Test of chemistry boundary conditions large-eddy simulations in urban areas

Large-Eddy Simulation (LES) allow to simulate pollutant dispersion at a fine-scale turbulence-resolving scale with explicitly resolved turbulent transport around building structures and in street canyons. The microscale urban climate model with atmospheric chemistry PALM-4U (i.e. PALM for Urban applications; Maronga et al., 2019, Met. Z., https://doi.org/10.1127/metz/2019/0909) has been developed within the collaborative project MOSAIK (Model-based city planning and application in climate change). With such a large-eddy simulation (LES) model, pollutant dispersion around buildings and within street canyons can be simulated, with explicitly resolving the turbulent transport in urban environments.

Cyclic boundaries are frequently applied in LES in order to obtain lateral boundary conditions for the turbulent quantities. In addition to the default cyclic boundary conditions, PALM-4U allows also time-dependent boundary conditions from regional models to account for variable weather conditions and regional scale pollutant transport. Turbulent fluctuations, which are not included in the boundary conditions from the regional simulation but are needed as additional boundary conditions for the LES model are produced by a turbulence generator (Maronga et al, 2019, GMDD, https://doi.org/10.5194/gmd-2019-103).

PALM-4U simulations with and without time dependent boundary conditions from regional simulations with WRF-Chem are performed for different setups in order to test the impact of the domain configuration. The simulations indicate that cyclic boundary conditions can lead to unrealistic accumulation of pollutants over urban areas with strong sources, which is not the case when time-dependent boundary conditions are applied. However, even though a turbulence generator is applied, explicit setting of time-dependent boundary conditions requires large model domains, in order to obtain fully developed turbulence within the domain of interest, increasing the computational demand of the simulation.