Impact of Large-Scale Circulations on Ground-Level Ozone Variability over Eastern China

Abstract

The seasonal and interannual variations in ground-level ozone across eastern China from 2014 to 2022 were strongly influenced by meteorological conditions and large-scale atmospheric circulations. We applied empirical orthogonal function (EOF) and singular value decomposition (SVD) analyses to explore these relationships. The EOF analysis identified three primary patterns of ozone variability: a dominant seasonal cycle over most of mainland China, an anti-correlation between northern and southern China during transitional seasons, and elevated springtime ozone concentrations in coastal regions. The SVD results further demonstrated that seasonal ozone variability was primarily driven by the annual radiation cycle across much of China. In contrast, the East Asian summer monsoon (EASM) was linked to the relatively low summer ozone levels observed in southern China. The anti-correlation between northern and southern China was associated with western Pacific subtropical high (WPSH) movement, which promoted sunny weather conditions and was conducive to ozone formation. Additionally, high springtime ozone levels in northern coastal regions were influenced by pollutant transport from continental cold high (CCH) events, while the cloud-free conditions and intense solar radiation in southern China contributed to elevated ozone concentrations.

1. Introduction

Ground-level ozone (O₃) has received widespread attention due to its harmful impacts on human health and ecosystems [1]. It is formed as a secondary air pollutant through photochemical reactions involving volatile organic compounds (VOCs), nitrogen oxides (NOx), and carbon monoxide (CO) [2]. However, the relationship between ozone and its precursors is complicated and remains uncertain [3,4]. In recent years, significant increases in ozone levels have been observed in China, with both ground-level ozone and total column ozone over eastern China showing notable upward trends, increasing by 0.23 ppbv/y and 0.28 DU/y, respectively [5,6]. Despite the Chinese government’s implementation of various air pollution control measures, addressing ozone pollution has proven challenging. Consequently, there is growing concern over the factors contributing to the increase in ground-level ozone, particularly those driving long-term trends.

The annual averaged ground-level ozone concentration in China could increase due to three main potential ways, even if there is no observed continuous increase in the concentration of precursors (VOCs and NOx). These potential ways include: (1) non-linear chemical effects resulting from the simultaneous emission control of ozone precursors, aerosols, and other species during photochemical reactions; (2) increased photolysis rates due to reduced aerosol light extinction and natural factors enhancing radiative flux; (3) altered atmospheric transport or less dilution driven by natural factors, such as boundary layer meteorology or large-scale circulation. The first two mechanisms are anthropogenic, while the third is influenced by natural variability.

Since 2013, significant progress has been made in reducing primary air pollutants in China, prompting substantial research into the first two mechanisms. For instance, some studies focus on the non-linearity of O3-VOCs-NOx reactions and how ozone formation regimes impact ozone exceedance events [5,7]. Although NOx control measures can reduce peak ozone levels, they may paradoxically elevate average ozone concentrations [8]. Indeed, nationwide NOx reduction since 2013 has led to an increase in mean ozone levels in China. However, this non-linear response to NOx reductions alone cannot fully explain the observed rise in ozone concentrations in China [9]. Reductions in aerosols, achieved through effective emission controls, may also contribute to elevated ozone levels by influencing both chemical reactions and atmospheric dynamics [9,10,11,12]. For example, using GEOS-Chem simulations, Li et al. estimated that aerosol reductions account for approximately 40% of the ozone increase over the North China Plain (NCP) by slowing the removal of hydroperoxy radicals (HO₂), thereby accelerating ozone formation [9,10,11]. Additionally, several studies have shown that reductions in aerosols can enhance ozone formation by increasing photolysis rates [12,13,14,15,16]. While extensive research has examined anthropogenic mechanisms, studies exploring the role of natural factors in constraining the long-term trend of ozone are still limited.

