An approach to measuring volumetric surface change with repeat airplane- and drone-derived lidar point clouds of eroding river banks

Repeat lidar precisely records how Earth's surface topography changes over time. However, 3D change between lidar point clouds is difficult to measure on steep, curved surfaces due to inherent directionality in both lidar instrument geometry and data processing methods. In this paper, we use two pairs of lidar point clouds to measure lateral river bank erosion along the ~200 m-wide Trinity River in Texas from September 2023 to February 2024. One pair of point clouds was obtained from an airplane, while the other was obtained with a drone-based system. We classify all four datasets using a custom workflow, which creates dense ground point clouds that improve data coverage on steep slopes and near slope breaks as compared to default classification methods. After classification, we show that drone-derived point clouds have ~2–8x more points on steep surfaces than airplane-derived point clouds. We then compute 3D change volumes between pairs of point clouds. To do this, we use horizontal M3C2 distances to identify points that changed between surveys, then use Poisson surface reconstruction to enclose volumes of the bank surface that eroded. Results are compared to volumes computed from channel edge movement and raster differencing. Each method estimates similar eroded volumes, but our reconstruction approach utilizes the full 3D geometry of the point clouds, showing individual channel bank failures and larger channel bend migration patterns. This approach could also be applied to measure fluxes of material on other laterally changing topographic surfaces, such as beach cliffs and glacier termini.