Abstract
The impact of traffic pollution on the health and safety of residents that live in roadside residential buildings has been a major concern for governments. This study investigated the spatial distributions of PM2.5 concentration due to road traffic emissions and put forward a spatial distribution model for the estimation of PM2.5 concentration (SDC) based on Machine Learning. Meanwhile, based on SDC model, the decrease in life expectancy (DLE) of residents was assessed. On-site monitoring of PM2.5 concentration was conducted on different floors of a typical residential building situated by the roadside. Computational Fluid Dynamics (CFD) simulation was conducted for the spatial distribution analysis of PM2.5 concentration, which was verified by measurements. The findings show the PM2.5 concentration was decreased from 74 μg/m^3to 43 μg/m^3within 0 to 120 m distance from the road, and was decreased from 73 μg/m^3 to 42 μg/m^3 within 0 to 60 m height. The DLE in these locations was up to 5.11 years. The concentration of PM2.5 was stabilized within 40 to 45 μg/m³ when the building height was above 60 m (roughly the 17th floor from the ground) and the distance was 120 m away from the road. The DLE in these locations was stabilized within 0.62 years to 0.91years. The SDC model was established to efficiently predict DLE of residents and PM2.5 concentration along roadside. These findings would facilitate the precaution guidelines making of urban pollution as well as the future planning of urban health and safety buildings.