The roughness parameters of shallow open-channel flows over very rough walls were investigated in the laboratory with particle image velocimetry (PIV) for two rough beds (λ_f= [0.2,0.4]) and three roughness-height to water-depth confinement or submergence ratios (α = [0.15,0.24,0.33]). The flow was measured both within the canopy and above without disturbing the flow by gaining complete optical access with specially developed techniques. This allowed the flow to be spatially resolved at the scale of the roughness elements so that the change of scale to ID flow parameters could be evaluated. In particular, the measurements enabled reliable estimates of the double-averaged mean and turbulence profiles to be obtained by minimizing and quantifying the usual errors introduced by limited spatial sampling, as well as temporal sampling. To estimate correctly the total shear stress or frinction velocity, the measurements show that the dispersive stresses are necessary, unlike often assumed. In order to better define and determine the roughness layer height, often introduced into the boundary condition for numerical models, a new methodology based on the measured spatial dispersion is used which takes into account the often unavoidable temporal sampling errors (Florens et al.,2013). The results reveal values well below the usual more ad hoc estimates. The ID double-averaged statistics are then used to investigate the effect of low relative submergence of the roughness elements on the roughness parameters and the logarithmic law. The logarithmic law is shown to persist for submergence ratios at least as high as 0.33, while the roughness sublayer largely extends into it. A dependence of the roughness length on relative submergence is observed, but not for the displacement height.
展开▼