The effect of anthropogenic emission, meteorological factors, and carbon dioxide on the surface ozone increase in China from 2008 to 2018 during the East Asia summer monsoon season

Despite the implementation of the Clean Air Action Plan by the Chinese government in 2013, the issue of increasing surface ozone (O₃) concentrations remains a significant environmental concern in China. In this study, we used an improved regional climate–chemistry–ecology model (RegCM-Chem-YIBs) to investigate the impact of anthropogenic emissions, meteorological factors, and CO₂ changes on summer surface O₃ levels in China from 2008 to 2018. Compared to its predecessor, the model has been enhanced concerning the photolysis of O₃ and the radiative impacts of CO₂ and O₃. The investigations showed anthropogenic emissions were the primary contributor to the O₃ increase in China, responsible for 4.08–18.51 ppb in the North China Plain. However, changed meteorological conditions played a crucial role in decreasing O₃ in China and may have a more significant impact than anthropogenic emissions in some regions. Changed CO₂ played a critical role in the variability of O₃ through radiative forcing and isoprene emissions, particularly in southern China, inducing an increase in O₃ on the southeast coast of China (0.28–0.46 ppb) and a decrease in southwest and central China (−0.51 to −0.11 ppb). Our study comprehensively analyzed O₃ variation across China from various perspectives and highlighted the importance of considering CO₂ variations when designing long-term O₃ control policies, especially in high-vegetation-coverage areas.

1 Introduction

O₃ is a strong oxidant detrimental to human health (Lu et al., 2020; Liu et al., 2018a) and vegetation growth (Monks et al., 2015; Wang et al., 2017a). Furthermore, it is a crucial active compound influencing the earth's radiative balance, with an effective radiative forcing of up to 0.47 W m⁻² in 2019 (IPCC, 2021). Tropospheric O₃ has garnered significant attention over the past few decades due to its crucial role in air quality and climate change (Duan et al., 2017; Li et al., 2019; Ashmore and Bell, 1991; Lu et al., 2018).

With the rapid development in China, emissions of O₃ precursors have been on the rise, leading to an annual increase in O₃ concentrations since the beginning of the 20th century (Liu and Wang, 2020a; Ma et al., 2016). Surface O₃ pollution has become a severe air quality concern in China (Verstraeten et al., 2016; Xu et al., 2018), particularly in major urban areas such as the North China Plain (NCP), Fenwei Plain (FWP), Yangtze River Delta (YRD), Pearl River Delta (PRD), and the Sichuan Basin (SCB) (Y. Wang et al., 2020; T. Wang et al., 2017a; Yin and Ma, 2020; Shen et al., 2019; Zhao et al., 2018; X. Wang et al., 2009). Although the Chinese government initialized the Clean Air Action Plan in 2013 to control air pollution, the concentration of O₃ precursors and PM_2.5 has significantly decreased (Zhai et al., 2019). However, surface O₃ concentrations continue to increase in major urban areas.

Recent studies have suggested that regional meteorological conditions influence surface O₃ levels through various pathways (Jacob and Winner, 2009; Shen et al., 2016; Lin et al., 2008). Modeling studies have shown that O₃ levels are sensitive to temperature, humidity, wind speed, mixing height, and other meteorological conditions (Pfister et al., 2014; Sanchez-Ccoyllo et al., 2006). For instance, temperature affects the chemical formation rate of O₃ (Lee et al., 2014), while precipitation reduces surface O₃ concentrations through wet removal (Fang et al., 2011). Additionally, the elevated planetary boundary layer (PBL) height enhances upward movement, resulting in lower surface O₃ concentrations (Haman et al., 2014). Therefore, long-term modeling of surface O₃ levels must consider changes in meteorological conditions.

CO₂ is the primary anthropogenic radiative force of the climate system (Gauss et al., 2003; Schimel et al., 2015). CO₂ can impact regional air temperature and precipitation, leading to changes in surface O₃ concentrations (Lu et al., 2013; Yang et al., 2014).

On the other hand, biogenic volatile organic compounds (BVOCs) are significant O₃ precursors, and isoprene is the primary species among BVOCs that vegetation emits (Zheng et al., 2009; Fiore et al., 2011). In most of China, O₃ is volatile organic compound (VOC) limited in the summer, especially in industrial cities (Li et al., 2018; Wu et al., 2018). Thus, it plays a significant role in modulating O₃ levels and positively correlates with O₃ concentrations in major urban areas of China. It is known that CO₂ can enhance vegetation's photosynthesis (Sun et al., 2013; Heald et al., 2009; Tai et al., 2013; Monson and Fall, 1989), which may directly increase isoprene emission (Rapparini et al., 2004). Based on the observation, Rosenstiel et al. (2003) found that the isoprene emissions of plants grew by about 21 % and 41 % when CO₂ reached 800 and 1200 ppm, respectively. However, Wilkinson et al. (2009) indicated that different vegetation types show varying responses in isoprene emission when CO₂ increases. Isoprene emission was decreased by 30 %–40 % in Populus tremuloides Michx but increased by about 100 % in Quercus rubra when CO₂ concentrations were grown (Sharkey et al., 1991). High concentrations of CO₂ may inhibit the emission of isoprene by reducing the activity of BVOCs synthetase or decreasing the synthesis of adenosine triphosphate (Possell et al., 2005). Guenther et al. (1991) also indicated that isoprene emissions were significantly reduced when CO₂ was increased from 100 to 600 µmol mol⁻¹. In summary, the impact of elevated CO₂ on isoprene emission may be positive or negative, mainly related to the relative size of the inhibition caused by elevated CO₂ and promotion by enhanced photosynthesis.