Han et al. identified the most influential meteorological factors in different regions over eastern China and established a quantitative linkage between the daily variability in summertime ground-level ozone and meteorology [17]. Using a multiple linear regression (MLR) model, Chen et al. demonstrated that natural factors, including solar radiation cycle, precipitation, surface temperature, and meridional wind at 600 hPa and 1000 hPa, could explain 27% of the upward ground-level ozone trends in southern China [18]. The large-scale circulations driving the change in these meteorological parameters could play an important role in the interannual variability in ozone pollution. Several studies have reported the impact of a single atmospheric circulation variation on the variability in ozone. For example, the atmospheric subsidence of the western Pacific subtropical high (WPSH) could lead to weather conditions that promote ozone formation during summertime pollution episodes over the Yangtze River Delta (YRD) region in China [19]. Statistical analysis of ozone concentrations in Korea has shown that the strengthened western North Pacific subtropical high (WNPSH) contributed to the observed increase in ozone levels during summer, mainly due to the unusual precipitation pattern [20]. Based on statistical analysis of ozone data collected across China, Zhao and Wang suggested that the daily variability pattern of summer ozone, with a north–south contrast, is associated with the variability in WPSH intensity [21]. Furthermore, as an important element of background atmospheric circulation in East Asia, the impacts of the East Asian summer monsoon (EASM) on ozone variation have also been widely studied in different regions [22,23]. For example, Li et al. emphasized that the interannual variability in the EASM can affect the spatial distribution of ozone in summer through lower tropospheric wind, cloud cover, and downward shortwave radiation, which influence the transport and chemical formation of ozone [24]. While these studies have provided valuable insights into how specific large-scale circulations influence ground-level ozone in China, there is still a lack of research investigating all dominant circulations and their relative impacts on the spatiotemporal variability patterns of ground-level ozone. Such research could help understand the connection between local air pollution variability and large-scale circulation on a mid- or long-term scale.

This paper utilized statistical methods to analyze ground-level ozone and meteorological data collected from 2014 to 2022 in eastern China. The goal was to understand the impact of meteorological factors and large-scale circulation patterns on ground-level ozone variability in eastern China at seasonal and interannual scales. The study involved using empirical orthogonal function (EOF) analysis to identify spatiotemporal patterns of ground-level ozone and circulation, and using singular value decomposition (SVD) analysis to investigate the driving circulations for seasonal and interannual ozone variability. This paper is structured as follows: Section 2 introduces the observational data and the main statistical methods used. The results and discussion section includes three parts: (a) An EOF analysis of ozone, investigating the dominant ozone variation patterns over eastern China; (b) An EOF analysis of meteorology, presenting the important circulation patterns in East Asia; and (c) An SVD analysis, analyzing the associations of ozone variation with meteorology and large-scale circulations.

2. Data and Methods

2.1. Data

The Chinese Ministry of Ecology and Environment (MEE) expanded its surface monitoring network in 2013, providing real-time data on six key pollutants (particulate matter, fine particulate matter, sulfur dioxide, nitrogen dioxide, carbon monoxide, and ozone) across China (http://www.mee.gov.cn/, last accessed: 14 October 2024). This study focused on the relationship between atmospheric circulation and ground-level ozone on seasonal and interannual scales. It analyzed monthly and daily average ozone records. Out of more than 2000 monitoring stations in eastern China (Figure S1), only 1273 with continuous daily records during 2014–2022 were used for the analysis. This selection was made to ensure the accuracy of the statistical analysis, and these stations were well distributed across eastern China, making them suitable for a nationwide study.

The meteorological data used in this study were the NCEP-NCAR Reanalysis 1 data provided by the NOAA PSL, Boulder, CO, USA (https://psl.noaa.gov, last accessed: 14 October 2024). The data include the geopotential height at 925 hPa (Hgt925), surface wind field (Wind), surface relative humidity (RH), vertical winds at 500 hPa (W500, with positive/negative values representing upward/downward winds), and net surface shortwave radiation (Rad), as well as total cloud cover (Tcc). The spatial resolutions of Hgt925, Wind, RH, and W500 were 2.5° × 2.5°; the others were Gaussian grid with 192 × 94 global grids.

