Four-year community-wide PM_2.5 exposure characterization using a low-cost sensor network in a rural valley influenced...

Exposure to fine particulate matter (PM_2.5) from woodsmoke is a national and global public health concern. While wood heat use is increasing in the Northeast U.S., exposure to woodsmoke in rural valleys in this region remains understudied. Low-cost sensors have recently emerged as a promising strategy to better assess PM_2.5 exposure in communities across the globe. However, real-world, long-term community air monitoring studies deploying low-cost sensors are lacking. Such studies are necessary to validate performance over time for the goals of assessing exposure trends and providing practical guidance to future community-scale projects. Here, we evaluated PM_2.5 community-wide exposure over four years in a rural, woodsmoke impacted community that deployed a Purple Air network. We determined significant differences between the PM_2.5 regulatory reference monitor and Purple Air PM_2.5 concentrations across the community, over multiple heating seasons, at both hourly (p < 0.01) and 8-h average time intervals (p < 0.001). PM_2.5 was especially elevated in the evenings (6 p. m.-2 a.m.), with a maximum 1-h average of 81.7 μg/m³ and a maximum 8-h average of 78.7 μg/m³. Consistent with other performance evaluations in the literature, we determined Purple Air sensors have low bias after correction (2.4 % Normalized Mean Bias Error [NMBE], 3.3 μg/m³ Root Mean Square Error [RMSE]). Co-located sensors were reliable in harsh winter conditions for up to three years. We suggest quality assurance, data management, correction model selection, and citizen science partnerships are critical considerations for sensor network deployments aiming to assess long-term exposure and health.