Numerous studies have concluded that anthropogenic emissions are the primary drivers of surface O₃ increases in different regions or years in China. Meanwhile, the effects of meteorological parameters can be substantial (Wang et al., 2019c; Lu et al., 2019; Dang et al., 2021; Liu and Wang, 2020a). For instance, Li et al. (2020) indicated that anthropogenic emissions were the primary cause of surface O₃ increase in China from 2013 to 2019. Liu and Wang (2020a) suggested that anthropogenic emissions play a dominant role in the O₃ variety in China, but the effects of meteorological conditions could be more significant in some regions. Han et al. (2020) analyzed the O₃ changes in summer and suggested that meteorology can explain about 43 % of that in eastern China.

Previous studies have mainly focused on the impact of anthropogenic emissions and meteorological factors on the rise of O₃ levels, with limited attention given to the role of CO₂ variations. However, due to the rapid socioeconomic growth in China and the subsequent surge in energy consumption, CO₂ emissions and concentrations have also increased significantly, particularly in the eastern coastal region (Lv et al., 2020; Ren et al., 2014). Furthermore, given the significant impact of CO₂ on O₃, it is crucial to evaluate the influence of changes in CO₂ concentration on the maximum daily 8 h average (MDA8) O₃ concentrations at the surface. Thus, a comprehensive assessment of the impact of anthropogenic emissions, meteorological factors, and CO₂ on surface O₃ is imperative.

Here, we employed an advanced regional climate–chemistry–ecology model to assess the impact of anthropogenic emissions, meteorological factors, and carbon dioxide variations during the summer monsoon period (May, June, July, and August) on surface O₃ levels. Our findings can facilitate the development of a comprehensive O₃ improvement strategy. Section 2 describes the methods and data, and the results and discussion are given in Sect. 3; finally, the conclusions are shown in Sect. 4.

2 Methods and data

2.1 Measurement data

We compared the simulated regional meteorological factors with the European Centre for Medium-Range Weather Forecasts Interim reanalysis data (ERA-Interim) at 37 vertical levels, which included temperature, relative humidity, and wind speed (Balsamo et al., 2015; Hoffmann et al., 2019). The observed surface O₃ was taken from the China National Environmental Monitoring Center (CNEMC), which had more than 1400 environmental monitoring stations in 2018 (Wang et al., 2018; Kong et al., 2021; Zheng et al., 2014). The World Data Centre for Greenhouse Gases (WDCGG) data (Liu et al., 2009; Li et al., 2017) were applied to evaluate the simulated surface CO₂ concentrations. The monitoring stations of O₃ and CO₂ are shown in Fig. 1.

Figure 1 Model domains for the RegCM-Chem-YIBs model. The regions with black boundaries are the North China Plain (34–41^∘ N, 113–119^∘ E), the Yangtze River Delta (30–33^∘ N, 119–122^∘ E), the Pearl River Delta (21.5–24^∘ N, 112–115.5^∘ E), the Sichuan Basin (28.5–31.5^∘ N, 103.5–107^∘ E), and the Fenwei Plain (33.5–39^∘ N, 106–113^∘ E) regions.

2.2 Model description

The RegCM-Chem-YIBs is a regional climate–chemistry–ecology model developed from the RegCM model. RegCM is a regional climate model initially developed by the International Centre for Theoretical Physics (ICTP) (Giorgi et al., 2012). Shalaby et al. (2012) integrated the Chem chemistry model into the RegCM model and incorporated the condensed version of the Carbon Bond Mechanism Z (CBM-Z) to enhance the model's capabilities. To further improve the model's performance, Yin et al. (2015) added a volatility basis set (VBS) scheme to simulate secondary organic aerosols (SOAs). Xie et al. (2020) further modified the model by incorporating CO₂ as a tracer, which is subject to regulation by sources, sinks, and atmospheric transport processes. The model represents the four sources and sinks of CO₂ as surface fluxes, including emissions from fossil fuels and biomass burning, air–sea CO₂ exchange, and terrestrial biosphere CO₂ fluxes. Additionally, the model incorporated the Yale Interactive Terrestrial Biosphere (YIBs), a land carbon cycle model that enables the simulation of ecological processes, including carbon assimilation, allocation, and autotrophic and heterotrophic respiration (Yue and Unger, 2015).

