An Analytical Study of Atmospheric Pollution Levels in the Maharashtra Region

Abstract

Air pollution represents a major environmental and public health concern in Maharashtra, driven by rapid urbanization, industrial growth, and increasing vehicular emissions. This study presents a region-specific statistical and forecasting assessment of atmospheric pollutants across six major regions - Amravati, Chhatrapati Sambhajinagar (Aurangabad), Konkan, Nagpur, Nashik, and Pune using long-term monitoring data obtained from the Maharashtra Pollution Control Board (MPCB). Station-level observations were standardized and aggregated into monthly regional averages for AQI, RSPM, SPM, NOx, SO2, and CO. Descriptive statistical analysis and correlation assessment indicate that particulate matter is the dominant contributor to air quality variability, with AQI exhibiting strong positive association with RSPM across regions. Marked spatial heterogeneity is observed, reflecting differences in emission sources, industrial intensity, population density, and regional meteorological conditions. Seasonal patterns further highlight the influence of atmospheric dispersion and climatic factors on pollutant concentration levels. To examine temporal dynamics and predictive capability, three univariate time-series forecasting frameworks Prophet, ARIMA, and ETS were implemented and comparatively evaluated using scale-dependent and percentage-based error metrics. The results demonstrate that forecasting accuracy varies across pollutants and regions. Models perform reliably for series characterized by clear trend and seasonal structure, particularly for AQI and particulate matter. However, pollutants exhibiting intermittent behavior or near-zero concentrations show unstable percentage-based error measures, underscoring statistical limitations of certain evaluation metrics in environmental applications.   Overall, the study establishes a statistically rigorous framework for regional air quality assessment and forecasting in Maharashtra. The findings highlight the importance of model selection, metric stability, and regional context in environmental time-series analysis, providing quantitative evidence to support informed regulatory planning and public health interventions.