Anthropogenic and natural causes for the interannual variation of PM_2.5 in East Asia during summer monsoon periods from 2008 to 2018

There was a significant difference in near-surface PM_2.5 across China after the implementation of the Clean Air Action Plan in 2013. This study used the regional climate-chemistry-ecosystem coupled model RegCM-Chem-YIBs to investigate interannual variations in PM_2.5 across East Asia from 2008 to 2018. The drivers of PM_2.5 variability were examined from anthropogenic and natural perspectives. Compared to 2008, PM_2.5 showed little variation during the pre-governance (PreG) period (2009–2013). However, during the post-governance (PostG) period (2014–2018), a substantial decline in PM_2.5 was simulated, particularly in the North China Plain (−36.76 µg m⁻³) and the Sichuan Basin (−33.96 µg m⁻³). Anthropogenic pollutant emissions were the primary drivers of PM_2.5 reductions, contributing −10.39 to −3.82 µg m⁻³ in the PreG period and −33.86 to −8.45 µg m⁻³ in the PostG period. The influence of meteorological conditions on PM_2.5 during the PreG period (−6.31 to 2.32 µg m⁻³) was comparable to that of anthropogenic pollutant emissions. Additionally, in the vegetation-rich region, the impact of CO₂ emission changes on PM_2.5 was comparable to that of anthropogenic pollutant emissions. Our study comprehensively examined the drivers of PM_2.5 concentration changes from 2008 to 2018. We highlight a significant intensification in the contribution of anthropogenic pollutant emissions and reveal that, in regions characterized by dense vegetation, changes in CO₂ concentrations exert a pronounced impact on PM_2.5 variations.

1 Introduction

PM_2.5 refers to fine particulate matter with an aerodynamic diameter less than or equal to 2.5 µm (Chen et al., 2018). Its sources include industrial emissions, vehicular exhaust, biomass burning, and secondary formation from atmospheric gases (Wu et al., 2020). Major chemical components of PM_2.5 include sulfates, nitrates, ammonium salts, organic carbon, elemental carbon, and heavy metals (van Donkelaar et al., 2019; Li et al., 2017a). PM_2.5 is one of the primary atmospheric pollutants in China (Fontes et al., 2017), posing significant risks to human respiratory health (Feng et al., 2016; Xing et al., 2016). Long-term exposure to PM_2.5 can lead to respiratory diseases such as chronic bronchitis, emphysema, and asthma (Kim et al., 2015; Pui et al., 2014; Xing et al., 2016). Additionally, PM_2.5 is critical as a short-lived species influencing atmospheric radiation processes (Hu et al., 2017). It affects the radiative energy balance of the Earth–atmosphere system by scattering or reflecting solar radiation (direct effect) (Wu et al., 2021) and altering cloud microphysical properties (indirect effect) (Wang et al., 2018a; Wu et al., 2021).

With China's rapid economic development, widespread PM_2.5 pollution became prevalent across the country in the early 21st century (Ma et al., 2016). In the most severely polluted urban areas, the annual average PM_2.5 concentration exceeded 100 µg m⁻³ (van Donkelaar et al., 2010). From 2000 to 2008, the national average PM_2.5 concentration in China was 49.4±14.2 µg m⁻³. In eastern China, the average concentration was 55.4±16.1 µg m⁻³, while the Beijing–Tianjin–Hebei region experienced average levels as high as 62.1±22.5 µg m⁻³. The Yangtze River Delta (YRD) saw an average concentration of 63.0±11.1 µg m⁻³, the Pearl River Delta (PRD) recorded an average of 52.4±5.8 µg m⁻³, and the Sichuan Basin averaged 61.6±13.4 µg m⁻³ (Wei et al., 2021). To mitigate the severe PM_2.5 pollution, China implemented the Clean Air Action Plan in 2013 (Li et al., 2019). This policy led to a significant nationwide decrease in PM_2.5 concentrations (Zhang et al., 2019), and there has been a notable improvement in air quality since that date (Vu et al., 2019; Li et al., 2018).

The variation in PM_2.5 concentrations is influenced by three key factors: anthropogenic pollutant emissions, meteorological conditions (Xiao et al., 2021), and carbon dioxide (CO₂) changes. Anthropogenic pollutant emissions encompass industrial production, transportation, and energy consumption (An et al., 2019), which release amounts of primary PM_2.5, as well as the precursors of secondary PM_2.5 such as volatile organic compounds (Kurokawa and Ohara, 2020) and nitrogen oxides (Wu et al., 2020; Zheng et al., 2021b; Kurokawa and Ohara, 2020). Consequently, reducing these emissions is essential for mitigating PM_2.5 concentrations, as they directly contribute to both the formation and persistence of particulate pollution (Zheng et al., 2018; Zhang et al., 2019).

