Chemically enhanced transboundary ozone pollution suppresses city-level emission control benefits

As an important worldwide environmental threat, ambient ozone pollution harmfully impacts public health [1, 2], crop production [3], and the surface ecosystem [4]. Despite substantial mitigation progress in developed regions, such as Europe and the United States, city-level ozone reduction remains a challenging task. In many cities in developing regions, the threat of ozone pollution has been increasingly pervasive [5]. Although city-level ozone mitigation is closely tied to achievements of the second (zero hunger) and third (good health and wellbeing) sustainable development goals of the United Nations, the effective and efficient control of city-level ozone pollution remains an unresolved issue.

Over past decades, the observed increase in surface summertime ozone pollution in cities in the North China Plain and Fenwei Plain (NCP + FP) has been substantially higher than those observed in other cities worldwide [5]. Furthermore, NCP + FP is characterized by its dense population and intensive agricultural activity. Thus, the Government of China has determined that ozone pollution must be reduced. For example, China's 14th Five Year Plan (FYP, 2021–2025) sets a specific target of at least a 10% reduction in ozone precursor emissions. To date, in NCP + FP cities, emission control has been conducted by the Government, which implemented collaborative '2 + 26' (simultaneous emission control in its 2 + 26 cities) and 'one city one policy' (city-specific emission control policies) programs. In 2021, the '2 + 26' program was extended to '60 cities' in the entire NCP + FP. Although these policies were adopted based on the success in PM_2.5 mitigation, they did not lead to ozone reductions in most cities [6]. In addition to anthropogenic emissions [7], massive emissions from natural sources (i.e. soil NO_x [2], biogenic volatile organic compounds (VOCs) [8]) further complicates ozone mitigation. Therefore, the effectiveness of current ozone mitigation policies must urgently be investigated to develop more effective strategies. Such an exploration could also serve as an important reference for other cities worldwide struggling with similar ozone threats.

Many previous studies on ozone mitigation of NCP + FP cities have attempted to design specific local emission reduction pathways for its individual cities or city clusters, such as Beijing [9], Jinan (a city in the NCP) [10], Xi'an (a city in the FP) [11] and Beijing–Tianjin–Hebei and surrounding areas [12–14]. An implicit assumption behind these studies is that the absolute benefit of local emission reduction would be determined by local emission control strategies and not be influenced by transboundary pollution. Previous and ongoing ozone mitigation policies in NCP + FP cities have been designed based on this assumption. However, these studies and policies have simplified impacts of transboundary transport.

In recent years, other studies have attempted to directly quantify the transboundary contribution on ozone pollution for individual cities or city clusters, such as Beijing [15], the NCP [16–18], and Xi'an. Another study [19] that focused on deep emission cuts (40%) by the mid-21st century preliminarily found that compared with only local emission control, nationwide emission control could be helpful for mitigating ozone in Beijing–Tianjin–Hebei and Yangtze River Delta regions. However, all these studies implicitly assumed that the amount of ozone contributed by local and transboundary (from outside the city or NCP + FP) sources could be (or approximately) linearly added. In fact, transported pollutants could also affect ozone chemical formation and alter the efficacy of the local control measures, as one previous study [20] has found that internationally transported pollutants could affect China's locally emitted pollution from PM_2.5 perspective. In all previous studies focusing on transboundary impacts on local ozone pollution, the embedded chemical interaction mechanism has not been mentioned and the authentic influence of the transboundary transport on ozone mitigation of NCP + FP cities remains unquantified. When local emissions are cut, the transboundary ozone contribution, including its chemical interactions with locally emitted pollutants, might change. Thus, the exact effects of transboundary transport between NCP + FP cities and from outside NCP + FP on the effectiveness of city- or region-specific ozone mitigation policies remain unclear.

