Air quality maps of EEA member and cooperating countries for 2023. PM10, PM2.5, O3, NO2, NOx and BaP spatial estimates and their uncertainties. ETC HE Report 2025/5.

ETC HE Report 2025/5 Air quality maps of EEA member and cooperating countries for 2023 Authors: Jan Horálek (CHMI), Leona Vlasáková (CHMI), Markéta Schreiberová (CHMI), Nina Benešová (CHMI), Philipp Schneider (NILU), Pavel Kurfürst (CHMI), Frédéric Tognet (INERIS), Ondřej Vlček (CHMI), Lucie Školoudová (CHMI) PM10, PM2.5, O3, NO2, NOx and BaP spatial estimates and their uncertainties ETC HE Report 2025/5 Cover design: EEA Cover picture: Concentration map of O3 indicator 93.2 percentile of maximum daily 8-hour means, 3-year mean 2021-2023. Units: µg/m3. (Map 3.1 right, this report.) Layout: EEA / ETC HE (CHMI) Publication Date: 21.11.2025 DOI: 10.5281/zenodo.17427294 Version: 1 EEA activity: Human health and the environment Legal notice Preparation of this report has been co-funded by the European Environment Agency as part of a grant with the European Topic Centre on Human Health and the Environment (ETC HE) and expresses the views of the authors. The contents of this publication does not necessarily reflect the position or opinion of the European Commission or other institutions of the European Union. Neither the European Environment Agency nor the European Topic Centre on Human Health and the Environment is liable for any consequences stemming from the reuse of the information contained in this publication. How to cite this report: Horálek, J., Vlasáková, L., Schreiberová, M., Benešová, N., Schneider, P., Kurfürst, P., Tognet, F., Vlček, O., Školoudová, L. (2025). Air quality maps of EEA member and cooperating countries for 2023. PM10, PM2.5, O3, NO2, NOx and BaP spatial estimates and their uncertainties (Eionet Report – ETC HE 2025/5). European Topic Centre on Human Health and the Environment. (https://doi.org/10.5281/zenodo.17427294) The report is available from https://www.eionet.europa.eu/etcs/all-etc-reports and https://zenodo.org/communities/eeaetc/?page=1&size=20. ETC HE coordinator: Stiftelsen NILU, Kjeller, Norway (https://www.nilu.com/) ETC HE consortium partners: Federal Environment Agency/Umweltbundesamt (UBA), Aether Limited, Czech Hydrometeorological Institute (CHMI), Institut National de l’Environnement Industriel et des Risques (INERIS), Swiss Tropical and Public Health Institute (Swiss TPH), Universitat Autònoma de Barcelona (UAB), Vlaamse Instelling voor Technologisch Onderzoek (VITO), 4sfera Innova S.L.U., klarFAKTe.U Copyright notice © European Topic Centre on Human Health and the Environment, 2025 Reproduction is authorized provided the source is acknowledged. [Creative Commons Attribution 4.0 (International)] More information on the European Union is available on the Internet (http://europa.eu). European Topic Centre on Human Health and the Environment (ETC HE) https://www.eionet.europa.eu/etcs/etc-he ETC HE Report 2025/5 3 Contents Contents ........................................................................................................................................ 3 Acknowledgements ....................................................................................................................... 5 Data availability ............................................................................................................................. 6 Summary ....................................................................................................................................... 7 1 Introduction ......................................................................................................................... 10 2 Particulate matter ................................................................................................................ 12 2.1 PM10 annual average .................................................................................................... 13 2.1.1 Concentration map ......................................................................................... 13 2.1.2 Population exposure ....................................................................................... 14 2.2 PM10 – 90.4 percentile of daily means ......................................................................... 18 2.2.1 Concentration map ......................................................................................... 18 2.2.2 Population exposure ....................................................................................... 19 2.3 PM2.5 annual average ................................................................................................... 23 2.3.1 Concentration map ......................................................................................... 23 2.3.2 Population exposure ....................................................................................... 24 3 Ozone ................................................................................................................................... 