Sensitivity Analyses of SwayLES: A Coupled 3-D Large-Eddy Simulation Tree-Sway Model

A coupled 3-D Large-Eddy-Simulation-Tree-Sway-Model (SwayLES) has been developed to study the aerodynamic interactions between tree-sway motions and coherent gusts in the canopy roughness sublayer (CRSL) in and above forest canopies. High-resolution large domain simulations of the fully (two-way) coupled version of SwayLES have been applied to a dense forest (vegetation area index - VAI = 6.5) and a sparse forest (VAI = 2) under two different geostrophic winds (40 m/s and 20 m/s) in neutral stratification. A companion four LES runs, in which the trees do not sway, have also been performed, in order to examine the effects of tree-sway on airflow characteristics in the CRSL and the outer layers of the Atmospheric Boundary Layer (ABL). It is shown that the impact of tree-sway on turbulence in the CRSL is not as significant as that due to the large difference in VAI which is primarily responsible for the difference in turbulent characteristics in the CRSL. However, these simulations require a substantial computational resource and take a long time to carry out on supercomputers at NCAR, which make it impractical to use them to conduct a systematic investigation of the effects of varying tree-sway mechanical properties. Hence, we adopt a one-way approach, in which we use 10 Hz time series of 3-D turbulent wind field of CRSL saved from the four fully-coupled SwayLES runs to drive tree-sway motions. This approach allows us to conduct a large set of sensitivity runs with a range of values of elasticity, damping coefficient and tree mass in a reasonable time on local UNIX workstations. Results of these sensitivity runs on simulated tree-sway characteristics and on aerodynamic drag forces acting on airflows are the focus of this presentation, including a comparison with measured tree-sway with similar natural frequencies and vegetation density. A goal of these simulations is to select representative combinations of tree-sway mechanical properties for the fully-coupled SwayLES runs to examine any potential differences from the one-way approach. Simulations completed to date show that much reduced damping is needed in order to match observed spectral peaks of both tree-sway displacement and tree-sway velocity. On the other hand, the simulations and observations appear to agree that the mass of tree plays a significant role on the characteristics of tree-sway spectra, particularly at frequencies of the dominant turbulent eddies in the CRSL.