A-priori laboratory study of subfilter-scale models for LES over heterogeneous land surfaces

Our understanding of subfilter-scale (SFS) physics in large-eddy simulations (LESs) of the atmospheric boundary layer must be improved to allow for more reliable simulations, especially over realistic surface conditions such as heterogeneous, rough terrain. The largest errors associated with LES occur due to its limited ability to parameterize the subfilter-scale fluxes that account for the effect of the unresolved (subfilter) scales on the resolved scales. The goal of this study is to validate and test current subfilter-scale models in a boundary layer over heterogeneous surface conditions using a unique, high-resolution data set that was acquired in a wind tunnel using stereoscopic particle image velocimetry (PIV). The data set consists of both horizontal (at z/δ = 0.05) and vertical velocity fields at three positions downstream (x/δ = 0.5, 1.5 and 2.5) of a rough-to-smooth surface transition. This data set is ideally suited for a priori evaluation of subfilter-scale variables. Two-dimensional spatial filtering is applied (without using Taylor's hypothesis) to evaluate the filtered velocity field, filtered strain rate tensor and SFS stress tensor. These subfilter-scale quantities are used to study the SFS transfer rate of energy, which is found to exhibit highly-intermittent behavior that includes occurrences of both positive (forward scatter) and negative (backscatter) values. Our results show that the intermittency of this SFS transfer rate of energy decreases with downstream distance away from the surface-roughness transition as does the turbulent kinetic energy of the flow. We perform a priori testing of various SFS models and find, among other results, a low correlation between the measured SFS stress and modeled SFS stress using an eddy-viscosity model. The results of this study provide a better understanding of current SFS models and provide guidance toward the development of improved SFS parameterizations over heterogeneous land surfaces.