National and transboundary contributions to surface ozone concentration across European countries

Abstract

Tropospheric ozone impacts health, climate, and ecosystems. Effective ozone mitigation policies are challenged by limited quantitative understanding of national versus transboundary contributions to surface ozone. This study uses a chemical transport model with a source apportionment algorithm to analyze ozone contributions across Europe from 2015 to 2017 during peak ozone season. We quantify country-level ozone production and imported ozone, distinguishing contributions from 35 European countries, neighboring countries, seas, and hemispheric influences. Results show substantial contributions from outside the 35 European countries, with hemispheric contributions playing a significant role. European contributions are crucial during high ozone episodes, especially from Germany, France, Italy, the UK, Poland, and Spain. Spain, northern Italy, and northwest France are identified as areas where national precursor reductions would be more effective in improving national air quality. Furthermore, 25 of the 35 European countries studied are net importers of cumulative ozone mass, with the Netherlands, Belgium, and the UK acting as major exporters. These findings highlight the need for comprehensive air quality policies and cross-border cooperation.

Introduction

Tropospheric ozone (O₃) stands as a pivotal air pollutant with far-reaching implications for human health¹, the vitality of sensitive vegetation and crops, and ecosystem productivity². Additionally, it plays a substantial role as a greenhouse gas³, contributing in 0.47 W m⁻² to the effective radiative forcing since pre-industrial times⁴. O₃ is a secondary air pollutant formed through nonlinear photochemical reactions in the atmosphere. These reactions involve precursors such as carbon monoxide (CO), methane (CH₄), nitrogen oxides (NO_x), and peroxy radicals generated by the photochemical oxidation of non-methane volatile organic compounds (NMVOCs), which emanate from both natural and anthropogenic sources^3,5. Several factors influence the concentration of O₃ at a given location, including proximity to the source of O₃ precursors, geographical location, meteorological conditions, stratospheric O₃ intrusions, and the transport of O₃ and its precursors from local to hemispheric scales^3,6,7. The meteorological conditions that favor surface O₃ include stagnant air masses, abundant sunlight, high temperatures, low humidity, light winds, and specific geographic features like valleys.

In Europe, despite the continued effort to reduce O₃ precursor emissions in the last decades, surface O₃ concentrations have not followed a proportional decreasing trend⁸ and persistently exceed the European Union standards and the World Health Organization (WHO) air quality guidelines. The European Air Quality Directive (AQD) (Directive 2008/50/EC) sets a target value of 120 µg/m³ for the maximum daily 8 h running mean concentration (MDA8), which cannot be exceeded more than 25 days per calendar year, averaged over 3 years. The recent 2021 WHO’s Air Quality Guidelines (AQGsv2021) sets a target MDA8 of 100 µg/m³ that should not be exceeded more than 3–4 days per calendar year, and 60 µg/m³ as the maximum 6-month running mean of MDA8¹. According to the European Environment Agency (EEA), about 95% of the urban population was exposed to O₃ concentrations above the AQGsv2021 and 12% above the European AQD target value in 2020⁹. O₃ is currently responsible for an estimated 24,000 yearly premature deaths, representing an increase of 40% over 2009⁹. About 93% of the reporting stations show concentrations above the AQGsv2021 for the MDA8 O₃ and the 6-month running mean of the MDA8 is met in only in 3% of the reporting stations¹⁰.

In the free troposphere, the lifetime of O₃ can range from weeks to months¹¹, enabling its long-range transport. This, coupled with the nonlinear nature of O₃ production dependent on the chemical regime, poses a considerable challenge in mitigating O₃ levels. Developing effective O₃ mitigation strategies from local to international scales necessitates a quantitative assessment of the source of O₃ across European countries. Regrettably, there is a lack of measurement-based source apportionment techniques for assessing the diverse origins of O₃¹². In this context, source-receptor methods within chemical transport models (CTMs) emerge as invaluable tools for comprehending O₃ contributions and their sources.

Several model approaches exist to establish the relationship between O₃ concentration and their emission sources¹³. The most commonly employed method is the perturbation method (“brute force”), which compares multiple simulations. These simulations assume emission reductions in specific sectors or regions of interest compared to a reference simulation where emissions remain unchanged. The difference between these simulations, adjusted by the assumed mission reduction, quantifies the contribution of the targeted emission sector or region. While this method allows understanding the sensitivity of O₃ to particular emission reductions, as a source contribution method it has limitations due to the non-linearity inherent in the formation of O₃¹⁴.

A second approach, known as the tagging method, involves tracking pollutants emitted throughout their lifetime, considering the physico-chemical processes they undergo in the atmosphere, including advection, vertical mixing, chemistry, and deposition. This method quantifies the contribution of a sector or region to the total concentration of a pollutant while accounting for nonlinear processes and mass conservation, aspects that provide more realistic estimates compared to the perturbation method. In situations where nonlinear processes are absent, both methods yield equivalent results. However, for pollutants such as O₃, these methods can provide significantly different outcomes in multiple scenarios^15,16. Source apportionment techniques enable the distinction of contributions by sectors or regions, including the differentiation between hemispheric, transboundary, and local sources of pollution. The principal limitation of the tagging method is that the results cannot reflect the emissions changes as in the perturbation method.

In the northern hemisphere, the primary emissions of O₃ precursors are concentrated in mid-latitudes, and large-scale transport of O₃ is largely driven by prevailing westerly winds. These winds facilitate the movement of O₃ from East Asia to North America, from North America to Europe, and from Europe to Central Asia¹¹. Numerous modeling studies have underscored the importance of long-distance transport, including hemispheric scales, in influencing O₃ levels across Europe. In many regions, the contribution from sources outside of Europe is more important than the local contribution to O₃ concentrations. For instance, ref. ¹⁷ conducted an extensive analysis using six air quality models to assess the transboundary contribution to O₃ levels on a continental level in the northern hemisphere. Their findings revealed that 45% to 65% (depending on the model) of the surface O₃ prevailing in Europe originates from intercontinental transport during the summer of 2010. Moreover, Lupasçu and Butler¹⁸ found that, in specific European regions, long-range transport could elucidate up to 45% of the total surface O₃ levels also during the summer of 2010. Romero-Alvarez et al.¹⁹ reported that the hemispheric contribution in regions like the United Kingdom averaged 71% from May to August in 2015. Similarly, ref. ²⁰ showed that the contribution from long-distance transport in Europe during the summer of 2018 averaged 65% in the entire study domain, with approximate values exceeding 80% in the Nordic countries and around 50% in Central Europe. Additionally, through an observation-based analysis, Derwent and Parrish²¹ estimated that ~85% of the annual 8-hourly maximum ozone levels in 2020 in the United Kingdom were attributable to long-range transport. Current estimates of the external contribution to O₃ in Europe show a notable range of variability, likely arising from diverse chemical transport models (CTMs), methodologies for source apportionment, input datasets, and interannual variability, among other factors, used across studies.

Shipping emissions also represent a notable source of O₃ in Europe. Lupasçu and Butler¹⁸ estimated their contribution at ~19% to summer O₃ levels along the Mediterranean coast in 2010, while ref. ¹² reported a 13% contribution during high O₃ episodes in 2012, particularly affecting the Mediterranean coastal areas of the Iberian Peninsula. Using a global model, ref. ²² showed the substantial impact of NOx emissions from shipping on O₃ over the Northern Hemisphere oceans, contributing around 5–10 ppb along the Atlantic coast and 5–20 ppb along the Mediterranean coast on average in 2010. Jonson et al.²³ and Schwarzkopf et al.²⁴ showcased how shipping emissions contribute to O₃ titration near major shipping lanes in the North and Baltic Seas. Jonson et al.²³ additionally reported that shipping contributes to higher O₃ levels along the Atlantic and Mediterranean coasts. Petetin et al.²⁵ conducted a sensitivity analysis examining the impact of various emissions scenarios on O₃ pollution in Spain, finding that a 20% reduction in shipping emissions resulted in a 44% decrease in exceedances of the MDA8 O₃ threshold set by the AQGsv2021. This underscores the crucial role of maritime traffic emissions, particularly along the Mediterranean coast.

