Search

The daily variations of near-surface ozone (O 3), one of the major air pollutants in China, depend on meteorological conditions. In this study, we analyzed the characteristics of O 3 pollution and the complicated interactions among the controlling meteorological factors. In China, O 3 pollution is becoming severe and high-concentration areas are expanding greatly. We found that O 3 was correlated to numerous meteorological variables including air pressure, temperature, relative humidity, wind speed, wind direction, sunshine duration, evaporation, and precipitation. Using generalized additive models (GAMs), we developed predictive models to forecast the maximum daily 8 h O 3 mass concentration for 334 cities in China. Different cities were influenced by different meteorological variables, with temperature (90.77%), relative humidity (68.15%), and sunshine hours (66.96%) being the top three influencing factors. In addition, we found that the influence of these meteorological factors on O 3 pollution was nonlinear and impacted by the interaction between variables. Considering individual relationships between each meteorological factor and O 3 , the average coefficient of determination (R2) of GAMs for the 334 cities was 0.64 ± 0.16. However, considering the interaction between variables, the average R2 for 334 cities of revised GAMs increased to 0.72 ± 0.15. The R2 was significantly improved in 96.72% of the cities after considering the interactions among the meteorological factors based on a ten-fold cross-validation method. Furthermore, we found that the GAM residuals were unbiased. As an example of using and interpreting the GAM results, we focused on a case study in Hangzhou during the G20 summit in 2016. During this period, nitrogen oxides (NO x) and volatile organic compounds were reduced by 51.67% and 49.00%, respectively, because of local emission controls. Despite these changes in precursor emissions, O 3 changed slightly during this period, suggesting that O 3 in Hangzhou at the time was not controlled by local precursor emissions, but was impacted by meteorological conditions. • GAMs could assess O 3 concentration by considering time variable, non-linear relationship, and meteorological interactions. • T, RH and sunshine hours are top three influence meteorological factors on O 3 concentration for 75.84% cities in China. • GAMs can capture the O 3 concentration accurately even if there were reductions of NOx and VOCs: 51.67% and 49.00%. [ABSTRACT FROM AUTHOR]