In a large eddy simulation of the atmospheric boundary layer, results near the surface suffer from major deficiencies of the subgrid-scale (SGS) model due to inherently insufficient resolution there. However, efforts to develop improved models have been partially hampered by the lack of experimental data. In this paper SGS stress is studied experimentally using two-dimensional velocity fields measured with a newly developed array technique that combines a sonic array in the lateral direction with Taylor’s hypothesis in the streamwise direction. Detailed analyses show that, under convective conditions, the subgrid velocities are statistically independent of the resolvable-scale horizontal velocities; thus the large-scale eddies advect the SGS eddies but do not directly interact with them. As a result, the SGS stress components that involve resolvable-scale horizontal velocity components have strong statistical dependence on these velocity components. These SGS stress components also have the largest variances. Other SGS stress components involve only velocities whose length scales are comparable to the filter scale and are statistically dependent on the resolvable-scale vertical velocity but are statistically independent of the resolvable-scale horizontal velocities. Furthermore, these two types of SGS stress terms are independent of each other. The present study suggests that the two types of SGS stress components are related to the dynamics at the largest scales and at the filter scale, respectively, and need to be modeled separately to capture their distinct statistical characteristics.
Corresponding author address: Dr. Chenning Tong, Department of Meteorology, The Pennsylvania State University, University Park, PA 16802-5013.