While hemispheric transport of O₃ plays a substantial role, it is essential to recognize the key contribution from regional or local photochemical O₃ formation, particularly in cases where concentrations surpass the thresholds defined by the European Union directive. For instance, ref. ¹² demonstrated that during a high O₃ episode in 2012 within the Iberian Peninsula, local contributions ranged from 24 to 54% of the MDA8 levels. Similarly, ref. ¹⁸ conducted a study during the summer of 2010, revealing that the maximum local contribution in a European region, such as the Po Valley, could account for up to 35% of the total O₃ concentrations. In specific areas, such as Southern Italy and Malta, the contribution of O₃ precursor emissions from the rest of Europe could represent up to 53% of the O₃ levels. These findings underscore the complexity of O₃ and highlight the need to consider both regional/local and transboundary factors. The convention for long-range transboundary air pollution (CLRTAP) has been diligently reporting transboundary pollution over Europe, encompassing SO₂, NO₂, particulate matter (PM), and O₃ (as seen in the EMEP annual reports²⁶). Specifically for O₃, recent reports calculate the potential reduction of specific metrics that would result from reduced emissions of NO_x and NMVOC, offering insights into the contributions of different countries. Also, the Copernicus Atmosphere Monitoring Service (CAMS) program of the European Union maintains the CAMS policy²⁷, a valuable resource for regulators. This program provides contribution information from various models, offering a comprehensive view of pollutant sources in different European cities.

In summary, achieving a comprehensive understanding of the relative significance of national and transboundary pollution contributions is crucial for informed policy development. Within the European context, this knowledge has the potential to empower policymakers in distinguishing the roles played by individual countries and in formulating effective strategies to enhance air quality throughout the continent. Previous research in this field has provided valuable insights into the influence of transboundary transport and local emissions on O₃ levels within various European countries and regions. Nonetheless, a substantial quantitative knowledge gap persists regarding the specific proportions of transboundary versus national contributions across European countries. The primary objective of this study is to contribute toward bridging this knowledge gap by employing a comprehensive approach. This study is the first work to track both national and imported O₃ across 35 European countries utilizing a CTM complemented with a source tagging method. The imported O₃ is further categorized by European country, neighboring nations, maritime sources in proximity to the European continent, and hemispheric sources. This investigation is centered on the high-ozone summer season of three consecutive years, spanning from June to August in 2015, 2016, and 2017. In contrast to prior studies, we also account for and examine the influence of meteorological interannual variability on the contributions obtained.

In the Methods section, we describe the experimental setup and the tagging method used in this study. In the Results section, we first evaluate the performance of the model against observational data and provide an overview of the diverse sources of O₃ in all the tagged European countries. Then, we direct our focus towards contributions necessitating international cooperation for reduction. Furthermore, this section delves into the impact of European emissions, the margin available to mitigate O₃ exceedances, and each country’s role in transboundary O₃ pollution. Finally, the last section summarizes our conclusions and outlines the regulatory implications arising from this study.

Results

We analyze the country-level O₃ production relative to the imported O₃ over Europe. We begin with a detailed evaluation of the simulated O₃ over Europe. The source contribution results are introduced with an overview of the mean contributions across the 35 European countries, followed by an analysis of the daily intracountry and interannual country variability of these contributions. Subsequent subsections delve into a more detailed examination of each contribution.

Evaluation of simulated ozone concentrations

The CALIOPE system underwent an extensive evaluation for its O₃ predictions, in accordance with the FAIRMODE recommendations outlined in the Methods section. All assessed monitoring stations achieved an MQI value below 1 for the MDA8 O₃ (mean = 0.25; 75th percentile = 0.27; 25th percentile = 0.21), signifying CALIOPE met the MQO set forth by FAIRMODE for O₃. Notably, the lowest MQI was found in Finland (0.16), while the highest was obtained in Romania (0.44). For a comprehensive overview, Fig. S2 in the Supplement shows a target diagram, including the data from all the stations aggregated by European country. All countries featured MQI values within the green circle, meaning that the MQO was met across the board.

The model showed an NMB of 1.5%, an NME of 15.9%, and an r of 0.72 over the summer of 3 years (Table 1). The mean results for each of the three years of these metrics by monitoring station are shown in Figs. S3, S4, and S5 in the Supplement. The model results meet the criteria of ref. ²⁸ in 80, 94, and 86% of the stations for NMB, NME, and r, respectively. The goal is met in 32, 53, and 36% of the studied stations, for NMB, NME, and r, respectively. A comprehensive breakdown of evaluation results by country and year is available in the Supplementary material Sect. S3 (Tables S3, S4, and S5). Despite the low variability between the three years (Table 1), there is a slightly positive bias in 2016 compared with the other years, mainly due to a higher overestimation in the surroundings of the BENELUX and a lower underestimation over central Europe. In general, our evaluation revealed a tendency toward underestimation in the model’s predictions for Germany, Czechia, Austria, and the north of Italy. Conversely, there is a general overestimation along the coastlines of Spain, Portugal, Belgium, Netherlands, and the United Kingdom. This overestimation might be attributed to the model’s limitations in capturing the intricate halogen chemistry responsible for O₃ depletion, particularly in coastal regions²⁹, and uncertainties in shipping emissions, among other aspects. Despite these deviations, the model exhibited high correlation coefficients across all European stations, with the exception of those situated in the Alps, where complex topography posed additional challenges.

Table 1 Statistical overview of MDA8 O3 concentration over June-July-August 2015, 2016, and 2017 for all rural background stations located below 1000 m above sea level over Europe

Additionally, the model captures well the increase in MDA8 O₃ concentrations between EMEP stations where the contributions from hemispheric transport O₃ is most dominant and those that are strongly impacted by national contributions, reflecting the ability of the model to represent local to regional photochemical O₃ enhancements (Sections S5 and S6 in the supplement). We also found a strong correlation between the observation-based estimates from Derwent and Parrish²¹ and our simulated values across 32 EMEP stations in northwestern Europe (R² = 0.85, see Sect. S6). Such comparison with observational data further substantiates the robustness of our modeling method to disentangle contributions from long-range transport and photochemical production driven by local and regional anthropogenic precursor emissions.

Mean MDA8 ozone contributions by country

During the analyzed summers, surface O₃ concentrations exhibit a pronounced latitudinal gradient across Europe, with levels increasing from north to south³⁰. At the national scale during the summers of 2015, 2016, and 2017, the countries’ mean MDA8 O₃ concentrations range from 60–90 µg/m³ in northern countries to 105–120 µg/m³ in southern countries (Fig. 1). The pie charts in Fig. 1 depict the O₃ relative contribution for each of the 35 labeled European countries (see Table S6 in Sect. S4 in the Supplement for the contribution in absolute values). These contributions, according to the distinctions outlined in the Methods section, are NATIONAL, EUC, SEA, NOEU35, and BCON. Notably, these contributions can vary significantly depending on the O₃ precursor emissions and the country’s geographical location, topography, and climate.

Fig. 1: Mean MDA8 O₃ contributions by country.

Mean modeled MDA8 O₃ concentration and percentage of the O₃ contribution over 35 European countries for summer 2015, 2016, and 2017. The contributions include the national O₃ contribution (NATIONAL, purple), the imported O₃ in each country originated from each of the remaining 34 countries (EUC, yellow), along with the contributions from hemispheric transport (BCON, gray), non-European neighboring countries (NOEU35, red), and maritime sources (SEA, blue).