28 3.1 Ozone – 93.2 percentile of maximum daily 8-hour means .......................................... 28 3.1.1 Concentration map ......................................................................................... 28 3.1.2 Population exposure ....................................................................................... 30 3.2 Ozone – peak season average of maximum daily 8-hour means ................................ 35 3.2.1 Concentration map ......................................................................................... 36 3.2.2 Population exposure ....................................................................................... 37 3.3 Ozone – SOMO35 and SOMO10 .................................................................................. 40 3.3.1 Concentration maps ........................................................................................ 40 3.3.2 Population exposure ....................................................................................... 42 3.4 Ozone – AOT40 vegetation and AOT40 forests ........................................................... 48 3.4.1 Concentration maps ........................................................................................ 48 3.4.2 Vegetation exposure ....................................................................................... 51 3.5 Ozone – Phytotoxic Ozone Dose (PODY) for crops and forest trees ............................ 55 3.5.1 Phytotoxic Ozone Dose maps ......................................................................... 57 4 NO2 and NOx ........................................................................................................................ 62 4.1 NO2 – Annual mean ...................................................................................................... 62 4.1.1 Concentration map ......................................................................................... 63 4.1.2 Population exposure ....................................................................................... 64 4.2 NOx – Annual mean ...................................................................................................... 67 4.2.1 Concentration map ......................................................................................... 67 5 Benzo(a)pyrene.................................................................................................................... 70 5.1 Benzo(a)pyrene – Annual mean ................................................................................... 70 5.1.1 Concentration map ......................................................................................... 70 5.1.2 Population exposure ....................................................................................... 72 6 Accumulated risks ................................................................................................................ 73 7 Exposure trend estimates .................................................................................................... 74 7.1 Human health PM10 and PM2.5 indicators .................................................................... 75 7.2 Human health and vegetation related ozone indicators ............................................. 77 ETC HE Report 2025/5 4 7.3 Human health NO2 indicator ........................................................................................ 80 7.4 Human health BaP indicator ........................................................................................ 81 List of abbreviations .................................................................................................................... 82 References ................................................................................................................................... 84 Annex 1 Methodology ................................................................................................................. 90 A1.1 Mapping methodology ................................................................................................. 90 A1.2 Calculation of population and vegetation exposure .................................................... 92 A1.3 Phytotoxic Ozone Dose above a threshold flux Y (PODy) calculation .......................... 94 A1.4 Methods for uncertainty analysis .............................................................................. 104 Annex 2 Input data .................................................................................................................... 106 A2.1 Air quality monitoring data ....................................................................................... 106 A2.2 Chemical transport modelling outputs ..................................................................... 108 A2.3 Other supplementary data ........................................................................................ 109 Annex 3 Technical details and mapping uncertainties ............................................................. 