[Mr. Matti]

Hello,

i have got a problem concerning the order of advection scheme near the surface or wall. The momentum equations are discretized by a 5th order finite difference scheme. Switching the vertical advection to a scheme of lower order (3rd order or 2nd) with a smaller stencil near the surface (to avoid additional ghost layers) yields to unrealistic high velocities near the bottom or rather in the layer where the switching occurs. It looks like discontinuity in that levels.

Is this a typical behavior of finite difference schemes applied to the momentum equations or is it maybe model-dependent? This problem concerning a LES code (maybe it helps).

Has anybody experience with this or can give me some advice to handle this?

Thanks