Characterizing the Surface Grain Size Distribution in a Gravel-Bed River Using UAV Optical Imagery and SfM Photogrammetry

Understanding the sediment grain size distribution in riverbeds is important prior to analyzing sediment transport, riverbed morphology, and ecological habitats. Previous studies have demonstrated the potential of grain size estimation by riverbed surface roughness through the linear relation between local grain size by manual sampling and corresponding percentile roughness by point cloud analysis of optical imagery. However, the applicability of linear relations is inherently limited, as they are typically derived from datasets that cover a relatively narrow range of local grain sizes. This study aims to develop an integrated grain size-roughness relation applicable across a broader spectrum of grain sizes. Investigations were conducted in four mountainous river reaches (R1, R2, R3, and R4) characterized by coarse grains and a broad grain size distribution (D16–D84: 2.3–525 mm) in Taiwan. High-precision 3D point cloud data were generated using UAV-SfM (Unmanned Aerial Vehicle-Structure-from-Motion) techniques for roughness metric calculation. First, the reach-scale linear correlation analysis was conducted between the grain sizes (Di, where i = 16, 25, 50, 75, and 84, respectively) and their corresponding percentile roughness (RHi) based on the data of Reach R1. Subsequently, all paired Di-RHi data on Reach R1 were used to derive an integrated power-law D-RH relation. Moderate to strong correlations (R2 = 0.60–0.94) were observed in linear Di-RHi relations. The integrated D-RH relation provides a reach-scale single, continuous relation, capturing the average tendency of the grain size-roughness correlations in Reach R1. Applicability tests at additional sites showed that it has better accuracy for coarser grains (D50–D84) than for finer grains. Future studies could extend this approach to a broader range of river segments to better delineate its applicability limits.