Development and application of a multi-scale modeling framework for urban high-resolution NO₂ pollution mapping

Abstract. Vehicle emissions have become a major source of air pollution in urban areas, especially for near-road environments, where the characteristics are difficult to capture by single-scale quality model due complex composition underlying surface. Here we developed hybrid CMAQ-RLINE_URBAN quantitatively analyze effects vehicle on roadside NO2 concentrations at high spatial resolution 50 m × m. To estimate influence various street canyons dispersion pollutants, machine-learning-based canyon flow (MLSCF) scheme was established based computational fluid dynamics and two machine learning methods. The results indicated that compared with Community Multi-scale Air Quality (CMAQ) model, improved underestimation concentration sites mean bias (MB) changing from −10 6.3 µg m−3. MLSCF obviously increased upwind within deep changes wind environment caused vortex. In summer, relative contribution vehicles Beijing areas 39 % average, similar CMAQ-ISAM (Integrated Source Apportionment Method) but it significantly decreased distance road centerline, freeways, reached 75 %.