A key result shown in Fig. 1 is the dominant contribution from long-range transport (BCON) to all receptor regions. BCON contributes strongly to O₃ in all European countries, ranging from 37% in Malta to 88% in Iceland of the total MDA8 O₃, consistent with prior studies^17,20. The second most important contribution comes from the transboundary transport between European countries (EUC), representing between 5% (in Iceland) to 41% (in Liechtenstein). However, the NATIONAL influence becomes as relevant as EUC in high-emitter countries (e.g., France and Spain), and the SEA contribution is dominant in coastal areas in the Mediterranean Sea and the Atlantic Ocean (e.g., Cyprus). In many coastal countries, shipping emissions account for a higher percentage than the NATIONAL influence, such as Malta, Cyprus or Denmark, among others. On the other hand, the NOEU35 contribution is notable only in countries bordering this region, such as Cyprus, Bulgaria, or Greece (20, 12, and 12%, respectively), and outweighs the EUC contribution in Cyprus. In contrast, countries like Iceland, Ireland, the Nordic countries, and the United Kingdom, are mainly influenced by BCON, ranging from 73 to 88% of their total O₃. In these countries, emissions from their local sources (NATIONAL) and their neighbors (EUC) represent a low percentage, below 10% in Iceland and Ireland, and below 19% in the Nordic countries and the United Kingdom, indicating limited scope for O₃ reduction through changes in precursor emissions at national and European levels. Indeed, ref. ¹⁹ reported similar contributions in the United Kingdom, where the majority of O₃ comes from hemispheric transport during May-August 2015. Nevertheless, the European contribution (EUC + NATIONAL) can account for over 44% of the total O₃ on average in some countries, with Italy, France, Germany, and Spain having the most considerable NATIONAL influence (averaging 16%). In contrast, the other countries’ NATIONAL contribution ranges between 3% in Denmark and 12% in Portugal. Small countries, like Montenegro, Luxembourg, and Liechtenstein, contribute less than 3% of the total O₃ with their own emissions.

Figure 2 presents a heatmap depicting the country-to-country contributions to the MDA8 O₃ concentration averaged over the 3 years. Germany, France, Italy, the United Kingdom, Spain, and Poland appear as the primary contributors to the other European countries. Indeed, in many countries, this contribution surpasses their NATIONAL contribution (e.g., France to Belgium, where France contributes 1.6 more than Belgium to itself). This results from the transport of national emissions to neighboring countries, facilitated by strong horizontal pollution transport, particularly in areas with substantial industrial activity. This makes contributions between European countries (EUC) a crucial source. Central European and BENELUX (Belgium, Netherlands, and Luxembourg) countries exhibit the highest EUC contributions, exceeding 38% of the MDA8 O₃. These countries also have a lower impact from BCON and SEA, which represent around 48 to 55% and 3 to 8%, respectively (Fig. 1). The NOEU35 contribution is relatively low in central European countries (below 3%), but reaches 10–12% in the eastern ones. These eastern receptor regions are quite heavily influenced by their neighbors (EUC), accounting for 20 to 30%. On the other hand, the SEA contribution in eastern European countries is relatively low (<5%), except for Greece (8%). The SEA’s influence is more notorious in Mediterranean countries, averaging 10%, and is particularly important in island countries, like Malta and Cyprus (26 and 15%, respectively). The influence of SEA in central European countries is not negligible, contributing 3.5–6% to O₃ concentration in all inland countries (Fig. 1).

Fig. 2: Country-to-country mean contributions of ground-based MDA8 O₃ (in µg/m³) for the studied 35 European countries over the summers of 2015, 2016, and 2017.

Diagonal corresponds to the NATIONAL O₃ contribution.

Previous studies^{11,17,18,19,20,22} consistently underscore the important role of long-range transport on O₃ levels in Europe, aligning with our study. However, the percentage attributed to long-range transport of O₃ varies across studies, where most of them range from 45 to 65%. Notably, Lupasçu and Butler¹⁸ generally report lower attribution to long-range transport and higher attribution to the local and rest of Europe’s contribution compared to our findings. This discrepancy may be attributed to differences in the tagging methodology employed by Lupasçu and Butler¹⁸, where O₃ formation is exclusively attributed to NOx emissions (see Tagging method and tagged regions in the Methods section). Shu et al.³¹ reports a similar trend when they compared different tagging methodologies, with a notable reduction of long-range transport contribution when only NOx emissions are tagged. However, ref. ¹⁹, with the same approach as Lupasçu and Butler¹⁸, in a more regional study in the United Kingdom, found results more aligned with our findings. Derwent and Parrish²¹ also showed results consistent with our findings, showing alignment between results derived from an observational-based analysis and modeled results (See Sect. S6 in the Supplement).

Daily and intracountry variability

We have seen so far the mean contributions from all sources by country. However, at a more regional level, the situation can exhibit considerable variations. The influence of each contribution can vary significantly during specific episodes or in different regions within each country, where photochemical processes and meteorological conditions exert a substantial impact. This can become important to understand the role of each source in each country, especially when it comes to the impact of the NATIONAL and EUC contribution during periods of high O₃ concentrations.

The daily variability, influenced by meteorological fluctuations, combined with intracountry variations for each contribution and country, is illustrated in Fig. 3. This figure comprises five panels, each corresponding to one of the studied source contributions (NATIONAL, EUC, SEA, NOEU35, and BCON). Within each panel, the whisker plots represent the distribution of MDA8 O₃ concentrations across grid cells and days for each country. The countries are arranged within each panel in descending order of their mean contribution. Furthermore, Fig. 3 provides insights into interannual variability at the country level. This is expressed as the difference between the maximum and minimum yearly (3-monthly) averages normalized by the three-yearly average. For a complementary perspective on these findings, please refer to Fig. S6 in the Supplement, where the same figure is presented in terms of absolute values.

Fig. 3: Distribution of model results by source contribution.

Each whisker plot shows the contributions from different sources: a NATIONAL, b EUC, c SEA, d NOEU35, and e BCON, over the simulated period and grid cells in percentage. The plots show the average values (green triangle), the median (yellow line), the lower extreme (p5), the lower quartile (p25), the upper quartile (p75), the upper extreme (p95), and the maximum value (red dot). The black continuous line indicates the difference between the maximum and minimum yearly (or 3-monthly) averages over the 3-year average. Note: the NOEU35 plot (d) uses a different scale due to the magnitude of the values.

As demonstrated by the statistical analysis in Fig. 3, it becomes evident that all source categories across all countries exhibit substantial variability and can become prominent contributors to MDA8 O₃ concentrations. For instance, on certain days and within specific locations, the NATIONAL source can emerge as the predominant O₃ contributor in nearly all countries. Notably, countries such as Italy, France, Germany, and Serbia can witness situations where up to 60% of the total MDA8 O₃ originates from their own emissions (Fig. 3a, depicted as red dots). For instance, the NATIONAL contribution in European countries increases by a factor of 2.7 on average, going from 1.9 in Spain to 3.9 in Luxembourg on the days with higher NATIONAL contribution (above the 95th percentile) compared to the mean overall period of study. This aligns with the findings of Lupasçu and Butler¹⁸, who reported an increase in the NATIONAL contribution for O₃ above the 95th percentile. However, it is important to note that the 75th percentile of the NATIONAL influence remains below 25% in the countries with a higher average contribution and drops below 5% in the countries with a lower average contribution (Fig. 3a).

EUC often emerges as the primary contributor to the total O₃ in numerous countries and can even surpass the influence of BCON in many instances. For example, in the Netherlands, EUC can account for concentrations exceeding 175 µg/m³, constituting 80% of the MDA8 O₃. Even in countries with fewer neighbors, such as Ireland, the EUC contribution remains substantial, reaching more than 58% of the MDA8 O₃ or 90 µg/m³ (see Fig. 3b and Fig. S6b in the Supplement).

Despite their lower average influence, both SEA and NOEU35 contributions can play a role in nearly all countries. The SEA contributions exhibit a fairly similar consistent pattern in several countries, with greater variability observed in the Mediterranean countries where coastal areas are more impacted (see Fig. S6c in the Supplement). Receptor countries may also experience substantial pollution intrusions from the NOEU35 region, making NOEU35 a noteworthy source of O₃ in almost all countries in sporadic situations. In some cases, for example, in Iceland or the United Kingdom, NOEU35 can contribute more than 20% of the MDA8 O₃ (Fig. 3d).

