Forecasting and Assessment of Air Quality Dynamics in Northeast India Using Machine Learning Models

Authors

Kumar Shubham, Department of Civil Engineering, National Institute of Technology Patna, Patna, Bihar-800005, IndiaFollow
Gopikrishnan T, Department of Civil Engineering, National Institute of Technology Patna, Patna, Bihar-800005, IndiaFollow
Anshuman Singh, Department of Civil Engineering, National Institute of Technology Patna, Patna, Bihar-800005, IndiaFollow

Publication Date

3-1-2025

Abstract

This study investigated air quality dynamics in Northeast India, a region with unique terrestrial features, including the Eastern Himalayas. While air quality varies across districts, pollution impacts the entire area. Northeast India’s rich ecology is crucial for Himalayan climate regulation. Robust air quality monitoring and pollution control are essential to preserve environmental balance. This work focused on forecasting emissions of aerosols, SO₂, NO₂, CO, HCHO, O₃, and CH₄ primarily associated with human activities. Utilizing data from the Tropospheric Monitoring Instrument (TROPOMI) satellite instrument from 2019 to 2023, a 9-mo forecast was conducted using 5 machine learning models: random forest, cubic regression, linear regression, quadratic regression, and k-nearest neighbors' algorithm (KNN) models. The effectiveness of models was evaluated through R², mean square error (MSE), and mean absolute error (MAE). The results showed a strong alignment between regional dynamics and models with low MSE and high R² values. Perpetual air quality monitoring is crucial for region-specific modeling and solutions. Gas concentration variations emphasize the need for regularly updated air quality reports. The random forest model was found to be most effective with high R² values: UV aerosol index (0.97 in Imphal, Aizawl), CO (0.96 in Imphal), NO₂ (0.92 in Gangtok), O₃ (0.98 in Gangtok), SO₂ (0.92 in Gangtok), and CH₄ (1.00 in Itanagar, Shillong). Correlation analysis with Central Pollution Control Board (CPCB) data showed notable results for Aerosol-PM2.5 (0.76 in Imphal) and Aerosol-PM10 (0.79 in Imphal). Findings from this study may help identify effective machine learning models for forecasting and assessing air quality.

Recommended Citation

Shubham, Kumar; T, Gopikrishnan; and Singh, Anshuman (2025) "Forecasting and Assessment of Air Quality Dynamics in Northeast India Using Machine Learning Models," The Philippine Agricultural Scientist: Vol. 108: No. 1, Article 4.
Available at: https://www.ukdr.uplb.edu.ph/pas/vol108/iss1/4