The ecological model (YIBs) was fully coupled into the regional climate–chemical model (RegCM-Chem) to reproduce the interactions between atmospheric composition and the ecosystem in the actual atmosphere (Xie et al., 2019). The meteorological factors and air components simulated by RegCM-Chem were input into the YIBs model to simulate vegetation physiological processes and calculate land surface parameters such as carbon dioxide flux, BVOCs emissions, and stomatal conductance. Conversely, the simulations of the YIBs model were fed back into the RegCM-Chem model to adjust the air qualities, temperature, humidity, circulation, and other meteorological fields. The RegCM-Chem-YIBs has been extensively applied to study surface O₃, PM_2.5, CO₂, the summer monsoon, and the interactions between air quality and the ecosystem (Zhuang et al., 2018; Pu et al., 2017; Xu et al., 2022; Xie et al., 2018; Ma et al., 2023).

The RegCM model offers a variety of physical and chemical parameterization options. Here, the climatological chemical boundary conditions were driven by the Model for Ozone and Related Chemical Tracers (MOZART). The gas-phase chemistry employed the CBM-Z scheme (Zaveri and Peters, 1999). For the boundary layer scheme, the Holtslag PBL approach was utilized (Khayatianyazdi et al., 2021). The Grell cumulus convection scheme was employed to simulate convective processes (Grell, 1993). The CCM3 radiation scheme and CLM3.5 land surface module were used in the model (Collins et al., 2006; Giorgi and Mearns, 1999; Decker and Zeng, 2009).

Figure 2 Changes in the surface MDA8 O₃ concentrations (units: ppb) during the summer monsoon period from 2009 (a), 2010 (b), 2011 (c), 2012 (d), 2013 (e), 2014 (f), 2015 (g), 2016 (h), 2017 (i), and 2018 (j) relative to 2008.

Table 1 The numerical experiment in this study.

2.3 Model improvements

2.3.1 Radiation

In the previous version of the RegCM-Chem-YIBs model, radiative calculations only accounted for changes in the spatiotemporal distribution of particulate matter. To simplify the radiation calculations, the atmospheric CO₂ and O₃ concentrations were assumed to be constant throughout the year. However, atmospheric CO₂ and O₃ are subject to modulation by various sources, sinks, physical processes, and chemical processes (Ballantyne et al., 2012; Wang et al., 2019a). Additionally, rapid urbanization in China has led to an annual increase in CO₂ and O₃ concentrations (Guan et al., 2021; Wei et al., 2022), with elevated concentrations and growth rates primarily distributed in the eastern regions where urbanization is most prominent (Shi et al., 2016; Wang et al., 2017b). To more accurately simulate the atmospheric radiation balance and East Asian monsoon climate, it is necessary to incorporate spatiotemporal variations of CO₂ and O₃ concentrations into the radiation module. Therefore, we included the varying CO₂ and O₃ concentrations simulated by the model in the radiation module to calculate the corresponding radiative forcing.

2.3.2 Photolysis

The photolysis process was simulated using the Tropospheric Ultraviolet and Visible (TUV) model, which is commonly used to compute photolysis rates in various models (Tie et al., 2003; Shetter et al., 2002; Borg et al., 2011). The TUV model employs input parameters such as zenith angle, altitude, ozone column, SO₂ column, NO₂ column, aerosol optical depth (AOD), single scattering albedo (SSA), and albedo, among others, to calculate photolysis rates (Singh and Singh, 2004). However, in the TUV module of the RegCM-Chem-YIBs model, AOD and SSA were held constant. This is problematic as accurate aerosol optical parameters, such as AOD and SSA, play a crucial role in the photolysis of O₃ (Lefer et al., 2003). To address this issue, we incorporated temporally and spatially varying AOD and SSA simulated by the RegCM-Chem-YIBs model into the photolysis rate calculations in the TUV module. This enabled us to accurately incorporate the extinction effect of the varying particles into the photolysis reaction, leading to more realistic simulations of air components and regional meteorology.

2.4 Emissions and Experiment settings

Anthropogenic emissions from 2008 to 2018 were obtained from the Multi-resolution Emission Inventory for China (MEIC), which has been compiled and maintained by Tsinghua University since 2010 (Zheng et al., 2018; Wang et al., 2014). CO₂ emissions and boundary conditions were derived from the NOAA CarbonTracker CT2019 dataset (Jacobs et al., 2021). The initial meteorological boundary data, such as temperature, relative humidity, and wind, are derived from the ERA-Interim reanalysis dataset with a horizontal resolution of 0.125^∘, a temporal resolution of 6 h, and 37 vertical levels (Liu et al., 2018b). The weekly mean sea surface temperature dataset was obtained from the National Ocean and Atmosphere Administration (NOAA) (Reynolds et al., 2002).

The simulation domain was illustrated in Fig. 1, with the target region centered at 36^∘ N and 107^∘ E and with a grid resolution of 60 km by 60 km. The model used 18 vertical levels, ranging from the surface to 50 hPa.

In the Base experiment, we incorporated interannual variations in anthropogenic emissions, meteorological fields, and CO₂ emissions. Meteorological conditions (CO₂ emissions) were kept constant at 2008 levels over 10 years, referred to as the SIM_MET=2008 (SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) experiment.