Meteorological conditions play a significant role in influencing near-surface PM_2.5 concentrations (Chen et al., 2020b; Xiao et al., 2021). Elevated temperatures can accelerate atmospheric chemical reactions (Mousavinezhad et al., 2021), including oxidation and photochemical processes, thereby promoting the formation of PM_2.5 (Zhong et al., 2018). In addition, moderate increases in temperature can significantly enhance the emissions of biogenic volatile organic compounds (BVOCs) by stimulating the activity of the synthase enzyme. However, when temperatures exceed the physiological tolerance threshold of plants, decreased enzyme activity or metabolic disruption may suppress emissions (Lindwall et al., 2016; Kleist et al., 2012). Therefore, temperature changes can influence atmospheric PM_2.5 concentrations by modulating the emissions of BVOCs. Precipitation aids in removing particulate matter from the atmosphere through wet deposition (Zhang et al., 2013), effectively reducing PM_2.5 pollution levels (Wu et al., 2018). Additionally, wind speed and direction are crucial factors in the transport and dispersion of particulate matter (Li et al., 2017b). Higher wind speeds facilitate the dispersion of particulate matter over a wider area, reducing its local accumulation and mitigating air pollution in specific regions (Li et al., 2017b; Zhang et al., 2018). The increase in planetary boundary layer height (PBLH) strengthens atmospheric upward motion (Ait-Chaalal et al., 2016), thereby reducing near-surface PM_2.5 concentrations (Pan et al., 2019).

Changes in CO₂ concentrations can influence PM_2.5 pollution levels through several mechanisms. Elevated CO₂ concentrations impact the atmospheric radiation balance, altering the distribution and intensity of precipitation (Cao et al., 2012), which directly affects PM_2.5 concentrations by influencing wet deposition rates (Zhang et al., 2022). Additionally, changes in CO₂ concentrations can affect vegetation photosynthesis and growth, which alter the emissions of BVOCs that can participate in atmospheric chemical reactions to form secondary organic aerosols and thereby impact atmospheric PM_2.5 concentrations (Sun et al., 2013, 2012). It is worth noting that elevated CO₂ concentrations may also directly inhibit BVOC emissions by reducing the activity of BVOC synthase enzymes (Heald et al., 2009; Pegoraro et al., 2004). Therefore, the impact of increased CO₂ on vegetation BVOC emissions can be either positive or negative, depending primarily on the relative strength of the inhibitory effect from enzyme suppression versus the stimulatory effect from enhanced photosynthesis (Sun et al., 2012). Isoprene is the most abundant species among BVOCs, so changes in CO₂ concentrations can indirectly affect near-surface PM_2.5 concentrations by influencing isoprene emissions from vegetation (Sun et al., 2013; Lin et al., 2013; Kramer et al., 2016).

Numerous studies have used statistical models and numerical simulations to investigate the impacts of meteorological conditions and anthropogenic pollutant emissions on PM_2.5 concentration changes in China. The results consistently indicate that changes in anthropogenic pollutant emissions are the primary driver of PM_2.5 variation. Zhang et al. (2019), using the WRF-CMAQ model at the national scale, found that meteorological conditions accounted for only 9 % of the total decline in PM_2.5 concentrations during 2013–2017 in China, suggesting that emission reductions were the dominant factor. Similarly, based on a multiple linear regression model, Chen et al. (2020a) reported that anthropogenic pollutant emission reductions contributed 73 %, 87 %, and 84 % to the PM_2.5 decline in the North China Plain, Yangtze River Delta, and PRD, respectively, while the contribution of meteorological conditions ranged from 10 % to 26 %. Cheng et al. (2019), also employing the WRF-CMAQ model, found that the decrease in PM_2.5 concentrations in Beijing over the same period was mainly attributable to local (65.4 %) and regional (22.5 %) emission reductions, with meteorological conditions accounting for only 12.1 %.

Current research primarily emphasizes the impact of anthropogenic pollutant emissions (Zheng et al., 2018) and meteorological changes on PM_2.5 concentrations (Zhang et al., 2019; Zhai et al., 2019), while the potential influence of CO₂ concentration changes on PM_2.5 pollution levels remains largely underexplored. Additionally, following the implementation of the Clean Air Action Plan in 2013, significant decreases in PM_2.5 concentrations were observed in China. Concurrently, CO₂ levels continued to rise (Xu et al., 2022), with the influence of CO₂ on PM_2.5 strengthening annually. Therefore, it is essential to analyze the evolution of PM_2.5 concentrations from 2008 to 2018 in detail, and attribute changes in PM_2.5 levels to every factor, such as anthropogenic pollutant emissions, meteorological conditions, and CO₂ variations.

2 Methods and data

2.1 Model description

In this study, we employed the coupled regional climate-chemistry-ecology model RegCM-Chem-YIBs (Xie et al., 2019, 2024). The RegCM-Chem component simulates key meteorological variables, including temperature, humidity, precipitation, and radiation, along with atmospheric pollutants including ozone and particulate matter (Shalaby et al., 2012). The YIBs (Yale Interactive terrestrial Biosphere) model focuses on simulating vegetation physiological processes, such as ozone-induced damage, photosynthesis, and respiration (Lei et al., 2020). Additionally, it computes important land surface parameters, including CO₂ flux, BVOC emissions, and stomatal conductance (Yue and Unger, 2015). The YIBs model employs a leaf-level BVOC emission scheme based on vegetation photosynthesis. Unlike the traditional MEGAN (Model of Emissions of Gases and Aerosols from Nature) model, this approach incorporates the influence of plant photosynthesis on BVOC emissions, making it more representative of actual plant physiological processes. In this scheme, leaf-level BVOC emission rates depend on the photosynthetic rate, leaf surface temperature, and intracellular CO₂ concentration (Yue and Unger, 2015; Lei et al., 2020; Yue et al., 2015).