The percentage contribution of BCON values exhibits considerable variability across most countries, ranging from 25% to nearly 100% of the MDA8 O₃ (5th and 95th percentile; Fig. 3e). However, when considering absolute values, the variability remains lower and relatively constant, where the difference between the interquartile range is less than 20 µg/m³ (Fig. S6e in the Supplement) in almost all the countries. This underscores the importance of other contributions, even though BCON controls O₃ background levels, where the lowest values (below the 5th percentile) represent more than 25% of the MDA8 O₃ in the country with less BCON contribution (MT).

The increased significance of the European contribution (NATIONAL + EUC) during episodes of high MDA8 O₃ is remarkable as shown in Fig. 4 and Fig. S7, where the episodic enhancement of these contributions (days with MDA8 O₃ above the 95th percentile) is depicted per country compared to average results. Notably, ref. ¹² also found an increase of NATIONAL O₃ during high O₃ episodes in Spain. While the NATIONAL contribution can increase by around 10% in Germany and Italy, the EUC emerges as a key contributor across most countries during such events. This is particularly evident in the United Kingdom, Ireland, and BENELUX, where EUC experiences enhancements between 22 and 29%. This rise in the European contributions is largely counterbalanced by a marked decrease in percentage by BCON, while SEA and NOEU35 exhibit marginal changes.

Fig. 4: Difference (%) in the MDA8 O₃ source contributions per country between O₃ episodes (days above 95th percentile of MDA8 O₃) and mean values.

On top of the figure, the value of the 95th percentile of MDA8 O₃ in each country is shown. The countries are arranged in descending order based on their average NATIONAL contribution, shown in Fig. 3a.

On the other hand, the interannual variation, as depicted by the continuous line in Fig. 3, is influenced by meteorological differences across the three years and the varying emissions considered each year. This impact is particularly pronounced when receptor regions are distanced from emission sources, making them susceptible to changes in air circulation that affect different source contributions. However, typically, the contribution from these sources remains relatively low. This phenomenon is strikingly evident in the case of NOEU35, which exhibits greater interannual variability compared to all other tagged sources (Fig. 3d). This variability in NOEU35 can be attributed to its inherently low values and the substantial distance from emission sources. Even minor shifts in air mass circulation or temperature patterns can exert a notable influence on this interannual variability. Interestingly, the variability of NOEU35 generally increases as one moves from southeast to northwest countries. A similar pattern is observed in the case of SEA, with inland countries experiencing greater interannual variability (exceeding 16%) compared to coastal countries (with less than 12% variability), except for those bordering the Adriatic Sea. Countries like Denmark and Belgium exhibit remarkably low SEA interannual variability, registering values not exceeding 0.5% (Fig. 3c).

The interannual variability in BCON is predominantly influenced by the general atmospheric circulation, resulting in relatively low fluctuations over the years, all remaining below 15% across all studied countries. Central European countries display higher interannual variability, while Nordic countries and Iceland exhibit variations lower than 3% (Fig. 3e).

Upon examining the European contribution (EUC), it becomes apparent that more remote countries generally show greater interannual variability, due to the effects of changing air mass patterns, affecting them more significantly than countries with nearby emission sources. For example, countries in close proximity to France, Germany, and Italy show lower interannual variability (Fig. 3b).

Finally, the NATIONAL contribution demonstrates interannual variability below 22% in all countries, with the exception of Malta and Liechtenstein, where their relatively small sizes contribute to high interannual variability, with values of 53 and 34%, respectively (Fig. 3a).

Overall, interannual variability typically exhibits a rather consistent and homogeneous pattern across most sources studied, particularly for contributions with the most substantial impact and regions in close proximity to emission sources. For such contributions and regions, variability is likely more influenced by variations in temperature from year to year, affecting photochemical processes on O₃.

Contributions from hemispheric transport, non-European neighboring countries, and maritime sources to local ozone

This section examines the spatial distribution of the contributions originating from sources other than the inland European national sources to the O₃ concentration across all the European countries. This encompasses contributions from BCON, SEA, and NOEU35, which are, to a large extent, beyond the control of individual countries and require international coordination, as exemplified by the CLTRAP agreement since 1979. Figure 5 presents the mean absolute (upper panel) and percentage (bottom panel) of BCON, SEA, and NOEU35. To ease understanding, the results are aggregated by administrative divisions of countries (NUTS), a standard developed and regulated by Eurostat (https://ec.europa.eu/eurostat/web/nuts/background, last access: 20 October 2023). The plot at the model resolution can be found in Sect. S6 of the Supplement.

Fig. 5: Contributions from hemispheric transport, non-European neighboring Countries, and maritime Sources to MDA8 O₃ concentrations in the 35 studied European countries.

The mean contributions of a BCON (hemispheric transport), b SEA (maritime sources), and c NOEU35 (non-European neighboring countries) to MDA8 O₃ concentrations are presented in µg/m³ (top panel) and percentage (bottom panel) for the summers of 2015, 2016, and 2017. Note that each plot uses a different scale.

BCON emerges as the most important external contributor affecting Europe, with SEA and NOEU35 following in importance. BCON exerts a substantial impact in all European regions, ranging from 48 µg/m³ in the less affected region (Bas-Rhin, France) to 77 µg/m³ in the most affected one (Aosta Valley, Italy) (Fig. 5a). The regions where the BCON influence is more pronounced can be attributed to several factors. Firstly, the proximity to the domain boundary results in a considerable intrusion of O₃ from the west, predominantly carried by the westerly winds (see Fig. S8). Over the ocean, O₃ experiences reduced dry deposition rates compared to land³². Consequently, as O₃-rich air masses reach the land, the O₃ contribution from the BCON begins to deposit at an accelerated rate. This, together with titration of O₃ in areas with high NOx emissions, can explain the diminishing influence of BCON while moving eastwards across Europe. For instance, in France, the influence of BCON diminishes as one moves further eastward, while the altitude remains relatively constant (5a). Conversely, extensive pollution from the east of our domain, originated from high emissions in western Asia, exerts a comparatively lower impact on Europe, due to atmospheric circulation patterns and the abrupt topography of Turkey which partly shields Europe to the east (Fig. S8).

Secondly, the high-altitude mountain areas such as the Alps, the Pyrenees, and the Apennines exhibit the highest BCON values in Europe (between 60 and 78 µg/m³ in the Alps; Fig. 5). This is due to their location in the free troposphere where the long-range hemispheric O₃ transport dominates.

Thirdly, a distinctive scenario unfolds in the Iberian Peninsula, where the Spanish plateau exhibits higher values compared to the Pyrenees. The Spanish plateau’s noteworthy feature is its elevated average altitude (660 m), allowing for an augmented contribution of O₃ from upper atmospheric layers due to the strong daily development of the Planetary Boundary Layer (PBL) during summer. The PBL’s considerable height, often exceeding 2000 m, fosters a so-called fumigation scenario, wherein high concentrations of O₃ from upper layers are mixed down to surface levels^7,12.

BCON contributions in percentage terms range from 87% in northern regions to 37% in Malta, reflecting a substantial decline in BCON influence from northwest to southeast (Fig. 5a, bottom panel). Southern European countries are characterized by high temperatures, intense solar radiation, limited precipitation, and stagnant weather conditions, all of which contribute to heightened local O₃ formation⁶. This explains why regions like the Spanish plateau, despite harboring high O₃ concentrations from BCON, exhibit lower percentage values compared to northern regions such as the United Kingdom. Central Europe displays the lowest BCON percentage (40–50%), emphasizing the pivotal role of EUC contributions in these areas, as elaborated upon in the preceding section.

