Revealing Hidden Episodic Sources in Complex Urban Atmospheres: A Case Study of Firework Events during the ASIA-AQ Campaign

Accurate source characterization of particulate matter requires the simultaneous analysis of chemical composition and the ability to distinguish between continuous background emissions and transient episodic events. Traditional source apportionment methods, such as Positive Matrix Factorization (PMF), often smooth out short-term spikes when applied to long-term datasets, effectively "averaging" high-intensity episodes into stable source profiles. This limitation poses a significant challenge in complex Asian urban atmospheres, where specific pollution events can dominate short-term air quality deterioration yet remain obscured in annual averages.

This study utilizes high-time-resolution measurements of PM_2.5 chemical composition collected during the 2024–2025 ASIA-AQ campaign in Southern Taiwan. We applied a dynamic source apportionment approach to resolve episodic sources that are typically difficult to identify using conventional long-term analysis.

Our analysis identifies distinct "firework-related emissions" factors that are strictly episodic. These factors were characterized by high loadings of Bismuth (Bi) and specific trace metal signatures. Results indicate that while these sources contribute minimally to the annual average PM_2.5 mass, they are the dominant contributors (> 50%) during specific pollution episodes. Failing to isolate these episodic factors leads to the misattribution of pollution mass to other continuous sources, such as traffic or industrial emissions.

This study demonstrates that relying solely on long-term average source apportionment may underestimate the health risks associated with acute exposure events. The proposed event-driven analysis framework provides a more accurate scientific basis for targeted control strategies in highly polluted environments.