Integrating Low-Cost Sensors with Dispersion Modelling for High-Resolution Insights into Urban Air Quality

Urban areas experience elevated air pollution levels which pose significant health risks. Reducing exposure to poor air quality and mitigating the associated negative health impacts requires informed policy measures. This study advances urban air quality modelling by developing an air quality model (baseline model) and further integrating measurements from a network of low-cost sensors and regulatory monitors into the model output (data fusion model). The resulting data fusion model provides accurate air quality data in high spatiotemporal resolution. The data fusion model showed higher PM_2.5 concentrations during evening hours and winter months, with a population-weighted exposure to PM_2.5 almost twice as high as predicted by the baseline model during these months. The models exhibited different spatial patterns, with the data fusion model showing a shift in peak concentrations from the city centre to residential areas, where levels were up to 10 µg/m³ higher than the baseline model. These differences are likely attributable to an underestimation of residential emissions in the baseline model. While both models were FAIRMODE compliant, the data fusion model showed a reduced bias for most monitoring stations compared to the baseline model. The data fusion model enabled a more accurate assessment of existing policies, specifically those aimed at reducing urban air pollution from solid fuel burning. Moreover, by identifying locations and sectors which contribute significantly to high levels of PM_2.5, the data fusion model supports the formation of targeted air quality policies. This enables cities to maximise reductions in air pollution and exposures, thereby safeguarding public health.