An investigation into ozone production based on statistical modeling in Jinan, China

This study analyzed the effects of atmospheric pollutants and meteorological factors on ozone (O₃) concentrations in Jinan using 2017-2023 daily data. Statistical analysis showed that Jinan's annual average O₃ concentration has an upward trend, with pronounced seasonal variations (higher in spring/summer) coinciding with more O₃ exceedance days. Agglomerative hierarchical clustering, employing Euclidean distance as the similarity measure, was applied to classify 2556 daily sea-level pressure fields into 6 types. Marked differences in O₃ concentrations were observed among these pressure field categories, with lower-pressure systems consistently associated with elevated O₃ levels. To identify the key drivers of O₃ exceedance events, the eXtreme Gradient Boosting (XGBoost) model was constructed, and its classification results were interpreted using the SHapley Additive exPlanations (SHAP) method. This analysis identified the top three meteorological determinants of O₃ exceedance as follows: 2-m air temperature, ultraviolet albedo for direct radiation, and relative humidity. Furthermore, to elucidate the chemical mechanisms governing O₃ formation, threshold regression analysis was performed for the period from May to September-recognized as the critical window for O₃ pollution in Jinan. The results demonstrated that nitrogen dioxide (NO₂) and formaldehyde (HCHO) promote O₃ formation through photochemical reactions. Notably, O₃ formation was found to be significantly intensified when the formaldehyde-to-nitrogen dioxide ratio (FNR) exceeded a threshold value of 0.893. By innovatively applying threshold regression to dissect the influence of FNR on O₃ formation, this study provides a robust scientific foundation for the formulation of targeted O₃ pollution control strategies in Jinan.