The changes in O₃ concentrations relative to 2008 between 2009 and 2018 were determined by comparing simulations of different years with 2008 in the Base experiment (Eq. 1). The impact of changed meteorological conditions on O₃ concentrations relative to 2008 was assessed by comparing results between SIM_MET=2008 and the Base experiment in the same year (Eq. 2). The contribution of changed CO₂ emissions was similarly estimated (Eq. 3). Finally, the influence of anthropogenic emissions was calculated by excluding the impact of meteorological factors and CO₂ from the changes in O₃ concentrations (Eq. 4). Table 1 shows the results of the numerical experiments.

$$\begin{array}{}\text{(1)}& {\displaystyle }\mathrm{\Delta }{\mathrm{O}}_i={\mathrm{Base}}_i-{\mathrm{Base}}_{\mathrm{2008}}\text{(2)}& {\displaystyle }\mathrm{\Delta }{M}_i={\mathrm{Base}}_i-{\mathrm{SIM}}_{i,\mathrm{MET}=\mathrm{2008}}\text{(3)}& {\displaystyle }\mathrm{\Delta }{\mathrm{C}}_i={\mathrm{Base}}_i-{\mathrm{SIM}}_{i,{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}\text{(4)}& {\displaystyle }\mathrm{\Delta }{E}_i=\mathrm{\Delta }{\mathrm{O}}_i-\mathrm{\Delta }{M}_i-\mathrm{\Delta }{\mathrm{C}}_i\end{array}$$

ΔO_i represents the changes in O₃ concentrations in the year i relative to 2008. Base_i represents the O₃ concentrations in the Base experiment in the year i. ΔM_i represents the changes in O₃ concentrations in the year i due to meteorological factors variations. SIM${}_{i,\mathrm{MET}=\mathrm{2008}}$ represents the O₃ concentrations in the SIM_MET=2008 experiment in the year i. ΔC_i represents the changes in O₃ concentrations in the year i due to CO₂ variations. SIM${}_{i,{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ represents the O₃ concentrations in the SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ experiment in the year i. ΔE_i represents the changes in O₃ concentrations in the year i due to anthropogenic emissions variations.

In this work, both meteorological and CO₂ boundary conditions were kept consistent in base and sensitivity studies. We did not consider the impact of boundary conditions on O₃ due to the following reasons. First, in general, the regional model was coupled with the global model to get a more realistic influence from the boundary. However, for long-term climate–chemistry modeling, such a coupling means a large computing resource. Second, the boundary conditions were derived from global models (Liu et al., 2017; Ban et al., 2014) and have to be prescribed in numerical experiments. Finally, fixed boundary conditions were widely used in some O₃ studies in China (Liu and Wang, 2020a, b; Wang et al., 2019b). Moreover, regional emissions are the primary source of surface O₃ in China, with contributions accounting for 80 % from May to August (Lu et al., 2019). Therefore, the impact of fixed boundary conditions can be ignored in the current stage.

Figure 3 The response of the surface MDA8 O₃ mixing ratios (units: ppb) to variations in meteorological conditions during the summer monsoon period in 2009 (a), 2010 (b), 2011 (c), 2012 (d), 2013 (e), 2014 (f), 2015 (g), 2016 (h), 2017 (i), and 2018 (j) relative to 2008.

Figure 4 The MDA8 O₃ (a–c, units: ppb), precipitation (d–f, units: mm d⁻¹), clouds (g–i, units: %), shortwave flux (j–l, units: W m⁻²), wind speed (m–o, units: m s⁻¹), temperature (p–r, units: K), and planetary boundary layer height (s–u, units: m) during the summer monsoon period in 2008 from the base simulations (the left column) and their responses due to variations in meteorological conditions in PreG (2009–2013, the central column) and PostG (2014–2018, the right column) relative to 2008.

3 Results and discussion

3.1 Model evaluation

The ability of RegCM to reproduce East Asian climate and air quality has been widely evaluated in recent years. Previous studies have demonstrated that RegCM was capable of the essential characteristics and interannual variations of air components and meteorological fields in East Asia (Xu et al., 2023; Ma et al., 2023; Zhuang et al., 2018). Given that the monitoring of near-surface O₃ levels by CNEMC was initiated only in late 2013, the monitoring sites in 2013 and 2014 were limited, and the monitoring period was disjointed. As a result, in this study, we compared the simulated meteorological fields, O₃, and CO₂ levels with observations only from 2015 to 2018.

Figures S1–S4 demonstrated that the RegCM-Chem-YIBs model effectively captured the spatial distribution and magnitude of temperature, humidity, and wind over East Asia at 500, 850, and 1000 hPa between 2015 and 2018. However, due to the complex terrain's influence on the lower atmosphere, most models show better results at higher levels (Zhuang et al., 2018; Anwar et al., 2019; Xie et al., 2019). Thus, the simulations at 500 hPa were more consistent with the reanalysis data. At 1000 hPa, the simulated wind speed was slightly higher than the reanalysis data in eastern China. This difference may be due to common deficiencies in meteorological models, such as insufficient horizontal resolution, initial and boundary conditions, and physical parameterizations (Cassola and Burlando, 2012; Accadia et al., 2007), particularly in areas with low wind speeds (Carvalho et al., 2012).