113 A3.1 PM10 ........................................................................................................................... 113 A3.2 PM2.5 .......................................................................................................................... 117 A3.3 Ozone ........................................................................................................................ 120 A3.4 NO2 and NOx .............................................................................................................. 127 A3.5 BaP ............................................................................................................................. 129 Annex 4 Concentration maps including stations ....................................................................... 131 ETC HE Report 2025/5 5 Acknowledgements The ETC HE task manager was Jan Horálek (CHMI, Czechia). The EEA project manager was Alberto González Ortiz. The external task ETC HE reviewer was Luca Pozzoli (NILU, Norway). The air quality monitoring data for 2023 were extracted from the AQ e-reporting database by María Colina and Jaume Targa (4sfera, Spain). The EMEP modelling data for benzo(a)pyrene for 2023 as prepared under the Meteorological Synthesizing Centre - East (MCS-E) were provided by Oleg Travnikov (EMEP, MCS-E, Slovenia). ETC HE Report 2025/5 6 Data availability The air quality maps presented in this report are available in the GeoTIFF format on the EEA geospatial data catalogue, see below. PM10 maps (annual average and 90.4 percentile of daily means), 2023: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/177d85ca-1664-4357-ad5e7e7d598c5774 (accessed 19 November 2025). PM2.5 map (annual average), 2023: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/57807f24-25dc-4df3-9468dcff8cee5468 (accessed 19 November 2025). O3 maps (93.2 percentile of maximum daily 8-hour means, peak season average of maximum daily 8hour means, SOMO35, SOMO10, AOT40 for vegetation, AOT40 for forests), 2023: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/0fd47f22-c899-4f76-979e53386b67068e (accessed 19 November 2025). NO2 map (annual average), 2023: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/317cf5a5-9909-46b2-bc170d40ff3138eb (accessed 19 November 2025). NOx map (annual average), 2023: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/a2c980b7-5632-4ff3-bf0cb75e627339cd (accessed 19 November 2025). The entire time series of the maps for all pollutants: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/82700fbd-2953-467b-be0a78a520c3a7ef (accessed 19 November 2025). ETC HE Report 2025/5 7 Summary Air quality concentrations maps of the member and cooperating countries of the European Environmental Agency (EEA)( 1 ) have been prepared for the year 2023. The maps are primarily based on air quality data as reported under the Ambient Air Quality Directives (EC, 2004, 2008). The mapping method ('Regression – Interpolation – Merging Mapping') follows a methodology developed earlier (Horálek et al., 2024, and references cited therein); it combines the monitoring data with outputs from a chemical transport model and other supplementary data such as land cover and satellite data. Population exposure Regarding PM10 (i.e. particulate matter with a diameter of 10 µm or less), concentrations and population exposure remained above the European Union (EU) and World Health Organisation (WHO) standards in large parts of Europe in 2023. Approximately 7% of the considered European population was exposed to concentrations above the EU PM10 annual limit value of 40 µg/m3 and 63% of the population was exposed to concentrations exceeding the WHO Air Quality Guideline (AQG) level of 15 µg/m3 in 2023. Furthermore, 16% of the population was estimated to live in areas where the 90.4 percentile of the PM10 daily means was above the EU limit value of 50 µg/m3. When it comes to PM2.5, approximately 0.2% of the considered European population (excluding Türkiye in this case of PM2.5) was exposed to concentrations above the EU PM2.5 limit value of 25 µg/m3 and 2.4% above the EU PM2.5 indicative limit value of 20 µg/m3, respectively. Moreover, 96% of the population was exposed to concentrations above the WHO AQG level of 5 µg/m3. The countries with the highest values of annual averages PM2.5 are located in south-eastern Europe. The population-weighted concentration of the PM10 indicators annual average and 90.4 percentile of daily means for the considered European population is estimated to be 20.4 µg/m3 and 34.5 µg/m3, respectively, in 2023. The populationweighted concentration of the PM2.5 annual average for the considered European population (without Türkiye) is estimated to be 10.4 µg/m3 in 2023. For ozone, it was estimated that 12% of the considered European population lived in areas where concentrations exceeded the ozone target value threshold (that is, the target value for just one year) based on 2023 data, and 10% based on the 3-year target value for 2021-2023. Countries with the highest values of the ozone indicator 93.2 percentile