Geographic sources of ozone air pollution and mortality burden in Europe

Ground-level ozone (O3) is formed in the troposphere through photochemical reactions involving nitrogen oxides (NOx) and volatile organic compounds (VOCs) from natural and anthropogenic sources. High ambient O3 levels, especially during the warm season, are linked to adverse respiratory outcomes and increased mortality. In Europe, exposure to current O3 levels is a major source of premature mortality, with >95% of the population exposed to levels exceeding WHO air-quality guidelines (daily maximum 8-hour average of 100 µg m−3), exacerbated by rising temperatures, urbanization, and demographic aging. O3 concentrations at a given location are strongly influenced by long-range transport of O3 and its precursors. Despite this, mitigation policies have largely focused on national or regional scales. The study aims to quantify the health impacts of transboundary-transported O3 across Europe, distinguishing national versus imported contributions and their associated mortality burden, focusing on O3 itself due to its longer atmospheric lifetime compared with its precursors.

Prior work demonstrates that O3 health impacts are substantial globally and in Europe, with time-series studies evidencing short-term associations between O3 and mortality. The European Environment Agency (EEA) reports widespread exceedances of WHO O3 guidelines and has produced country-level O3-attributable mortality estimates. Studies in southwestern Europe indicate imported O3 contributes 46–68% of daily surface O3, underscoring transboundary influences. However, a comprehensive Europe-wide assessment attributing O3 pollution and mortality to distinct geographic sources has been lacking. Literature also highlights intercontinental transport of O3 and precursors and the influence of international trade on transboundary pollution-related health burdens.

Design: A continent-wide health impact assessment for warm seasons (May–September) of 2015–2017 across 813 contiguous regions in 35 European countries. Data: Weekly all-cause mortality counts (weeks 18–39) and annual population estimates from Eurostat; modeled O3 concentrations from the CALIOPE air-quality system. Air-quality modeling: CALIOPE (18 km grid over Europe) integrates WRF-ARW v3.6 (meteorology; initial/boundary from ERA-Interim), HERMES v3 (anthropogenic emissions; CAMS-REG-AP v4.2), MEGAN v2.0.4 (biogenic emissions), and CMAQ v5.0.2 (chemical transport; boundary conditions from CAMS global analysis). Source apportionment: Implemented a tagging method within CMAQ to track O3 and its precursors (NOx, VOCs) by geographic source through transport, chemistry, and deposition. O3 formation regime (NOx- vs VOC-limited) determined by H2O2:HNO3 ratio (<0.35 VOC-sensitive; >0.35 NOx-sensitive). Contributions categorized as national, other 34 European countries, other non-EU-35 within domain, ocean/sea (maritime), and outside domain (hemispheric). Period modeled: May–September 2015–2017. Statistical analysis and health impact: Used meta-analytic log-linear exposure–response coefficient β = 0.00018 (95% CI 0.00012–0.00024) per 1 µg m−3 increase in daily maximum 8-hour average O3. Attributable fraction AF(x,d) = 1 − exp(−β × O3(x,d)). Calculated weekly AF averages per grid, then population-weighted AF per region using 1-km population grids (2015). Weekly attributable numbers AN(r,w) = N(r,w) × AF(r,w), where N is regional weekly deaths. Summed over warm season; apportioned AN by O3 source contributions; 95% CIs derived using β bounds. Sensitivity analysis: Assumed a safe threshold at 70 µg m−3 by excluding days below 70 µg m−3 and centering the association at 70 µg m−3; repeated attribution. Model evaluation: Compared modeled O3 to EEA rural background stations (<1,000 m a.s.l.) with >75% hourly data completeness. Performance: Pearson r = 0.66 ± 0.11; normalized mean bias 5.49% ± 9.25; normalized RMSE 20.46% ± 5.7. Followed FAIRMODE MQI; model met MQI ≤ 1 at 100% of stations, satisfying modeling quality objectives.