Therefore, by focusing on summertime (June–July–August) ozone mitigation in NCP + FP (figure S1), we examine the effectiveness of collaborative emission control among and beyond its 60 cities. September and January, representative of fall and winter, are included for sensitivity analysis. We employ an average daily maximum 8 h average (MDA8) ozone, consistent with current regulations and ozone-related public health estimation [21]. We employ the nested GEOS-Chem model to simulate surface ozone after comprehensively evaluating, including surface measurements for simulated ozone (S5.1 in supplementary material) and NO₂ (S5.2), observations collected from literature for simulated NMVOCs (S5.3), satellite datasets for simulated ozone chemical regime (S5.4), and ground-based multi-axis differential optical absorption spectroscopy observations for simulated NO₂ vertical profiles (S5.5). We combine two groups of nested GEOS-Chem simulations, with the first for 2015 through 2020 focusing on historical transboundary transport (0.5° latitude × 0.625° longitude), and the second for a representative year of 2019, when ozone exposure and ozone-related premature deaths peaks [6] (0.25° latitude × 0.3125° longitude). For 2019, we have more detailed scenarios and the analyses in the main text are mainly based on them. More information about model set-up and scenario design is available in S1, S2 and table S1, with emission datasets in S4 of supplementary material. As a sensitivity test, WRF-Chem is also employed to test the consistency between multi-models (S3).

In this study, we, first, examine the effects of local ozone mitigation strategies within NCP + FP by assigning either a uniform or city-specific reduction ratio to each of the 60 NCP + FP cities. Then, we quantify the effects of nationwide emission control on city-level ozone mitigation over NCP + FP. Furthermore, we examine how transboundary transport from outside NCP + FP could be chemically enhanced with local emission reduction, explore the underlying chemical mechanism, and reveal the corresponding suppression impact on the effectiveness of local emission reduction.

Ozone pollution is severe over NCP + FP. Figure 1(a) shows simulated MDA8 ozone concentrations in summertime 2019 in BASE simulation. The regional average MDA8 ozone is 80.0 ppbv, and the highest value reaches 94.0 ppbv. Most cities exceed the Chinese standard (160 μg m⁻³, which is equivalent to 82.3 ppbv at 298.15 K and 1013.25 hPa), and the rest exceed the guidelines of the World Health Organization (100 μg m⁻³, equivalent to 51.5 ppbv). This is also the case for ozone pollution from 2015 through 2020 (figure S2).

Zoom In Zoom Out Reset image size

A uniform 20% reduction in anthropogenic emissions (LCx20 in table S1) over NCP + FP could only slightly mitigate ozone pollution (by 2.6% on regional average). Note that the natural emissions (i.e. natural (non-fertilizer) soil NO_x, biogenic NMVOCs) remain constant. Due to the nonlinearity of ozone chemical reactions, such a uniform emission cut even worsens ozone pollution over two northeastern cities of NCP + FP (figure 1(b)). This deterioration is more severe in June and August, especially in September (figure S3). Here, we follow previous studies [2, 22] to analyze the ozone chemical regime using the ratio of hydrogen peroxide (H₂O₂) to nitric acid (HNO₃) concentrations (referred to as H₂O₂/HNO₃ hereafter). In BASE (figure S4), the average H₂O₂/HNO₃ is below 0.2 around Tangshan and Tianjin, indicating a NO_x-saturated regime, in which excess NO_x emissions would suppress ozone formation and could further titrate ozone [23]. This is consistent with those in previous observational and model-based studies [2, 24] and explains ozone deterioration in figure 1(b). The simulated chemical regime is also validated using satellite observations with high spatial consistency (figure S5); more information about this validation is available in S5.4 of supplementary material.

Furthermore, to examine the effects of 'one city, one policy' initiative, we assign city-specific reduction ratios to NCP + FP cities (figure 2(a)) in LCxHIGH and LCxLOW scenarios. In LCxHIGH scenario, the city with the highest anthropogenic emissions of ozone precursors (expressed as NO_x+ NMVOCs) is assigned with the highest reduction ratio, and it is the opposite in LCxLOW, as detailed in supplementary material S1 and table S1. Figures 2(b) and (c) show relative differences in anthropogenic emissions (of NO_x and NMVOCs), H₂O₂/HNO₃, ozone production efficiencies (OPEs; calculated by dividing ozone chemical production rate by NO_x emissions [2]), ozone chemical lifetimes (estimated by dividing ozone mass by ozone chemical loss rate), and MDA8 ozone for all 60 cities in LCxHIGH and LCxLOW compared to those in LCx20, respectively. Note that we use the chemical production and loss rates of O_x (comprising ozone and species with which ozone rapidly cycles [25]) from GEOS-Chem to approximate those of ozone, since 95% of O_x is ozone.

