Modelling chemistry and transport in urban street canyons: Comparing offline multi-box models with large-eddy simulation

Abstract Computational fluid dynamics models are resource-intensive, particularly when complex chemical schemes are implemented, and this computational expense limits their use in sensitivity analyses. We propose a flexible multi-box model that permits spatial disaggregation of sources and depositions to simulate the transportations and distributions of chemical species in street canyons with any aspect ratios for which a large eddy simulation (LES) of the flow exists. The spatial patterns of reactive species in the multi-box simulations are in good agreement with those from the LES, especially for the deep canyon from which air escapes more slowly. The overestimation of the LES simulation worsens somewhat due to segregations when the chemistry of volatile organic compounds (VOCs) is included but the overall pattern is captured in a modelling framework. By reducing computational costs by several orders of magnitude, the multi-box model allows more sensitivity testing than the LES, and is an effective approach to investigate spatial pattern of fast non-linear chemistry or microphysics at the street scale.