A localized dynamic model for large-eddy simulation of the neutrally buoyant atmospheric boundary layer
AuthorAnderson, William C.
MetadataShow full item record
The combination of Geostrophic forcing and an ABL height of O(1000 m) in the atmospheric boundary layer (ABL) leads to a Reynolds number (Re) of O(10^8). Re's of this magnitude indicate turbulence with an excessive number of scales of motion, or degrees of freedom. The computational power required for explicit representation of all scales (to the Kolmogorov scale) in such flows is far beyond that which is currently available (even when massively parallel computing facilities are employed); from this, the method of large-eddy simulation has emerged. In this methodology, a filtering operation separates the scales of motion into resolved- (R-) and subgrid-scale (SG-S) motions. The R-S motions are typically large and anisotropic (owing to their interaction with the boundary conditions), whilst the SG-S motions are small. The R-S motions are solved explicitly using the filtered Navier-Stokes (N-S) equations -- SG-S motions are parameterized. Parameterization of the SG-S fluid motions has been, and remains, the topic of a considerable research effort. A new SG-S model is presented, namely the LDTKE model (dynamic point-to-point computation of SG-S turbulent kinetic energy with a 1-equation model). 1-equation refers to a prognostic equation for TKE (Turbulent Kinetic Energy). The model is applied to LES of the neutrally buoyant ABL. Many highly sophisticated dynamic (tuning-free) SG-S models have recently been developed, however we still observe adhoc averagaing/clipping. TKE-based SG-S parameterizations have been extensively used, although often they're based on constant coefficients -- the varaint presented here combines a completely dynamic modeling procedure with point-to-point computations. The only constraints imposed on LDTKE is that the eddy-viscosity may not be negative (attempts to include TKE backscatter in LDTKE caused numerical instabilities).