Zoom In Zoom Out Reset image size

As shown in figures 2(b) and (c), city-level changes in MDA8 ozone for LCxHIGH and LCxLOW are essentially the same as those in LCx20. The detailed changes in these three scenarios compared to BASE are shown in figure S6. For cities that are assigned higher reduction ratios than in LCx20, additional mitigation benefits are suppressed via the transitioned chemical regime, which shifts toward NO_x-sensitive situations (increasing H₂O₂/HNO₃) and enhanced ozone formation from residuals (from residue anthropogenic plus natural emissions) and transboundary transported precursors (increasing OPE). Additionally, the chemical lifetime of ozone is extended. Taking Tangshan (an industrial city indicated by dark purple) in LCxHIGH for example, compared to LCx20, its NO_x and NMVOCs are additionally reduced by about 11.5% and 12.0% respectively, which are accompanied with H₂O₂/HNO₃ increased by 6.1%, OPE increased by 10.3% and lifetime of O₃ enhanced by 3.9% (figures S6 and 2(b)). Increased OPE is generally because when anthropogenic NO_x and NMVOCs are additionally reduced, the total ratio of NMVOCs to NO_x (anthropogenic + natural) would increase due to massive biogenic VOCs emissions.

We tracked the changes in ozone chemical reactions (detail information in S2 of supplementary material and table S2). As shown in figure S7, the hydroxyl radical would react more with NMVOCs accelerating ozone formation [26]. The extended ozone lifetime is due to the decreasing chemical loss with NO_x and NMVOCs resulting from the emission reduction (figure S7). Thus, ozone locally formed and transported from other cities could stay in the atmosphere for longer periods and thus compensated for the decline in ozone. Therefore, the MDA8 O₃ change over Tangshan is only 0.1% compared to LCx20 (figure S6). Conversely, cities that are assigned lower reduction show opposite trends.

Overall, while city-level emissions in LCxHIGH and LCxLOW scenarios differ from those in the LCx20 by up to ±10%, the resulting relative differences in MDA8 ozone concentrations remain within ±1.0% for the summer average (figure 2) and for each summer month as well as September (figure S8).

As sensitivity tests, we further employed another air quality model WRF-Chem to simulate ozone pollution over NCP + FP under LCx20 and LCxHIGH scenarios (detailed in supplementary material S3). The WRF-Chem simulations were conducted only for June 2019 to save compute costs, with the same emission set-ups as those in GEOS-Chem simulations. As shown in figure S9, the relative differences in MDA8 ozone between LCx20 and LCxHIGH remain within ±1.5% (figure S9(c)), suggesting the feature is robust among air quality models.

This result means that even if cities had specific emission reduction plans, ozone mitigation benefits beyond a universal emission cut would be suppressed by the nonlinear ozone chemistry and cross-city transboundary transport.

Transboundary ozone attributable to anthropogenic sources over the rest of mainland China substantially contributes to ozone pollution over NCP + FP and is quantified based on the difference between BASE and NLCx100 scenarios. In summertime 2019, averaged over NCP + FP, the transboundary MDA8 ozone is approximately 8.5 ppbv (figure 3). According to tagged ozone simulations (detailed in the supplementary material S2), the transboundary MDA8 ozone consists of 9.4 ppbv produced outside of NCP + FP (direct contribution; figure S10(a)) and about −0.9 ppbv which is due to the transported precursors (indirect contribution; figure S10(b)). The latter implies a suppression effect on the local ozone production process. The transboundary contribution for each summer month and September ranges from 5.3 ppbv to 9.5 ppbv (figure S11). On the summertime average, over cities in southern NCP + FP, the transboundary ozone could reach 23.6 ppbv and account for 33.2% of the total MDA8 ozone. Furthermore, the transboundary anthropogenic ozone (8.5 ppbv) is approximately four times higher than the ozone mitigation benefit of a 20% reduction in local emissions (2.1 ppbv) and is close to half of the ozone mitigation benefit of a 100% cut in local emissions (20 ppbv, figure 5(a)). The minimum city-level average transboundary MDA8 ozone (4.9 ppbv) is higher than the maximum ozone mitigation benefit of a 20% local reduction (3.5 ppbv), as shown in figure S12. Summertime transboundary MDA8 ozone over 2015–2020 ranges from 6.9 ppbv to 9.0 ppbv, with the proportion to total MDA8 ozone from 8.8% to 11.5% (figure S13).

