Calibration of low-cost sensor data for ambient PM2.5 monitoring across urban and rural settings in South Central Uganda

Air pollution is a leading risk factor for the global burden of disease. There are limited data from Sub-Saharan Africa (SSA). This study addresses significant data gaps in air quality monitoring in South-Central Uganda. Using a near-reference grade monitor and a network of 27 low-cost air quality monitors in the rural Rakai region and urban cities of Masaka and Kampala, we developed a locally tailored calibration model for ambient PM_2.5 concentrations. Using calibrated data across all low-cost monitors, we then examined spatiotemporal trends in ambient PM_2.5 concentrations. The calibration model, which adjusts for relative humidity, demonstrated robust performance with an R² of 0.92, a root mean squared error (RMSE) of 3.83 µg/m³ and bias of -0.39 µg/m³ compared to an RMSE of 20.37 µg/m³ and bias of 15.96 µg/m³ with the raw data. Mean PM_2.5 concentrations during the dry and wet seasons were 22.1 µg/m³ and 12.7 µg/m³ in rural areas, and 37.7 µg/m³ and 23.2 µg/m³ in urban areas, respectively. We observed diurnal variations, with PM_2.5 levels peaking in the early morning (6 a.m.-9 a.m.) and late evening (6 p.m.-10 p.m.), correlating with peak traffic hours. Low-cost sensors can enhance air quality research in regions like South-Central Uganda that lack air pollution data. Calibration tailored to local conditions can improve measurement accuracy of ambient PM_2.5 concentrations.