Distributed LES-based drag parameterisation for heterogeneous urban neighbourhoods

The form and density of buildings modify the air flow and momentum exchange within cities, and therefore strongly affect local wind, temperature, humidity and pollution. Numerical weather prediction (NWP) models currently do not account for heterogeneity in their surface layer parameterisation. Regional models represent buildings, if at all, based on quantities such as plan and frontal area indices, and parameterise their impact at the lowest grid level, even though buildings can protrude a significant height into the atmospheric boundary layer.

To investigate how to parameterise urban roughness in NWP, we analysed high-resolution building-resolving large eddy simulations (LES) with the uDALES model over a range of heterogeneous urban neighbourhoods. The simulation setups have a similar building density and frontal aspect ratio, but vary in complexity with different building heights, plan areas and street geometries. Using the LES data we developed a parameterisation model that describes the vertical distribution of building drag inside a heterogeneous urban canopy. The drag force exerted on the atmosphere represents the momentum loss in the urban canopy due to buildings and can be incorporated as additional stress term in the momentum equations. The parameterisation represents the spatial heterogeneity effects in a one-dimensional vertical function, and links the building drag force to the heterogeneity of building layouts. A characterisation of the vertical and horizontal heterogeneity of built-up neighbourhoods is used as model input.