Terrestrial Laser Scanning for Measuring Stream Bank Erosion within Legacy Sediments: Data Processing and Analysis Methods

Land clearing for agricultural purposes following European settlement of America resulted in upland erosion rates 50-400 times above long-term geologic rates in much of the North Carolina Piedmont region. A considerable amount of the eroded sediment was subsequently aggraded on floodplains and impounded in the slackwater ponds behind milldams. This trapped "legacy" sediment is commonly mistaken for natural floodplain deposition and has remained largely unrecognized as a potential source of accelerated sediment erosion contributing to modern water quality impairment. In this study, terrestrial laser scanning (TLS) is utilized to monitor stream bank evolution along a reach that has breached a former millpond. Due to the unique surface geometry and orientation of the stream bank, vegetation occlusion, and true 3D structure of the point cloud, a systematic data processing approach is implemented to compute the change in sediment volume between repeat TLS surveys. The processing approach consists of the following four steps: 1) segmentation of the stream bank point cloud; 2) transformation of the point cloud such that the xy plane is parallel to the trend of the bank; 3) filter vegetation by selecting local lowest point within a grid cell; 4) smooth high frequency noise 5) generate bare earth digital elevation model (DEM). From the DEMs, change in volume was quantified for a 13 m x 3.5 m section of the stream bank providing an estimate on erosion rates and slumping between surveys. The major mechanisms for the observed changes are freeze-thaw events and fluvial entrainment. To evaluate the surface evolution between the distinct sedimentary layers (legacy vs non-legacy) that comprise the stream bank, elevation change is modeled as a continuous trivariate function z = f(x,y,t) where x,y is horizontal location, t is time, and z is a first-surface referenced elevation. Hence, z=0 for all x,y at t=0, time of first survey. The filtered, transformed, and first-surface referenced point clouds for each survey are merged into a single point cloud that is then interpolated into a 3D space-time voxel model. The 3D cube captures the relative change of the stream bank surface elevation (z) across the temporal span of the surveys providing a novel representation for visualization and analysis. For example, an isosurface extracted from the voxel representation shows locations with the same elevation change relative to the initial survey. The geometry of the isosurface indicates different mechanisms of bank surface evolution.