Large-Eddy Simulations on the Effects of Drag Force of Trees : A real case study

As acoustic propagation, or pollen dispersion, a lot of applications need large-eddy simulations (LES) to accurately represent the influence of the boundary layer on this phenomena. In LES, the scales between the model and the vegetation are of the same order (~ 1m). A roughness approach, totally valid only if the first model level is above the canopy, is not sufficient to describe the flow above and within the vegetation. Thus a classical drag force formulation has been implanted in the Meso-scale Non Hydrostatic model Meso-NH (Lafore et al., 1998). It consists in adding drag terms, to the momentum and subgrid turbulent kinetic energy equations, driven by the foliage density in the grid mesh.

Firstly, an inter-comparison between Meso-NH simulations and results from Dupont (Dupont, 2008b) has been done. It shows that Meso-NH, in the same manner as ARPS, is able to reproduce the principal characteristics of the turbulent flow over and within homogeneous canopy measured by Su and al. (1998).

Then, three 3D simulations under neutral, stable and unstable atmosphere have been done over a plane-forest-plane pattern to see the sensibility to the drag force approach to these situations. Effectively, as we expect, in stable atmosphere, the turbulence generation due to the wind drag by the trees, has a very important impact on the observed flow near the canopy.

Finally, a real complex terrain with heterogeneous cover (grass, forest, crops) has been studied. Simulations, under stable and unstable atmosphere, have been done in order to compare to results from the experimental acoustic campaign of Lannemezan2005. They were conducted with three grid-nested domains centered around Lannemezan (France) using 50, 10m and 2m horizonals resolutions. The vertical resolution was around 1m. Even if before the implementation of drag force approach, the results obtained for wind and temperature were in good agreement with the experiment, a positive impact of these modification can be observed. This contribution is essential to better represent the acoustic propagation, as it will be carried on with an acoustic propagation model and an example of this application will be shown.

Due to the low computation time of this method, the drag force approach seems necessary to be used in real case every time that the vertical resolution is finer than the vegetation height.