Estimating Floodplain Vegetative Roughness using Drone-Based Laser Scanning and Structure from Motion Photogrammetry

Abstract:

High-resolution drone laser scanning (DLS) and structure from motion (SfM) photogrammetry-derived vegetation heights were compared at the Virginia Tech StREAM Lab to determine Manning’s roughness coefficient. Two forms of a calibrated approach and a calculated approach were utilized to estimate roughness in raster form from the two data sets (DLS & SfM photogrammetry). These rasters were then inputted into a two-dimensional (2D) hydrodynamic model (HEC-RAS). A calculated approach considers plant characteristics to determine vegetative roughness, while a calibrated approach adjusts roughness values by using field data, from a velocity probe in the floodplain, to determine roughness. Model simulations were compared to seven actual high-flow events during the fall of 2018 and 2019 using measured field data (velocity sensors, groundwater well heights, marked flood extents). All models were not significantly different to water surface elevations from our 18 wells in the floodplain (p > 0.05). There was a decrease in RMSE (-0.02 m) when comparing calculated and calibrated models. A decrease in RMSE for DLS compared to SfM (-0.01 m) was also observed. This increase might not justify the increased cost of a DLS setup over SfM (~150,000 vs ~2,000 USD), though future studies are needed. These results will help improve hydrodynamic modeling efforts, which are becoming increasingly important for management and planning in response to climate change, specifically in regions were high flow events are increasing.