The efficiency of depth-integrating samplers in sampling the suspended sand load in gravel bed rivers

Abstract

The depth-integrating sampler approach for measuring the velocity-weighted mean concentration of suspended sand in rivers was compared with the point-sampler approach whereby vertical profiles of velocity and sediment concentration are measured then integrated. The aim was to determine, for sand suspension in gravel bed rivers, the uncertainty induced when the depth-integrating sampler traverses the near-bed zone of high sand concentration too quickly to average-out the dominant fluctuations in sand concentration associated with turbulence. Depth-integrated and point samples, velocity profiles, and turbulence measurements were collected from three gravel bed rivers at various flood stages. For two rivers, the sand and silt-clay fractions of the suspended load were determined, while at the third river four separate sand size fractions and the silt-clay fraction were analysed. The results showed that the agreement at-a-vertical between the two approaches improved exponentially as a function of the shear velocity to fall speed ratio, u ∗ /w , ranging from up to ±70% for u ∗ /w < 4 down to about ±5% for u ∗ /w > 30 . The exponential trend is consistent with diffusion models for suspended sediment vertical distributions which predict that the mixing increases as a function of the Rouse number w/(B ku ∗ ). Thus when the u ∗ /w ratio is close to 1 the suspended sand load is concentrated near the bed and a depth-integrating sampler will only sample it for a small fraction of its total traverse time; conversely, with a large u ∗ /w the sand will be mixed over the flow depth and it will be sampled for the same time by a depth-integrating sampler as by a point sampler. The agreement between the two approaches appeared to saturate at about ±5% at high values of u ∗ /w in response to temporal variations in sediment concentration and flow velocity longer than the sampling time of either sampler, i.e. of the order of 1 min or more.