Exploring the impacts of key drivers on ground-level ozone in Southeast China using machine learning

Abstract

As a major atmospheric contaminant, ground-level ozone (O3 ) necessitates comprehensive analysis of the determinants influencing its concentration. Although recent researches have greatly improved our understanding of O3 formation in urban areas, the role of biogenic interactions in densely vegetated ecosystems remains unclear. In this study, high resolution measurements of ground-level O3 and environmental parameters were carried out in Fujian province, with the highest forest coverage in China. Key drivers that contribute to O3 formation were identified and quantified using an extreme gradient boosting model. Surface downward solar radiation, PM2.5, relative humidity, and air temperature and forest coverage were the most important variables in the O3 prediction. Especially, these findings revealed a significant positive relationship between forest coverage and O3 concentration. Forest coverage leads to an increase of O3 concentration up to 11.1 mu g/m3. Understanding the roles of these variables prove essential to determine the reason why O3 formation is driven by meteorological conditions and biogenic sources in Southeast China, and suggest the challenge of air pollution control with climate warming in the future.