DATA-DRIVEN URBAN AIR QUALITY ASSESSMENT AND NITROGEN DIOXIDE PREDICTION USING MACHINE LEARNING FOR ENVIRONMENTAL MONITORING
Keywords:
Nitrogen dioxide, Urban air quality, Machine learning, XGBoost, Environmental monitoringAbstract
SuNitrogen dioxide (NO₂) pollution continues to be a significant environment problem as it shows high level-spatial, seasonal, meteorological and co-pollutant dependency. This study aimed to meet the requirements of accurate, interpretable and operationally useful NO₂ predictions for urban environmental monitoring. Based on the Beijing Multi-Site Air Quality dataset, a data-driven framework that combines descriptive air-quality assessment, chronological development-validation-locked-test partitioning, Ridge Regression, Random Forest, Gradient Boosting, XGBoost, Extra Trees, SHAP-based interpretation, and exceedance detection with thresholds was developed. The mean NO₂ concentration was 50.6 µg/m3, with the highest mean in winter (62.4 µg/m3), and the highest station mean in Wanliu (65.3 µg/m3). With the lowest MAE (7.26µg/m³), RMSE (11.63µg/m³), and R2 (0.901), XGBoost had the best test performance when it was locked into the test set. The model was found to have a sensitivity of 86.23%, specificity of 99.70% and F1-score of 85.28% for hourly exceedances greater than 200 µg/m³. The framework brings together temporal validation, explainability and exceedance assessment to provide a reproducible approach to link predictive modelling with environmentally relevant choices for monitoring across urban stations. This study proves that interpretable ensemble learning can aid in precise prediction of NO₂ and real-world urban environmental monitoring.
References
BP. Sicard, E. Agathokleous, S. C. Anenberg, A. De Marco, E. Paoletti, and V. Calatayud, “Trends in urban air pollution over the last two decades: A global perspective,” Science of the Total Environment, vol. 858, Art. no. 160064, 2023, doi: 10.1016/j.scitotenv.2022.160064.
M. Wang, H. Li, S. Huang, Y. Qian, K. Steenland, Y. Xie, S. Papatheodorou, and L. Shi, “Short-term exposure to nitrogen dioxide and mortality: A systematic review and meta-analysis,” Environmental Research, vol. 202, Art. no. 111766, 2021, doi: 10.1016/j.envres.2021.111766.
P. Huangfu and R. Atkinson, “Long-term exposure to NO₂ and O₃ and all-cause and respiratory mortality: A systematic review and meta-analysis,” Environment International, vol. 144, Art. no. 105998, 2020, doi: 10.1016/j.envint.2020.105998.
S. Huang et al., “Long-term exposure to nitrogen dioxide and mortality: A systematic review and meta-analysis,” Science of the Total Environment, vol. 776, Art. no. 145968, 2021, doi: 10.1016/j.scitotenv.2021.145968.
J. Song, Y. Wang, Q. Zhang, W. Qin, R. Pan, W. Yi, Z. Xu, J. Cheng, and H. Su, “Premature mortality attributable to NO₂ exposure in cities and the role of built environment: A global analysis,” Science of the Total Environment, vol. 866, Art. no. 161395, 2023, doi: 10.1016/j.scitotenv.2023.161395.
X. Chen, L. Qi, S. Li, and X. Duan, “Long-term NO₂ exposure and mortality: A comprehensive meta-analysis,” Environmental Pollution, vol. 341, Art. no. 122971, 2024, doi: 10.1016/j.envpol.2023.122971.
M. Méndez, M. G. Merayo, and M. Núñez, “Machine learning algorithms to forecast air quality: A survey,” Artificial Intelligence Review, vol. 56, pp. 10031–10066, 2023, doi: 10.1007/s10462-023-10424-4.
M. Jamei, M. Ali, A. Malik, M. Karbasi, E. Sharma, and Z. M. Yaseen, “Air quality monitoring based on chemical and meteorological drivers: Application of a novel data filtering-based hybridized deep learning model,” Journal of Cleaner Production, vol. 374, Art. no. 134011, 2022, doi: 10.1016/j.jclepro.2022.134011.
S. Abirami and P. Chitra, “Regional air quality forecasting using spatiotemporal deep learning,” Journal of Cleaner Production, vol. 283, Art. no. 125341, 2021, doi: 10.1016/j.jclepro.2020.125341.
