Air Quality Study of Three Non-attainment Zones of West Bengal: A Modelling Approach

Abstract

West Bengal has seven of 132 air pollution non-attainment zones identified by Central Pollution Control Board (CPCB) in 2019. This study examines Paschim Bardhaman, comprising three of these industrial areas. Data was collected on Particulate Matter (PM₁₀), Nitrogen Dioxide (NO₂), and Sulphur Dioxide (SO₂) from West Bengal Pollution Control Board (WBPCB) for Asansol, Raniganj, Benachity, Angadpur, and PCBL More. Time series forecasting was performed using Seasonal ARIMA (SARIMA), Exponential Smoothing (ETS), Artificial Neural Networks (ANN), Trigonometric Seasonality, Box-Cox Transformation, ARMA Errors, and Trend and Seasonal Components (TBATS). Best-fit model was determined through Time Series Cross-Validation (TSCV) and Hold-out validation methods, ANN was identified as the optimal model for all study sites. Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) comparisons validated the accuracy of ANN. Normalized Mean Squared Error (NMSE) values for PM₁₀ ranged from 0.153 to 0.489, for NO₂ from 0.118 to 0.347, and for SO₂ from 0.009 to 0.318 across the sites. Results highlight superiority of ANN due to its minimal overprediction and underprediction. Raniganj exhibited the highest PM₁₀ levels, followed by Benachity. Projections indicate Benachity will have the highest NO₂ levels, followed by Asansol and Raniganj. SO₂ concentrations were within acceptable limits. PM₁₀ is the primary pollutant at all sites, with significant variations in concentrations across the cities, none of which meet prescribed levels of CPCB. This highlights the immediate need for effective mitigation strategies.