Figure 5b reveals that SEA makes a substantial contribution, particularly in coastal regions, especially those near major maritime routes extending from Cyprus to the Baltic Sea. Along the Mediterranean coast and Portugal, the SEA impact is more pronounced than on the Atlantic coast. This difference can be attributed to the favorable conditions for O₃ formation and accumulation in the Mediterranean, contrasting with the titration effect caused by elevated NOx concentrations at Europe’s Atlantic principal ports (Amsterdam, Hamburg, Rotterdam, and London), as Jonson et al.²³ and Schwarzkopf et al.²⁴ pointed out. In the latter, the SEA O₃ concentrations remain below 14 µg/m³ (Fig. 5b). Lupasçu and Butler¹⁸ found similar results, emphasizing the significance of shipping activities along the Mediterranean coast. The transport of precursors from the sea or the ocean to the more inland regions significantly influences O₃ in many regions (5–10%), with only central and eastern European regions displaying minimal impact (less than 5%). Fig. S8 in the Supplement illustrates the high O₃ concentrations in the Mediterranean Sea and the lower concentrations in the Atlantic Ocean. This discrepancy primarily stems from higher temperatures and air mass stagnation in the Mediterranean Sea, shielded by orographic features. This also accounts for the elevated concentrations observed in islands such as Malta, Cyprus, and Greece.

The contribution of NOEU35 is notably lower in comparison to the other external contributions. Countries within the NOEU35 region and in close proximity to Europe, such as Turkey, Russia, Ukraine, and Belarus, contribute between 5 to 20% of the total O₃ in the neighboring European regions. Cyprus is mainly affected by emissions from Lebanon, Israel, the north of Egypt, and the south of Turkey, representing about 25% of the total O₃ in Cyprus. Concentrations along the Turkish coast and within Middle Eastern regions inside of our domain are around 30 µg/m³ (Fig. S8 in the Supplement), slightly higher than levels found in countries like Germany or France, which range from around 16 to 24 µg/m³. This underscores the relative significance of emissions from Turkey and the Middle East in comparison to Europe’s principal contributors. Southern Spain and Italy can be influenced by emissions from north African countries such as Morocco, Algeria, and Tunisia (about 5 to 10%), although their impact is less pronounced than that of NOEU35 in the eastern regions due to lower emissions, distance to these countries, atmospheric circulation and topography (Atlas mountain).

Contributions from other European countries

European countries are implementing various abatement plans to improve air quality following the general guidelines established at the European Commission level and at a international scale with the Gothenburg Protocol. However, the effectiveness of these national mitigation measures can be limited by the NATIONAL contribution and that of its neighboring European countries (EUC). In this section, we explore the relationship between each country’s national contribution and that of its neighboring European countries.

Figure 6 shows the percentage contribution of the main contributors to European O₃ levels, including Germany, France, Italy, United Kingdom, Poland, and Spain (Fig. 2). Each country’s contribution depicted in the figure encompasses its impact on the overall MDA8 O₃ levels throughout Europe, including its own contribution. The same data were presented at model resolution and absolute values in Fig. S9 of the Supplement. Furthermore, these countries are the main contributors to total NOx and NMVOCS anthropogenic emissions in Europe, highlighting the relationship between anthropogenic emissions and O₃ contribution (see Table S7 in the Supplement). Each of these countries contributes between 10 and 30% to their national O₃ levels, and their contributions are distributed relatively uniformly across their respective territories. Yet, certain regions exhibit a more pronounced NATIONAL impact. The Po Valley, for instance, stands out as the European region with the highest NATIONAL contribution, accounting for more than 30% and reaching concentrations of 40 µg/m³. This region is known for its high atmospheric pressure conditions and strong emissions³³, leading to air mass stagnation and elevated O₃ levels. The Rhone Valley also experiences a notable impact from its own emissions, with contributions up to 21%. Here, pollution tends to be channeled between the French Massif Central and the Alps. In the case of Germany and Poland, their own influences are similarly distributed, with the highest contributions located in central regions. Still, in Germany, there is a higher concentration in the southern part of the country (see Fig. S9 in the Supplement). This can be attributed to the natural barrier posed by the Alps and the accumulation of pollutants. In the United Kingdom and Spain, national emissions are more decentralized, with the highest O₃ levels occurring in the southeast of the island in the case of the United Kingdom. These results are aligned with the findings of ref. ¹⁹ during May-August 2015 in the United Kingdom, which reported an average NATIONAL contribution of 13%.

Fig. 6: Contribution in the percentage of the main countries with the highest MDA8 O₃ levels in Europe, average of summer 2015, 2016, and 2017.

Each country’s contribution depicted in the figure encompasses its impact on the overall MDA8 O₃ levels throughout Europe, including its own contribution.

The impact of major contributor countries on their closest neighbors is substantial. Transboundary pollution is intensified in nations subjected to strong prevailing winds, such as Germany, the United Kingdom, Poland, and the northern regions of France, while it is constrained by geographical barriers in others, notably in Spain and Italy. Among these contributor countries, Germany, France, and the United Kingdom have the most pronounced impact on European regions. The Great European Plain, flanked by the Alps and the Carpathian Mountains, allows these countries to make substantial contributions to areas in northeastern Europe, ranging from 2 to 8% (Fig. 6).

Poland’s influence, primarily shaped by westerly winds, predominantly affects the NOEU35 region. However, northern winds can transport pollution from Poland to the southeastern regions of Europe, contributing 2 to 6% (Fig. 6). Italy and Spain contribute less to central and northern Europe, with Mediterranean countries having a higher impact due to the Azores anticyclone, which separates them from the westerly Atlantic winds. Mountainous barriers, including the Pyrenees, Massif Central, and Alps, further influence this limited contribution. Italy also exhibits a relatively low contribution to southeastern Europe (below 8%), primarily due to the topography of these regions. In the Rhone Valley, pollution is funneled mainly from France and the United Kingdom toward Spain and Italy when prevailing winds blow from the North, resulting in elevated transboundary pollution levels along the coasts of Spain and Italy (see Fig. S9).

National versus European contributions

In this section, we aim to identify both the main regions in Europe where there is more potential to address high O₃ levels through national abatement plans and those that are more dependent on abatement plans in other European countries. Note that such potential could be limited by other factors not addressed in this section, such as whether the best available technologies are already in place, the cost of deploying specific mitigation actions, or whether the achieved reduction would actually contribute to meeting objective targets. To assess this, we have employed an indicator, the ratio of the NATIONAL to the EUC contribution, [NATIONAL]/[EUC]. Regions with ratios exceeding one are dominated by emissions within their own country (Fig. 7a). In Fig. 7b, we present the primary country contributor for each NUTS region, irrespective of the extent of the NATIONAL influence. Our analysis reveals a substantial gradient in ratio values, both latitudinally and longitudinally. These variations can be attributed to prevailing wind patterns and differing conditions favoring O₃ formation across Europe³⁴.

Fig. 7: National versus European contributions.

a Ratio of the average concentrations by nuts between the NATIONAL contribution and the contribution of the other European countries (EUC). b Principal country contributor in each nut, without considering the national contribution.

The Nordic countries exhibit relatively low O₃ production and NOx emissions, resulting in minimal national photochemical O₃ formation, indicated by a ratio of less than 0.5. In the southern regions of the Nordic countries, emissions are somewhat higher (Table S7), but they still receive a substantial contribution from transboundary pollution, leading to ratios that are close to 0.0 (Fig. 7a). The United Kingdom plays a major role in influencing the ratios in the southern Nordic regions (Fig. 7b).