The RegCM-Chem and YIBs models exchange variables every 6 min, facilitating dynamic coupling between regional climate, atmospheric chemistry, and ecosystem processes. The RegCM-Chem-YIBs model simulated both primary and secondary PM_2.5 emissions, including dust, black carbon, organic carbon, sulfates, nitrates, and ammonium. The structure of the model is shown in Fig. 1.

In the RegCM-Chem-YIBs model, changes in CO₂ concentrations affect PM_2.5 primarily via two mechanisms: first, CO₂-induced radiative forcing alters the atmospheric radiation balance, leading to shifts in temperature, precipitation, and boundary-layer structure that modulate PM_2.5 formation, transport, and removal (Li and Mölders, 2008; Matthews, 2007); and second, through the YIBs module, changes in CO₂ concentration modulate photosynthetic activity and stomatal behavior, altering BVOCs emissions that undergo atmospheric photochemical oxidation to form secondary organic aerosols, a significant fraction of PM_2.5 (Kergoat et al., 2002; Kellomaki and Wang, 1998).

Figure 1 Framework of the RegCM-Chem-YIBs model.

2.2 Model configurations

The study area covers the entire East Asian region, with a horizontal grid resolution of 60 km, centered at 36° N and 107° E. A terrain-following coordinate system was used vertically (Bleck and Benjamin, 1993), dividing the atmosphere into 18 layers from the surface to 50 hPa.

Anthropogenic pollutant emissions data were obtained from the Multi-resolution Emission Inventory for China (MEIC v1.4) developed by Tsinghua University (Geng et al., 2024). Surface CO₂ flux data were sourced from the National Oceanic and Atmospheric Administration (NOAA) CarbonTracker CT2019 dataset, which includes contributions from fossil fuel combustion, biomass burning, and ocean–atmosphere CO₂ exchange (Peters et al., 2007). Meteorological fields were derived from ERA-Interim reanalysis (Balsamo et al., 2015), while sea surface temperature data were taken from NOAA's weekly mean dataset (Huang et al., 2021). The model employed the Grell cumulus parameterization scheme, the CCM3 radiation scheme, the Holtslag PBL scheme for boundary layers, the CBM-Z mechanism for meteorology and chemistry, and the TUV photochemistry scheme.

2.3 Experiment settings

The numerical experiments are presented in Table 1. The SIM₂₀₀₈ experiment represents the baseline conditions for the year 2008. In the SIM_Base experiment, interannual variations in meteorological fields, CO₂ emissions, and anthropogenic pollutant emissions (excluding CO₂ emissions) were considered for simulations spanning the period 2009–2018, representing the baseline conditions for those years. Additionally, the SIM_MET=2008 and SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ experiments were designed, where meteorological fields and CO₂ emissions were fixed at their 2008 levels, while simulations were conducted for the 2009–2018 period. The simulation period spans from April to August each year. From this, the results from May to August, corresponding to the East Asian Summer Monsoon (EASM) period, were selected for analysis.

Changes in PM_2.5 concentrations were attributed to three main factors: anthropogenic pollutant emissions, meteorological conditions, and CO₂ variations. By comparing the simulation results from different years in the SIM_Base experiment to SIM₂₀₀₈ (SIM_Base−SIM₂₀₀₈), we quantified changes in PM_2.5 concentrations relative to 2008 for the period 2009–2018. To evaluate the impact of meteorological conditions on PM_2.5 concentrations, we compared the results of the SIM_Base experiment with those of the SIM_MET=2008 experiment for the same year (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$). Similarly, the contribution of CO₂ emission changes to PM_2.5 variations was assessed by comparing the SIM_Base experiment with the SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ experiment (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) in the same year. The contribution of anthropogenic pollutant emissions was then determined by subtracting the effects of meteorological and CO₂ emission changes from the total PM_2.5 variation.

It is noteworthy that, as a principal greenhouse gas, CO₂ modifies meteorological parameters – such as radiation, temperature, and precipitation – which in turn influence PM_2.5 levels. In the comparison between experiments SIM_Base and SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$), all meteorological changes derive solely from variations in CO₂ emissions, a mechanism fundamentally different from the meteorological influences identified in experiments SIM_Base and SIM_MET=2008 (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$).

Table 1 The numerical experimental in this study.

2.4 Model evaluations

Observed PM_2.5 data were obtained from the China National Environmental Monitoring Center. This study used hourly PM_2.5 concentrations during the summer monsoon period (1 May to 31 August) from 2015 to 2018. A total of 366 monitoring stations across Chinese cities, selected based on data completeness and representativeness, were used for model validation. The locations of these stations are shown in Fig. S5. CO₂ observations were sourced from the World Data Centre for Greenhouse Gases, including all seven sites in East Asia: Waliguan, Korea Tae-ahn Peninsula, Ulaanbaatar in Mongolia, Lulin, Yonagunijima, Cape D'Aguilar (Hong Kong), and King's Park. Detailed station locations are shown in Fig. S6. Reanalysis data for temperature, wind fields, and relative humidity were obtained from the ERA-Interim dataset.

