Evaluation on Impact of Spring Festival Fireworks and Firecrackers Setting-off on Yancheng City Based on a Bayesian-Optimized XGBoost Model
To address the significant increase in fine particulate matter (PM2.5) and its chemical component concentrations caused by the concentrated setting-off of fireworks and firecrackers during the Spring Festival in Yancheng City, this study introduced the Bayesian-optimized XGBoost model (BO-XGBoost) based on PM2.5, particulate component, and meteorological observation data. The model simulated non-setting-off baseline concentrations using meteorological factors as independent variables, enabling quantitative assessment of setting-off contributions. Results showed that the concentrated setting-off exerted significantly differentiated effects on various pollutants. Among water-soluble ions, K+ and Mg2+ were core characteristic tracers, with concentrations reaching 15.89 and 20.08 times baseline levels on Lunar New Year's Eve, directly reflecting high-intensity emissions. Secondary conversion ions such as SO4^2- and NO3^- showed sustained high contributions on both Lunar New Year's Eve and the fifth day of the first lunar month, reflecting cumulative effects of atmospheric chemical transformation. Among inorganic elements, K, Ba, and Sr were core characteristic tracers, with concentrations showing explosive growth on Lunar New Year's Eve, serving as direct fingerprints of fireworks. Elements such as Pb and Mn were also significantly affected, reflecting direct heavy metal emissions. Temporal comparisons indicated that emission intensity on Lunar New Year's Eve was significantly higher than on the fifth day, with increased proportional contribution of secondary conversion processes on the fifth day. The study achieved accurate quantification of setting-off contributions through a data-driven model, clarifying pollution fingerprint characteristics and temporal differentiation patterns, providing scientific basis for air quality management and policy optimization during the Spring Festival.