Similar dynamics are observed in the Baltic States, where Poland emerges as the primary influencer. Germany and Poland share comparable scenarios regarding their contribution to O₃ levels. Despite being surrounded by neighboring countries with high O₃ contributions, both countries maintain relatively high national ratios. Germany’s ratio is close to 1, while Poland’s ratio ranges between 0.25 and 0.5 along its borders and approaches 1 in the vicinity of Warsaw. The border areas of both countries have a higher transboundary contribution due to their proximity to other states, resulting in lower ratios of around 0.25 to 0.75 (Fig. 7a). This highlights the importance of the national contribution of both Germany and Poland to O₃ levels in their respective regions. Moving to southeastern Europe, ratio values fall between 0.0 and 0.75 in most regions, with lower values closer to the borders. Countries along the Adriatic Sea experience favorable O₃ formation conditions, but their national emissions remain relatively low, and Italy’s contribution significantly impacts many regions (7b). In some areas, Italy’s contribution surpasses the national influence. Further to the east, countries such as Romania, Bulgaria, and Greece, feature higher ratio values ranging from 0.25 to 1 in the southeastern part of Europe. The orography of these regions provides protection from the primary sources of European O₃. Conversely, Czechia, Austria, Slovakia, Hungary, Switzerland, and BENELUX countries have the lowest ratio values, indicating that their national contributions are less important than the transboundary transport of pollution from other European countries. However, the main O₃ hotspots in these countries still offer opportunities for reducing O₃ levels with national abatement plans (Fig. 7a).

The analysis reveals that Western European regions tend to have higher ratio values, indicating a relatively small influence of cross-border pollution and a higher O₃ formation from national sources. An important aspect is the country size, where larger countries will tend to have larger ratios since O₃ production is more likely to take place within the national boundaries. Among European countries, Spain has the highest national influence on its O₃ levels, with a ratio close to 2 in its central and northern regions (Fig. 7a). Coastal and bordering regions, mainly affected by the pollution from France and Portugal, exhibit lower ratios of around 1 (Fig. 7b). However, the northwest region of Spain, which is close to Portugal, displays remarkably high ratio values despite its lower national contribution, owing to the dominant influence of northern winds. The characteristic poor ventilation conditions and enhanced photochemistry during summer are key factors to explain the high ratio values in Spain^7,12. In contrast, northern Italy has a national contribution greater than any other European region (Fig. 3a), yet its ratios (ranging from 1 to 1.75) are lower than those observed in central Spain. This phenomenon is a consequence of multiple contributing countries. While none of the contributions in northern Italy yield high concentrations independently due to the natural barrier of the Alps, the cumulative effect of all these sources renders transboundary pollution in this area quite substantial. Most regions in France and the United Kingdom show ratios above 1, with a similar pattern (Fig. 7a). The southeast of the United Kingdom and the Rhone Valley regions in France show high NATIONAL influence, which is balanced by substantial transboundary pollution. In contrast, the northern and western regions of the United Kingdom, along with the western part of France, also exhibit high ratios due to the diminished influence of transboundary pollution.

Cumulative imported and exported ozone among European countries

This section analyses the balance between the imported and exported cumulative total mass (Tg) of O₃ at the surface across European countries, independent from individual NATIONAL contributions and contributions from the BCON, NOEU35, and SEA. This provides an overall synthesized view of the cumulative contribution of each European country to the surface O₃ in Europe. The cumulative mass of imported O₃ (expressed in Tg) is averaged over the three studied years and it was calculated by summing up the hourly resolution non-NATIONAL O₃ mass contributions over all first model layer grid cells of a specific country over the summer months. Similarly, the exported mass was quantified by accumulating the O₃ mass attributed to a specific country across all the grid cells over the rest of the countries. So, in both cases, the cumulative O₃ mass is an estimate of the impact of O₃ during its lifetime at the surface until it is removed from the atmosphere through any sink process (e.g., chemistry loss, wet or dry deposition). Henceforth, we use the term import/export as the cumulative mass of O₃ imported/exported to/from a specific country.

As expected, countries with larger extent, such as Germany, France, the United Kingdom, Italy, Poland, and Spain, are more susceptible to both being influenced by neighboring countries (imported mass) and exerting a substantial influence in return (exported mass). All in all, 25 countries act as net importers, while only 10 as net exporters of O₃ (Fig. 8a).

Fig. 8: Cumulative imported and exported ozone among European countries.

a Mean cumulative mass of O₃ imported-exported by country over the summer of 2015, 2016, and 2017. b Normalization of the imported and exported mass by the area of each country.

The major net importers of O₃ across Europe are Greece and Italy (net import balance above 0.5 Tg), each with very distinctive situations. Greece mainly imports O₃ (marginal export contribution) primarily due to its geographical location surrounded by high-emitter countries in the north and concentrated national emissions in the south of the country. On the contrary, Italy, despite being somewhat protected by the orography in the north (as discussed in the National versus European contributions section), imports large amounts of O₃ mainly from France, Germany, Spain, and the United Kingdom (Fig. 2), while still exports large amounts of O₃ towards the Mediterranean neighbors. Italy is the country with the largest import contribution in Europe. The remaining net-importing countries are more balanced between O₃ imports and exports.

On the other hand, the leading net exporters of O₃ in Europe are the United Kingdom, Germany, and France (above 0.5 Tg). While Germany and France are characterized by substantial import and export components, the United Kingdom primarily acts as a net O₃ exporter in Europe (over 1 Tg of exported vs 0.1 Tg of imported O₃ mass). The geographical localization of the United Kingdom (mainly affected by Atlantic Ocean air masses) and being one of the largest O₃ precursor emitters in Europe (see Table S7) explain this unique result. The three countries are characterized by being the major NOx emitters in Europe mainly due to their extent and economic activity.

Conversely, the Netherlands and Belgium are net exporters of O₃, despite their low extent and reduced import/export contribution. Their rather high NOx emissions (see Table S7) and the influence of the United Kingdom may result in a considerable O₃ loss through titration, consequently affecting their O₃ budgets. Furthermore, the Netherlands and Belgium exhibit very low NATIONAL contributions to their overall O₃ levels (see Mean MDA8 ozone contributions by country section), indicating that most of the O₃ influencing their neighboring regions originate from their own O₃ precursor emissions.

Finally, for a fair comparison of the O₃ balance among the 35 European countries, we compute a normalized imported and exported mass using an intrinsic characteristic of each country, its surface area (Fig. 8b). This normalization enables a comparison of how much O₃ a country exports or imports relative to others. Note that normalizing by area only affects the absolute values but not the import-export balance’s sign. Overall, the net budget per unit area remains within a similar range across all countries, ±0.3 Tg/m². Despite this overall consistency, smaller countries like Malta and Liechtenstein show notably unbalanced budgets, importing significantly more than other European countries. Conversely, the most socioeconomically advanced countries present a rather balanced import/export budget of O₃, considering their size, falling within the range of most European countries. Netherlands, Belgium, and the United Kingdom emerge as leading exporters of O₃ according to their size (Fig. 8b). Notably, smaller countries with higher emissions stand out in this ratio. On the other hand, despite their substantial size and potential for emitting biogenic precursors, the Nordic countries show minimal export of O₃ and its precursors.

Conclusions

Our study has estimated the transboundary and national contribution in 35 European countries for O₃ using the CMAQv5.0.2.- ISAM tagging method within the CALIOPE air quality modeling system. We tagged the national O₃ contribution (NATIONAL) for each of the 35 countries, along with the imported O₃ in each country originated from each of the remaining 34 countries (EUC), the region neighboring non-EU countries (NOEU35), the region covering oceans and seas (SEA), and the boundary conditions of our domain (BCON). To enhance the representativeness of the results, we focused the study on the summer of three years (2015, 2016, and 2017), highlighting the main sources of O₃ in Europe and the cross-border contribution between European countries.

First of all, we show that all identified sources of O₃ precursors (NATIONAL, EUC, NOEU35, SEA, and BCON) play a crucial role in specific regions and days across all countries, highlighting the need for regulatory efforts at hemispheric, European, and national levels. However, the impact of the European contribution (NATIONAL + EUC) takes more relevance in high MDA8 O₃ episodes (95th percentile), where the European contribution increases by 13% on average in all the countries and can increase up to 30% in the Netherlands and the United Kingdom.

Differences between weather and emissions during the 3 years studied show relatively low interannual differences in most countries and across all sources. The largest uncertainties come from the most distant sources and have a relatively low effect on the total MDA8 O₃ percentage. Nonetheless, spatial and temporal variability has been found in the different sources in each country, requiring analysis at a more specific administrative level, such as NUTS 2 or 3. Conducting the study at the administrative level provides greater visibility into O₃ issues in each country to decision-makers.