As shown in Table 2 and Figs. S1–S6, the SIM_Base experiments reproduce 2015–2018 PM_2.5 and CO₂ concentrations with high correlations and low biases relative to observations, while their simulated meteorological fields closely match reanalysis data. Overall, the RegCM-Chem-YIBs model effectively captures the fundamental characteristics and temporal trends of meteorological factors, PM_2.5, and CO₂ concentrations in East Asia (Ma et al., 2023a, b).

Table 2 Evaluations of the near-surface CO₂ and PM_2.5 in East Asia.

RMSE: root mean square error; R: correlation coefficient.

3 Results and discussion

3.1 PM_2.5 variation

Changes in PM_2.5 concentrations from 2009 to 2018 relative to 2008 were quantified by comparing simulation results from each year in the SIM_Base experiment with SIM₂₀₀₈ (SIM_Base−SIM₂₀₀₈). Figure 2 illustrates the changes in near-surface PM_2.5 concentrations across East Asia from 2009 to 2018. PM_2.5 concentrations are notably higher in the North China Plain, northeastern China, and eastern China (Shanghai, Jiangsu, Zhejiang), largely driven by industrial emissions, vehicle exhaust, coal combustion, and dust from human activities (Wang et al., 2017). In contrast, regions in western China (Yunnan, Gansu, Xinjiang) exhibit lower PM_2.5 levels due to limited industrial activity, lower population density, and more favorable meteorological conditions (low water vapor content, lower temperatures, and weak solar radiation are unfavorable for the formation of secondary aerosols such as sulfates, nitrates, and organic aerosols) (Wei et al., 2021; Xue et al., 2020). Developed cities and industrial centers like the PRD and Fuzhou (Fujian Province) continue to encounter challenges related to PM_2.5 pollution. Moreover, the Sichuan region, characterized by its enclosed basin geography and high population density, also experiences high PM_2.5 pollution levels (Wang et al., 2018b). From 2009 to 2013, PM_2.5 concentrations in China remained relatively stable, with levels averaging around 90 µg m⁻³ in the North China Plain and the Sichuan Basin (SCB). However, following the implementation of the Clean Air Action Plan in 2013, PM_2.5 levels significantly declined nationwide. By 2018, concentrations had dropped to below 50 µg m⁻³ across much of the country.

Figure 2 Near-surface PM_2.5 concentrations (µg m⁻³) over East Asia during the EASM period from 2009 (a) to 2018 (k) (SIM₂₀₀₈). Key regions are highlighted by black boxes, including the North China Plain (NCP), Fenwei Plain (FWP), Yangtze River Delta (YRD), Pearl River Delta (PRD), and Sichuan Basin (SCB).

Figure 3 and Table 3 present the changes in PM_2.5 concentrations relative to 2008 across East Asia from 2009 to 2018. Since 2008, most regions in China have seen varying degrees of PM_2.5 reduction. During the pre-governance (PreG) period (2009–2013), the largest decrease occurred in the YRD, with a reduction of 14.77 µg m⁻³, followed by the SCB and PRD, where concentrations dropped by 10.59 and 8.69 µg m⁻³, respectively. In contrast, the Fenwei Plain (FWP) and PRD experienced smaller changes, with reductions of less than 3 µg m⁻³. PM_2.5 concentrations across China significantly decreased after the implementation of the Clean Air Action Plan in 2013. The most notable reductions were simulated in the North China Plain and SCB, where PM_2.5 concentrations dropped by 36.76 and 33.96 µg m⁻³, respectively. In the FWP and YRD, PM_2.5 concentrations decreased by 22.16 to 27.89 µg m⁻³. In contrast, the PRD saw a smaller reduction, with levels decreasing by just 8.03 µg m⁻³. This may be attributed to the region's significant influence from the summer monsoon and relatively lower impact from anthropogenic pollutant emissions. Further analysis of these factors will be conducted in subsequent sections.

Table S1 shows that the mean PM_2.5 trend over China during the PreG (2009–2013) and post-governance (PostG; 2014–2018) periods was −1.84 and −2.90 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$, respectively. These values are consistent with the findings of Silver et al. (2025), who reported a PM_2.5 trend of −2.47 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$ for 2014–2017 in China based on ground-based observations. Similarly, Lin et al. (2018) reported PM_2.5 trends of −0.65 and −2.30 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$ for 2006–2010 and 2011–2015 in China, respectively. Using satellite remote sensing data, Ma et al. (2019) found declines of 1.03 and 4.27 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$ for 2010–2013 and 2013–2017 in China, respectively. The high-resolution Chinese air quality reanalysis developed by Kong et al. (2021) using data assimilation techniques indicated a more pronounced decline of −5.80 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$ for PM_2.5 from 2013 to 2018 in China. In addition, Silver et al. (2018), based on multi-source data, reported a trend of −3.40 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$ for 2015–2017 in China. Therefore, our simulation accurately captures the observed PM_2.5 trends over China from 2008 to 2018, providing a robust foundation for subsequent attribution analyses.

Overall, before 2013, near-surface PM_2.5 concentrations across China showed little variation. However, after 2013, a significant reduction in PM_2.5 pollution levels was simulated nationwide. Changes in PM_2.5 concentrations were attributed to three main factors: anthropogenic pollutant emissions, meteorological conditions, and CO₂ variations. The following sections analyze each factor's contribution to the changes in PM_2.5 concentrations from 2008 to 2018.

