Two-decade surface ozone (O3) pollution in China: Enhanced fine-scale estimations and environmental health implications

Surface ozone (O3) has become a primary pollutant affecting urban air quality and public health in mainland China. To address this concern, we developed a nation-wide surface maximum daily average 8-h (MDA8) O3 concentration dataset for mainland China (ChinaHighO3) at a 10-km resolution with a start year of 2013, which has been widely employed in a wide range of studies. To meet the increasing demand for its usage, we have made important enhancements, including the development of a more advanced deep-learning model and the incorporation of major source updates, such as 1-km surface downward shortwave radiation and temperature directly from satellite retrievals, as well as a 1-km emission inventory. Additionally, we have extended the temporal coverage dating back to 2000, increased the spatial resolution to 1 km, and most importantly, notably improved the data quality (e.g., sample-based cross-validation coefficient of determination = 0.89, and root-mean-square error = 15.77 μg/m3). Using the substantially improved new product, we have found dynamic and diverse patterns in national surface O3 levels over the past two decades. Peak-season levels have been relatively stable from 2000 to 2015, followed by a sharp increase, reaching peak values in 2019 and subsequently declining. Additionally, we observed a large relative difference of 12 % in peak-season surface O3 concentrations between urban and rural regions in mainland China. This disparity has greatly increased since 2015, particularly in the Beijing-Tianjin-Hebei and Pearl River Delta regions. Notably, since 2000, nearly all of the population across mainland China (> 99.7 %) has resided in areas exposed to surface O3 pollution exceeding the World Health Organization (WHO) recommended long-term air quality guideline (AQG) level (peak-season MDA8 O3 = 60 μg/m3). Moreover, the short-term population-risk exposure to daily surface O3 pollution has shown a significant increasing trend of 1.2 % (p