Meteorologically adjusted regression analysis of urban air quality in Dhaka, Bangladesh, during pre-, during-, and post-lockdown periods (2017-2023).

This study investigates long-term air quality variability in Dhaka, Bangladesh, across pre-lockdown, lockdown, and post-lockdown phases of the COVID-19 pandemic (2017-2023). Unlike prior pollutant-specific or satellite-based analysis, this work integrates ground-based pollutant records with meteorological datasets to extricate anthropogenic influences from climatic variability. A methodological framework combining meteorological adjustment with multiple linear regression (MLR) modeling was applied to evaluate the relative effects of traffic activity and atmospheric conditions on major urban pollutants. The lockdown period provided a unique natural experiment, revealing the sensitivity of urban air quality to reduced human activity, while post-lockdown rebounds underscored the persistence of structural emission sources. The lockdown period was associated with substantial reductions in primary pollutants, whereas ozone exhibited an increasing trend linked to altered atmospheric chemistry. Seasonal analysis identified winter as the most polluted period due to unfavorable dispersion conditions and elevated combustion-related emissions. Regression results demonstrated that meteorological variables, particularly temperature, exerted strong control over particulate matter variability, while traffic activity remained strongly associated with nitrogen dioxide (NO 2 ) concentrations. Overall, the study advances regression-based urban air quality assessment by explicitly incorporating meteorological adjustment and provides a scientific basis for data-driven and meteorologically informed air quality management in rapidly urbanizing megacities.