Comparison of terrestrial lidar, SfM, and MBES resolution and accuracy for geomorphic analyses in physical systems that experience subaerial and subaqueous conditions

Authors: ROWLEY, TAYLOR - Louisiana State University; Ursic, Michael - Mick; KONSOER, KOREY - Louisiana State University; Langendoen, Eddy; MUTSCHLER, MIKE - Seahorse Geomatics Inc; SAMPEY, JOSHUA - Seahorse Geomatics Inc; POCWIARDOWSKI, PAWEL - Norbit Subsea

Submitted to: Geomorphology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/23/2020
Publication Date: 4/1/2020
Citation: Rowley, T., Ursic, M.E., Konsoer, K.M., Langendoen, E.J., Mutschler, M., Sampey, J., Pocwiardowski, P. 2020. Comparison of terrestrial lidar, SfM, and MBES resolution and accuracy for geomorphic analyses in physical systems that experience subaerial and subaqueous conditions. Geomorphology. 355: 107056. https://doi.org/10.1016/j.geomorph.2020.107056.
DOI: https://doi.org/10.1016/j.geomorph.2020.107056

Interpretive Summary: The magnitude of forces eroding the boundary of a river strongly depends on local roughness or geomorphic features such as topographic variations of various spatial length scales. Recent advances in lower-cost subaqueous and subaerial remote sensing techniques have made high-resolution, centimeter-scale topographic datasets more available to geoscientists. Combined data sets collected using both subaqueous and subaerial techniques are often combined for subsequent analysis, but little is known about the potential accuracy and resolution discrepancies of geomorphic features remotely sensed with different techniques. ARS scientists in Oxford, MS, in collaboration with researchers from Louisiana State University and industry compared the accuracy of subaerial techniques, terrestrial lidar and structure-from-motion (SfM), and a subaerial technique, multibeam echo sounding (MBES), for a variety of topographic surfaces: dune, hummocks, cobble, and gravel. After thorough assessment of each method, lidar has lowest root-mean-square error for overall surface detection. SfM has less than 1 cm error, and on average the MBES surfaces have 1.96 cm error. As a result, lidar and MBES field datasets can be combined and compared to detect both geomorphic features and their temporal adjustment, however caution must be used and consideration of the features of interest must be taken into account.

Technical Abstract: This study provides a systematic comparison of lidar, multibeam echo sounding (MBES), and structure-from-motion (SfM) derived datasets collected over a fixed surface contained within a large outdoor flume converted to a 15 x 2.5 x 1.5 m tank. The surface incorporated various morphological features and structures for measurement technique comparison over varying surface type and geometries. Two lidar scans and two photograph sets (used in SfM) capture the subaerial surface within the experimental tank before and after the tank was filled for MBES measurements. MBES surveys at varied frequencies (200 and 400 kHz), swath widths (90 and 150°), and phase/amplitude bottom detection mode (0 and 84%) settings were conducted along longitudinal transects. Postprocessing of the data delivers three-dimensional point clouds that are georeferenced via RTK-GNSS measurements of control points within the experimental tank. All datasets are compared to the first lidar dataset (lidar1), herein defined as the reference surface. Root mean square error (RMSE) is calculated for each dataset resulting in 0.3 - 3.37 cm error between the three systems (under the experimental conditions). Datasets are then detrended to determine how each instrument detects individual geomorphic features at various resolutions. The wide swath width of the MBES generated the largest difference between surfaces, which may be attributed to echoing from the cement tank walls. Error found between techniques suggest that caution should be used when resolving textures less than 3.5 cm with MBES, but is dependent on distance and point spacing during the survey.