Figure 3 Changes in near-surface PM_2.5 concentrations (µg m⁻³) during the EASM period from 2009 (a) to 2018 (j) relative to 2008 in East Asia (SIM_Base−SIM₂₀₀₈).

Table 3 Changes in near-surface PM_2.5 concentrations (µg m⁻³) during the EASM period from 2009 to 2018 relative to 2008 in the North China Plain (NCP), Fenwei Plain (FWP), Yangtze River Delta (YRD), Pearl River Delta (PRD), and Sichuan Basin (SCB) (SIM_Base−SIM₂₀₀₈).

3.2 Contribution of meteorological conditions

The impact of meteorological conditions variations on PM_2.5 concentrations were assessed by comparing SIM_Base results with those from SIM_MET=2008 for the same year (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$). As shown in Fig. 4, during the PreG period, precipitation increased by 2–4 mm d⁻¹ in China's eastern coastal and western inland regions, while it decreased by approximately 2 mm d⁻¹ in central China. This increase in precipitation facilitates the reduction of near-surface PM_2.5 concentrations through wet deposition. Consequently, trends in PM_2.5 concentrations are inversely related to precipitation: concentrations decreased by 2–16 µg m⁻³ in the eastern coastal and western inland regions, while they increased by 4–8 µg m⁻³ around 110° E in central China. Additionally, in northeastern and southwestern China, wind speeds increased by 1 to 2 m s⁻¹, contributing to the reduction of PM_2.5 concentrations. In contrast, decreased wind speeds in southeastern and central China facilitated the accumulation of PM_2.5. During the PostG period, the significant increase in temperature (Fig. 4l) promoted the formation of PM_2.5, leading to an expansion of the areas where PM_2.5 concentrations increased. Overall, PM_2.5 concentrations have decreased in the eastern coastal and western inland regions but increased in the central area of China.

Table 4 indicates that in the North China Plain (NCP) region precipitation increased by 0.58 to 0.6 mm d⁻¹ and wind speed rose by 0.17 to 0.26 m s⁻¹ during the PreG and PostG periods, resulting in a decrease in near-surface PM_2.5 concentrations of 1.6 to 4.01 µg m⁻³. In the FWP region, PM_2.5 concentrations increased by 1 to 2.31 µg m⁻³, which was associated with a rise in temperature of 0.1 to 0.46 K and a significant decrease in PBLH of 108.5 to 15.3 m. In the YRD region, the increase in wind speed of 0.48 to 1.02 m s⁻¹ facilitated a reduction in PM_2.5 concentrations by 0.43 to 0.61 µg m⁻³. Conversely, in the PRD region, reduced precipitation combined with increased temperature contributed to an increase in PM_2.5 concentrations, ranging from 0.11 to 1.49 µg m⁻³. In the SCB region, PM_2.5 concentrations rose by 0.29 µg m⁻³ during the PreG period, which can be linked to a significant decrease in PBLH of 136.5 m. In the PostG period, PM_2.5 concentrations decreased by 1.14 µg m⁻³, which can be attributed to an increase in precipitation (0.37 mm d⁻¹) and a decrease in temperature (0.14 K).

Figure 4 The PM_2.5 (a–c, µg m⁻³), precipitation (d–f, mm d⁻¹), wind speed (g–i, m s⁻¹), temperature (j–l, K), and planetary boundary layer (PBL) height (m–o, m) during the EASM period in 2008 (left), and their mean changes due to meteorological variations in the PreG (2009–2013, center) and PostG (2014–2018, right) periods relative to 2008. PreG-2008 and PostG-2008 represent the average annual differences between experiments SIM_Base and SIM_MET=2008 (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$) for the periods 2009–2013 and 2014–2018, respectively.

Table 4 Impact of meteorological condition changes on PM_2.5 (µg m⁻³), precipitation (mm d⁻¹), wind speed (m s⁻¹), near-surface temperature (K), and planetary boundary layer (PBL) height (m) during the EASM period in the PreG (2009–2013) and PostG (2014–2018) periods relative to 2008. PreG and PostG represent the average annual differences between experiments SIM_Base and SIM_MET=2008 (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$) for the periods 2009–2013 and 2014–2018, respectively.

3.3 Contribution of CO₂

The contribution of CO₂ emission changes to PM_2.5 variability was quantified by comparing the SIM_Base experiment with the SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ experiment (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) within the same year. As shown in Fig. 5, following the ongoing urbanization and industrialization, CO₂ concentrations across East Asia rose by 2–10 ppm during both the PreG and PostG periods, with a sharper increase in the PostG period. CO₂ influences atmospheric PM_2.5 concentrations both through its radiative effects on precipitation and by altering BVOC emissions from vegetation. Overall, CO₂ changes contributed to PM_2.5 variations across East Asia from 2008 to 2018, ranging from −4 to 6 µg m⁻³. PM_2.5 pollution levels generally increased in the PreG period, while reductions were more common in the PostG period.

