Spatial Modeling of Ambient PM2.5 and Ozone Exposure in Western Massachusetts Using a Low-Cost Air Quality Monitoring Sensor Network

DONG GAO, Jiarong Qi, Alexander De Jesus, Kerry Kelly, Sarita Hudson, Krystal Godri Pollitt, Yale University

     Abstract Number: 585
     Working Group: Identifying and Addressing Disparate Health and Social Impacts of Exposure to Aerosols and Other Contaminants across Continents, Communities, and Microenvironments

Abstract
The impact of ambient fine particulate matter (PM_2.5) and ozone exposure on public health is well recognized, particularly in areas with high asthma prevalence. The Pioneer Valley region in Western Massachusetts, reported to be one of the most challenging places to live with asthma in the US, faces limitations in assessing air pollutant exposure due to the sparse distribution of regulatory air quality monitoring stations. To address this gap and evaluate the spatiotemporal variability of PM_2.5 and ozone exposure in fine scale, we leveraged emerging low-cost sensors (TELLUS AirU and PurpleAir) and engaged community residents to develop a community-based air quality monitoring network across environmental justice communities, through a partnership between local community groups, city officials, non-for-profit organizations, health care facilities, academic researchers, and residents in the cities of Springfield, Chicopee, and Holyoke. Eighty sensors were strategically deployed across the region to capture spatial variations in PM_2.5 concentrations, including twenty metal oxide sensors for ozone monitoring. Prior to deployment, the sensors were calibrated against reference instruments to derive correction factors for each sensor. A correlation was identified between sensors, with coefficients of determination R² of 0.98~0.99 for PM_2.5 measurements and greater than 0.85 for reducing and oxidizing gases measurements, demonstrating similar level of accuracy and reliability of sensors in real environmental scenarios. For post-deployment calibration, we derived long-term PM_2.5 and ozone calibration models with over one-year data, using multiple linear regression and the random forest algorithm, respectively. These models were applied to the sensors throughout the study region to provide pollutant estimates at fine spatiotemporal scales. Integrating low-cost sensors into citizen-science based air monitoring program has promising applications to resolve monitoring disparity and capture hotspots to inform the community of the potential burden on health.