Quantifying changes in bed state and grain mobility in steep bedrock rivers through repeat UAV photogrammetry surveys

Abstract:

Bedrock rivers are responsible for eroding bedrock and transporting sediment, though have a wide range of channel morphology and sediment grain size even within an individual river. In steep landscapes, hillslope mass wasting can lead to boulder deposits where individual clasts protrude near the level of high-water flows. Quantifying both how these large clasts can be fluvially transported and how they limit transport of finer sediment is difficult without field observations. Here, we use a series of UAV and handheld photogrammetry surveys to characterize channel morphology and sediment distribution over 14 kilometers of bedrock river channel in Taiwan’s Eastern Central Range at 1-5 cm resolution. These surveys were repeated between 2015 and 2020, and capture bed state across a range of discharges including typhoons in 2016 and 2019.

We use a range of techniques to trace sediment dynamics and monitor how each grain size fraction changes between these repeat surveys, including 3D point cloud differencing, tracking individual boulders, mapping cover of different grain size fractions, and monitoring spatiotemporal patterns of topographic roughness. At the patch scale, sediment aggrades and erodes up to 1-2 meters between surveys, though this is not spatially uniform. The magnitude of boulder mobility is tied to peak flow, with boulders up to 5 meters in diameter being fluvially transported in peak typhoon flows. Finer grained sediment (gravel and cobbles) are locally stored behind boulders after major flows, decreasing roughness and boulder protrusion. The subsequent mobilization and reorganization of gravel and cobbles changes the spatial patterns of roughness along the channels. These high-resolution methods allow us to bridge observations of sediment transport from grain scale processes to reach scale patterns and can be used to inform modeling on how coarse sediment distributions evolve over time.