Application of LiDAR-derived DEMs and their by-products for geomorphic assessment from stream-reach to watershed scales: Horseshoe Run, WV

The application of Digital Elevation Models (DEMs) for geomorphic assessment at a range of scales has increased over the past two decades. Before the advent of LiDAR, DEMs were generated using topographic maps or aerial photographs. The resulting grid-based DEMs generally had coarse resolutions ranging from 10 m to 30 m. Research has shown these resolutions to be acceptable when investigating large areas, but fail to provide adequate resolution for finer scale geomorphic assessment. LiDAR technology allows the creation of much finer resolution DEMs, ranging from 0.5 m to 6 m, depending on the sensor used. The Canaan Valley Institute created LiDAR point data for the Horseshoe Run watershed, West Virginia, during leaf-off conditions in 2006, to support ongoing stream restoration efforts. The LiDAR data allowed creation of 0.5 m grid-spacing DEM with an estimated vertical accuracy of approximately 15 cm. This high resolution DEM provided an optimal dataset for assessing geomorphic parameters at a variety of spatial scales, and allowed creation of exceptionally detailed shaded relief maps that were an effective tool for mapping mass movements, complex fluvial terrace sequences, and floodplains. Using the 0.5 m LiDAR-derived DEM, 21 stream channel cross-sections and 2 stream-reach longitudinal profiles were extracted, using the interpolate line and profiler tool in ArcMap 9.2, and compared to field based surveys. Bankfull channel width, maximum depth, mean depth and cross-sectional area were calculated for each cross-section for both surveyed and LiDAR-derived data. Channel slopes were calculated for each longitudinal profile for both datasets. The results indicate a systematic underestimation of channel depth and overestimation of channel width. Twenty of 21 LiDAR-derived cross-sectional areas were within 10% relative error of ground-surveyed cross sections. In order to assess the use of LiDAR-based resolution DEMs at a courser scale, the 0.5 m DEM was resampled to 2 m, 5 m, and 10 m resolutions. Longitudinal profiles for Horseshoe Run and its major tributaries were created using the interpolate line and profiler tool in ArcMap 9.2. Each longitudinal profile was then smoothed using a 6th order polynomial. The results indicate that the 2 m DEMs provide the best fit when smoothed using the 6th order polynomial. The 10 m DEM produced longitudinal profiles with significant noise.