Seminar- Accuracy in LES Prediction of the Atmospheric Surface Layer: the SFS Stress Model, the Surface Stress Model, and Numerical Friction

In previous analysis we have presented a theory to address a basic difficulty that has plagued large-eddy simulation of turbulent boundary layers: the inability to predict the law-of-the-wall (LOTW) in mean velocity in the surface layer at high Reynolds numbers. Brasseur & Wei (Phys. Fluids 22) developed a theory that explains the source of the difficulty and a framework within which LES can be designed to rectify the problem. The essential difficulty lies in nonphysical frictional content within the discretized dynamical system and the extent to which that frictional content interferes with the inertial scaling that underlies LOTW. Practical models replace inertial flux with dissipative dynamics, so part of this frictional content enters through the subfilter-scale (SFS) model. However, this spurious content also enters through the surface stress model, the structure of the grid and dissipation within the numerical algorithm. I shall review briefly the essential nature of the difficulty and our proposed framework to overcome the error. I shall then discuss accuracy in context with the role of the SFS and surface stress models and prediction of the von Kármán constant, and the consequences of dissipation in the numerical algorithm. To demonstrate the effects we compare spectral LES with finite volume LES of the atmospheric boundary layer and rough-wall turbulent channel flow.