Spatiotemporal evolutions and drivers of ground-level ozone in China (2015–2020): A GTWR-Kriging approach

Ground-level ozone pollution has emerged as a primary environmental challenge in China. An accurate and high-resolution analysis of ground-level ozone concentrations is crucial for effectively mitigating pollution and achieving sustainability goals. However, previous studies had inherent limitations in fulfilling this requirement from both data and method perspectives. Additionally, comprehensive analyses of spatiotemporal evolution and factors influencing ozone pollution are scant, particularly those based on accurate pollution delineation. This study introduces a Geographically and Temporally Weighted Regression-Kriging (GTWR-Kriging) model to address spatial and temporal non-stationarity and thus enhance model prediction performance. Model validation and comparison demonstrate that the GTWR-Kriging model approaches the estimation accuracy of general machine learning models while surpassing traditional linear models. Importantly, it maintains strong interpretability regarding factors influencing ozone pollution. This study identifies that CO2 anthropogenic emissions, 10 meter V wind component, and leaf area index with low vegetation, surface net thermal radiation, total precipitation, and grassland are primary drivers of ozone pollution. Leveraging ozone-related big data, this study extends the GTWR-Kriging model nationally, generating high-resolution (1km × 1km) maps of ground-level ozone concentrations across China from 2015 to 2020. A spatiotemporal analysis across 337 prefectural cities identifies four Primary Prevention and Control Regions for Ozone Pollution. Theoretically, the development of the GTWR-Kriging model enriches the literature for accurate and high-resolution ozone studies and environmental science. Practically, empirical insights from the 1km×1km ozone maps and influencing factors support tailored and evidence-based policy-making for effective ozone pollution prevention and control.