of maximum daily 8-hour means are situated in southern and central Europe. Additionally, 98% of the considered European population lived in areas where the ozone concentration was above the WHO AQG level of 60 µg/m3 (the peak season average of maximum daily 8-hour means). The population-weighted concentration of the O3 indicator 93.2 percentile of maximum daily 8-hour means for the considered European population is estimated to be 108.1 µg/m3 in 2023. Approximately 3% of the considered European population was exposed to NO2 concentrations above the EU annual limit value of 40 µg/m3 in 2023. Except for Türkiye, the majority of population lived well below the limit value in 2023. However, about 67% of the considered European population including Türkiye was exposed to annual average concentrations above the WHO AQG level of 10 µg/m3. The population-weighted concentration of the NO2 annual average for the considered European population is estimated to be 15.5 µg/m3 in 2023. Based on the experimental map of benzo(a)pyrene (BaP), it is estimated that 8.5% of the considered European population live in areas where BaP concentrations are above the threshold 1.0 ng/m3. The highest BaP concentrations are shown in Poland, north-eastern Czechia, west Balkan, eastern Po Valley ( 1 ) The EEA member countries are the 27 Member States of the European Union (EU-27), plus Iceland, Lichtenstein, Norway, Switzerland, and Türkiye. The EEA cooperating countries are Albania, Bosnia and Herzegovina, Kosovo under the UN Security Council Resolution 1244/99, Montenegro, North Macedonia, and Serbia. In addition, three microstates (Andorra, a country which voluntary reports air quality data, Monaco and San Marino) are also presented in this report. ETC HE Report 2025/5 8 in northern Italy and some populated locations in central and south-eastern Europe, Latvia and Finland. The population-weighted concentration of the BaP annual average for the considered European population (without Türkiye) is estimated to be about 0.4 ng/m3 in 2023. Figure S.1 illustrates the population exposure to different air pollutants, for the whole considered European population (excluding Türkiye in the cases of PM2.5 and BaP). Figure S.1 Percentage of the population (%) exposed to different values of PM10 (annual average and 90.4 percentile of daily means), PM2.5 (annual average), O3 (93.2 percentile of maximum daily 8-hour means, SOMO35 and peak season average of maximum daily 8hour means), NO2 (annual average) and BaP (annual average) in 2023 Accumulated risks Out of the total population of 560 million in the mapped area (including Türkiye), 5% (30.1 million) people live in areas where two or three of the most frequently exceeded air quality standards (PM10 daily limit value, O3 target value and NO2 annual limit value) as specified in EC (2008) are exceeded; 0.1% (0.5 million) people live in areas where all these three standards are exceeded. The worst situation in 2023 was observed in Cyprus, Greece, Türkiye, where less than 0.05%, 4% and 0.2% lived in areas where all three standards are exceeded, respectively. Vegetation exposure Based on 2023 data, ozone concentrations (AOT40 for vegetation) are above the EU target value threshold (the TV for just one year) for the protection of vegetation (EU, 2008) in about 12% of the agricultural areas and above the EU long-term objective in 91% of the agricultural areas. Ozone concentrations (AOT40 for forests) are above the critical level for the protection of forests in about 73% of the forested areas. Relating to 5-year mean 2019-2023, 16% of all European agricultural land including Türkiye has been exposed to ozone concentrations above the target value of 18 000 µg/m3·h. Considering the long-term objective of 6 000 µg/m3·h, the total European area (including Türkiye) in excess was 90%. In 2023, POD6 values for wheat above the highest CL (i.e. for protein yield) were observed across a large continuous area of Europe, covering western, central and south-western regions. For potatoes, POD6 values exceeded the CL for tuber yield in large parts of western and central Europe. In contrast, exceedances for tomatoes were limited to a few coastal areas, consistent with previous years. ETC HE Report 2025/5 9 The CL of POD1 for beech was exceeded in almost the entire European mapped area, with the exception of large areas of the Iberian Peninsula and Türkiye and many smaller areas throughout the mapped area. The CL of POD1 for spruce was exceeded almost throughout the entire mapped European area, with the exception of large areas in northern Europe and a larger area in Romania. In a limited number of cases, concentrations of NOx are above the EU critical level. However, since most of these cases happen in urban areas, this is only relevant if there is vegetation in those areas. Changes over time Since 2005, maps for most of the pollutants have been prepared in an overall consistent way (although the mapping methodology has been subject to continuous improvement). This enables an analysis of