Multilinear regression analysis of PM2.5 in Kampala and Fort Portal cities: Effects of meteorological factors and lagged pollution

Rapid urbanization across Sub-Saharan Africa intensifies fine particulate matter (PM 2.5 ) pollution, yet the combined effects of meteorology and pollution persistence remain poorly understood. This study investigates the spatiotemporal variability of PM 2.5 in Kampala (urban) and Fort Portal (semi-urban), Uganda, using daily observations from October 2021 to January 2024. Calibrated low-cost AirQo sensor data were integrated with meteorological parameters, including temperature, humidity, wind speed, wind direction, and precipitation, as well as one-day lagged PM 2.5 , to develop enhanced multilinear regression (MLR) models. Results revealed strong seasonal contrasts, with mean dry-season concentrations in Kampala (38.3 μgm -3 ) and Fort Portal (32.9 μgm -3 ) exceeding World Health Organization and NEMA-Uganda limits. Model performance varied by city, explaining up to 57% of daily PM 2.5 variability in Kampala and 80% in Fort Portal. The inclusion of lagged PM 2.5 significantly improved model accuracy, highlighting persistence effects under stagnant meteorological conditions. Wind rose analysis showed that southerly and westerly winds enhanced pollutant transport, particularly during dry months, suggesting potential transboundary contributions to Fort Portal’s pollution burden. Although the models performed well during dry seasons, predictive power declined in wet seasons due to rainfall-induced washout effects not fully captured by linear formulations. These findings emphasize the importance of meteorological drivers and pollution persistence in shaping urban air quality and support data-driven interventions such as emission control, traffic management, biomass burning reduction, and regional cooperation to protect public health in rapidly urbanizing African cities. • Daily PM 2 . 5 levels in Kampala and Fort Portal showed clear seasonal cycles with the highest concentrations occurring during dry periods. • Annual background PM 2 . 5 was higher in Kampala due to stronger anthropogenic emissions and urban activity. • Multilinear regression models explained up to 80% of daily PM 2 . 5 variability in Fort Portal and 57% in Kampala. • Wind speed, relative humidity, and precipitation consistently reduced PM 2 . 5 while temperature showed location specific effects. • One day lagged PM 2 . 5 was a strong predictor in both cities, indicating significant pollution persistence. • Wind patterns revealed directional transport of pollution, including regional influences toward Fort Portal. • Calibrated low cost sensors provided reliable long term PM 2 . 5 monitoring suitable for data limited African environments.