Use of CFD modeling for estimating spatial representativeness of urban air pollution monitoring sites and suitability of their locations

A methodology to estimate the spatial representativeness of air pollution monitoring sites is applied to two urban districts. This methodology is based on high resolution maps of air pollution computed by using Computational Fluid Dynamics (CFD) modelling tools. Traffic-emitted NO2 dispersion is simulated for several meteorological conditions taking into account the effect of the buildings on air flow and pollutant dispersion and using a steady state CFD-RANS approach. From these results, maps of average pollutant concentrations for January-May 2011 are computed as a combination of the simulated scenarios. Two urban districts of Madrid City were simulated. Spatial representativeness areas for 32 different sites within the same district (including the site of the operative air quality stations) have been estimated by computing the portion of the domains with average NO2 concentration differing less than a 20% of the concentration at each candidate monitoring site. New parameters such as the ratio AR between the representativeness area and the whole domain area or the representativeness index (IR) has been proposed to discuss and compare the representativeness areas. Significant differences between the spatial representativeness of the candidate sites of both studied districts have been found. The sites of the Escuelas Aguirre district have generally smaller representativeness areas than those of the Plaza de Castilla. More stations are needed to cover the Escuelas Aguirre district than for the Plaza de Castilla one. The operative air quality station of the Escuelas Aguirre district is less representative than the station of the Plaza de Castilla district. The cause of these differences seems to be the differences in urban structure of both districts prompting different ventilation.