Figures S5 and S6 demonstrated that the model accurately reproduced the observed increase in surface CO₂ and O₃ from 2015 to 2018, with high correlation coefficients ranging from 0.39 to 0.74 (Table 2). The model effectively captured the high concentrations of O₃ in major urban areas such as the NCP, the YRD, the PRD, the SCB, and the FWP, while also successfully reproducing the gradient in CO₂ concentrations between eastern and western China. However, the model slightly underpredicted MDA8 O₃ concentrations (−4.02 to −3.21 ppb) and overestimated CO₂ levels (3.32–7.07 ppm). These discrepancies are mainly attributed to uncertainties in the emissions inventory (Hong et al., 2017). Overall, the simulated meteorological factors and surface CO₂ and O₃ concentrations were deemed acceptable.

Table 2 Evaluations of the surface CO₂ (units: ppm) and MDA8 O₃ (units: ppb) during the summer monsoon period in East Asia.

OBS: observation; SIM: simulation; MB: bias; RMSE: root mean square error; R: correlation coefficient. MDA8 O₃: the maximum daily 8 h average O₃.

Table 3 The changes of MDA8 O₃ over the North China Plain, Fenwei Plain, Yangtze River Delta, Pearl River Delta, and Sichuan Basin during the summer monsoon period from 2009 to 2018 relative to 2008.

3.2 Ozone variation from 2008 to 2018

Figure S7 illustrates the mean seasonal MDA8 O₃ concentrations in East Asia during the summer monsoon period from 2008 to 2018. High O₃ concentrations appeared in eastern China, which can be attributed to increased emissions, high temperatures, high humidities, and intense radiation in the region (Gao et al., 2020; Mousavinezhad et al., 2021; Wei et al., 2022). Surface O₃ increased annually in most of China between 2008 and 2018, with megacity clusters experiencing a more significant increase.

We conducted a regional analysis of surface O₃ increases in five target regions: the NCP, the YRD, the PRD, the SCB, and the FWP. In 2018, the surface MDA8 O₃ concentrations averaged 74 ppb in the NCP region, while the other areas had lower concentrations (ranging from 42 to 67 ppb in the FWP, YRD, PRD, and SCB). The lower surface O₃ levels in the SCB and FWP regions were attributed to lower anthropogenic emissions. The YRD and PRD regions were more affected by meteorological factors, with the East Asian summer monsoon bringing in cleaner air and precipitation from the sea, leading to lower air pollution concentrations (He et al., 2012). The spatial distribution and increasing trend of surface MDA8 O₃ concentrations in China were consistent with previous studies (Li et al., 2020; Shen et al., 2022).

Figure 2 and Table 3 illustrate the changes in surface MDA8 O₃ concentrations from 2009 to 2018 relative to 2008. The surface MDA8 O₃ concentrations in China increased drastically over the past decade, particularly in 2017 and 2018 (6.79–32.03 ppb). We divided the period from 2009 to 2018 into two phases based on the Clean Air Action Plan implemented in 2013: the pre-governance period (PreG, 2009–2013) and the post-governance period (PostG, 2014–2018). The surface MDA8 O₃ concentration increased significantly in NCP (18.42 ppb), followed by SCB (11.21 ppb), FWP (10.9 ppb), and the YRD (10.07 ppb), while it increased slightly in PRD (4.94 ppb) in PostG relative to 2008. Our results were consistent with previous studies by Lu et al. (2020), Ma et al. (2016), and Mousavinezhad et al. (2021).

Table 4 Response of the MDA8 O₃ mixing ratios (units: ppb), precipitation (units: mm d⁻¹), cloud fraction (units: %), shortwave flux (units: W m⁻²), wind speed (units: m s⁻¹), temperature (units: K), and planetary boundary layer height (units: m) to the changes in meteorological conditions over the North China Plain, Fenwei Plain, Yangtze River Delta, Pearl River Delta, and Sichuan Basin during the summer monsoon period in PreG (2009–2013) and PostG (2014–2018) relative to 2008.

3.3 The effect of meteorology in the 2008–2018 ozone increase

Overall, the meteorological variations from 2008 to 2018 were unfavorable for the O₃ increase during the EASM (East Asia summer monsoon) period, as illustrated in Fig. 3.

Based on Fig. 3 and Table 4, it is evident that meteorological conditions had a significant impact on surface MDA8 O₃ in the NCP and FWP regions during the PostG period (−0.09 to −0.04 ppb) compared to the PreG period (−1.41 to −0.88 ppb). In the SCB region, the impact of meteorological fields was relatively weak (−0.41–0.71 ppb), attributed to the basin topography and stable atmospheric conditions. However, in the eastern and southeastern coastal areas of China, due to the significant influence of the EASM, the impact of meteorological conditions may be more critical than that of anthropogenic emissions. For instance, in the YRD and PRD regions, meteorological conditions significantly changed O₃ levels (−1.29–1.3 ppb) compared to anthropogenic emissions (0.81–0.87 ppb) in 2013, indicating the significant influence of meteorological conditions on surface O₃.

