Numerical Modelling of Neutral Boundary-layer Flow Across a Forested Ridge

Forest canopies have been shown to alter the dynamics of flows over complex terrain. Deficiencies have been found when tall canopies are represented in numerical simulations by an increase in roughness length at the surface. Methods of explicitly modelling a forest canopy are not commonly available in community numerical-weather-prediction models. In this work, such a method is applied to the community Weather Research and Forecasting model. Simulations are carried out to replicate a wind-tunnel experiment of neutral boundary-layer flow across a forested ridge. It is shown that features of the flow, such as the separated region on the lee slope of the ridge, are reproduced by the roughness-length or canopy-model methods. Shear at the top of the ridge generates turbulence that spreads vertically as the flow moves downstream in both cases, but is elevated to canopy top where a canopy model is used. The roughness-length approach is shown to suffer several deficiencies, such as an over-prediction of mean wind speeds, a lack of turbulence over flat forested ground, and an insufficient vertical extent of turbulence at all locations of the domain studied. Sensitivity to the horizontal resolution of the simulation is explored. It is found that higher resolution simulations improve reproduction of the mean flow when modelling the canopy explicitly. However, higher resolutions do not provide improvements for the roughness-length case and lead to a reduction in the horizontal extent of the separated region of flow on the lee slope of the ridge.