBTCX orthomosaic was 0.110 m (RMSE=0.123 m). The vertical difference in target locations from the lidar DEM was smaller, with a bias of only 0.086 m (RMSE=0.099 m). The mean elevation difference at the transect points between the lidar and SfM was 0.054 m (RMSE=0.095 m). This level of accuracy in both the lidar and SfM falls within JALBTCX standards of +/-0.196 m, indicating that SfM provides a timely, affordable way to map small regions of interest with great accuracy.