Abstract
The onset of sediment motion in rivers is important for predictions of river stability, and for the design of hydraulic structures and river restoration projects. Considerable uncertainties in calculations and measurements of the onset of motion exist. Most calculations of the onset of sediment motion do not explicitly include turbulence effects but recent studies have suggested that impulse, the product of the duration and magnitude of drag forces that are greater than a critical value, is likely to cause grain movement. We explore if sediment motion can be systematically explained with instantaneous drag forces and impulses. In a series of 26 flume experiments, we measured instantaneous pressures and velocities on a mobile test grain for which the precise timing of motion was known. We used these measurements to calculate drag forces and impulses in a range of possible ways. Impulse and drag forces were concluded to cause particle motion in a given experiment if their highest measured values occurred during grain motion rather than during any time when the test grain was stable. Use of the measured upstream velocity profile instead of a single point velocity provided calculated drag forces and impulses that better corresponded to the onset of particle motion. The correlations of drag forces and impulses with particle motion were also greatly dependent on the selected drag coefficient, implying that field applications of impulse may need to consider the effects of grain shape and orientation instead of simply assuming spherical particles. Out of all the various drag force and impulse parameters we tested, an impulse that incorporated a decreasing resisting force during particle rotation out of its pocket explained the greatest percentage (88%) of observed grain motions. The start of grain rotation could not be explained by impulse for 12% and 17% of particle motions when we used velocity and pressure data, respectively, to calculate impulse. This suggests that either the onset of particle motion may be sometimes driven by another flow parameter, or that typically measured velocity and pressure data used to calculate impulse may not adequately capture the spatial variation in flow structure around a grain. A temporally variable drag coefficient could in theory indirectly account for some of these spatial and temporal variations in grain-scale flow. Use of a temporally variable drag coefficient did not improve the performance of impulse in explaining particle motion, implying that a more complex flow parameter that accounts for spatial flow patterns may sometimes be needed. Understanding how sediment fluxes in rivers are related to applied shear stresses is imperative for improving restoration efforts or minimizing loss of property though urbanized reaches. Bedload equations often predict inaccurate sediment fluxes partly because of uncertainty in the shear stresses that cause the start of sediment motion (critical Shields stresses). Although often assumed to be a constant value, the critical Shields stress can increase with greater channel slope, which has been potentially explained by a wide variety of processes. To fully understand this phenomenon, we conducted a series of flume experiments through a range of slopes in which we measured the critical Shields stress and near-bed flow velocity at the onset of motion of a mobile test grain with a fixed pocket geometry. We used two bed configurations, one with and one without large immobile grains, to explore the effects of large boulders on critical Shields stresses. Contrary to previous studies that have shown a general increase in critical Shields stress with greater channel slope or relative roughness, the critical Shields stress in our experiments only increased when (1) boulders were added, and (2) the boulder tops began to emerge from the flow. Otherwise, critical Shield stresses remained roughly constant with greater slope or relative roughness. We tested many of the previously hypothesized reasons for critical Shields stress increases with slope and found that none could fully explain our experimental observations. We hypothesize that in our data and in natural rives, many of the observed changes in critical Shields stress are caused by decreases in boulder submergence and increases in boulder concentration with higher slope. These changes in boulder properties drive previously unaccounted for complex variations in the flow structure that affect the onset of motion of finer, more mobile particles. The critical Shields stress can also vary between grain sizes, and can be predicted with hiding functions, which describe how grain mobility is affected by the underlying grain size distribution, such that the ratio of the grain size (Di) to the median bed grain size (D50) determines the critical Shields stress for the ith grain size. A patch is defined as an area of the bed that is occupied by grains of distinct size distribution, where its boundary is defined by a clear change in grain size distribution indicating the neighboring patch. The relative mobility of grains throughout a reach have been studied, but the effect of local variation of grain sizes between patches on grain relative mobility is largely unknown. We explore the effects of patch-scale grain size variability on the degree of sediment mobility by developing hiding functions for different patch types within the Erlenbach torrent (Brunni, Switzerland). To determine hiding functions for each patch type, we used: (i) the D84 of the mobile tracer grain size distribution from each patch type for 10 storm events (discharges of 0.17 to 2.1 cms), by monitoring the movement of painted and RFID tagged tracer grains from the most prevalent patch types, and (ii) the median local shear stress on each patch type for each discharge modeled using the quasi-3D FaSTMECH model. We also measured in-situ protrusions (vertical distance a grain extends relative to near-by grains upstream) and calculated friction angles (the angle a grain must rotate through for mobilization) for all grain sizes present on each patch type, which are local grain-scale parameters that can lead to the size-selective entrainment that hiding functions often describe. We observed protrusion to be greater for larger grains, but for the same grain size (Di) protrusion was higher on finer patches. However, all patches have about the same relation between relative grain size (Di/D50) and dimensionless values of protrusion (protrusion/grain height) and friction angle.