Improving Hydraulic Modeling Through High-Resolution Mapping of River Habitat and Geomorphic Features with UAVs, Structure from Motion, and Machine Learning

Abstract Submission: For over a century, the Los Angeles River has faced environmental degradation due to urbanization and flood control. Recent decades have witnessed a surge in public interest in river restoration while maintaining flood protection for nearby communities. The success of ecosystem restoration and management hinges on our understanding of how alterations in flow impact water supply, quality, habitat resilience, and flood risk. Current LA River modeling relies on simplistic trapezoidal cross-section assumptions from construction plans, which may prove insufficient for the more organic shapes outlined in project proposals. Additionally, the costs and time associated with collecting high-quality data present challenges. Our study addresses these needs by advancing efficient mapping and hydraulic modeling of the river: 1) We developed a cost-effective mapping approach using Unmanned Aerial Vehicles (UAVs), Structure from Motion (SfM), and machine learning. Using a compact RGB camera, the UAV captured ultra-high-resolution images (GSD = 1.6 cm) of a 2-mile soft-bottom section of the river. This study was conducted in both 2023 and 2024, allowing us to track changes in vegetation over two years and improve classification accuracy; 2) We created hydraulic models using HEC-RAS software based on UAV-derived data. With HEC-RAS 2D modeling, we conducted detailed two-dimensional flow modeling and derived maximum steady and unsteady flow rates for the channel. We also compared our model simulations with data from a nearby water gauge, Arroyo Seco. The model was calibrated by adjusting Manning’s n coefficients to better match observed water velocities. The results show that flow rate outputs closely aligned with data from the Arroyo Seco water gauge. Our approach provides valuable insights into flood risk assessments and the evolving river channel. These findings could guide future river restoration projects and inform decisions that promote sustainable urban river environments.

Learning Objectives/Expected Outcome (Optional) :