Probabilistic Machine Learning Early Warning for Urban PM2.5 in SEA Cities

Baharuddin Mide; Dhimas Tribuana; Usman  Sattar; Dayanti Dayanti

doi:10.29207/resti.v10i7199

Authors

Baharuddin Mide Universitas Ichsan Sidenreng Rappang
Dhimas Tribuana Akademi Sekretari Manajemen Indonesia Publik https://orcid.org/0009-0002-8504-0740
Usman Sattar Politeknik Bombana
Dayanti Dayanti Universitas Patria Artha

DOI:

https://doi.org/10.29207/resti.v10i7199

Keywords:

air quality monitoring, early warning, exceedance prediction, machine learning, PM₂.₅

Abstract

Air pollution, particularly fine particulate matter (PM₂.₅), poses a critical threat to public health in rapidly urbanizing regions. Reliable early-warning systems are essential for mitigating exposure risks, yet challenges remain in cities with heterogeneous sensor coverage and event frequency. This study aimed to develop and evaluate a probabilistic, portable across cities early-warning framework for PM₂.₅ exceedances in Southeast Asia, focusing on Jakarta, Singapore, and Bangkok. Using a staged experimental design (Exp-A through Exp-E), we integrated regression-based back-casts with classification-based exceedance alerts, applied variant selection across thresholds and horizons, and validated for operational readiness through model freezing, documentation, and online simulation. Results showed that Jakarta achieved near-perfect exceedance prediction up to 6-hour horizons (F1 ≈ 0.99), Singapore maintained strong performance at short horizons (F1 ≈ 0.91 at 2–3 hours), while Bangkok yielded moderate but actionable signals at very short horizons (F1 ≈ 0.62 at 1 hour). Regression components provided stable situational awareness, and online smoothing reduced false alarms by approximately 15–20% without degrading performance. The framework demonstrated that calibrated exceedance probabilities can serve as an effective basis for city-level air quality alerts, with reliability strongly influenced by data density and event prevalence. This work contributes a reproducible, transparent, and computationally efficient approach that bridges machine learning innovation with practical environmental management. The findings emphasize the importance of horizon-specific calibration and adaptive strategies, offering both theoretical insights and practical value for policymakers in urban air quality governance.

Downloads

Download data is not yet available.

References

[1] WHO, “WHO Global Air Quality Guidelines: Particulate Matter (PM₂.₅ and PM₁₀), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide,” World Health Organization, Geneva, 2021. [Online]. Available: https://www.who.int/publications/i/item/9789240034228

[2] S.-A. Logothetis, G. Kosmopoulos, O. Panagopoulos, V. Salamalikis, and A. Kazantzidis, “Forecasting the Exceedances of PM2.5 in an Urban Area,” Atmosphere (Basel)., vol. 15, no. 5, p. 594, May 2024, doi:10.3390/atmos15050594.

[3] Y. Li et al., “Impacts of estimated plume rise on PM 2.5 exceedance prediction during extreme wildfire events: a comparison of three schemes (Briggs, Freitas, and Sofiev),” Atmos. Chem. Phys., vol. 23, no. 5, pp. 3083–3101, Mar. 2023, doi:10.5194/acp-23-3083-2023.

[4] B. Peng-in, P. Sanitluea, P. Monjatturat, P. Boonkerd, and A. Phosri, “Estimating ground-level PM2.5 over Bangkok Metropolitan Region in Thailand using aerosol optical depth retrieved by MODIS,” Air Qual. Atmos. Health, vol. 15, no. 11, pp. 2091–2102, Nov. 2022, doi:10.1007/s11869-022-01238-4.

[5] B. Haryanto, B. Jalaludin, A. Asyary, N. Roestandy, and F. Nugraha, “Associations Between Ambient PM2.5 Levels and Children’s Pneumonia and Asthma During the COVID-19 Pandemic in Greater Jakarta (Jabodetabek),” Ann. Glob. Health, vol. 91, no. 1, Feb. 2025, doi:10.5334/aogh.4623.

[6] X. Sun, C. E. Ivey, K. R. Baker, A. Nenes, N. P. Lareau, and H. A. Holmes, “Confronting Uncertainties of Simulated Air Pollution Concentrations during Persistent Cold Air Pool Events in the Salt Lake Valley, Utah,” Environ. Sci. Technol., vol. 55, no. 22, pp. 15072–15081, Nov. 2021, doi:10.1021/acs.est.1c05467.

[7] N. Wu et al., “Development of a high-resolution integrated emission inventory of air pollutants for China,” Earth Syst. Sci. Data, vol. 16, no. 6, pp. 2893–2915, Jun. 2024, doi:10.5194/essd-16-2893-2024.

