Vertical velocity distributions in sand-bed alluvial rivers

An experimental study of flows over bedforms was conducted in two alluvial river systems; the Kankakee River and the Missouri River with the purposes of (1) evaluating the applicability of laboratory scale results to field-scale problems, (2) characterizing the spatially averaged (longitudinal) mean velocity profile to evaluate available velocity-distribution models, and (3) characterizing the spatially averaged Reynolds stress distribution and evaluating available shear-partitioning models. Three separate data sets were collected (two from the Missouri and one from the Kankakee Rivers) with detailed velocity data measured at numerous locations along bedforms.; The large-river-scale data reflect similar flow characteristics to those of laboratory-scale flows. Turbulence production may be more appreciable in the outer region of a large river flow than for flume flows. In general, the Reynolds stress distributions at laboratory- and field-scales are similar, with the both flow scales having the highest shear stresses in the zones of flow separation and along the shear layer.; The velocity-distribution models of Smith and McLean (1977) and Nelson and Smith (1989A) were evaluated using the field data from this study, as well as the field data of Smith and McLean (1977) and Kostascuk and Villard (1996). In an effort to provide a simple field approach for estimating the reachwise spatially averaged vertical-velocity profile, the standard velocity-defect model also was also evaluated, accounting for the wake effect through knowledge of the bulk Richardson number. The velocity-defect model mean flow velocities were within 2% of the measured values and estimated spatially averaged point velocities were within 10% for z/H > 0.1.; Shear partition models of Einstein and Nelson-Smith were evaluated with the field data collected in this research. Einstein's method had an absolute average percent error of 11.3%, while the Nelson-Smith method had an absolute average percent error of 22.2% when the smaller bedforms were used as the geometry in the Nelson-Smith method.