The potential of micro air sensors for democratizing air quality monitoring and mirroring air pollution at schools using citizen science

Abstract

The growing concern over children's exposure to fine particles in the school environment underscores the urgent need for accessible, localized air quality monitoring solutions. The expanding global use of air sensors for air quality monitoring in previously undermonitored environments motivates our investigation into their capacity to address this need for monitoring at schools. This study therefore harnesses citizen science methodologies using air sensors to address critical research gaps in conventional air quality monitoring to empower communities to actively participate in understanding and addressing air pollution that directly affects children's health and academic performance. We deployed two air sensors measuring PM_2.5, one each at two schools in Ghana: Salvation Army Senior High School (SASHS) and West Africa Senior High School (WASHS). Prior to deployment at the schools, we conducted a colocation study at the University of Ghana to establish sensor performance. Based on this colocation, we use a multilinear regression model to correct the raw air sensor data, improving the coefficient of determination from 0.91 to 0.92 and the mean absolute error from 3.08 μg m⁻³ to 1.69 μg m⁻³ respectively. Following this, the sensors were deployed at the schools from June 17, 2024 to May 25, 2025. We observed a bimodal peak at the two schools: morning and evening rush hour peaks at WASHS of 35 μg m⁻³, and 28 μg m⁻³ morning and nearly 45 μg m⁻³ evening peaks at SASHS. Hourly, maximum concentrations at the two schools vary significantly. Source components were attributed to wind speed-direction and background activities at each dominant wind sector. PM_2.5 pollution was attributed to biomass burning at SASHS and traffic at WASHS, with natural sources such as the harmattan impacting both schools. At the two schools, observations exceeded the World Health Organization Air Quality Guideline for daily PM_2.5 exposure threshold by 73%. Secondly, we tested the preference of students to specific visualization tools in the “openair” package to support location-specific air quality management by selecting timePlot, timeVariation, windRose, and polarCluster functions to visualize the reported data. Participants were able to link the observations to potential source components, and used these insights to propose school-specific mitigation strategies such as clean cooking initiatives, tree planting, proper waste management, and environmentally friendly public transportation. Overall, this demonstrates the applicability of the air sensor data to determine pollutant patterns and sources for the school environments, as well as the accessibility of the data to students to support mitigation planning.