SwayLES: a coupled large-eddy simulation tree-sway model

A tree sway model is coupled with a high resolution (2m x 2m x 2m) 3-D large-eddy simulation (LES) of the atmospheric boundary layer (ABL) in a numerical domain of (1 km, 1 km, 1.536 km) to explicitly solve for the 2-D sway motions of 62,500 individual trees (625 stems per hectare) driven by turbulent gusts in the canopy roughness sublayer (CRSL). The coupled model was applied to simulate a sparse (vegetation area index VAI = 2) and a dense (VAI = 6.5) forest with mean canopy heights of 20 m. Many simulated tree sway characteristics resemble those measured in a large-scale field experiment at an AmeriFlux forest site in Howland, Maine, USA. These characteristics include spectral peaks at the natural vibration frequencies, complex looping feature in 2-D sway displacements, asymmetry in the probability density function (PDF) with large positive skewness of the streamwise displacements and quasi-symmetry in the PDFs with small skewness of the crosswind displacements and of sway velocities in both the streamwise and crosswind directions. The spatial patterns of two-point correlations of simulated tree sway displacement closely match those of simulated turbulent velocity at the canopy top, both of which indicate that turbulent integral length scales are greater over the sparse forest than the dense forest in the streamwise and the crosswind directions. This result demonstrates the potential utility of on-going field measurements of a large array of tree sway displacements for characterizing the horizontal scales of coherent gusts in a real forest. However, the two-point correlations of simulated tree sway velocity significantly differ from those of the simulated turbulent velocity, which agrees with similar simulations of waving alfalfa crops. A somewhat surprising result is the opposite effects of tree-sway on simulated airflow in the roughness sublayer above the forest such that tree-sway enhanced and reduced turbulence over the sparse and dense forests, respectively. However, tree-sway was shown to have little influences on turbulent length scales in the CRSL which is largely influenced by vegetation area density.