2.2. Empirical Orthogonal Function

Empirical orthogonal function (EOF) analysis, first proposed by Pearson in 1902, is a method used to decompose variable fields that change with time (A(n, t)) into pairs of orthogonal spatial patterns and corresponding time series (Equation (1)). The spatial patterns, also known as loadings, eigenvectors, or EOFs, illustrate the geographical distribution characteristics of the variable field. Time series, on the other hand, are expansion time coefficients known as scores, eigenvalues, or principal components (PCs). The EOF modes are ranked in decreasing order based on the percentage of the variance they account for. Hence, the original dataset can be reconstructed by using the first several significant modes (k) according to the variance contribution (Equation (1)). EOF analysis is a widely used and powerful method for analyzing dominant modes and was first introduced to meteorology by Fukuoka in 1951 [25,26]. It is still extensively used in atmospheric and meteorological science [27].

$$\mathrm{A}(\mathrm{n},\mathrm{t})\approx {\sum }_{\mathrm{i}=1}^{\mathrm{k}}{\mathrm{E}\mathrm{O}\mathrm{F}}_{\mathrm{i}}\left(\mathrm{n}\right){\mathrm{P}\mathrm{C}}_{\mathrm{i}}\left(\mathrm{t}\right)$$

(1)

In this study, EOF analysis was performed on ground-level ozone and meteorological parameters. The daily averaged ground-level ozone data were from 1273 sites in eastern China, while the meteorological data were retrieved within the latitude and longitude range of [5°–50° N, 95°–155° E], which covers most of eastern Asia. To present the changes in the original dataset more intuitively, we normalized the PCs in this paper. Then, the normalized PCs were regressed with the original matrix to obtain the new EOFs. In this way, EOFs can represent the change in the original matrix when PCs change one unit.

2.3. Singular Value Decomposition

Singular value decomposition (SVD) is a powerful mathematical technique used to decompose a matrix into several component matrices that may contain data of interest or meaningful information [28]. In atmosphere science, SVD is often applied to analyze the coupling or relationship between two fields and to study the spatial and temporal patterns of their dominant interaction [28,29].

Specifically, by combining the normalized ozone data and any meteorological data into one matrix (X) and performing SVD, we can decompose it as follows:

$$\mathrm{X}=\mathrm{U}\mathrm{S}{\mathrm{V}}^{\mathrm{T}}\approx {\sum }_{\mathrm{i}=1}^{\mathrm{k}}{\mathrm{S}\mathrm{V}}_{\mathrm{i}}{\mathrm{u}}_{\mathrm{i}}{\mathrm{v}}_{\mathrm{i}}^{\mathrm{T}}$$

(2)

where U and V are the orthogonal matrices whose columns are the X’s left and right singular vectors, respectively. S is a diagonal matrix containing the singular values (SVs) of X, arranged in descending order. Therefore, each singular vector pair, u_i and v_i (from U and V, respectively), describes a mode of co-variability between ozone and the analyzed meteorological fields. The corresponding singular value (SV_i) quantifies the strength of this mode, with higher singular values indicating stronger correlations. Similarly to the EOF analysis, we determined the dominated correlated patterns between ozone and meteorological factors by using the first several significant modes (k) (Equation (2)).

3. Results

3.1. Spatiotemporal Patterns of Ground-Level Ozone

Figure 1 shows the ground-level ozone’s EOF results (first three leading modes). The top panels display the spatial characteristics (O3_EOF1–O3_EOF3), and the bottom panel shows their associated time coefficients (O3_PC1–O3_PC3, performing a 30-day moving average for showing the seasonal variability). These three leading modes account for 63.1% of the total variance.

