Impact of Spatiotemporal Factors on Exposure to PM2.5 as Residents move between Residential, Commercial and Recreational areas

Abstract

BACKGROUND AND AIM: Reduction of annual average PM2.5 concentrations by 2.5μg/m³ is associated with a 3.5% reduction in all-cause mortality. However, population exposure is typically assessed assuming that a person spends their full day at their residential address but numerous studies have demonstrated that PM2.5 can have small-scale spatial variations. Consequentially, the impact of spatial movement in a person’s day is often not represented in epidemiology studies. METHODS: A network of 50 low-cost sensors were deployed in Allegheny county (PA) and used for prediction modelling of PM2.5 using spatial and temporal variables. The county was gridded and categorized into 4 land cover areas (residential, commercial, recreational and other) at 50m resolution. Daily concentrations were predicted at each grid location for 2017 using land-use random forest (LURF) models. Weight-based probability was assigned to each residential area, with medium- and high-density areas two and three times as probable as low-density areas, respectively. Using probability-based sampling with 100,000 iterations, the predicted concentrations were computed for various population groups (children, working adults, non-working adults and seniors) to account for variations in movement patterns, with different behavior patterns assigned for weekdays and weekends. Weekday and weekend combinations were then proportionately summed to compare estimated annual exposure of the residents using static (residential) vs movement-based models. RESULTS:Daily concentrations were predicted to be higher in all non-residential areas when compared to residential areas {(baseline scenario)}. The average exposure to PM2.5 for all population groups were higher than the baseline scenario, with exposure of a working adults being the highest (~0.2μg/m³ higher than baseline). CONCLUSIONS:Findings of this study can be used to get a more accurate representation of exposure. Additionally, combined with epidemiology studies on impact of PM2.5 concentrations, it can help us determine the reduction in exposure and mortality rate due to behavioural changes. KEYWORDS: PM2.5, population exposure, air pollution, low cost sensors