Evaluating AQI limitations and proposing an enhanced urban air quality index integrating black carbon and methane across diverse urban climates

Abstract

India's air pollution crisis reflects not only rising emissions but also limitations in measurement and policy systems. The National Air Quality Index (NAQI) maintained by the Central Pollution Control Board (CPCB) currently excludes black carbon (BC) and methane (CH4), two critical short-lived climate pollutants with significant environmental relevance. This study evaluates the implications of integrating these pollutants into air quality assessment using six years (2019-2025) of weekly data for eight pollutants across three cities representing diverse geographical, climatic, and urbanization contexts, including Delhi, Indore, and Gangtok. Seven machine learning models, such as Random Forest, Gradient Boosting, XGBoost, LightGBM, Support Vector Regression, Multilayer Perceptron, and Ensemble, were developed and compared for pollutant prediction. Gradient Boosting showed the most consistent performance (R2 = 0.929 for NO2 in Indore and R2 = 0.930 for PM10 in Delhi), while ensemble approaches performed better under the complex mountainous conditions of Gangtok. Pollutant correlation analysis revealed distinct city-specific emission patterns, including a multi-source combustion profile in Delhi, a notable negative correlation between O3 and BC in Indore (r = 0.600), and solid-fuel dominance in Gangtok. A comparison between the conventional NAQI and the proposed Extended AQI (EAQI) indicated that the existing NAQI system systematically underestimates pollution levels. The findings support incorporating BC and CH4 into India's AQI system to improve air quality assessment, climate-resilient planning, and public health advisories. © 2026 Elsevier B.V.