Table 5 presents a detailed analysis of the five target regions. In northern China, particularly the NCP and FWP regions, limited vegetation coverage means CO₂ impacts surface PM_2.5 concentrations mainly through precipitation changes. In the PostG period, precipitation increased by 0.06–0.13 mm d⁻¹, lowering PM_2.5 concentrations by 0.98–1.3 µg m⁻³. Similarly, in the SCB, precipitation rose by 0.21–0.64 mm d⁻¹, reducing PM_2.5 concentrations by 0.49–0.73 µg m⁻³ in the PreG and PostG periods. However, in the YRD and PRD regions, where vegetation coverage is higher, CO₂ primarily impacts PM_2.5 concentrations by modulating BVOC emissions. The impact can be either positive or negative (Possell et al., 2005) depending primarily on the balance between the inhibitory effects on synthase activity and the stimulatory effects of enhanced photosynthesis (Wilkinson et al., 2009). In the YRD region, isoprene fell by 0.32–0.58 µg m⁻³ during both periods, while precipitation rose by 0.09–0.13 mm d⁻¹, collectively reducing PM_2.5 by 0.02–0.05 µg m⁻³. In the PRD region, isoprene concentrations increased significantly by 0.31–0.92 µg m⁻³, while precipitation decreased by 0.33–1.02 mm d⁻¹. Consequently, PM_2.5 concentrations rose by 0.31–1.13 µg m⁻³ during both the PreG and PostG periods.

Figure 5 The PM_2.5 (a–c, µg m⁻³), CO₂ (d–f, ppm), precipitation (g–i, mm d⁻¹), and isoprene (j–l, µg m⁻³) during the EASM period in 2008 (left) and their mean changes due to CO₂ emission variations in the PreG (2009–2013, center) and PostG (2014–2018, right) periods relative to 2008. PreG-2008 and PostG-2008 represent the average annual differences between experiments SIM_Base and SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) for the periods 2009–2013 and 2014–2018, respectively.

Table 5 Impact of CO₂ emission changes on PM_2.5 (µg m⁻³), CO₂ (ppm), precipitation (mm d⁻¹), and isoprene (µg m⁻³) during the EASM period in the PreG (2009–2013) and PostG (2014–2018) periods relative to 2008. PreG and PostG represent the average annual differences between experiments SIM_Base and SIM${}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$ (SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) for the periods 2009–2013 and 2014–2018, respectively.

3.4 Contribution of anthropogenic pollutant emissions

The contribution of changed anthropogenic pollutant emissions to PM_2.5 variation was determined by removing the effects of meteorological and CO₂ emission changes from the total variation. Figure 6 illustrates a significant downward trend in PM_2.5 concentrations across East Asia since 2008. During the PreG period, PM_2.5 levels decreased by an average of 5 to 10 µg m⁻³ over East Asia. Following the implementation of the Clean Air Action Plan in 2013, a marked reduction in PM_2.5 concentrations was simulated. The most substantial decreases, of approximately 60 µg m⁻³, occurred in the NCP and SCB regions. Anthropogenic pollutant emissions emerged as the primary drivers of this decline, with their spatial distribution and magnitude of impact closely corresponding to the overall changes in PM_2.5 concentrations. In contrast, the effects of changing meteorological conditions and CO₂ emissions on PM_2.5 levels in East Asia were relatively minor, ranging between −5 and 5 µg m⁻³. Meteorological conditions have reduced PM_2.5 concentrations in the eastern coastal and western regions of China, while increasing them in the central region. In the PostG period, the extent of PM_2.5 concentration increases has expanded. The impact of CO₂ emission changes on PM_2.5 levels shows different trends in the PreG and PostG periods. In the PreG period, changes in CO₂ emissions primarily led to an increase in PM_2.5 concentrations. However, in the PostG period, the rise in CO₂ concentrations began to have a negative impact, leading to a reduction in PM_2.5 concentrations.

Figure 6 The total changes in PM_2.5 concentrations (All, SIM_Base−SIM₂₀₀₈), and the changes in PM_2.5 attributed to variations of anthropogenic pollutant emissions (Emis, All-Met-CO₂), meteorological conditions (Met, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$), and CO₂ emissions (CO₂, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) during the EASM period in the PreG (2009–2013, left) and PostG (2014–2018, right) periods relative to 2008.

Based on Fig. 7 and Table 6, PM_2.5 concentrations in the NCP region decreased by 5.28 µg m⁻³ during the PreG period and by 33.86 µg m⁻³ in the PostG period. Anthropogenic pollutant emissions were the primary driver of these changes. During the PreG period, the influence of meteorological conditions on PM_2.5 was comparable to that of anthropogenic pollutant emissions, with changes in meteorology contributing −4.01 µg m⁻³ and emissions contributing −5.28 µg m⁻³. However, in the PostG period, the impact of meteorological factors diminished to −1.6 µg m⁻³, indicating that anthropogenic pollutant emissions became the predominant factor in the reduction of PM_2.5 concentrations. In contrast, the effect of changes in CO₂ emissions on PM_2.5 levels was relatively minor, ranging from −1.3 to 0.6 µg m⁻³.

The situation in the FWP region is similar to that of the NCP region, with anthropogenic pollutant emissions as the primary driver of reduced PM_2.5 concentrations. During the PreG and PostG periods, the contributions of anthropogenic pollutant emissions to PM_2.5 levels were −5.75 and −22.18 µg m⁻³, respectively. In contrast, meteorological conditions contributed to an increase in PM_2.5 concentrations, with a contribution of 2.32 µg m⁻³ in the PreG period, comparable to the impact of anthropogenic pollutant emissions. Meanwhile, the influence of CO₂ emissions on PM_2.5 levels was relatively minor.

