Many research and management applications in river science require accurate estimates of one or more percentiles from the grain-size distribution of subsurface bed material. These include the estimation of bedload fluxes using most common formulae, the evaluation of bed material as a source of fine sediment, and the characterization of streambed substrates for spawning success. The standard criteria for collecting subsurface fluvial samples demand large sample masses, which are difficult to obtain in many field situations. Consequently the criteria may be ignored or, where the criteria are followed, substantial time and effort is expended to collect relatively few samples. Grain-size information may therefore be of a low quality, and restricted sampling may leave important spatial and temporal patterns of variation undetected. This study uses a coupled field and laboratory approach to evaluate existing bulk-sampling recommendations for a set of sediment mixtures that are representative of natural gravel beds. Based on an analysis of bias and precision of a suite of grain size percentiles, matrix-supported sediments tend to be characterized more easily than framework-supported sediments for a given range in grain size. Percentiles from bimodal distributions are more difficult to estimate accurately over the full range compared to unimodal distributions, especially near the bimodality gap. In general, most percentiles from most mixtures achieve negligible levels of bias and good precision with sample masses that are substantially less than those currently recommended. In the absence of a fuller understanding of the relations between sampling performance and sediment characteristics it would be imprudent to abandon these criteria. However, where a project requires a large number of samples, a sampling experiment or incremental field sample could prove to be a cost-effective tool for reducing overall effort by establishing specific criteria for the sediments under examination.

You do not currently have access to this article.