As urbanization progresses, more comprehensive and advanced methods are required to analyze the modifications of urban microclimate. Street level air pollution due to vehicular exhausts is of concern in the urban environment and is significantly affected by thermal stratification, yet numerical studies of dispersion under realistic surface heating are scarce. To address this shortcoming, a detailed indoor-outdoor building energy model (TUF-IOBES) is employed to compute heat fluxes from street and building surfaces, which are then used as boundary condition for a PArallelized Large-Eddy Simulation Model (PALM). Series of fluid flow and thermal field simulations are performed for an idealized, compact mid-rise urban environment (λp=0.29, λf=0.25 with no vegetation). In comparison with previous studies, our model considers the transient non-uniform surface heating caused by solar insolation and inter-building shadowing, while coupling the indoor-outdoor heat transfer, flow field and passive pollution dispersion. Traffic emissions are characterized as a city-wide near-ground volume source. A horizontal (based on the temperature difference between the two walls) and a vertical (based on the difference between the street temperature and roof level air temperature) Richardson number (Ri_h and Ri_v) are used to characterize atmospheric instability and solar tilt with respect to the wind direction. The validity of this choice of non-dimensional numbers is investigated through simulations with different wind speed and surface radiative properties, but the same sets of Richardson numbers. The pollutant Exchange Rate (PCH) from the horizontal and vertical ventilating faces of the canyon, and the pollutant concentration at pedestrian level (z=1.5-2 m) are examined in order to quantify the effect of surface heating on air quality as a function of time of day and Ri_h and Ri_v. Local ambient conditions (wind speed) and surface material properties (ground and wall albedo) are modified to span a realistic range of buoyancy parameters. Quantifying the change in pollutant dispersion in street canyons based on these parameters can ultimately inform urban designers on the impact of their design on air quality, human health and comfort.

Understanding transport mechanisms for momentum or scalar between urban surfaces and the atmosphere is one of essential issues so as to predict flow and scalar distributions attributed to complicated urban climates. One of methods to relate these vertical distributions and surface effects, namely forcings such as momentum sink or scalar source/sink, are known as similarity theories. For example, under various conditions of thermal stratification, the vertical gradient of velocity or scalar can be expressed by the surface flux of momentum or scalar and the dimensionless shear or scalar function within the inertial sub-layer, where the flow and scalar distributions are horizontally homogeneous, as known as Monin-Obukhov similarity theory. In the case of neutral conditions, especially, more detailed investigation have been done in order to quantify the effects of surface geometries on aerodynamic parameters in velocity profile, such as roughness length or displacement height by means of both field measurements and wind-tunnel experiments. Moreover, similarity of the parameters between scalar and momentum has been also confirmed to be possibly applicable for urban surfaces.

In addition to the similarity theory in the inertial sub-layer, several features of flow distribution and turbulent statistics have been reviled in the layer directly influenced by roughness elements, namely the roughness sub-layer and canopy layer. For instance, it is known that velocity profile satisfies logarithmic law even within the roughness sub-layer, if spatial-averaged quantities are taken, even though, the layer does not satisfy the premise of the similarity theory in the inertial sub-layer. In addition, there are characteristic vertical trends on turbulent statistics; sweep events, or strong downward motions with high momentum fluid, are dominant near roughness elements whereas ejection events, or strong upward motions with low momentum fluid, dominate sweep events in upper part of the roughness sub-layer, being consistent with the tendency in the inertial sub-layer. Moreover, the visualization of turbulent structures around these events based on numerical simulations shows the existences of hair-pin and head-down hair-pin like vortices in the roughness sub-layer.

In contrast to these extensive understanding of characteristics of turbulent structure casing momentum transport, and consequently resulting spatially averaged fields of statistics, little of features for scalar dispersion are known especially in the roughness sub-layer. Non-uniformity of scalar distribution seems more easily generated in actual urban area than that of momentum, because scalar source or sink are usually rather scattered, meaning the situation in which horizontal homogeneity can be assumed is more rare for scalar. Although a few studies have revealed that there is similarity between scalar and momentum transport to some extent also in the roughness sub-layer, holistic understanding of scalar transport in the layer and the canopy layer with horizontally non-uniform flow distribution has not been reached yet.

Therefore, we have performed a large-eddy simulation on scalar dispersion from continuous area surface located on the floor of a rough surface. As the roughness condition, a cubical array in staggered layout is selected in order to generate both of the roughness sub-layer and the canopy layer. The scalar boundary layer can develop to streamwise direction and the height can reach up to four times of height of the cubes approximately. The purposes of this study are summarized as follows: (1) to quantify the streamwise characteristics of scalar boundary layer, including trends of concentration and turbulent flux for scalar, (2) to apply quadrant analysis so as to estimate the contribution of updraft or downdraft to total scalar turbulent flux, and then to discuss the similarity and dissimilarity of momentum and scalar in terms of flux decomposition, and (3) to visualize conditionally averaged flow filed which may strongly affect updraft or downdraft with scalar.

In this study an analysis of the exchange velocity (Ue) in a real street canyon with trees is presented. By recalling that Ue is a bulk velocity scale that can be used as a surrogate for the complex transfer processes between the canopy and the overlying atmosphere, we attempt to quantify it by means of experimental investigations and numerical simulations.

Specifically, we analyse the effect of trees on local meteorological variables through in situ measurements and Computational Fluid Dynamics (CFD) simulations. Measurements are taken in a neighbourhood of a medium-size Mediterranean city (Lecce, IT) with street canyons of aspect ratio H/W~1 (where H is the average building height and W is the average width of the street). Numerical simulations are performed with the commercial Computational Fluid Dynamics (CFD) FLUENT model with Reynolds Stress Model (RSM) for the treatment of turbulence. The effect of trees in the numerical simulations has been evaluated by considering them as a porous medium using a Leaf Area Index (LAI) measured in situ by a ceptometer. An extension of the current formulation for Ue given by the ratio between the pollutant flux at roof level through the exchange surface and the difference between the spatially-averaged pollutant concentration within the urban canopy and the background concentration is considered to include the direct dependence on the LAI.

Results provide evidence that flows and turbulent structures vary within the street canyon with and without trees. It is shown that Ue may diminish of factor about 10 in presence of trees with respect to the tree-free cases. This finding has clear influence on ground level concentration which may be substantially under-predicted if this effect is neglected. In this presentation we will show how the knowledge of Ue maps for a given city correlated with the local meteorology can be used for a more effective urban planning. This study also aims to stimulate the development of new numerical models which explicitly account for the effect of trees in urban areas.