In Figure 1a, the O3_EOF1 values are mainly positive fors stations in mainland China, indicating that ground-level ozone in most of mainland China followed a consistent temporal variation pattern (as represented by O3_PC1 in Figure 1d). According to the principle of the EOF analysis, when O3_PC1 > 0 (or O3_PC1 < 0), the ground-level ozone concentration in mainland China was higher (or lower) than average. The variation in O3_PC1 in Figure 1d shows a pronounced seasonal pattern, indicating that when O3_PC1 peaked in late spring and summer or reached its lowest point in winter, the ozone concentration also reached its maximum or minimum. Therefore, O3_EOF1 and O3_PC1 described the seasonal variation and the basic averaged spatial pattern of ground-level ozone over most of eastern China, explaining approximately 47.4% of the total ozone variance.

Upon examining Figure 1a closely, it is clear that the O3_EOF1 values decreased from north to south, suggesting that the seasonal variation amplitude of ozone and the “single peak (unimodal)” pattern were more significant in northern China (NorthC) than in southern China (SouthC). These findings were consistent with previous local and regional studies. For example, in NorthC, the ozone concentration in Shandong Province, Xi`an, and the Beijing–Tianjin–Hebei region showed unimodal seasonal patterns, with peak ozone levels typically occurring in June and low ozone levels in December or January [30,31,32]. In the middle latitudes, the ozone in the Yangze-River Delta (YRD) region also showed a unimodal trend, but the highest ozone appeared in May, earlier than NorthC [33]. In some low-latitude areas of China, the seasonal variation in the amplitude of ozone was smaller, showing a bimodal seasonal pattern with peak ozone values appearing both in spring and autumn, particularly in autumn [34].

In light of the negative O3_EOF1 values in Taiwan and some coastal stations in southeastern China (Figure 1a), we comprehensively compared the ozone variations at various sites. In Figure S2a, the ozone time series between sites with positive and negative O3_EOF1 values in the Pearl River Delta (PRD) region in southeastern China were compared. The comparison suggests that their seasonal trends were similar, but lower summer ozone concentrations were measured at the negative sites (Figure S2a). Moreover, we conducted a comparison of summer ozone levels in Taiwan (with negative O3_EOF1 values), Fujian (coastal mainland China, with positive O3_EOF1 values of a lesser magnitude), and central China (with positive O3_EOF1 values). Our findings reveal that the standardized summer ozone levels in Taiwan and Fujian were notably lower than those in central China, with interannual variations. This further validated the significant role of the remarkably low summer ozone concentration in leading to small or negative O3_EOF1 values. This intriguing seasonal variation pattern in Taiwan and southeastern China may be attributed to the oceanic air dilution effect under prevailing summer monsoon southerlies or the accompanying cloudy weather, which was not conducive to ozone formation [35,36].

In Figure 1b, the O3_EOF2 values in SouthC were strongly positive, while they were negative in NorthC. The corresponding O3_PC2 values (Figure 1d) showed a regular annual bimodal trend and generally reached their peak in transitional seasons (March to May and October, respectively), explaining 10.8% of the total variance. Specifically, when the values of O3_PC2 peaked in the transitional season, the ground-level ozone concentration in SouthC increased, especially in autumn (~10 ppbv). When O3_PC2 dropped to negative values in summer, the stations in SouthC with positive O3_EOF2 values accompanied a decrease in ozone, while the stations in NorthC with negative O3_EOF2 values saw an increase in ozone levels. It can be inferred that the second mode of ozone EOF represents an anti-correlation pattern of the intraseasonal/interannual variability between NorthC and SouthC.

The third mode accounts for 4.89% of the total variance. As depicted in Figure 1c, O3_EOF3 indicates high ozone levels in both the coastal areas of NorthC and SouthC, with O3_PC3 showing positive values in spring and autumn. Further analysis of the ozone observations in spring and autumn revealed that this EOF mode generally represents high ozone levels in coastal NorthC in spring (Figure S3a) and high ozone levels in coastal SouthC in both spring and autumn (Figure S3b). The former has also been reported in previous studies [22,37], and the latter confirms the extremely high ozone levels in the autumn in SouthC.