J. Gu, B. Yang, M. Brauer, and K. M. Zhang, “Enhancing the evaluation and interpretability of data-driven air quality models,” Atmospheric Environment, vol. 246, Art. no. 118125, 2021, doi: 10.1016/j.atmosenv.2020.118125.
H. Liao, Y. Li, M. Wu, and H. Chen, “Air quality prediction by integrating mechanism model and machine learning model,” Science of the Total Environment, vol. 899, Art. no. 165646, 2023, doi: 10.1016/j.scitotenv.2023.165646.
M. Vega García and J. L. Aznarte, “Shapley additive explanations for NO₂ forecasting,” Ecological Informatics, vol. 56, Art. no. 101039, 2020, doi: 10.1016/j.ecoinf.2019.101039.
S. Long, X. Wei, F. Zhang, R. Zhang, J. Xu, K. Wu, Q. Li, and W. Li, “Estimating daily ground-level NO₂ concentrations over China based on TROPOMI observations and machine learning approach,” Atmospheric Environment, vol. 289, Art. no. 119310, 2022, doi: 10.1016/j.atmosenv.2022.119310.
C.-L. Wu, H.-D. He, R.-F. Song, X.-H. Zhu, Z.-R. Peng, Q.-Y. Fu, and J. Pan, “A hybrid deep learning model for regional O₃ and NO₂ concentrations prediction based on spatiotemporal dependencies in air quality monitoring network,” Environmental Pollution, vol. 320, Art. no. 121075, 2023, doi: 10.1016/j.envpol.2023.121075.
C. Tao, M. Jia, G. Wang, Y. Zhang, Q. Zhang, X. Wang, Q. Wang, and W. Wang, “Time-sensitive prediction of NO₂ concentration in China using an ensemble machine learning model from multi-source data,” Journal of Environmental Sciences, vol. 137, pp. 30–40, 2024, doi: 10.1016/j.jes.2023.02.026.
O. A. Alawi, H. M. Kamar, A. Alsuwaiyan, and Z. M. Yaseen, “Temporal trends and predictive modeling of air pollutants in Delhi: A comparative study of artificial intelligence models,” Scientific Reports, vol. 14, Art. no. 30957, 2024, doi: 10.1038/s41598-024-82117-z.
S. Mandal, S. Boppani, V. Dasari, and M. Thakur, “A bivariate simultaneous pollutant forecasting approach by Unified Spectro-Spatial Graph Neural Network (USSGNN) and its application in prediction of O₃ and NO₂ for New Delhi, India,” Sustainable Cities and Society, vol. 114, Art. no. 105741, 2024, doi: 10.1016/j.scs.2024.105741.
Q. He, K. Qin, J. B. Cohen, D. Li, and J. Kim, “Quantifying uncertainty in ML-derived atmosphere remote sensing: Hourly surface NO₂ estimation with GEMS,” Geophysical Research Letters, vol. 51, no. 18, Art. no. e2024GL110468, 2024, doi: 10.1029/2024GL110468.
N. Ahmad, C. Lin, A. K. H. Lau, J. Kim, T. Zhang, F. Yu, C. Li, Y. Li, J. C. H. Fung, and X. Q. Lao, “Estimation of ground-level NO₂ and its spatiotemporal variations in China using GEMS measurements and a nested machine learning model,” Atmospheric Chemistry and Physics, vol. 24, pp. 9645–9665, 2024, doi: 10.5194/acp-24-9645-2024.
S. Wang, S. Zhang, D. Wang, and W. Li, “Spatial-and-local-aware deep learning approach for ground-level NO₂ estimation in England with multisource data from satellite-based observations and chemical transport models,” International Journal of Applied Earth Observation and Geoinformation, vol. 139, Art. no. 104506, 2025, doi: 10.1016/j.jag.2025.104506.
L. Xu, B. Xu, Y. Sun, and Y. Cao, “Interpretable and scalable deep learning for urban NO₂ prediction via multisource data,” Transportation Research Part D: Transport and Environment, vol. 149, Art. no. 105039, 2025, doi: 10.1016/j.trd.2025.105039.
M. A. Javadi, M. Zare Shahneh, and Z. Amiri, “Integration of Sentinel-5P satellite data and machine learning for spatiotemporal prediction of NO₂ in Delhi: Impacts of COVID-19 lockdown,” Atmospheric Pollution Research, vol. 17, no. 1, Art. no. 102702, 2026, doi: 10.1016/j.apr.2025.102702.
S. Chen, “Beijing Multi-Site Air Quality,” UCI Machine Learning Repository, 2017, doi: 10.24432/C5RK5G.
