Urban population currently forms the largest percentage of total human population in recorded history. The United Nations reported that 54% of people lived in urban environments in 2015, which is slated to increase to 68% by 2050. Preparing for this massive shift requires anticipating possible health hazards to a metropolitan population, and urban meteorology forms a distinct part of this anticipation. Understanding the impact of large urban cores and central business districts (CBDs) on air quality and urban dispersion will help city planners work towards neighborhoods with effective and safe removal of potentially harmful pollutants, like PM2.5. This requires a framework to understand how pollutants are dispersed in an urban canopy.

Advances in Large Eddy Simulations (LES) in recent years facilitate studying this dispersion in more detail. Entire CBDs can be resolved within the domain of the LES, making it very attractive for urban meteorology. This study aims to utilize LES to quantify scalar plumes in an idealized urban canopy at atmospheric Reynolds numbers. A suite of LES was run over idealized urban geometry (cuboids), featuring both staggered and aligned geometry and identical plan and frontal area fractions. Non-Gaussian plume behavior was found in the near source region (x/H<12), although the urban geometry was found to lose its influence on the plume as distance downstream from the source increased. Evidence of street channeling on the plume moments was also found, namely in the form of excess positive kurtosis (K) values (leptokurtic) with an in-street scalar source. In addition to the plume statistics, the behavior of instantaneous scalar concentration at a point (relevant for an individual’s exposure to pollutants) is of great interest. Probability distribution functions (PDFs) of scalar concentration, joint PDFs of concentration and velocity fluctuations, and PDFs of time periods that concentration exceeds a set threshold were created to characterize the local behavior of scalar concentration and how it differs from average plume behavior. Non-Gaussian, exponential PDFs were found away from the plume mean centerline, yielding evidence of intermittent instantaneous scalar behavior on the edges of the plume. These findings emphasize the importance of exercising caution in the near source region when using operational Gaussian dispersion models.