- Average O3 concentration across countries (2015–2017 warm seasons): 101.9 µg m−3 (range 76.7 in Finland to 130.1 in Malta), with a south-to-north decreasing gradient. - Total O3-attributable deaths (May–September, 2015–2017): 114,447 (95% empirical CI 76,539–152,108), corresponding to 72.0 (95% eCI 48.1–95.6) annual deaths per 1 million inhabitants. - Highest mortality burdens in absolute numbers in Germany, Italy, France, the UK, Spain and Poland; highest mortality rates in southeastern Europe (Bulgaria, Serbia, Croatia, Hungary, Greece, Romania). - Source contributions to O3-attributable mortality: • Imported (non-national) O3 accounted for 88.3% of all O3-attributable deaths (intercountry range 83.0–100%). • Hemispheric sources (outside the study domain) were responsible for 56.7% of total O3-attributable mortality (range 42.5–87.2%). • Other European countries contributed 20.9% (range 5.1–40.0%). • Ocean/sea (maritime) sources contributed 7.2% (range 0–24.1%), notably higher in Malta (24.1%) and Cyprus (14.0%). - Major exporting countries of transboundary O3-related mortality: France (causing 4,003 deaths in other countries during warm seasons 2015–2017) and Germany (3,260). Substantial bilateral impacts include, for example, France’s contributions to Luxembourg (32.3%), Switzerland (29.3%), Belgium (24.4%), Liechtenstein (20.2%), Spain (16.8%) and Germany (16.3%); Germany’s contributions to Luxembourg (24.2%), Czechia (23.3%), Netherlands (21.5%), Denmark (20.3%), Austria (19.9%), Belgium (17.8%) and Poland (17.2%). - National versus imported balance: Southwestern European countries (Spain 53.7%, France 47.1%, Portugal 46.2%) had comparatively larger national contributions and smaller imported:exported ratios. - Sensitivity analysis (70 µg m−3 threshold): O3-attributable deaths reduced by about threefold to 36,523 (95% CI 24,382–48,633), or 23.0 (95% CI 15.3–30.6) annual deaths per 1 million; relative source apportionment changed minimally.

The study directly quantifies the national and imported geographic sources of O3 and links them to mortality across Europe, showing that the vast majority (88.3%) of O3-attributable deaths arise from imported O3, predominantly hemispheric transport (56.7%). This addresses the central question by demonstrating that O3-related health burdens are largely transboundary, rendering purely national mitigation strategies insufficient. The contributions from other European countries (20.9%) and maritime sources in Mediterranean states further emphasize the need for coordinated international actions. Policy implications include: (i) adopting pan-European and global strategies to meet WHO air-quality guidelines; (ii) considering a nitrogen Emission Control Area in the Mediterranean to reduce shipping NOx emissions; and (iii) maintaining strong local actions because local contributions can dominate during high-O3 episodes and to reduce exported O3. Comparisons with EEA estimates indicate higher burdens in this study when no threshold is applied, reflecting current evidence against a definitive safe threshold; when adopting a 70 µg m−3 threshold, estimates fall below EEA values, highlighting methodological sensitivities. Climate warming is expected to intensify O3 formation and some precursor emissions (e.g., biogenic VOCs), potentially increasing health impacts, linking air-quality improvements with climate mitigation. The results underscore the value of source-apportionment modeling for designing effective, multi-scale O3 control policies.

Across Europe, most O3-attributable mortality is driven by imported O3, especially hemispheric contributions, with additional substantial transboundary effects from other European countries and maritime sources. Achieving WHO air-quality guidelines and reducing premature deaths requires coordinated policies beyond national borders, including EU-level and global strategies. Quantitative attribution linking local health impacts to geographic sources should guide mitigation planning. Future research should refine source apportionment by economic sector and activity within each country, evaluate maritime emission control measures in the Mediterranean, and incorporate climate–air quality interactions to inform long-term policies.

- Only acute (short-term) mortality effects were assessed; potential chronic effects remain uncertain. - A single exposure–response function was applied uniformly across countries/regions, not accounting for population-specific variability in vulnerability. - Years of life lost (YLL) were not estimated due to lack of age-specific mortality data and limited generalizability of existing short-term O3–YLL functions (primarily from China). - Economic costs of O3-attributable deaths were not quantified. - Uncertainty from air pollution modeling was not propagated into health impact estimates. - Source-apportionment methods, while avoiding some limitations of sensitivity analyses, have their own uncertainties; alternative schemes could be explored in future work.