[8] L. Huang et al., “Recommendations on benchmarks for numerical air quality model applications in China – Part 2: Ozone and uncertainty analysis,” Atmos. Chem. Phys., vol. 25, no. 7, pp. 4233–4249, Apr. 2025, doi:10.5194/acp-25-4233-2025.

[9] Z. Liu, Z. Fang, and Y. Hu, “A deep learning-based hybrid method for PM2.5 prediction in central and western China,” Sci. Rep., vol. 15, no. 1, p. 10080, Mar. 2025, doi:10.1038/s41598-025-95460-6.

[10] M. Teng et al., “72-hour real-time forecasting of ambient PM2.5 by hybrid graph deep neural network with aggregated neighborhood spatiotemporal information,” Environ. Int., vol. 176, p. 107971, Jun. 2023, doi:10.1016/j.envint.2023.107971.

[11] Z. Gao et al., “Predicting PM2.5 levels and exceedance days using machine learning methods,” Atmos. Environ., vol. 323, p. 120396, Apr. 2024, doi:10.1016/j.atmosenv.2024.120396.

[12] M. Karl, S. Acksen, R. Chaudhary, and M. O. P. Ramacher, “Forecasting system for urban air quality with automatic correction and web service for public dissemination,” Int. J. Digit. Earth, vol. 17, no. 1, pp. 1–22, Dec. 2024, doi:10.1080/17538947.2024.2359569.

[13] World Resources Institute, “Working Together for Cleaner Air in Southeast Asia: Inside the SEA-AQ Community of Practice,” World Resources Institute, 2023. [Online]. Available:https://www.wri.org/update/working-together-cleaner-air-southeast-asia-inside-sea-aq-community-practice

[14] Health Effects Institute and Institute for Health Metrics and Evaluation, “State of Global Air 2024,” HEI, 2024. [Online]. Available:https://www.stateofglobalair.org/sites/default/files/documents/2024-06/soga-2024-report_0.pdf

[15] IQAir and Greenpeace, “2024 World Air Quality Report,” IQAir, 2024. [Online]. Available:https://www.greenpeace.org/static/planet4-chile-stateless/2025/03/edf90b7a-2024_world_air_quality_report_vf.pdf

[16] M. S. S. Sunder, V. A. Tikkiwal, A. Kumar, and B. Tyagi, “Unveiling the Transparency of Prediction Models for Spatial PM2.5 over Singapore: Comparison of Different Machine Learning Approaches with eXplainable Artificial Intelligence,” AI, vol. 4, no. 4, pp. 787–811, Sep. 2023, doi:10.3390/ai4040040.

[17] M.-G. Kim, S.-Y. Kim, K. Lee, P. Kim, H. Kim, and H.-J. Song, “Development and interpretation of PM2.5 estimation model for the Seoul Metropolitan Area using machine learning and explainable AI,” Atmos. Pollut. Res., vol. 16, no. 11, p. 102672, Nov. 2025, doi:10.1016/j.apr.2025.102672.

[18] D. A. Safira, H. Kuswanto, and M. Ahsan, “Improving the Forecast Accuracy of PM2.5 Using SETAR-Tree Method: Case Study in Jakarta, Indonesia,” Atmosphere (Basel)., vol. 16, no. 1, p. 23, Dec. 2024, doi:10.3390/atmos16010023.

[19] I. Zandi, A. Jafari, and A. Lotfata, “Enhancing PM2.5 Air Pollution Prediction Performance by Optimizing the Echo State Network (ESN) Deep Learning Model Using New Metaheuristic Algorithms,” Urban Science, vol. 9, no. 5, p. 138, Apr. 2025, doi:10.3390/urbansci9050138.

[20] C. M. Rosales, J. R. Bratburd, S. Diez, S. Duncan, C. Malings, and P. Pant, “Open Air Quality Data Platforms for Environmental Health Research and Action,” Curr. Environ. Health Rep., vol. 12, no. 1, p. 27, Aug. 2025, doi:10.1007/s40572-025-00487-6.

[21] OpenAQ, “OpenAQ API documentation, version 3,” [Online]. Available: https://docs.openaq.org/api. [Accessed: Aug. 14, 2026].

[22] Meteostat, “Hourly data,” Meteostat Developers, [Online]. Available: https://dev.meteostat.net/data/timeseries/hourly. [Accessed: Aug. 14, 2026].

[23] G. Ke et al., “LightGBM: A Highly Efficient Gradient Boosting Decision Tree,” Adv. Neural Inf. Process. Syst., vol. 30, 2017, [Online]. Available:https://proceedings.neurips.cc/paper/2017/file/6449f44a102fde848669bdd9eb6b76fa-Paper.pdf

[24] B. Zadrozny and C. Elkan, “Transforming classifier scores into accurate multiclass probability estimates,” in Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining, New York, NY, USA: ACM, Jul. 2002, pp. 694–699. doi:10.1145/775047.775151.

