Advancing PM2.5 Source Apportionment through Dispersion Normalized PMF: A Comprehensive Study across India

Authors:

Abstract
The recent IPCC report highlights PM2.5 as one of the most important contributors to air pollution and its health impacts. Determining the sources of air pollution, quantifying their contributions to ambient pollutant levels, and characterizing their spatial and temporal patterns are all important steps in developing effective mitigation strategies (Hopke, 2016). Source apportionment studies, especially those using receptor models, are critical for this purpose. Of these, Positive Matrix Factorization (PMF) is a widely applied methodology. The current work evolves standard PMF (S-PMF) by Tapper and Paatero, 1994 based on the implementation of dispersion-normalized PMF (DN-PMF) suggested by Dai et al., 2020 that includes meteorologically-influenced enhancement towards increasing local source emissions resolution. DN-PMF preserves source information otherwise obscured by variable atmospheric conditions, yielding refined and distinct source profiles. This research investigates the spatial and seasonal variability of PM2.5 sources across three Indian cities—Bhopal (MSL: 486 m), Mesra (MSL: 517 m), and Mysuru (MSL: 759 m)—as part of the COALESCE network in 2019. A Multiple Seasonal-Trend Decomposition using LOESS (MSTL) was applied to mixed layer height time series to decompose into their daily, weekly, and trend components. Stationarity tests were performed using the Augmented Dickey-Fuller with a p-value