During past decades China experienced rapid industrialization, which brought economic advantages, but also caused significantly deteriorated air quality. According to WHO guidelines, ozone (O₃) and fine particulate matter (PM_2.5) are the chief hazards of China's air pollution. Over the past decade, China launched the toughest-ever clean air policy in the country. These efforts, by reducing emissions of primary pollutants such as volatile organic compounds (VOCs), nitrogen oxides (NO_X) and others, improved PM_2.5 air quality. However, O₃ continued to increase. This contrasts with the generally decreasing O₃ trend in Europe and North America due to continuous reductions of VOCs and NO_X emissions since the 1990s. The causes of the contrasting trends and adverse O₃ pollution trends in China are still not well understood, representing a grand challenge for O₃ pollution control.

This study is based on August measurements in Beijing from 2006 to 2020, which is the most comprehensive, long-term observational data set of O₃ and its precursors VOCs and NO_X that is available in China. Using the dataset, we show that O₃ concentrations increased before 2014, and then decreased in the following years (Figure 1B). By removing the impact of meteorological variability on O₃, we obtain a normalized O₃, which still shows a non-monotonic trend (Figure 1D). The OH reactivity of VOCs (VOC^R) and NO_X (NO_X^R) show consistently decreasing trends (Figure 1A), confirming the effects of emission controls. The trend of the ratio between VOC^R and NO_X^R (VOC^R/NO_X^R) is similar to that of normalized O₃; both remained approximately constant before 2014, and then decreased rapidly after 2014 (Figure 1C).

Figure 1. Variations in ozone and OH reactivity of precursors derived from observations in Beijing, August between 2006 and 2020. Variations in daytime (6:00-19:00) averages of VOC^R (including primary hydrocarbons plus carbon monoxide), NO_X^R and column concentrations of NO₂ (A), MDA8 O₃ concentrations (B), the ratio of VOC^R to NO_X^R (C), and normalized MDA8 O₃ concentrations (D). Dashed lines show linear regression fits to observations in periods before and after 2014. Model-simulated MDA8 O₃ concentrations are included in (B).

To gain a deeper insight, we investigated the detailed radical cycling and photochemistry by applying an advanced box model based on the updated Master Chemical Mechanism v3.3 chemistry scheme (MCM 3.3). The box model simulation shows that OH increased continuously during 2006-2020, giving an approximate doubling (Figure 2A). The main drivers of the hydroxyl radical (OH) change are the increase of photolysis frequencies, decrease in NO_X emissions, decrease in VOC emissions, and reduced heterogeneous uptake (Figure 2B). The OH turnover rate induced by VOCs ([OH]×VOC^R) shows non-monotonic rends, which is consistent with the trend of observed O₃ concentrations (Figure 2C). This demonstrates that the change of ambient O₃ over the entire 2006 to 2020 period is consistent with our understanding of O₃ photochemistry and OH cycling. Further analysis suggests that VOC^R/NO_X^R affects the O₃ trend primarily through changing the primary production of atmospheric radicals, rather than changing the radicals’ chain length.

Figure 2. Variations in model-simulated OH concentrations and its driving forces in Beijing. (A) Variations in simulated daytime average OH concentrations (red dots) and measured and simulated (MVK+MACR)/isoprene ratios (blue dots and circles) in Beijing in August between 2006 and 2020. Dashed lines show exponential regression fits to trends. (B) Apportionment of the causes of the change in OH derived from box model sensitivity analysis. (C) Variations in OH turnover rate induced by VOCs (OH×VOC^R). Dashed lines show linear regression fits to trends in periods before and after 2014. (D) Correlation of OH×VOC^R with observed MDA8 O₃ and daytime O_X concentrations.

To propose a precise O₃ mitigation strategy, we simulate the combined effect of VOCs and NO_X on normalized O₃ and total OH turnover rate induced by all reactive gases (including VOCs, NO_X and SO₂) (Figure 3). The Beijing environment during 2006-2020 was overall in the VOC-limited regime for both O₃ and total OH turnover rate, with a trend of moving toward to the transition regime. Figure 3 clearly shows that the turning point of the O₃ trend appears in 2014 due to a faster decrease in VOC^R/NO_X^R after 2014 than before 2014. Future projection scenarios show that reactive VOCs abatement should receive near-future priority for O₃ pollution control in China, followed by further NO_X controls.

Figure 3. Isopleth plots for (A) ozone and (B) total OH turnover rate as a function of NO_X^R and VOC^R. Red solid circles represent the average levels of MDA8 O₃ and daytime OH turnover rate for each year during 2006-2020, and the solid lines indicate fits to the trends during 2006-2014 and 2014-2020. The yellow dotted lines indicate the VOC^R level from biogenic emissions. The blue and magenta arrows indicate future routes for VOCs and NO_X abatement that mitigate ozone effectively in the short term (2020-2025) and long term (after 2025), respectively.

In summary, we find that the different O₃ trends can be well explained and predicted by a novel parameter, the reactivity ratio of VOCs to NO_X. Our results show that reducing the reactivity ratio is the most effective way to control O₃ pollution in China in the near future. With this approach, optimal emission reduction roadmaps can be derived directly from measurements of reactivity ratios, avoiding uncertainties of models and emission inventories.

For more detailed information, see our article “Ozone pollution mitigation strategy informed by long-term trends of atmospheric oxidation capacity” in Nature Geoscience.