In the YRD region, anthropogenic pollutant emissions are the primary driver of reduced PM_2.5 concentrations. Due to its location in eastern China, the YRD region is more affected by the EASM, resulting in more pronounced effects of changing meteorological conditions on PM_2.5 levels compared to the NCP and FWP regions. During the PreG period, the impact of meteorological conditions on PM_2.5 concentrations reached as high as −6.31 µg m⁻³.

In the PRD region, changes in anthropogenic pollutant emissions have contributed to a reduction in PM_2.5 concentrations, ranging from −8.45 to −3.82 µg m⁻³. However, changes in meteorological conditions and CO₂ emissions have led to increases in PM_2.5 levels, ranging from 0.11 to 1.49 µg m⁻³. Similar to the YRD region, the effects of changing meteorological conditions on PM_2.5 concentrations are significant, peaking at 1.49 µg m⁻³ during the PreG period. Located in southeastern coastal China, the PRD's rich vegetation cover enhances the impact of CO₂ emission changes on PM_2.5 concentrations. During the PreG period, the influence of CO₂ emission changes on PM_2.5 levels reached 1.13 µg m⁻³, comparable to the effect of anthropogenic pollutant emissions (−3.82 µg m⁻³). In the PostG period, the impact of CO₂ emission changes (0.31 µg m⁻³) surpassed that of meteorological conditions (0.11 µg m⁻³).

In the SCB region, the basin topography results in relatively minor effects of meteorological conditions and CO₂ emission changes on PM_2.5 levels, with contributions ranging from −1.14 to 0.29 µg m⁻³ during both the PreG and PostG periods. In contrast, anthropogenic pollutant emissions are the primary drivers of reduced PM_2.5 concentrations, exerting an impact of −32.09 µg m⁻³ during the PostG period.

Figure 7 The total changes in PM_2.5 concentrations (All, SIM_Base−SIM₂₀₀₈) for the North China Plain (NCP), Fenwei Plain (FWP), Yangtze River Delta (YRD), Pearl River Delta (PRD), and Sichuan Basin (SCB) during the EASM period in the PreG (2009–2013) and PostG (2014–2018) periods relative to 2008, along with the variations in PM_2.5 due to anthropogenic pollutant emissions (Emis, All-Met-CO₂), meteorological conditions (Met, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$), and CO₂ emission (CO₂, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) changes.

Table 6 Changes in total PM_2.5 concentrations (ALL, SIM_Base−SIM₂₀₀₈) and the impacts of anthropogenic pollutant emissions (Emis, All-Met-CO₂), meteorological conditions (Met, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{\mathrm{MET}=\mathrm{2008}}$), and CO₂ emission (CO₂, SIM${}_{\mathrm{Base}}-{\text{SIM}}_{{\mathrm{CO}}_{\mathrm{2}}=\mathrm{2008}}$) variations on PM_2.5 concentrations (µg m⁻³) during the EASM period in the PreG (2009–2013) and PostG (2014–2018) periods relative to 2008.

3.5 Attribution of changes in PM_2.5

Figure 8 illustrates that PM_2.5 concentrations remained relatively stable across the five regions during the PreG period. However, in the PostG period, following the implementation of the Clean Air Action Plan, significant reductions in PM_2.5 concentrations were simulated in the NCP, FWP, YRD, and SCB regions, while the PRD region showed the smallest decrease.

Anthropogenic pollutant emissions are the primary factor driving PM_2.5 concentration reductions across the five regions, with their impact increasing linearly over time. In the PreG period, meteorological conditions had a relatively stronger influence on PM_2.5 levels, occasionally surpassing the effects of anthropogenic pollutant emissions. For example, in 2013, the meteorological and emission impacts on PM_2.5 in the NCP region were −17.35 and 4.49 µg m⁻³, respectively. Similarly, in the FWP region from 2013 to 2015, meteorological impacts ranged from −16.9 to −15.36 µg m⁻³, while emissions affected PM_2.5 concentrations between −15.8 and −2.27 µg m⁻³. The influence of meteorology also exceeded that of emissions in the YRD region during 2011–2012 and in the PRD region in 2010. Even in the SCB region, where meteorological impacts on PM_2.5 were relatively minor, meteorological effects in 2010 (8.59 µg m⁻³) were comparable to emissions (−14.67 µg m⁻³).

The influence of CO₂ emission changes on PM_2.5 levels was generally minor, but in the densely vegetated PRD region could be comparable to the effects of emissions and meteorology. The influences of CO₂ emissions, anthropogenic pollutant emissions, and meteorology on PM_2.5 are −0.25 to 3.11, −6.19 to −1.47, and −0.5 to 3.11 µg m⁻³, respectively from 2009 to 2013.

