Effect on flow structure of sand deposition on a gravel bed: Results from a two-dimensional flume experiment

Particle imaging velocimetry (PIV) was used to quantify the entire two-dimensional flow field over a series of fixed test beds in a laboratory flume. The test beds used in the experiments provide a two-dimensional representation of gravel beds with contrasting roughness. The twodimensional form of the beds, combined with the nonintrusive nature of PIV, allowed the velocity field to be quantified right down to the bed surface. The three beds used were designed to simulate the progressive effect of sand deposition on a gravel bed with median grain size (D50) of 25 mm. Such a situation is common in bimodal beds, but, as compared with gravel and sand cases, little work has been done to investigate the interaction between the variable roughness of a bimodal bed and the flow structure above it. The results demonstrate that as the effective roughness decreases the reverse flow often found in the lee of gravel particles is eliminated. Furthermore, near-bed velocities increase, while shear stress and turbulent kinetic energy decrease. Turbulent properties are also diminished higher up in the profile, although this is not the case for the mean downstream velocity, which remains unaffected except at the bed. Quadrant analysis reveals that Q2 and Q4 events become less frequent around high points as effective roughness decreases. All these responses to the changing bed conditions are most pronounced in areas where the effective roughness height (h) approaches 10 mm or less (h/D50 = 0.4). Such a situation is often found immediately upstream of gravel-sand transitions in natural rivers. It is hypothesized that the hydraulic response outlined here provides a detailed physically based explanation for the gravel-sand transition; the combination of reduced bed shear stress and Q2/Q4 events around gravel particles will result in greater mobility of the sand fraction relative to gravel. These experiments thus appear to demonstrate that there is a fundamental threshold between gravel bed and sand bed states that has not previously been quantified.