The three leading modes mentioned above all pass the significance test proposed by North et al. [38]. Their combined variance reached 63.1%. This study did not explore the remaining modes further due to their limited contributions and lack of clear patterns. In the following sections, we focus on examining the major atmospheric circulations and investigating their associations with the ozone variation patterns.

3.2. Dominant Circulation Patterns Derived from EOF Analysis

To identify the dominant large-scale circulation patterns, we applied EOF analysis to the surface wind field (Wind) and geopotential height at 925 hPa (Hgt925). The first three modes are presented in Figure 2. EOF1 revealed the most significant circulation patterns in eastern China, specifically the seasonal cycle of the East Asian Monsoon (EAM), characterized by northerly winds in winter and southerly winds in summer (Figure 2a). This was accompanied by lower pressures in summer and higher pressure in winter (Figure 2a). EOF2 depicted a cyclone system over the western Pacific, with strong northeastern winds impacting SouthC. In contrast, the winds were weak in NorthC (Figure 2b). Based on PC2, the most significant positive and negative phases occurred in autumn and spring, respectively (Figure 2d). This mode helps to illustrate the movement of the western Pacific subtropical high (WPSH), which was located at low latitudes before summer, then moved northward during summer and autumn and was anomalous at low latitudes. EOF3 also showed a low-pressure cyclone system but located around Japan. This indicates that there is a land–sea pressure contrast pattern in winter (positive phase of PC3), while it usually turns to a negative phase in spring, which may be related to the eastward area of cold high systems.

We also conducted an EOF analysis on several other related meteorological factors. Despite the driving factors of these parameters being more complex than the wind field, the primary components derived from the EOF analysis were quite similar. For example, the first EOF modes of surface shortwave radiation (Rad), total cloud cover (Tcc), and surface relative humidity (RH) were consistently dominated by seasonal variations (Figures S4–S6). Moreover, these EOF results also identified anti-correlations between SouthC and NorthC, including the third mode of Rad (Figure S4c) and RH (Figure S6c) and the second mode of Tcc (Figure S5c).

3.3. SVD Analysis

In order to investigate the potential connections between ozone and large-scale atmospheric circulations, we applied singular value decomposition (SVD) analysis to the combined matrix of ozone and meteorological variables. The coupled SVD modes allowed us to identify strong correlated patterns between ozone and meteorological factors.

Figure 3 illustrates the SVD results of ozone and surface wind field. It revealed that the first three SVD modes of ozone (Figure 3a–c) closely resembled the first three empirical orthogonal functions (EOF1 to EOF3) of ozone (Figure 1a–c), accounting for almost 98% of the total variance. The correlation coefficients between singular values (SV1–SV3, Figure 3g) of ozone and surface wind fields were 0.8, 0.5, and 0.44 for each mode, respectively. The SVD results between ozone and surface net shortwave radiation (Figure 4) also revealed close correlations, with the first three SVD modes explaining approximately 99.5% of the total variance. The correlation coefficients of singular values (SV1–SV3, Figure 4g) between ozone and radiation for each mode reached 0.86, 0.53, and 0.46, respectively.

