Open cbegeman opened 4 years ago
Since log law of the wall holds in a spatially-averaged sense, Bou-Zeid et al. (2015) filter the resolved velocity near the wall by 2 times the grid size. The current version of the code has some averaging but perhaps not enough: https://github.com/xylar/palm_les_lanl/blob/c293074463bb0bda5a6796182b56793a1d163d91/trunk/SOURCE/surface_layer_fluxes_mod.f90#L1054-L1055 (these terms are used to define the surface stress). What do you think about averaging u(i,j,k),u(i+1,j,k),u(i,j+1,k),u(i+1,j+1,k) for the surface stress calculation instead (and likewise for v)? It's probably a minor effect but I don't think it would hurt to implement.
u(i,j,k),u(i+1,j,k),u(i,j+1,k),u(i+1,j+1,k)
I agree, more spatial averaging would likely make some sense. I think the averaging that you point to is just to move from the respective edges to cell centers, so it's not really intended for smoothing purposes.
Since log law of the wall holds in a spatially-averaged sense, Bou-Zeid et al. (2015) filter the resolved velocity near the wall by 2 times the grid size. The current version of the code has some averaging but perhaps not enough: https://github.com/xylar/palm_les_lanl/blob/c293074463bb0bda5a6796182b56793a1d163d91/trunk/SOURCE/surface_layer_fluxes_mod.f90#L1054-L1055 (these terms are used to define the surface stress). What do you think about averaging
u(i,j,k),u(i+1,j,k),u(i,j+1,k),u(i+1,j+1,k)
for the surface stress calculation instead (and likewise for v)? It's probably a minor effect but I don't think it would hurt to implement.