Flow and Pollutant dispersion from a line source in 3D urban street canyons having different spatial morphologies

Street canyons are a common feature in many urban environments. They are also arguably the most polluted region of the urban boundary layer due to the presence of heavy traffic and limited penetration of fresh wind for ventilation. Hence a detailed understanding of the processes taking place within the canyon is crucial to ensuring good air quality in urban areas. In this study, we investigate the turbulent flow and pollutant dispersion in four model street canyons having different morphologies, by conducting time-resolved particle image velocimetry (PIV) measurements in a wind tunnel. The street canyons analysed have different aspect ratios (the ratio of the height of the street canyon to its width) and were surrounded by higher or lower buildings, i.e. by street canyons with a higher or lower aspect ratio. First, we introduce an approach for determining pollutant concentrations from PIV data and show that the method can be reliably used to measure the planar pollutant fluxes. Differences in the concentration and flow fields at different planes were observed, thus indicating the importance of considering the three dimensionality of the canyons. The turbulent scalar flux was found to play a dominant role in pollutant transport at the roof level for canyons surrounded by buildings having the same aspect ratio as the canyon. Conversely, advection dominates at roof level when a canyon is surrounded by buildings having different aspect ratios.  Quadrant analysis of the momentum and scalar flux reveals high correlations between ejections and sweeps and ventilation processes at the roof level. We apply the dynamic mode decomposition (DMD) technique, a data-driven algorithm, to extract dynamically relevant coherent structures of the flow and pollutant concentrations from the data. The results of DMD show that roof and ground level coherent structures play a crucial role in the ventilation of the street canyons.