Assessment of Vertical Mesh Refinement in Concurrently Nested Large-Eddy Simulations Using the Weather Research and Forecasting Model

Abstract

o facilitate multiscale simulation using the Weather Research and Forecasting model, vertical mesh refinement for one-way concurrent nested simulation was recently introduced. Grid refinement in the vertical dimension removes the requirement of different grid aspect ratios on the bounding versus the nested domain, such that results from refinement in the horizontal directions only, and thereby can also reduce numerical errors on the bounding domain arising from large aspect ratios in the presence of complex terrain. Herein, the impacts of vertical grid refinement on the evolving downstream flow in nested large-eddy simulations are evaluated in relation to other model configuration choices, including turbulence subfilter-scale (SFS) stress models, mesh configuration, and alternative methods for calculating several near-surface flow parameters. Although vertical nesting requires coarsening of the vertical grid on the bounding domain, leading to a smaller range of resolved turbulence scales in the nest?s lateral boundary conditions, parameter values within the nested domains are generally only minimally impacted, relative to nesting using the same vertical grid on each domain. Two dynamic SFS models examined herein generally improved simulated mean wind speed, turbulence kinetic energy, stresses and spectra, on both domains, and accelerate equilibration rates within nested domains, relative to two constant coefficient models. A new method to extrapolate horizontal velocity components to near-surface locations at nested domain lateral boundaries, and a correction to the calculation of deformation elements near the surface, are each shown to slightly alter mean parameter values, yet only minimally impact equilibration rates within the nested domain.