changes in exposure over time. Apart from minor methodological changes, a major change was introduced for PM10 and PM2.5 since the 2017 maps, taking into account air quality in urban traffic areas. For some pollutants, maps of several years are not available. The evolution of the population-weighted concentrations (as a measure of population exposure), expressed as relative change from 2005, is shown in Figure S.2. For absolute numbers, see Chapter 7. It can be seen that the population-weighted concentrations of PM10, PM2.5 and NO2 show in 2023 the lowest values in the nineteen-year time series. Figure S.2 Population-weighted concentration of PM10, PM2.5 and NO2 (annual mean), and ozone (SOMO35) in 2005-2023, and agricultural-weighted concentration of ozone indicator AOT40 for vegetation in 2005-2023, expressed as relative change (index, 2005 = 100) The PM population-weighted concentrations show a steady decrease of about 0.6 µg/m3 per year for PM10 annual average and 0.5 µg/m3 per year for PM2.5 annual average. For the ozone populationweighted concentration (expressed as SOMO35) no trend is observed for the period 2005-2023, due to the year-to-year variability. Also, no trend is observed for the agricultural-weighted concentration, in terms of AOT40 for vegetation. The NO2 population-weighted concentration (in terms of annual average) shows a decrease of about 0.7 µg/m3 per year over the period 2005-2023. ETC HE Report 2025/5 16 Figure 2.1 Percentage of the population (%) exposed to different PM10 annual averages (µg/m3) at country level, 2023 It is estimated that 63% and 36% of population of the considered mapped area has been exposed to annual average concentrations above the WHO Air Quality Guideline levels of 15 µg/m3 (WHO, 2021a) and the new 2030 EU limit value of 20 µg/m3 (EU, 2024), respectively. The same is true for 56% and 25% of the EU-27 population. Approximately 7% of population of the considered mapped area has been exposed to concentrations above the EU annual limit value (ALV) of 40 µg/m3; the same is the case for 0.02% of the EU-27 population. No population has been exposed to concentrations above the ALV in 36 out of 41 mapped countries. A limited fraction of the population (up to 7%) has been exposed to concentrations above the ALV in Greece, North Macedonia, Bosnia and Herzegovina and Cyprus (in ascending order). 46% of the population has been exposed to concentrations above the ALV in Türkiye. However, as the current mapping methodology tends to underestimate high values (see Annex 3, Section A3.1), the percentage of population exposed to concentrations above the ALV will most likely be underestimated. Additional population exposure above the ALV could therefore be expected in countries like Albania, Bosnia and Herzegovina, Cyprus, Greece, North Macedonia, Serbia and Türkiye where a relatively large fraction (ca. 20-40%) of the population lives in areas with concentration levels 30-40 µg/m3. The population-weighted concentration of the annual average for 2023 for the considered European population (including Türkiye) is estimated to be 20.4 µg/m3 and 16.8 µg/m3 for the EU-27 only. Values decreased by 1.8 and 2.4 µg/m3 compared to the previous 5-year mean 2018-2022, respectively. When assessing the absolute change in individual countries, the steepest decrease was found in Montenegro (6.8 µg/m3), the highest increase was estimated in Cyprus (1 µg/m3). Figure 2.2 shows, for the whole mapped area, the population frequency distribution for exposure classes of 1 µg/m3. The highest population frequency can be seen for classes between 9 and 25 µg/m3. One can see a quite continuous strong decline of population frequency for classes between 25 and 30 µg/m3 and a mild decline for classes beyond 40 µg/m3. ETC HE Report 2025/5 17 Figure 2.2 Population exposure frequency distribution, PM10 annual average, 2023. The WHO AQG level (15 µg/m3) is marked by the green line, the 2030 EU limit value (20 µg/m3) is marked by the yellow line, the current EU LV (40 µg/m3) is marked by the red line Note: Apart from the population distribution shown in the graph, it was estimated that 0.14% of population lived in areas with PM10 annual average concentrations between 75 and 230 µg/m3. Figure 2.3 shows for individual countries the PM10 annual average concentration to which the population in that country was exposed in 2023. It can be seen that, as already pointed out, the countries with the highest mean exposures to PM10 annual average are located in the central and south-eastern parts of Europe. Figure 2.3 PM10 annual average concentrations to which the population per country was exposed in 2023. The WHO AQG level (15 µg/m3) is marked by the green line, the 2030 EU limit value (20 µg/m3) is marked by the yellow line, the current EU limit value (40 µg/m3) is marked by the red line. Note: For each country, the box plot shows the concentration to which a percentage of the population was exposed: 50% in the case of the black marker, 25% and 75% in the cases of the box’s edges, 2% and 98% in