Although the correlations found by SVD analysis may not necessarily imply a causal relationship, we can still uncover the association between ozone and meteorology based on their physical and chemical relationships. Notably, the first coupled mode of ozone and wind field (Figure 3a,d), as well as those of ozone and net shortwave radiation (Figure 4a,d), indicate that seasonal variations in the EAM and shortwave radiation were significantly correlated with large-scale variations in ground-level ozone concentrations (Figure 3g and Figure 4g). These findings were consistent with previous studies that have highlighted the roles of the EAM and solar radiation in driving fluctuations in ground-level ozone. The impact of solar radiation is well documented, as it determines the photolysis rates associated with ozone photochemical production [24]. In SouthC, the summer radiation was lower than NorthC, resulting in lower ozone levels (Figure 4a,d). Meanwhile, the EAM influenced ground-level ozone levels primarily through its seasonal wind field patterns and associated meteorological conditions. Specifically, SouthC was significantly affected by the southerlies of the East Asian summer monsoon (EASM), which bring warm, moist oceanic air masses to SouthC (Figure 3d). This resulted in higher relative humidity (Figure S7d), more convective conditions (Figure S8d), and cloudier skies (Figure S9d). Clouds in this region can reduce the amount of solar radiation reaching the surface, thereby inhibiting ozone formation, and can also absorb ozone and its precursors, as well as oxidant radicals, further suppressing ozone production [39,40]. Therefore, our results suggest that the lower ozone concentration in SouthC in summer may be due to the cloudy weather conditions, which the EASM influenced. In contrast, the significant seasonal variation pattern of ozone in NorthC was related to the seasonal cycle in radiation.

Additionally, this study emphasized the relationship between ozone and the EASM on an interannual scale. As shown in Figure 5a, the annual O3_PC1 exhibited a similar trend to the strength of the EASM, suggesting that ground-level ozone was relatively lower (or higher) in most of eastern China when the EASM was weaker (or stronger). This finding aligned with previous research demonstrating that the EASM significantly influenced the interannual ozone levels through changes in wind patterns, which was critical for controlling the transport of ozone and its precursors across China [41]. Specifically, during weaker-EASM years, reduced inflow of ozone from west China and increased eastward outflow lead to more ozone being transported out of China, which effectively lowers average ozone levels across China in weaker-monsoon years. Conversely, in stronger-EASM years, the diminished eastward outflow allows for greater retention of ozone within China, resulting in comparatively higher average ozone concentrations. Such a transboundary transport mechanism was demonstrated to be the most dominant factor controlling ozone concentrations in weak/strong-EASM years in China [41,42].

The second SVD mode shows an anti-correlation between NorthC and SouthC. Similarly to the EOF2 of the surface wind field (Figure 2b), the SVD2 of surface wind (Figure 3e) also illustrated the movement of the WPSH system. This indicates that the NorthC and SouthC anti-correlation may be due to the variation (i.e., movement or strength) in the WPSH, which was further supported by the similar anti-correlation patterns reflected in the SVD2 of vertical winds at 500 hPa (W500, Figure S8e), total cloud cover (Tcc, Figure S9e), and shortwave radiation (Figure 4e). The underlying mechanism involved the atmospheric subsidence and sunny weather conditions associated with the WPSH, resulting in more solar irradiance reaching the near-surface and thus rapid ozone formation. This phenomenon was most significant in the autumn in SouthC, which had high ozone levels. Moreover, previous studies have suggested that the strength, movement, and western edge of the WPSH have undergone interannual changes [43]. Some studies reported that precipitation responses to the interannual variability in the WPSH were opposite between SouthC and NorthC [44,45]. Therefore, we used a WPSH strength index defined by the accumulative enhancement of geopotential height above the 5880 gpm isoline at 500 hPa averaged over the region in the range of [10°–40° N, 80°–180° E] to study its relevance with O3_PC2 at the annual scale. Figure 5b plots the annual O3_PC2 series and the opposite of the WPSH strength index (-WPSHI), which exhibited good agreement. The negative correlation means that when eastern China experienced a stronger WPSH that year, ozone enhancement was observed in NorthC, while a lower ozone level was observed in SouthC; the opposite relationship also holds. Understanding the interannual correlation between ozone and circulation patterns increases the possibility of predicting ozone levels on long-term scale.

