Evaluation of a CFD-based approach to estimate pollutant distribution within a real urban canopy by means of passive samplers.

The distribution of pollutants is spatially heterogeneous within urban streets making difficult to build a realistic concentration map. In this paper, a methodology based on computational fluid dynamic modeling with Reynolds-averaged Navier-Stokes approach is used to compute maps of concentration for a period of several weeks. The methodology is evaluated by comparing simulation results against experimental data from two different campaigns where a large number of passive samplers deployed in an area with heavy vehicular traffic in Madrid (Spain). The evaluation shows that the methodology is able to reproduce the general pattern of several-week averaged pollutant distribution in an urban area with heavy vehicular traffic, resolving the spatial variability up to a resolution of 1-2m. In addition, the model results fit satisfactorily the time evolution of the pollutant concentration measured at an air quality station deployed in the studied area. However, problems were detected close to zones with complex emissions patterns (tunnels, street forks, etc.), where the model compared poorly against passive sampler measurements. A preliminary assessment of the uncertainties induced in the numerical methodology due to consider NO2 as non-reactive pollutant under winter conditions indicates that it would be an acceptable approach for this particular case study. Overall, our analysis contributes to raise the confidence in that approached similar to the one presented in this study can be adopted for dealing with several aspects of the air quality management such as air quality assessment, optimization of the location of measurement stations, and the evaluation of air pollution reduction strategies.