Figure 5 Simulated responses of surface MDA8 O₃ mixing ratios (units: ppb) to the variations in CO₂ emissions during the summer monsoon period in 2009 (a), 2010 (b), 2011 (c), 2012 (d), 2013 (e), 2014 (f), 2015 (g), 2016 (h), 2017 (i), and 2018 (j) relative to 2008.

Figure 6 The simulated averaged MDA8 O₃ (a–c, units: ppb), CO₂ (d–f, units: ppm), precipitation (g–i, units: mm d⁻¹), and isoprene mixing ratios (j–l, units: µg m⁻³) in 2008 from the base simulations (the left column) and their changes due to variations in CO₂ emissions in PreG (2009–2013, the central column) and PostG (2014–2018, the right column) relative to 2008.

Our findings are consistent with previous studies. Liu and Wang (2020a) reported a decrease in O₃ in Shanghai from 2013 to 2017 due to changes in meteorological conditions. Chen et al. (2019) and Liu and Wang (2020a) also suggested that changed meteorological conditions had a negative impact on O₃ formation in the NCP and FWP regions and that the influence of meteorology on surface-level O₃ decreased in PostG. In addition, Cheng et al. (2019) found that the effects of meteorological conditions on long-term O₃ variations were less than 3 %, which is similar to our study.

As we know, the formation of surface O₃ is promoted by rising temperatures (Steiner et al., 2010). However, increased surface temperatures can also intensify turbulence within the planetary boundary layer (PBL), increasing PBL height (Guo et al., 2016). This increase in PBL height, coupled with the enhanced upward motion, can transport near-surface pollutants to the upper atmosphere, reducing their concentration in the lower atmosphere (Gao et al., 2016). Additionally, the upward motion can also facilitate cloud formation and precipitation, resulting in a reduction of near-surface atmospheric pollutants via precipitation washout (Yoo et al., 2014).

We have improved the accuracy of O₃ photodissociation rate calculations by including varying AOD and SSA in the TUV module, as described in Sect. 2.3.2. As a result, the increase in cloud cover reduced the shortwave radiation flux and photochemical formation rates of near-surface O₃, leading to decreased formation. Thus, the increase in near-surface temperature is often accompanied by an elevation in PBL height, enhanced cloud cover, precipitation, and reduced shortwave radiation. Moreover, higher wind speeds can enhance the dispersion of O₃ (Gorai et al., 2015).

The variations of MDA8 O₃, precipitation, clouds, shortwave flux (SWF), wind speed, temperature, and PBL height are presented individually in Fig. 4. The increase in SWF can accelerate O₃ formation through photochemistry (Jiang et al., 2012; Lelieveld and Crutzen, 1990). Therefore, the increased cloud fraction reduced surface O₃ by decreasing shortwave radiation, especially in NCP, FWP, YRD, and SCB in the PreG period (−10.63 to −1.6 W m⁻²). Furthermore, the enhanced precipitation in these regions (0.37–1.81 mm d⁻¹) reduced surface O₃ levels significantly. The significant increase in wind speed (0.17–0.26 m s⁻¹) also contributed to the reduction of surface O₃ in the NCP region (Table 4).

Another crucial factor is the elevated surface temperature (0–5 K), which intensified upward motion and raised the planetary boundary layer (PBL) height (0–500 m) across much of East Asia. Consequently, the increased temperature and PBL height could disperse surface-level O₃, thereby reducing its concentration.

3.4 The effect of CO₂ in the 2008–2018 ozone increase

The surface O₃ in southern China, which includes the YRD, PRD, and SCB regions, was characterized by high precipitation, temperatures, and vegetation cover and was significantly impacted by CO₂ (Fig. 5). Figure 6e and f demonstrate a marked rise in CO₂ levels across East Asia, particularly in eastern China, which was attributable to extensive human activity.

CO₂ is a significant driver of climate change and alterations in biogenic emissions. As shown in Fig. 6b and c, the impact of CO₂ on O₃ levels varies across locations, with a positive effect of 0.28–0.46 ppb along the southeastern coast of China but a negative influence of −0.51 to −0.11 ppb in southwest and central China. CO₂ affects O₃ concentration by influencing both precipitation and isoprene emissions. In western and central China, CO₂ primarily affects O₃ concentration through its impact on precipitation (Table 5). Elevated CO₂ concentrations lead to increased precipitation (0.06–0.64 mm d⁻¹) in the FWP and SCB regions, resulting in a decrease in surface O₃ (up to −0.51 ppb). In eastern and southern coastal China, where vegetation is abundant, CO₂ has a greater impact on isoprene emissions. In the YRD region, decreased isoprene (−0.58 to −0.32 µg m⁻³) and increased precipitation (0.09–0.13 mm d⁻¹) reduced MDA8 O₃ levels (0.09–0.14 ppb). In PRD, increased isoprene levels (0.31–0.92 µg m⁻³) and decreased precipitation (−1.02 to −0.33 mm d⁻¹) led to the enhancement of MDA8 O₃ (0.28–0.46 ppb).

