Conversions of surface grain-size samples collected and recorded using different procedures

Information about the grain-size distribution of the surface layer of sediment exposed on river beds is often critical in studies of fluvial hydraulics, geomorphology and ecology. A variety of sampling and analysis techniques are in common usage which produce grain-size distributions that are not directly comparable. This paper seeks to explore the appropriate conversions between different types of surface grain-size sampling methods. This is particularly timely in the light of increasingly widespread use of automatic and semiautomatic image-based measurement methods, the comparability of which with conventional measurement methods is relatively poorly constrained. For conversions between area-bynumber (paint-and-pick) and grid-by-number (pebble count) samples, the empirically-derived conversion factor (±2.2) was found to be greater than that predicted by the Kellerhals and Bray model (±2), but the errors associated with using the value predicted by the model were small (3.8% in mm). For conversions between areal samples recorded by count and weight, the empirically-derived conversion factor was approximately ±2.9, but the use of the value predicted by the Kellerhals and Bray model (±3) resulted in only small errors (5.2% in mm). Similarly, for conversions between image-based grain-size distributions recorded in area-bynumber and grid-by number form, the emipirically-derived conversion factor was ±1.9, but the using the model value of ±2 resulted in only small errors (4.1% in mm). Although these results are specific to the datasets analysed, the variety of sedimentary conditions included gives us confidence that the results are representative.