Similarly, the third SVD mode of the ozone and surface wind field (Figure 3f) suggests that the high ozone concentration in NorthC may be linked to the southerly winds, which bring highly polluted air masses from the YRD region. Huang et al. have reported the ozone enhancement mechanisms influenced by regional transport [46]. The abnormal southerlies over the coastal area of NorthC may be associated with the frequent continental cold high (CCH) activities in this period. Liao et al. provided evidence that southward-moving CCH systems, not continuously replenished with cold air masses, could be blocked in eastern China [47]. As a result, the coastal region is under the control of an anticyclonic system, which is conducive to ozone pollution. Additionally, in SouthC, such as Taiwan, the high ozone levels are likely influenced by the transport of pollutants from northeast Asian continental outflow (Figure 3f), which is associated with cold air outbreaks and circulations along the east boundary of the CCH [48]. By coupling long-term ozone observations and regional-scale chemical transport model (CTM) simulations, Chiang et al. reported that the northeast Asian continental outflow, which often happens under the influence of a NorthC high-pressure system, contributes over 30% of the ozone mixing ratio in North Taiwan [49]. They also found that the frequency of the occurrence of NorthC high-pressure systems influenced the frequency of high-ozone events in Taiwan. Regarding solar radiation, it is likely an important factor influencing the springtime ozone levels in SouthC, as suggested by Figure 4f, because the weather is usually sunny in this period before the breaking of the EASM, and the solar radiation reaches its maximum (Figure S4b).

The SVD analyses above have identified the EASM, WPSH, and CCH as the essential large-scale circulation systems that influence the interannual and intraseasonal variation in ozone over eastern China. These systems modulated the ozone concentration through physical and chemical pathways, particularly by influencing wind patterns and solar radiation.

4. Conclusions

The variability in ground-level ozone in eastern China was influenced by meteorological factors and large-scale circulation systems at different time scales. This study specifically focused on the seasonal and interannual time scales. Through EOF analysis, three spatiotemporal patterns of ground-level ozone over eastern China from 2014 to 2022 were identified: (1) the dominant seasonal variation in most of eastern China, (2) the anti-correlation between SouthC and NorthC during spring and autumn, and (3) elevated springtime ozone levels in coastal regions.

By combining SVD analysis and the underlying mechanisms, this study uncovered the associations of ozone patterns with meteorological factors and large-scale circulation systems. First, our findings indicate that the seasonal variation pattern of ozone was primarily controlled by the seasonal cycle of solar radiation, particularly in NorthC, as solar irradiance greatly impacted the photochemical production of ozone, whereas the summer monsoon-related southerlies in summer could lead to weather conditions that were not conducive to ozone formation in SouthC, hence lowering the ozone levels. Second, the movement of the western Pacific subtropical high (WPSH) accompanied by weather conditions conducive to ozone formation played a crucial role in the NorthC-SouthC anti-correlation mode in high-ozone seasons. Third, this study also found that springtime ozone increment in coastal areas of eastern China was impacted by the continental cold high (CCH) system, particularly in NorthC and Taiwan, mainly through the regional transport of pollutants. The high springtime ozone levels in SouthC were influenced by weather conditions that were conducive to ozone formation.

On a long-term scale, the interannual variation in the East Asian summer monsoon (EASM) could influence ozone via transport effects. Ground-level ozone was relatively lower (or higher) in most of eastern China when the EASM was weaker (or stronger). Additionally, the WPSH also influenced the anti-correlation of NorthC–SouthC on an interannual scale: when eastern China experienced a stronger WPSH that year, ozone enhancement was observed in NorthC, while a lower ozone level was observed in SouthC; the opposite relationship also holds.

In conclusion, this study provides important insights into the influence of large-scale circulation patterns on seasonal and interannual ozone variability in eastern China, offering valuable perspectives for improving air quality management and informing policy development in the context of ongoing climate change. Nevertheless, we still need to emphasize that these results should be interpreted with caution regarding long-term ozone predictability, since the analysis in this study used data obtained over a relatively short period (2014–2022, less than a decade). Future studies incorporating longer observational records will be essential for a more comprehensive examination of the effects of large-scale circulation patterns on ozone variability, establishing a stronger foundation for long-term predictions and broadening the applicability of our findings.