Abstract
Air pollution poses significant health and environmental challenges,particularly in rapidly urbanizing regions. Delhi-National Capital Regionexperiences air pollution episodes due to complex interactions betweenanthropogenic emissions and meteorological conditions. Understanding the causaldrivers of key pollutants such as $PM_{2.5}$ and ground $O_3$ is crucial fordeveloping effective mitigation strategies. This study investigates the causallinks of anthropogenic emissions on $PM_{2.5}$ and $O_3$ concentrations usingpredictive modeling and causal inference techniques. Integratinghigh-resolution air quality data from Jan 2018 to Aug 2023 across 32 monitoringstations, we develop predictive regression models that incorporatemeteorological variables (temperature and relative humidity), pollutantconcentrations ($NO_2, SO_2, CO$), and seasonal harmonic components to captureboth diurnal and annual cycles. Here, we show that reductions in anthropogenicemissions lead to significant decreases in $PM_{2.5}$ levels, whereas theireffect on $O_3$ remains marginal and statistically insignificant. To addressspatial heterogeneity, we employ Gaussian Process modeling. Further, we useGranger causality analysis and counterfactual simulation to establish directcausal links. Validation using real-world data from the COVID-19 lockdownconfirms that reduced emissions led to a substantial drop in $PM_{2.5}$ butonly a slight, insignificant change in $O_3$. The findings highlight thenecessity of targeted emission reduction policies while emphasizing the needfor integrated strategies addressing both particulate and ozone pollution.These insights are crucial for policymakers designing air pollutioninterventions in other megacities, and offer a scalable methodology fortackling complex urban air pollution through data-driven decision-making.