Stähly, SeverinFriederich, HeideDetert, Martin2017-05-312017-05-312017-05-31201710.1061/(ASCE)HY.1943-7900.0001286https://infoscience.epfl.ch/handle/20.500.14299/138048WOS:000399665000005The comparability of grain sizes emerging from different methods are discussed, including image-based grain-size analysis. Waterworked gravel-bed surfaces from laboratory and field experiments are analyzed in detail. Grain sizes estimated using freely available object-detection software are compared with grains measured with calipers by hand. On the basis of laboratory and field data, the pebble dimensions determined by square-hole sieving are demonstrated to underestimate real pebble dimensions by a factor of 0.83–0.86, and pebble dimensions derived from images underestimate the pebble-count measurements by a similar amount. Thus, for the present extensive data set, the software-detected grain sizes can be directly compared with grain sizes gained by square-hole sieving, as they are of the same order of magnitude. These results support future wide-spread use of image-based sieving for grain-size distribution analyses for both hydraulic research and engineering.Field measurementGrain-size distributionGravel bedImage processingObject detection.text::journal::journal article::research article