Impact of Local Terrain Features on Urban Airflow

Past work has shown that coupling can exist between atmospheric air flows at street scale (O(0.1 km)) and city scale (O(10 km)). It is generally impractical at present to develop high-fidelity urban simulations capable of capturing such effects. This limitation imposes a need to develop better parameterisations for meso-scale models but an information gap exists in that past work has generally focused on simplified urban geometries and assumed the buildings to be on flat ground. This study aimed to begin to address this gap in a systematic way by using the large eddy simulation method with synthetic turbulence inflow boundary conditions to simulate atmospheric air flows over the University of Southampton campus. Both flat and realistic terrains were simulated, including significant local terrain features, such as two valleys with a width about 50 m and a depth about average building height, and a step change of urban roughness height. The numerical data were processed to obtain averaged vertical profiles of time-averaged velocities and second order turbulence statistics. The flat terrain simulation was validated against high resolution particle image velocimetry data, and the impact of uncertainty in defining the turbulence intensity in the synthetic inflow method was assessed. The ratio between realistic and flat terrains of time-mean streamwise velocity at the same ground level height over a terrain crest location can be >2, while over a valley trough it can be <0.5. Further data analysis conclusively showed that the realistic terrain can have a considerable effect on global quantities, such as the depth of the spanwise-averaged internal boundary layer and spatially-averaged turbulent kinetic energy. These highlight the potential impact that local terrain features (O(0.1 km)) may have on near-field dispersion and the urban micro-climate.