The effects of bed roughness, isolated from those of bed permeability, on the vertical transport processes across the sediment–water interface (SWI) are not well understood. We compare the statistics and structure of the mean flow and turbulence in open-channel flows with a friction Reynolds number of 395 and a permeability Reynolds number of 2.6 over sediments with either regular or random grain packing at the SWI. The regular sediment interface is formed by cubic packing of spheres aligned with the mean-velocity direction. It is shown that, even in the absence of any bedform, the subtle details of the particle roughness alone can significantly affect the dynamics of turbulence and the time-mean flow. Such effects translate to large differences in penetration depths, apparent permeabilities, vertical mass fluxes and subsurface flow paths of passive scalars. The less organized distribution of mean recirculation regions near the interface with a random packing leads to a more isotropic form-induced stress tensor. The augmented wall-normal form-induced fluctuations play a significant role in increasing mixing and wall-normal mass and momentum exchange.

We use a three-dimensional, unstructured grid hydrodynamic model to examine circulation and exchange in the Saginaw Bay-Lake Huron system during the summer months for three consecutive years (2009–2011). The model was tested against ADCP observations of currents, data from a Lagrangian drifter experiment in the Saginaw Bay, and temperature data from the National Data Buoy Center stations. Mean circulation was predominantly cyclonic in the main basin of Lake Huron with current speeds in the surface layer being highest in August. Circulation in the Saginaw Bay was characterized by the presence of an anticyclonic gyre at the mouth of the outer bay and two recirculating cells within the inner bay. New estimates are provided for the mean flushing times (computed as the volume of the bay divided by the rate of inflow) and residence times (computed as e-folding flushing times based on dye concentration modeling treating the bay as a continuously stirred tank reactor) for Saginaw Bay. The average flushing time (over the 3 months of summer and for all 3 years) was 23.0 days for the inner bay and 9.9 days for the entire bay. The mean e-folding flushing time was 62 days (2 months) for the inner bay and 115 days (3.7 months) for the entire bay for the summer conditions examined in this work. To characterize the behavior of river plumes in the inner Saginaw Bay, trajectory data from GPS-enabled Lagrangian drifters were used to compute the absolute diffusivity values in the alongshore and cross-shore directions.