[25] A. Niculescu-Mizil and R. Caruana, “Predicting good probabilities with supervised learning,” in Proceedings of the 22nd international conference on Machine learning - ICML ’05, New York, New York, USA: ACM Press, 2005, pp. 625–632. doi:10.1145/1102351.1102430.

[26] C. Bergmeir, R. J. Hyndman, and B. Koo, “A note on the validity of cross-validation for evaluating autoregressive time series prediction,” Comput. Stat. Data Anal., vol. 120, pp. 70–83, Apr. 2018, doi: 10.1016/j.csda.2017.11.003.

[27] M. Niu, Y. Zhang, and Z. Ren, “Deep Learning-Based PM2.5 Long Time-Series Prediction by Fusing Multisource Data—A Case Study of Beijing,” Atmosphere (Basel)., vol. 14, no. 2, p. 340, Feb. 2023, doi:10.3390/atmos14020340.

[28] E. Koçak, “Comprehensive evaluation of machine learning models for real-world air quality prediction and health risk assessment by AirQ+,” Earth Sci. Inform., vol. 18, no. 3, p. 447, Sep. 2025, doi:10.1007/s12145-025-01941-7.

[29] R. S. Pathak, V. Pathak, and A. Rai, “A novel attention-based deep learning model for accurate PM2.5 concentration prediction and health impact assessment,” J. Atmos. Sol. Terr. Phys., vol. 274, p. 106583, Sep. 2025, doi:10.1016/j.jastp.2025.106583.

[30] A. Houdou et al., “Interpretable Machine Learning Approaches for Forecasting and Predicting Air Pollution: A Systematic Review,” Aerosol Air Qual. Res., vol. 24, no. 1, p. 230151, 2024, doi:10.4209/aaqr.230151.

[31] P. Patel et al., “A systematic study on PM2.5 and PM10 concentration prediction in air pollution using machine learning and deep learning model,” Environmental Chemistry and Ecotoxicology, vol. 7, pp. 1401–1415, 2025, doi:10.1016/j.enceco.2025.07.001.

[32] D. C. Pendergrass et al., “Continuous mapping of fine particulate matter (PM 2.5 ) air quality in East Asia at daily 6 × 6 km 2 resolution by application of a random forest algorithm to 2011–2019 GOCI geostationary satellite data,” Atmos. Meas. Tech., vol. 15, no. 4, pp. 1075–1091, Mar. 2022, doi:10.5194/amt-15-1075-2022.

[33] A. Utku et al., “Advancing Air Quality Monitoring: Deep Learning-Based CNN-RNN Hybrid Model for PM2.5 Forecasting,” Atmosphere (Basel)., vol. 16, no. 9, p. 1003, Aug. 2025, doi:10.3390/atmos16091003.

[34] X. Bai, N. Zhang, X. Cao, and W. Chen, “Prediction of PM 2.5 concentration based on a CNN-LSTM neural network algorithm,” PeerJ, vol. 12, p. e17811, Aug. 2024, doi:10.7717/peerj.17811.

[35] S. Shen, C. Li, A. van Donkelaar, N. Jacobs, C. Wang, and R. V. Martin, “Enhancing Global Estimation of Fine Particulate Matter Concentrations by Including Geophysical a Priori Information in Deep Learning,” ACS ES&T Air, vol. 1, no. 5, pp. 332–345, May 2024, doi:10.1021/acsestair.3c00054.

[36] X. Feng et al., “A hybrid model for enhanced forecasting of PM2.5 spatiotemporal concentrations with high resolution and accuracy,” Environmental Pollution, vol. 355, p. 124263, Aug. 2024, doi:10.1016/j.envpol.2024.124263.

[37] A. Salcedo-Bosch, L. Zong, Y. Yang, J. B. Cohen, and S. Lolli, “Forecasting particulate matter concentration in Shanghai using a small-scale long-term dataset,” Environ. Sci. Eur., vol. 37, no. 1, p. 47, Mar. 2025, doi:10.1186/s12302-025-01068-y.

[38] Y. Hu, Y. Ding, and W. Jiang, “Geographically Aware Air Quality Prediction Through CNN-LSTM-KAN Hybrid Modeling with Climatic and Topographic Differentiation,” Atmosphere (Basel)., vol. 16, no. 5, p. 513, Apr. 2025, doi:10.3390/atmos16050513.

[39] T. S. Anggraini, H. Irie, A. D. Sakti, and K. Wikantika, “Machine learning-based global air quality index development using remote sensing and ground-based stations,” Environmental Advances, vol. 15, p. 100456, Apr. 2024, doi:10.1016/j.envadv.2023.100456.

[40] G. Attanayake et al., “An RF-CNN pipeline for predicting PM2.5 concentration in Sri Lanka,” J. Hazard. Mater. Adv., vol. 19, p. 100782, Aug. 2025, doi:10.1016/j.hazadv.2025.100782.