Decadal Reversal of Global Surface Ozone Pollution

Ambient ozone (O ₃ ) pollution poses a critical threat to global environmental and public health. We reconstructed two decades (2000–2021) of historical daily surface O ₃ concentrations across global land from space using deep learning. 91% of land area exceeded the World Health Organization (WHO) air quality guideline (AQG) for annual maximum daily 8-hour average (MDA8) O ₃ concentrations of 30 ppb, affecting 90% of the world’s population. 96% of the land area and the entire population experienced exceedances of the WHO daily AQG level (50 ppb). We identified decadal trend reversals in population-weighted O ₃ concentrations across 90% of countries, with turning points clustered around 2011; 74% followed a distinct “decline–increase” trajectory. While over half of land areas showed significant declines in exceedance days prior to 2011, these improving trends reversed sharply thereafter, affecting 71% of the global population and indicating a widespread reversal in acute O ₃ exposure. A parallel reversal pattern was observed in mortality trends attributable to long-term O ₃ exposure, with a rising health burden linked to deteriorating O ₃ pollution in most nations. Explainable artificial intelligence analysis reveals that this trend reversal reflects a global rebalancing of driving factors. Meteorological factors continue to explain most variance (68%). Anthropogenic influences have grown in over half of the countries since the turning points. These findings call for strengthened precursor controls in densely populated regions and enhanced climate adaptation in areas where atmospheric conditions dominate O ₃ levels.