A comparison of surface sampling methods for coarse fluvial sediments

This is the authors' abstract. We don't add key points for this paper.

In order to characterize variability associated with sampling coarse fluvial sediment, surface grain‐size distributions were characterized at eight sample sites within gravel‐ to boulder‐bed channels. Four methods were used: (1) a random walk, (2) a sampling grid spaced at an interval equal to the intermediate diameter of the largest clast in the sampling area, (3) a sampling grid spaced at one half the interval of method 2, and (4) a randomly chosen subsection within which a patch count of most clasts at least partially exposed at the surface was conducted. At each site, six replicates of each method were performed, with each replicate consisting of 100 clasts. Sampling was performed by a single operator at six of the sites and by multiple operators at two of the same sites and at two additional sites. The research objectives were to characterize variability among replicates of a method, among methods, and among operators. Sample sites were divided into relatively well sorted (inclusive graphic standard deviation <1.4) and poorly sorted (≥1.5). The first three methods produce statistically indistinguishable values of D50 and D84 for both well‐sorted and poorly sorted sites when performed by a single operator. Multiple operators may produce statistically different population measures (D50, D84, sample distributions, and variance) using any of the four methods at either type of site. Because of operator‐induced variability, we caution against direct comparisons of grain‐size distributions between channels or within a single channel through time if more than one operator performs the sampling.