Fig. 1. Distribution of ozone monitoring stations and TOAR stations in China

Fig. 2. Schematic diagram of grid matching

Fig. 3. Validation of ozone model accuracy. (a) ET model's sample-based cross-validation accuracy; (b) ET-XGBoost model's sample-based cross-validation accuracy; (c) ET-XGBoost model's station-based cross-validation accuracy; (d) ET-XGBoost model's time-based cross-validation accuracy; (e) additional site verification accuracy

Fig. 4. Annual validation of ozone model performance

Fig. 5. Quarterly applicability accuracy validation of the ozone model. (a) Spring; (b) summer; (c) autumn; (d) winter

Fig. 6. Accuracy validation of the applicability of ozone model historical data

Fig. 7. Accuracy validation of regional scale applicability of ozone models. (a) Eastern; (b) central; (c ) western

Fig. 8. Comparison of ozone model estimates with site values. (a) Day scale; (b) monthly scale; (c) annual scale

Fig. 9. Average monthly MDA8 O₃ values from 2001 to 2020. (a)–(l) From January to December

Fig. 10. Average monthly MDA8 O₃ values in China and five urban agglomerations from 2001 to 2020

Fig. 11. Annual average MDA8 O₃ values from 2001 to 2020. (a)–(t) From 2001 to 2020

Fig. 12. Average annual MDA8 O₃ values in China and five urban agglomerations from 2001 to 2020

Fig. 13. Average annual MDA8 O₃ values for 2001–2010 (a) and 2011–2020 (b), and the difference in MDA8 O₃ for the first 10 years of the latter decade (c)