Handout (6.3 MB)
The experiments were performed in the atmospheric wind tunnel of the LHEAA with a terrain model in which the height of the obstacles used to represent the urban and the vegetation canopy is the same. The urban terrain was modelled by an idealized array of cubes arranged in a staggered manner with a plan area density of 25%, over a fetch of 360h where h = 50mm is the height of the cubes. This setup has been shown to enable the generation of a well-developed turbulent boundary layer in fully-rough regime, representative of the flow over an urban canopy (Rivet et al., 2012, Perret & Rivet, 2013). The vegetation canopy was modelled using an array of cylinders of height h and of diameter dr = 4mm, arranged to obtain a canopy density of n = 980 cylinders/m2 and a frontal area per unit volume a = n.dr =3.92m -1. The frontal area index of this staggered arrangement is λ = 0.39. Perret & Ruiz (2013) showed that the flow generated over this obstacle arrangement was representative of that in a vegetation canopy. The difference of terrain nature is characterized by a moderate value of the parameter M =log ( z01/z02)=-0.7 . This increase in roughness length is accompanied by a small decrease (20%) of the displacement height. The flow downstream of the roughness change was investigated using two-component Particle Image Velocimetry (PIV) performed in a vertical plane aligned with the main flow over a longitudinal distance of 18h. Three sets of 4000 vector fields (of dimensions 6h×4h ) were recorded successively to cover the investigated region and enable the computation of one- and two-point statistics above the canopy (Dong, 2012).
In the proposed study, the roughness change is characterized by a rapid adjustment of the mean flow confined to the 5h-long region just downstream the transition location where the flow engulfs into the vegetation canopy. The effect of the transition on the mean flow above the canopy has been found to be very weak, which did not allow the detection of an internal boundary-layer. However, the terrain change is found to strongly affect the characteristics of the turbulence such as the Reynolds stresses, the skewness and flatness of both the longitudinal and the vertical velocity components. Based on quadrant analysis, the number of ejection and sweep events and their relative contribution to the total shear-stress are influenced in the region between 0 and 5h downstream the terrain transition. Analysis of the two-point correlation coefficients of the velocity shows a rapid adjustment of the large-scale structures of the flow, the shape of the correlation becoming independent of the longitudinal location (whereas it depends on the relative distance to the obstacles in the case of the urban terrain). The associated integral length-scales are found to be influenced by the transition and roughly follow the growth of the internal boundary-layer that can be defined using the Reynold shear-stress only when points close to canopy are considered. It is thus demonstrated that, the 5h-long region downstream the terrain change, characterized by an engulfment of the flow into the vegetation canopy, is where most of the characteristics of the flow are affected and that the flow starts redeveloping downstream.