In some years, the impact of changed CO₂ can be as significant as or even surpass that of anthropogenic emissions and meteorology (Fig. 10). For example, in 2013, CO₂ caused an increase of 0.95 ppb in MDA8 O₃ in the YRD region, which exceeded that of anthropogenic emissions (0.87 ppb). Similarly, in the PRD region in 2012, the effect of CO₂, anthropogenic emissions, and meteorology was 1.41, 1.77, and 1.95 ppb, respectively. Even in the NCP in 2010, the impact of CO₂ (0.75 ppb) was comparable to that of anthropogenic emissions (1.5 ppb). In summary, CO₂ has a significant impact on surface O₃ concentrations by influencing radiation and isoprene emissions, with more prominent effects in regions with abundant vegetation.

Table 5 Simulated responses of MDA8 O₃ mixing ratios (units: ppb), CO₂ mixing ratios (units: ppm), precipitation (units: mm d⁻¹), and isoprene mixing ratios to the changes in CO₂ emissions over the North China Plain, Fenwei Plain, Yangtze River Delta, Pearl River Delta, and Sichuan Basin in PreG (2009–2013) and PostG (2014–2018) relative to 2008.

3.5 The effect of anthropogenic emission in the 2008–2018 ozone increase

Finally, we calculated the anthropogenic emissions' effect on the 2008–2018 O₃ increase. Figure S8 and Table S1 illustrate that the levels of PM_2.5, PM₁₀, SO₂, CO, and OC emissions remained consistently high during the PreG period. However, a linear decrease in emissions was observed after the implementation of the Clean Air Action Plan in 2013. Prior to 2013, the emission of VOCs increased steadily but subsequently stabilized. Similarly, the emission of nitrogen oxides (NO_x) exhibited an upward trend before 2013, but since then, the emissions have shown a linear decrease, with each subsequent year exhibiting lower levels of NO_x emissions. In comparison to other species, the emissions of ammonia (NH₃) remained relatively stable from 2008 to 2018. Our analysis results of the emissions of different species align with those of Zheng et al. (2018), who computed the changes of each species in the MEIC inventory from 2010 to 2017.

Figure 7 Simulated responses of the surface MDA8 O₃ mixing ratios (units: ppb) to variations in anthropogenic emissions in 2009 (a), 2010 (b), 2011 (c), 2012 (d), 2013 (e), 2014 (f), 2015 (g), 2016 (h), 2017 (i), and 2018 (j) relative to 2008.

Figure 8 Changes in the simulated surface MDA8 O₃ mixing ratios (units: ppb) from the base simulation (All, a, b); those due to variations in anthropogenic emissions (Emis, c, d), meteorological conditions (Met, e, f), and CO₂ emissions (CO₂, g, h) in PreG (2009–2013, the left column) and PostG (2014–2018, the right column) relative to 2008.

Figure 7 illustrates that anthropogenic emissions have caused a notable increase in surface O₃ levels across most of China, particularly in megacity clusters. The impact of anthropogenic emissions on O₃ concentration ranged from 2.33 to 18.51 ppb in the five target regions.

Figure 8 and Table 6 illustrate that the changes in surface O₃ caused by anthropogenic emissions are similar in magnitude and spatial distribution to the changes in the Base experiment. This suggests that anthropogenic emissions were the dominant factor driving the increase in surface O₃ in China from 2008 to 2018. Notably, a high-impact center of anthropogenic emissions was simulated in northern China, with the NCP region experiencing the most significant increase in surface O₃ (4.08–18.51 ppb), followed by the FWP, YRD, and SCB regions (4.10–11.5 ppb). In the PRD region, anthropogenic emissions led to a slight enhancement of O₃ by 2.33–5.74 ppb.

The role of anthropogenic emissions increased linearly from 2008 to 2018, despite the implementation of the Clean Air Action Plan in 2013 (Table 6). For example, anthropogenic emissions significantly increased surface MDA8 O₃ in the NCP region (4.08 ppb in PreG and 18.51 ppb in PostG). Similarly, FWP experienced increases of 5.15 and 11.5 ppb in the PreG and PostG periods, respectively. In the SCB region, the surface MDA8 O₃ was mainly affected by variations in anthropogenic emissions due to the high levels of anthropogenic emissions in the complex basin topography. In the YRD and PRD regions, anthropogenic emissions resulted in changes to O₃ of 2.33–11.17 ppb.

The reasons for this characteristic are multiple. Before 2013, the continuous increase in VOCs and NO_x emissions (Fig. S8b, c) facilitated the rise of O₃ levels. Following the implementation of the Clean Air Action Plan in 2013, the emissions of VOCs and NO_x were regulated. However, with the decrease in PM_2.5 levels, direct radiation increased, and scattered radiation decreased (Fig. 9), thereby promoting the photochemical formation of O₃ (Bian et al., 2007). In addition, the reduced NO emission weakened the titration effect (Fig. S8b), thus increasing surface O₃ (Li et al., 2022).