Zoom In Zoom Out Reset image size

Furthermore, we examine the effect of collaborative 20% emission reductions beyond NCP + FP cities (i.e. over the entire country) by contrasting the BASE and ALLx20 simulations (table S1). The ozone benefit is 3.6 ppbv (figure 1(c)), which is 72% more beneficial than the effect of a 20% cut within only NCP + FP (figure 1(b)). Furthermore, in contrast to figure 1(b), collaborative 20% emission reductions mitigate ozone pollution with no city exhibiting deteriorated ozone pollution for summer average (figure 1(c)) and each summer month (figure S3). In September, it could also contribute to reducing the area of ozone degradation (figure S3). Thus, for effectively reducing city-level ozone pollution and avoiding the NO_x reduction trap [23], transboundary ozone must be controlled.

Figure 4 shows average anthropogenic transboundary MDA8 ozone concentrations at different anthropogenic emission levels in NCP + FP cities, including no emission reduction (BASE) and uniform 20% (LCx20) and 100% (LCx100) emission. In the latter two scenarios, anthropogenic emissions outside NCP + FP cities are identical to those in BASE. With local emissions reduced, the transboundary ozone is enhanced, for the regional average, by 68% in the LCx100 compared to the BASE scenario (from 8.5 to 14.3 ppbv). For each summer month (figures S15 and S14), the enhancement percentage to BASE transboundary ozone is 57%, 69%, and 80% respectively, and can reach 94% in September. Spatially, the enhanced transboundary ozone can reach 9.7 ppbv (figure S14(c)) over southern cities. The enhancement is even larger than the BASE transboundary ozone over many cities with its ratio to BASE transboundary ozone larger than 100% (the right column of figure S14). According to tagged ozone results (figure S10), direct and indirect contributions increase by 0.7 ppbv and 5.0 ppbv respectively, with about 86% of enhanced transboundary ozone attributable to the indirect part.

Zoom In Zoom Out Reset image size

The enhancement of transboundary MDA8 ozone is attributable to the nonlinearity in ozone chemical reactions. First, by reducing all the local anthropogenic emissions, the local ozone chemical regime shifts toward a NO_x-sensitive regime, as indicated by the H₂O₂/HNO₃ ratio (represented by solid-colored circles in figure 4) increasing from 0.87 to 3.9. Therefore, residual (e.g. non-fertilizer-related soil emissions [27]) and transboundary transported precursors could form ozone more efficiently, as indicated by the nearly quadrupled OPE (represented by circles with respect to the right y-axis), due to enhanced reactions with NMVOCs species (figure S16). Additionally, the lifetime of ozone on regional average is extended by 71% (represented by the circle size in figure 4), due to the decreasing chemical losses with NO_x and NMVOC species (figure S16). Consequently, ozone produced locally and transported from outside NCP + FP could linger in the atmosphere for longer periods and be transported over greater distances.

A sensitivity test with other reduction ratios (LCxothers in table S1) shows that the chemical enhancement in transboundary MDA8 ozone is ubiquity (figures S17 and S18). With the deepening of local emission reduction, the enhanced transboundary ozone would increase, with the enhanced percentage to BASE transboundary ozone increasing from 1.2% under LCx5, to 1.9% under LCx10, to 4.1% under LCx20, to 57.1% under LCx100.

Note that there are significant differences between the chemically enhanced transboundary MDA8 ozone revealed by this study and the NO_x reduction trap (i.e. NO_x titration effect) [23, 28, 29] in terms of chemical mechanisms. The enhancement in transboundary MDA8 ozone is still significant over cities where with the NO_x-sensitive regime (figure S5). Controlling the transboundary transport would be helpful for avoiding the NO_x reduction trap in city-level ozone mitigation pathways (figure 1(c)).

The historical transboundary enhancement from 2015 through 2020 under LCx100 is always significant (figure S19), with the percentage ranging from 70.0% to 84.6%.

Due to the unintended enhanced transboundary ozone, ozone mitigation benefits of local emission reductions are suppressed. Figure 5(a) shows the modeled decline in MDA8 ozone when all the local anthropogenic emissions are removed and all emissions outside NCP + FP remain constant. This is referred to as the 'actual benefit' of local emission cuts. Figure 5(b) is similar to figure 5(a) but with anthropogenic emissions outside NCP + FP completely removed. This is referred to as the 'ideal benefit' of local emission cuts. The average regional actual and ideal benefits are 20.0 and 25.8 ppbv respectively, indicating that the benefits from local emission reductions are suppressed by as much as 5.8 ppbv. The suppression reaches 9.7 ppbv around southern cities in NCP + FP (figure 5(c)). The proportion of the suppression relative to the ideal benefit is approximately 23% on average and nearly 50% over southern NCP + FP (figure 5(d)). The suppression effect is significant for each month (figure S20).

