Coupling fluvial‐hydraulic models to predict gravel transport in spatially variable flows

Abstract

This study investigated spatial-temporal variations of shear stress and bed load transport at three gravel bed river reaches of the Williams Fork River, Colorado. A two-dimensional flow model was used to compute spatial distributions of shear stress (τ) for four discharge levels between one third of bankfull (Q bf ) and Q bf . Results indicate that mean τ values are highly variable among sites. However, the properties of the mean-normalized distributions of τ are similar across sites for all flows. The distributions of τ are then used with a transport function to compute bed load transport rates of individual grain size fractions. Probability distributions of the instantaneous unit-width transport rates, q b , indicate that most of the bed load is transported through small portions of the bed with high τ. The mean-normalized probability distributions of q b are different among sites for all flows except at Q bf , when the distributions overlap. We also find that the grain size distribution (GSD) of the bed load adjusts with discharge to resemble the grain size distribution of the subsurface at Q bf . We extend these results to 13 locations in the basin, using the mean-normalized distributions of shear stress and measured subsurface grain sizes to compute bed load transport rates at Q bf . We found a remarkably similar shape of the q b distribution among sites highlighting the basin-wide balance between flow forces and GSD at Q bf and the potential to predict sediment flux at the watershed scale.

Influence of Boulder Concentration on Turbulence and Sediment Transport in Open‐Channel Flow Over Submerged Boulders

“…However, with the small boulder spacing, the PDF of the near‐bed shear stress no longer follows a normal distribution but has a higher density in the region of τ ml / τ t 0 > 1. This agrees with the field observations of Segura and Pitlick (), who observed that 54–58% of the bed area in the river experiences local shear stress higher than the reach‐averaged shear stress.…”

“…With the aim of taking the spatial variability of the bed shear stress into consideration, equation is further modified by (1) using the local bed shear stress acting on the mobile sediments τ ml instead of the reach‐averaged shear stress τ m and (2) integrating the local bed load transport rate over the bed area of mobile sediments instead of using A m / A t . This modification is similar to those studies incorporating shear stress distributions into sediment transport equations (Monsalve et al, ; Segura & Pitlick, ; Yager & Schmeeckle, ). The FLVB formula is thus modified as in which where the subscript l stands for “local.” Bed load rates from equations , , and are computed and compared to reveal the relative accuracy of the three formulas in predicting bed load transport rates.…”

“…In addition to the difficulty of partitioning the total shear stress into skin friction and form drag, the local skin shear stress is also difficult to measure and/or predict. Segura and Pitlick () and Monsalve et al () used a 2‐D flow model to predict the spatially varying shear stress distribution, based on the (empirical) drag coefficient C d and the depth‐averaged flow velocity. Their results show the ability to reproduce the water surface elevation as compared to the observed data, and they successfully predict the shear stress distribution in a natural river.…”

The influence of a vegetated bar on channel-bend flow dynamics

“…We calibrated channel characteristics (bed roughness specified as Cd and lateral eddy viscosity, LEV) and considered them fixed after calibration (Table 1). We used a constant Cd, an approach that has been shown elsewhere to perform comparably to variable roughness in FaSTMECH (e.g., Segura and Pitlick, 2015). We set Cd to minimize the root mean square error (RMSE) of modelled water surface elevation (WSE) versus WSE measured in the field from 2011-2015, over a range of 5 calibration flows.…”

“…1; see Supplement for more detail). The RMSE ranges obtained through calibration are consistent with values reported in other studies that have used FaSTMECH (e.g., Legleiter et al, 2011;Mueller and Pitlick, 2014;Segura and Pitlick, 2015), providing 10 confidence in model performance. Relaxation coefficients were set to 0.5, 0.3, and 0.1 for ERelax, URelax, and ARelax, respectively, through trial and error.…”