We have identified that emissions outside EU-tagged countries are substantial, where the BCON controls the MDA8 O₃ background (37 to 88%) of all European countries, showing the importance of hemispheric contributions on the entire European continent. The influence of the NOEU35 region can become important on certain specific days, where southeastern Europe is particularly affected by NOEU35’s contribution (10 to 20%). On the other hand, the SEA contribution can become considerable in most of the coastal regions of Europe (up to 25% on average), especially on the Mediterranean coasts. Nevertheless, maritime intrusion towards inland Europe is not negligible (5–10% on average). The study suggests that implementing different NECA (Nitrogen emission control area) at the European level could significantly reduce O₃ in Europe, especially in the Mediterranean coastal regions, aligning with ref. ²⁵.

Looking at the contributions of the European countries, we can determine that Germany, France, Italy, the United Kingdom, Poland, and Spain are the main contributors to European MDA8 O₃. Improving pollution control technologies in those countries would have a potential reduction of MDA8 O₃ concentration in summer in those countries and throughout Europe at once. Indeed, cross-border contributions (EUC) are relevant in all countries, reaching more than 50% on certain days in most European countries. Despite that, in some regions of certain countries, such as Spain, France, the United Kingdom, and Italy, the national contribution may be more important than the EUC contribution. In these regions, changes in emissions at the national level will have a more important impact on total O₃ reduction, and they have more room for action to reduce O₃ by applying policies at a regional and national level (Spain, north of Italy, and northwest of France). On the other hand, most European regions are controlled by emissions not produced in their country, leaving little room for improvement by making national policies. Especially in the regions of central Europe, where the cross-border contribution is very important.

The comparison between the cumulative imported and exported O₃ mass shows how 25 out of the 35 European countries studied are net importers of O₃, and only 10 are net exporters. The United Kingdom, Germany, and France are the countries with the highest net export balance, while Greece and Italy have the highest net import balance. This balance shows the cross-border contribution of each country and the legislative responsibility they could have. Generally, there is a compensation of the O₃ mass with the surface area of the main emitting countries of precursors of O₃. This is primarily due to their larger surface area, leading to comparatively little impact on their neighboring countries. Netherlands, Belgium, and the United Kingdom are the main exporters of O₃, and small countries, such as Malta and Liechtenstein, are receiving much pollution considering their area. These findings highlight the countries that need to be more ambitious in designing O₃ pollution mitigation strategies to achieve more significant reductions in European O₃ levels.

Future work should study not only the contributions of countries among themselves but also the contributions of the different emission sectors. In addition, to evaluate more effectively at the legislative level, the different contributions on days with O₃ exceedance should be studied. On the other hand, this work has not distinguished between biogenic and anthropogenic contributions; future studies should consider the tagging of biogenic contributions by country. This could help us understand the intrinsic uncertainties of the CMAQ-ISAM tagging method, where the O₃ contribution is attributed to the limiting regime. The new version of the CMAQ-ISAM³¹ allows changes in the apportionment schemes. We will be able to obtain different approaches that would help us better understand the contribution of O₃ sources to Europe and the main differences with similar studies that do not use the CMAQ-ISAM.

Additionally, the findings and the results of this study hold significant potential for broader applications. They can serve as valuable inputs for health studies, enabling a deeper understanding of the health impacts associated with specific ozone sources, as demonstrated in ref. ³⁵. By identifying the predominant sources and their contributions to O₃ levels, researchers and policymakers can better assess the health risks to populations in different regions, ultimately leading to more targeted public health interventions and policies. Furthermore, these results can be used to establish relationships between economic data and source apportionment studies. Indeed, in Sect. S9 of the supplement, we show a strong relationship between the mass cumulative O₃ export and the gross domestic product – purchase power parity (GDP–PPP), suggesting a new avenue for research and future investigations despite the complex O₃ formation mechanism and the multiple variables that could be involved.

This work has quantified the source contribution in 35 European countries with a perspective focused on decision-makers and future abatement plans that different European countries could establish. Source apportionment studies point out the main sources of O₃, but further sensitivity analysis has to be done to predict the response to emission reductions. Overall, the study highlights the need for continued EU efforts in emission reduction and cross-border cooperation to effectively address the issue of O₃ pollution in Europe.

Methods

Air quality model

We used the CALIOPE regional air quality modeling system to simulate the surface O₃ concentrations over Europe. A detailed description of CALIOPE is available in previous publications (refs. ^36,37,38; and references therein). CALIOPE is a comprehensive system that integrates several components, including the meteorological model WRF-ARWv3.6³⁹, the emission model HERMESv3⁴⁰, the biogenic emission model MEGANv2.0.4^41,42, and the CTM CMAQv5.0.2⁴³. In this study, CALIOPE was configured to cover Europe at a horizontal resolution of 18 × 18 km² with 38 sigma layers from the ground up to 50 hPa in altitude, where the mid-layer height of the lowest model level is set to 20 m above ground level. Approximately 11 layers characterize the planetary boundary layer (PBL). Meteorological initial and boundary conditions were sourced from the European Center for Medium-range Weather Forecast interim reanalysis (ERA-Interim) at a spatial resolution of 79 × 79 km². The meteorological runs featured a daily 12-h spin-up, followed by a 23-h period for chemical transport simulations. Boundary conditions for reactive gases and aerosols were derived from the Copernicus global analysis (CAMS), with a spatial resolution of 80 × 80 km². We note that the CAMS analysis is notable for its assimilation of data from various satellite sensors, including OMI, SBUV, GOME-2, MLS, OMPS for O₃ assimilation, IASI and MOPITT for CO, and OMI for NO₂. Consequently, our approach ensures that the contribution of hemispheric sources to ozone levels in Europe is substantially constrained by observational data, thereby enhancing the reliability and accuracy of our modeling outcomes. CMAQv5.0.2 was configured with the CB05 gas-phase mechanism, extended with an updated toluene mechanism with active chlorine chemistry (CB05TUCL⁴⁴), and the sixth generation CMAQ aerosol mechanism, which includes the ISORROPIA II thermodynamic equilibrium module (AERO6⁴⁵).

Anthropogenic emissions were processed by HERMESv3, using data from the CAMS-REG-APv4.2 emission inventory developed under the Copernicus Programme⁴⁶, with a resolution of 0.1° × 0.05°. Table S7 in the Supplement shows the NOx and NMVOC anthropogenic emissions by country. Biogenic emissions were computed using MEGANv2.0.4, with updated emissions factors from MEGANv2.1 to estimate VOCs and NOx emissions from vegetation. This estimation considered temperature and solar radiation from the WRF-ARWv3.6 model. For additional CALIOPE configuration details, please refer to the Sect. S1 in the Supplement.

Tagging method and tagged regions

O₃ source apportionment tagging methods employ a tracer species system to track O₃ and its precursors for specific emission categories (such as emission sectors) and/or geographical regions. Different source apportionment schemes have been proposed depending on the attribution perspective. These options include attribution based on the reactant stoichiometry of all the species (1)³¹, full attribution to NOx reactants (2)^{18,19,22,31,47,48}, full attribution to VOCs reactants (3)^{22,31,48,49,50}, attribution to NOx and VOCs species (4)^15,31,51, and attribution to NOx or VOCs reactants based on the ratio of chemical production of H₂O₂/HNO₃, which characterizes the chemical regime (5)^12,52,53. Each of these approaches has its unique characteristics and limitations, with different implications for air quality management and outcomes³¹. For example, option 2 may result in an overestimation of NOx sources, as it attributes O₃ solely to NOx reactants. Conversely, option 3 may predominantly assign O₃ to biogenic emissions due to the significant amount of biogenic VOCs emitted compared to other sectors. Additionally, methods 1, 4, and 5 may have larger hemispheric contributions compared to methods 2 and 3, mainly due to the loss of label of NOx precursors as more reactions occur. In these cases, the high amount of hemispheric sources of O₃ or precursors could lead to a higher attribution to this source, as can be seen in Shu et al.³¹. In this work, we used an approach based on the limiting precursor of O₃ (5). Here, O₃ production is attributed to the limiting precursor responsible for O₃ formation.