Zoom In Zoom Out Reset image size

The suppression effects with other local reduction ratios are shown in figures S21 and S22. Although the suppression effect is relatively small with mild reductions (i.e. 0.1 ppbv under LCx5), it still could suppress upto 20% (18.4%–23.7%; right column in figures S21 and S22) of the ideal benefit. In January, representative for wintertime, transboundary transport would enhance the ozone deterioration from 1.2 ppbv (figure S23(b)) to 1.5 ppbv (figure S23(a)). This means that transboundary transport from outside NCP + FP could enhance the ozone deterioration in winter by 41% on the regional average (figure S23 (d)).

This implies that regardless of the season, whether in the early or later stages of emission reduction, transboundary ozone and its chemical enhancement are crucial for effective ozone mitigation.

Overall, this study suggests that local collaborative city-level emission control within NCP + FP might not be ideal for mitigating ozone pollution. Even implementing a range of city-specific emission cuts based on their anthropogenic emission levels, the additional benefits compared to those of uniform emission cuts would be suppressed by cross-city transboundary transport and the nonlinearity of ozone chemical processes. Sole local emission control could be inadequate for mitigating ozone. Since, when local emissions are reduced, transboundary ozone from outside NCP + FP would be chemically enhanced and further compromise the effectiveness of local emission reductions, owing to the transitioned chemical regime and extended lifetime of ozone. The absolute benefits resulting from local emission reduction would be expressed due to the existence of transboundary transport, whether in the early or later stages of emission control. Thus, collaborative emission control must be implemented both within and beyond NCP + FP for more successful ozone mitigation in NCP + FP cities.

China's carbon neutrality pathway is expected to be accompanied by massive changes in nationwide economic [30] and energy-consumption structures [30], and end-of-pipe control technologies [31]. Combined with the increasing emphasis on ozone pollution, this will lead to substantial changes in both the total amount and spatial distribution of emissions [32]. Additionally, climate change will influence the natural emissions [27], chemical formation of ozone [33], and regional transboundary transport pathways. Driven by these multiple factors, transboundary ozone transport and its chemical enhancement will likely significantly change in the future. Therefore, further studies on the temporal evolution of transboundary ozone for ozone polluted cities and city clusters will provide useful information for improving ozone mitigation strategies in a timely manner as a support and supplement for China's carbon neutrality pathway.

According to this study, even for city clusters that have very high local anthropogenic emissions, transboundary transport from other regions, including the interactions with locally emitted pollutants, is important for local ozone mitigation. This finding has global relevance. In the United States, the Ozone Transport Commission [34], which comprises 13 states, aims to mitigate ozone pollution in northeastern and mid-Atlantic cities. In Europe, the Gothenburg Protocol [35] sets emission ceilings for major pollutants in member nations. The extent of success of these collective mitigation efforts depends on how its transboundary pollution interacts with locally released pollution. Cities in developing countries, such as India, face similar issues [36]. In cities that have high emissions but less financial and/or technological capabilities, external aids could be mutually beneficial by reducing the amount of pollution exerted upon and transported from these less affordable places.

This study has several limitations. First, it mainly focuses on summer. Although we have included September and January for sensitivity analysis, systematic exploration in other months would be supportive for month-specific policies for ozone control. Second, although this study only explores the unidirectional transboundary impact exerted upon NCP + FP, the reverse transboundary transport also originates from NCP + FP. Due to different pollution levels, emission amounts, topographical conditions, etc., transboundary impacts could be different for each pair of cities/regions.

To that end, the importance of transboundary ozone and its chemical enhancement in conjunction with local emission control, as found in this study, sheds light on collaborative efforts required for mitigating ozone in China and can be valuable for other regions that face similar challenges of severe ozone pollution [1].

This work is funded by the National Natural Science Foundation of China (Grant 42330709) and supported by High-performance Computing Platform of Peking University. We thank Dr. Yugo Kanaya for providing MAX-DOAS measurements at Cape Hedo and Fukue stations. We thank Dr. Yiwen Hu for providing WRF-Chem simulations.

All data that support the findings of this study are included within the article (and any supplementary files).

The authors declare that they have no conflict of interest.

LC led the study. LC designed the study with input from JTL. HK provided the satellite-based NO_x emission data. LC wrote the manuscript with input from JTL and ST. All authors commented on the manuscript.