Flow and turbulence structure across the ripple–dune transition: an experiment under mobile bed conditions

Current knowledge of flow and turbulent processes acting across the sand bed continuum is still unable to unequivocally explain the mechanism(s) by which ripples become dunes. Understanding has been improved by comparative high-resolution studies undertaken over fixed bedforms at different stages in the continuum. However, these studies both ignore the role of mobile sediment and do not examine flow structure during the actual transition from ripples to dunes. The aims of the paper are: (i) to describe flow and turbulence characteristics acting above mobile bedforms at several stages across the transition; and (ii) to compare these data with those arising from experiments over fixed ripples and dunes. Laboratory experiments are presented that examine the turbulence structure across seven distinct stages of the transition from ripples to dunes. Single-point acoustic Doppler velocimeter sampling at three flow heights above a developing mobile boundary was undertaken. Time-averaged statistics and the instantaneous quadrant record reveal distinct changes in flow structure either side of the change from ripples to dunes. Initially, shear-related, high-frequency vortex shedding dominates turbulence production. This increases until two-dimensional (2D) dunes have formed. Thereafter, turbulence intensities and Reynolds stress decline and three-dimensional dunes exhibit values found over 2D ripples. This is the result of shear layer dampening which occurs when the topographically-accelerated downstream velocity increases at a faster rate than flow depth. Activity at reattachment increases due to high velocity fluid imparting high mass and momentum transfer at the bed and/or wake flapping. Suspended sediment may also play a role in turbulence dampening and bed erosion. Ejections dominate over sweeps in terms of event frequency but not magnitude. Strong relationships between inward interactions and sweeps, and ejections and outward interactions, suggest that mass and momentum exchanges are dependent upon activity in all four quadrants. The results contradict the notion present in most physical models that larger bedforms exhibit most shear layer activity. Consequently an improved model for the ripple–dune transition is proposed.