High-resolution monthly assessment of population exposure to PM2.5 and its relationship with socioeconomic activities using multisource geospatial data

Understanding the spatiotemporal dynamics of population exposure to PM2.5 (PEP) and its relationship with socioeconomic activity (SEA) is crucial to reduce exposure risks and health dangers. However, few studies have investigated the dynamic variations of PEP within large regions at high spatiotemporal resolution; further, the impact mechanism between PEP and SEA remains largely unclear. Therefore, we estimated highly accurate PM2.5 concentrations in the Hunan province, China, using the Boruta and random forest (RF) algorithms and evaluated high-spatiotemporal-resolution PEP based on the estimated PM2.5 and obtained population data. Nighttime light data were used as a proxy of SEA to analyze the relationship between PEP and SEA. The results revealed that the Boruta-RF model predicted PM2.5 with fewer errors than the RF and stepwise multiple linear regression models, with the mean root-mean-square error reduced by 6.18% and 11.15%, respectively. The monthly PM2.5 concentrations in 2015 showed a U-shaped curve, with the entire provincial population exposed to monthly mean concentrations > 15 mu g/m(3). Heavier PM2.5 pollution tended to occur in densely populated areas, particularly in winter months. Using both fine-scale PM2.5 and population data improved the reliability of monthly PEP assessments and avoided over- and under-responses. Moreover, the PEP risk exhibited a unimodal structure, with a peak in January, at which point the urban-rural difference in PEP was the greatest. Further, PEP was positively influenced by SEA, with clear spatial spillover effects. SEA had an active impact on PEP during festivals and holidays, with the greatest consistency between the two occurring in November. These findings provide crucial insights for managing PM2.5 pollution.