Adjustment Needed for Helley-Smith Bedload Samples Collected at low Transport Rates on Coarse Gravel Beds

River restoration frequently requires measurements of bedload transport as part of a site specific rating curve or as input for model calibration. A Helley-Smith type (HS) sampler is commonly used for these measurements, but it can have biased results. Several studies have shown that HS samplers deployed directly on a coarse gravel bed interact with the bed sediment. The interaction may take the form of perching on top of large particles (which prevents collection of particles moving in close contact with the bed), dislocating and scooping pebbles, creating scour around the sampler, and sucking sediment into the sampler due to a hydraulic efficiency >1. This interaction can lead to an assessment of transport rates that are either too high or too low. By contrast, studies that deployed a HS sampler on a sill report good correspondence with transport rates collected using other devices. This study examined whether elimination of direct bed contact with a HS sampler improves its sampling accuracy. Paired bedload samples were collected (either simultaneously or immediately following each other) with a 3- inch, thin-walled HS sampler in two mountain gravel-bed streams. One set of samples was collected conventionally by placing the sampler for 2 minutes each onto 15 evenly spaced locations across the stream directly on the bed. The other set was collected by placing the sampler for 5 minutes each onto 6 metal ground- plates installed flush with the streambed at about even intervals across the stream, i.e., mimicking placement of a HS sampler onto a sill. The time of ground contact per cross-section (15 x 2 min vs. 6 x 5 min) was the same for both sets of samples. Results from both streams showed that the HS sampler deployed on ground-plates measured smaller transport rates than the one deployed directly on the bed. The difference was most pronounced for the lowest flows and more pronounced for gravel than for sand bedload. At 50% of bankfull flow, gravel transport rates obtained when the sampler was deployed on the bed was 3 times greater than on the ground-plates at one stream and 150 times greater at the other. At near bankfull flow, transport rates from both deployments yielded similar results. The difference in sampling results can be explained as a function of transport rates. Both deployment versions yielded similar results for transport rates larger than 10 g/m/s because bed interactions contribute only a minor amount to the total sample volume. At low transport rates, these interactions dominate the sampling outcome, such that at rates of 1 and 0.1 g/m/s, the HS deployed on the bed yielded transport rates 1 and 3 orders of magnitude above those collected with the HS sampler deployed on ground-plates. Because deployment of the HS on ground-plates eliminates bed interaction and its distorted transport rates, an adjustment factor (correction function) is proposed for transport rates measured with the HS sampler set directly on a coarse gravel bed. A correction of sampling results measured during low transport would enable more accurate calibration of site specific models when only low flow measurements are available.