We applied CMAQ coupled with the Integrated Source Apportionment Method (CMAQ-ISAMv5.0.2⁵⁴) to quantify the contributions from selected countries or regions. The CMAQ-ISAM method used in this work assumes that the O₃ formation is governed by either a NO_x-limited or a VOC-limited chemical regime. In this framework, at any given time and location all O₃ production is attributed to the precursor that is limiting the reaction. Depending on the regime, either the NO_x or the VOC tracers associated with each of the tagged sources are proportionally attributed to the total O₃. The chemical regime that controls O₃ production is determined using the ratio PH₂O₂/PHNO₃ (ratio between the production rates of hydrogen peroxide and nitric acid), where a ratio below 0.35 designates a VOC-sensitive regime, while a ratio above 0.35 signifies a NOx-sensitive regime⁵⁵. However, there may be uncertainties associated with the relative proportion of NOx and VOCs simulated by the model and the threshold value⁵⁶. Nevertheless, the impact of the specific threshold value adopted in our study is somewhat limited since O₃ production is attributed by country. This approach operates as a mass-balance method, wherein the total O₃ concentration within each grid cell equals the summation of its tagged tracers. Table S1 in the Supplement lists the tagged species involved in O₃ formation within the CB05 extended mechanism⁴⁴.

We configured ISAM to discern O₃ and its precursors NO_x and VOCs, according to their countries and regions of origin across Europe and beyond. As depicted in Fig. 9, our tagging approach encompasses 37 distinct regions: the 35 European countries (EU35), a region grouping neighboring non-EU35 countries in the simulation domain (NOEU35), and a region covering the oceans and seas in the simulation domain (SEA). In addition, we attribute contributions from the CAMS chemical boundary conditions (BCON) and initial conditions (ICON). Notably, the ICON contribution becomes negligible after the model’s first 15 spin-up days of numerical integration. For the purposes of this study, we do not distinguish between natural and anthropogenic emission sources. The tagging strategy outlined above allows the quantification of the national O₃ contribution (NATIONAL) for each of the 35 countries, along with the imported O₃ in each country originated from each of the remaining 34 countries (EUC), NOEU35, SEA, and BCON. Hereinafter, we use the acronyms introduced in Fig. 9 to reference a specific country or tagged region. The results are discussed with respect to the regulatory MDA8 O₃ metric.

Fig. 9: O₃ tagged regions.

Non-European countries (NOEU35), ocean and sea (SEA) and 35 countries (EU35): Albania (AL), Austria (AT), Belgium (BE), Bulgaria (BG), Switzerland (CH), Cyprus (CY), Czechia (CZ), Germany (DE), Denmark (DK), Estonia (EE), Greece (EL), Spain (ES), Finland (FI), France (FR), Croatia (HR), Hungary (HU), Ireland (IE), Iceland (IS), Italy (IT), Liechtenstein (LI), Lithuania (LT), Luxembourg (LU), Latvia (LV), Montenegro (ME), Malta (MT), Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Serbia (RS), Sweden (SE), Slovenia (SI), Slovakia (SK), United Kingdom (UK). The model domain is shown in Fig. S1.

Study period

Our study focuses on the period with the highest photochemical formation of O₃ in Europe, which typically occurs during the warmer months, from June to August. To assess interannual variability with the current availability of consistent emission inventories, we conducted simulations for three consecutive years: 2015, 2016, and 2017. During this three-year study period, Europe experienced warm summers, with average temperature anomalies of +0.6 °C, +0.9 °C, and +0.7 °C, respectively, relative to the 1981–2010 reference period (Copernicus, https://climate.copernicus.eu/climate-bulletins,last access: 22 October 2023). Specifically, in 2015, a heatwave swept across Western Europe at the end of June and the beginning of July, with significant temperature anomalies in central Europe throughout the rest of the summer. The summer of 2016 was notably the hottest of the three years in Europe, particularly in regions such as Russia and around the Black Sea. In contrast, 2017 witnessed predominantly below-average temperatures in the northern regions and above-average temperatures in southern Europe. Over the 3 years, Europe faced numerous episodes of O₃ concentrations surpassing the target value established for the protection of human health, as outlined in the EEA. These exceedances were particularly prevalent in Mediterranean countries and central Europe. According to data from EEA, in 2015, 41% of the reporting stations recorded O₃ concentrations above the target value, while in 2016 and 2017, these percentages were 17 and 20%, respectively (EEA reports^30,57,58). The study of the summers in 2015, 2016, and 2017 holds particular significance due to the observed temperature anomalies during these consecutive warm years and the widespread exceedances of O₃ target values across Europe.

Observations and model evaluation

To evaluate the accuracy of our simulated MDA8 O₃ surface concentrations, we compared them with air quality measurements sourced from the AQ (Air Quality) eReporting database of the EEA⁸. We ensured data quality using the Globally Harmonized Observational Surface in Space and Time (GHOST⁵⁹;). GHOST is a project developed at the Barcelona Supercomputing Center to facilitate the quality-assured comparison between observations and models and their reproducibility. The quality assurance flags used in this process are detailed in Sect. S2 of the Supplement. Our assessment used 455, 460, and 467 O₃ monitoring stations for the summers of 2015, 2016, and 2017, respectively (see Supplement Sect. S3). To maintain data integrity, we applied a minimum hourly data availability threshold of 75% within a day (equivalent to 18 out of 24 h). We used rural background stations located below 1000 m above sea level, aligning with the spatial scales represented by the model resolution. Due to the lack of reporting stations during the study period, we were unable to evaluate the model over five counties, namely Albania, Bulgaria, Iceland, Liechtenstein, and Montenegro.

We employed several complementary methodologies to assess the model’s ability to simulate MDA8 O₃ during the summer period. First, we adopted the benchmarking approach developed by the Forum for Air Quality Modeling (FAIRMODE), which promotes standardized model usage among EU Member States, with emphasis on model applications under the European Air Quality Directives. The evaluation procedure is described in the Guidance Document on Modeling Quality Objectives and Benchmarking⁶⁰. This framework defines a Modeling Quality Indicator (MQI) as a statistical indicator calculated based on model results and measurements to assess the model’s reliability at each station. If the MQI is equal to or less than 1, the Modeling Quality Objective (MQO) is met, indicating that the model error is within an acceptable range compared to observational uncertainty. The calculation of the MQI is detailed in Sect. S3 in the supplementary material.

Second, we employed metrics recommended by ref. ²⁸ for MDA8 O₃, establishing goal and criterion values for specific statistics, including the Normalized Mean Bias (NMB; goal: ±5%; criteria: ±15%), the normalized mean error (NME; goal: 15%; criteria: <25%) and the correlation coefficient (r; goal: >0.75; criteria: >0.50). The goal values represent the statistical benchmark that the best models can attain, while the criteria values outline the expected results for most models.

Additionally, we specifically assessed the model’s performance at stations where the simulated hemispheric contribution is at its peak, as well as at stations where the model exhibits the greatest national influence (Sect. S5). For this assessment, we used the remote and regional background European monitoring and evaluation programme (EMEP) stations⁶¹ within the AQ (air quality) eReporting database of the EEA. This approach allows scrutinizing the model’s ability to capture background conditions and local/regional photochemical enhancements, respectively. This comparison also uses as a reference the Copernicus Programme CAMS global reanalysis (CAMSRA), which combines both model and observation-based information to provide a comprehensive estimation of the atmospheric conditions^62,63.

Finally, we compared our results against the observation-based analysis of ref. ²¹, where the contributions of long-range transport and regional photochemical enhancements (RPE) to O₃ were estimated for EMEP stations in northwestern Europe (see Sect. S6 for details in the Supplement).