Characterizing riverbed morphology and roughness: computational workflow integrating UAV-based 3D mapping and flow modelling

Understanding the hydrodynamics of river channels and predicting their evolution within the landscape relies on a thorough examination of flow resistance. Flow resistance dictates the distribution of shear stress along channel boundaries, as well as patterns and intensity of erosion on riverbeds and banks. In steep mountain streams, characterized by high relative roughness, flow resistance is predominantly influenced by form drag. Mountain streams often exhibit complex bed morphologies, due to the presence of large immobile boulders, sediment aggregates, and unique channel configurations. Such structures create intricate three-dimensional flow patterns that modify lift and drag forces on sediment, which in turn impact water velocities and sediment transport within the stream. In order to improve hydrodynamic models and better describe bed morphology and roughness, we are exploring new methods for determining bed roughness as well as developing new metrics capable of accounting for clustering, directionality, and water-sediment interactions, while distinguishing between sediment-driven and bedrock-imposed morphological features.

We used an uncrewed aerial vehicle (UAV)-based structure-from-motion (SfM) analysis to determine centimeter-level accurate measurements of 3D river channel morphology, an essential step in studying channel geometry and bed roughness. Utilizing these advancements, we are developing a fully automated workflow to characterize the geometry of natural river reaches, including shape metrics, hydraulic geometry estimates, grain size distribution and various roughness parameters. A standardized workflow to quantify bed roughness across diverse channels will facilitate the development of comprehensive databases, enable comparisons between field sites, and promote the application of measured roughness parameters in flow resistance equations. We analyze hundreds of meters of river channels from Taiwan and New Zealand and present the results obtained through our workflow. The workflow consists of processing raw drone footage into high-resolution 3D models, applying both custom and third-party code to extract and analyze roughness parameters, and conducting two-dimensional depth-averaged flow modelling to derive additional hydrodynamic parameters.