Our attribution analysis of PM_2.5 concentration changes is mainly consistent with previous studies, which have indicated that variations in anthropogenic pollutant emissions were the primary driver of PM_2.5 changes in China during the period 2013–2017, with meteorological conditions contributing approximately 9 %–26 % (Zhang et al., 2019; Chen et al., 2020a; Cheng et al., 2019). In our study, relative to 2008, the average contribution of anthropogenic pollutant emissions during the PreG period was 89.08 %, while meteorological conditions contributed 16.45 %. In the PostG period, following the implementation of the Clean Air Action Plan, the influence of anthropogenic pollutant emissions further increased to 96.26 %, whereas the contribution of meteorological conditions declined to 1.60 %. This finding underscores that the impact of changes in anthropogenic pollutant emissions on PM_2.5 concentrations was markedly enhanced after 2013. Notably, changes in CO₂ emissions had a significant impact on PM_2.5 levels, contributing −5.46 % during the PreG period and 2.14 % during the PostG period, with the latter effect surpassing that of meteorological conditions.

Figure 8 Changes in total PM_2.5 concentrations from 2008 to 2018 (ALL, red line) and the contributions of anthropogenic pollutant emissions (Emis, black line), meteorological conditions (Met, blue line), and CO₂ emission changes (CO₂, green line) to PM_2.5 concentrations (Units: µg m⁻³).

3.6 Uncertainties

The uncertainties in the MEIC emission inventory primarily arise from activity data, emission factors, spatial and temporal allocation methods, and the implementation status of pollution control measures (Hong et al., 2017; Zheng et al., 2021a), all of which may affect the accuracy of simulation results. Future improvements can be achieved by employing more refined and accurate emission inventories.

In addition, the use of a 60 km low-resolution grid limits the ability to represent local topography and physical processes, thereby introducing simulation errors (Harris et al., 2016; Ringler et al., 2013). Given that this study employs a fully coupled regional climate–chemistry–ecology model with extended simulation periods (three sets of 10-year simulations) and a broad regional scope (covering the entire East Asia region), computational resource constraints necessitated the use of 60 km grids. Numerous studies have employed the RegCM-Chem-YIBs model at a 60 km grid resolution to systematically analyze PM_2.5, O₃, CO₂, and the regional climate over East Asia (Ma et al., 2023a, b; Xu et al., 2023; Gao et al., 2021). These demonstrate its robustness and reliability in simulating East Asian atmospheric and climatic conditions. Future studies could enhance simulation accuracy by increasing computational resources and employing higher-resolution grids.

4 Conclusions

This study employed numerical experiments with the RegCM-Chem-YIBs model to analyze the interannual variability of near-surface PM_2.5 concentrations in East Asia from 2008 to 2018. The analysis examines the drivers of annual PM_2.5 changes in detail, focusing on three key factors: anthropogenic pollutant emissions, meteorological conditions, and CO₂ concentration changes.

Compared to 2008, PM_2.5 concentrations in East Asia exhibited minimal change during the PreG period, with most areas showing variations between −10 and 5 µg m⁻³. In contrast, following the implementation of the Clean Air Action Plan, PM_2.5 concentrations decreased significantly (the PostG period). This reduction was especially notable in the NCP and the SCB regions, with declines of 36.76 and 33.96 µg m⁻³, respectively.

Anthropogenic pollutant emissions are the primary driver of the decline in PM_2.5 concentrations in East Asia, with their impact on PM_2.5 levels increasing linearly over time. During the PreG and PostG periods, the contributions of anthropogenic pollutant emissions to PM_2.5 concentrations in the NCP, FWP, YRD, PRD, and SCB regions ranged from −10.39 to −3.82 and −33.86 to −8.45 µg m⁻³, respectively.

Changes in meteorological conditions have led to decreased PM_2.5 concentrations along China's eastern coastal and western inland regions, while increasing PM_2.5 levels in central areas. During the PreG period, the influence of these meteorological changes on PM_2.5 concentrations was comparable to that of anthropogenic pollutant emissions, ranging from −6.31 to 2.32 µg m⁻³.

CO₂ indirectly influences PM_2.5 concentrations by affecting precipitation and isoprene emissions from vegetation. In the sparsely vegetated NCP and FWP regions, CO₂ impacts near-surface PM_2.5 primarily through changes in precipitation. Conversely, in the vegetation-rich PRD region, CO₂ affects PM_2.5 concentrations mainly by altering isoprene emissions, with an impact comparable to that of anthropogenic pollutant emissions. From 2009 to 2013, the effects of anthropogenic pollutant emissions and CO₂ changes on PM_2.5 ranges are −0.25 to 3.11 and −6.19 to −1.47 µg m⁻³, respectively.

In summary, PM_2.5 concentrations in East Asia have significantly declined since 2013, primarily driven by changes in anthropogenic pollutant emissions. During several years of the PreG period, variations in meteorological conditions affected PM_2.5 levels to a degree comparable to that of anthropogenic pollutant emissions. However, following the implementation of the Clean Air Action Plan in 2013, the influence of anthropogenic pollutant emissions increased significantly, while the impact of meteorological factors diminished considerably. This simulation underscores the critical importance of stringent air pollution control measures in mitigating PM_2.5 concentrations. Moreover, we highlight that in regions with dense vegetation cover, changes in CO₂ emissions play a noteworthy role in regulating PM_2.5 levels, with the average effect during the PostG period even surpassing that of meteorological conditions. Given the sustained rise in CO₂ levels in recent years, it is imperative to integrate the modulatory effects of CO₂ into PM_2.5 simulating models and control strategies.