Our results are consistent with previous studies by Wang et al. (2019b) and Liu and Wang (2020b), which also showed the dominant and almost linear role of anthropogenic emissions in the increase in O₃ from 2013 to 2015 in four major Chinese cities (Beijing, Shanghai, Guangzhou, and Chengdu).

Figure 9 The variations of the surface direct radiation (a, b, units: W m⁻²) and diffuse radiation (c, d, units: W m⁻²) in the PreG (2009–2013, a, c) and PostG (2014–2018, b, d) period relative to 2008.

Figure 10 Interannual variations of the surface MDA8 O₃ mixing ratios (units: ppb) in the summer monsoon period (ALL) and the responses of variations in anthropogenic emissions (Emis), meteorological conditions (Met), and CO₂ emissions (CO₂) in the (a) North China Plain, (b) Fenwei Plain, (c) Yangtze River Delta, (d) Pearl River Delta, and (e) Sichuan Basin in 2008–2018 relative to 2008.

Table 6 Simulated response of the MDA8 O₃ mixing ratios (units: ppb) to the changes in anthropogenic emissions (Emis), meteorological conditions (Met), and CO₂ emissions (CO₂) over the North China Plain, Fenwei Plain, Yangtze River Delta, Pearl River Delta, and Sichuan Basin in PreG (2009–2013) and PostG (2014–2018) relative to 2008.

3.6 Attribution analysis of ozone changes in 2008–2018

Finally, we presented an attribution diagram depicting the changes in O₃ concentration from 2008 to 2018. The total variation in O₃ concentration can be attributed to the combined effects of meteorological changes, changes in CO₂ concentration, and anthropogenic emissions (Fig. 10).

The primary driver of the O₃ concentration variation from 2008 to 2018 was the changes in anthropogenic emissions, particularly in regions with high emissions, such as the NCP and FWP. Although the Clean Air Action Plan was implemented in 2013, it did not reduce the contribution of anthropogenic emissions to the O₃ increase. Even in the PostG period, with the development of urbanization and industrialization, the impact of changed anthropogenic emissions on O₃ has gradually become more prominent than changed meteorology and CO₂. The contribution of changed meteorology to O₃ was generally negative in the five regions, with a more significant impact in the YRD and PRD regions. This may be attributed to their proximity to the ocean and susceptibility to the summer monsoon influence. Changes in CO₂ concentration affected O₃ concentration by altering radiation and isoprene emissions, with a more significant impact in the YRD and PRD regions where vegetation was abundant. In some years, it even surpassed the effects of anthropogenic emissions. Therefore, we suggest that the influence of CO₂ concentration changes on O₃ concentration should be considered in regions with high vegetation coverage.

4 Conclusions

In this study. First, we improved the RegCM-Chem-YIBs model regarding the photolysis of O₃ and the radiation effect of CO₂ and O₃. Second, we assessed the impacts of anthropogenic emissions, meteorological factors, and CO₂ on the surface O₃ increase in China during the EASM season from 2008 to 2018.

In the NCP and FWP regions. The increased surface O₃ was primarily attributed to the changes in anthropogenic emissions (4.08–5.15 ppb in PreG and 11.5–18.51 ppb in PostG). Furthermore, the impact of anthropogenic emissions has increased linearly, despite the Clean Air Action Plan being implemented in 2013. In contrast, the effects of meteorological factors and CO₂ on O₃ were weak during the study period.

In the YRD and PRD regions. Ignoring the principal contributions of anthropogenic emissions, CO₂ significantly impacted the O₃ variations (−0.14–0.46 ppb). The varied CO₂ led to surface MDA8 O₃ changes of −0.09 to −0.14 ppb in the YRD and 0.28–0.46 ppb in the PRD by modulating the isoprene emissions and precipitation. On the other hand, the meteorological conditions played a more significant role in surface O₃ than in the NCP, FWP, and SCB regions, resulting in a decrease in MDA8 O₃ from 2008 to 2018 (−4.42–3.25 ppb).

In the SCB region. The increase in surface O₃ from 2008 to 2018 was primarily driven by variations in anthropogenic emissions. The effect of meteorological conditions was weak due to the high level of emissions and basin topography. However, the changes in CO₂ significantly impacted surface O₃ levels and were unfavorable to O₃ formation during the study period (−3.0 ppb in PreG and PostG).

In conclusion, anthropogenic emissions dominated the O₃ increase in China from 2008 to 2018, and the effects of meteorological conditions on surface O₃ could be more significant in some regions. Furthermore, we emphasize the significance of CO₂ emissions, particularly in southern China, as a critical contributor to O₃ variations. Therefore, it is vital to consider CO₂ variability in future predictions of O₃ concentrations. Such consideration would be helpful for designing long-term O₃ control policies.