the cases of the whiskers’ edges. ETC HE Report 2025/5 18 2.2 PM10 – 90.4 percentile of daily means The Ambient Air Quality Directive (EC, 2008) describes the PM10 daily limit value (DLV) as “a daily average of 50 µg/m3 not to be exceeded more than 35 times a calendar year”. This requirement can be evaluated by the indicator 36th highest daily mean, which is in principle equivalent to the indicator 90.4 percentile of daily mean. However, for measurement data these two indicators are equivalent only if no data is missing, which is in general not the case. As shown in de Leeuw (2012), the additional uncertainty related to incomplete time series is substantially smaller when using percentile values instead of the x-th highest value. Furthermore, the Air Quality Directive requires the use of the 90.4 percentile when random measurements are used to assess the requirements of the PM10 DLV. As in the previous reports since the maps for 2014 (Horálek et al., 2017a), the PM10 daily means are expressed as the 90.4 percentile instead of the formerly used 36th highest daily mean. The analysis does not include comparisons against the WHO AQG levels and the revised LV2030 for daily means, due to the different pollutant aggregations (i.e. corresponding 99 and 95.1 percentile of daily means, respectively). 2.2.1 Concentration map Map 2.3 presents the final combined map, where red and purple marked areas indicate values of the 90.4 percentile of daily means above 50 µg/m3 (i.e. concentrations above the DLV of 50 µg/m3 on more than 35 measurement days). The similar mapping procedure as in the case of the annual average is used. The mapping details and the uncertainty analysis are presented in Annex 3. Large areas with concentrations above the DLV are observed in northern Italy (i.e. the Po Valley) and in southern, southeastern and eastern parts of Türkiye. Urban and surrounding areas with concentrations above the DLV are observed in Balkan countries and Cyprus. In general, the south-eastern, the south and the central parts of Europe appear with higher concentrations and population-weighted concentrations than the western and the northern parts. Map 2.3 Concentration map of PM10 indicator 90.4 percentile of daily means, 2023 The relative mean uncertainty (relative RMSE or RRMSE) of the final combined map of the 90.4 percentile of PM10 daily means is 26% for rural and 30% for urban background areas including Turkish ETC HE Report 2025/5 19 stations. The mean uncertainty for the map without Türkiye is 20% for both rural areas and 22% for urban background areas (Annex 3, Section A3.1). Thus, the mapping uncertainty of this is at the similar level as in the case of the PM10 annual average. Map 2.4 presents the difference between 2023 and the 5-year mean 2018-2022 for the 90.4 percentile of daily means for PM10. Orange to red areas show an increase of PM10 concentration in 2023, while blue areas show a decrease. Map 2.4 Difference in concentrations between 2023 and the 5-year mean 2018-2022 for PM10 90.4 percentile The situation is similar to that of the annual average value of PM10. The highest increases are observed in Türkiye, parts of southern and south-eastern Europe with the highest increases in Cyprus, Greece and parts of Italy (Sardinia, Sicilia and south of Italy). On the other hand, there are decreases in the most of remaining countries, mainly central Europe with the most notable decrease in Poland, parts of some Balkan countries with the most notable decrease in western Romania, and in Po Valley in Italy. 2.2.2 Population exposure Table 2.2 and Figure 2.4 give the population frequency distribution for a limited number of exposure classes to PM10 indicator 90.4 percentile of daily means for individual countries, for five European regions, for EU-27, and for the total considered area. More detailed Table 2.2 also presents the 2023 population-weighted concentration and the difference in the population-weighted concentrations between 2023 and the 5-year mean 2018-2022 according to Equation A1.7. In 2023 about 16% of the considered European population (including Türkiye) and 3% of the EU-27 population are estimated to live in areas with PM10 concentrations above the EU daily LV of 50 µg/m3. ETC HE Report 2025/5 20 Table 2.2 Population exposure and population-weighted concentrations, PM10 indicator 90.4 percentile of daily means, 2023 Country ISO Population (inhbs·1000) PM10 - perc90.4, exposed population (%) Population-weighted concentration < 20 20-30 30-40 40-50 50-75 > 75 2023 5-year mean Diff. Albania AL 2 762 0.2 11.6 12.4 18.4 57.5 48.7 48.2 0.6 Andorra AD 80 14.5 85.4 0.1 26.0 31.3 -5.4 Austria AT 9 105 20.8 74.8 4.4 23.2 27.6 -4.4 Belgium BE 11 743 2.2 88.5 9.3 27.3 32.4 -5.2 Bosnia and Herzegovina BA 3 194 0.0 12.4 20.9 16.6 36.4 13.6 51.3 64.0 -12.7 Bulgaria BG 6 448 0.7 17.8 42.2 36.4 2.9 37.2 45.3 -8.2 Croatia HR 3 851 1.3 38.6 34.3 24.2 1.6 33.1 40.9 -7.8 Cyprus CY 1 419 2.4 14.0 80.2 3.5 55.6 52.8 2.8 Czechia CZ 10 828 8.2 80.6 11.2 25.8 34.8 -9.0 Denmark (incl. Faroe Islands) DK 5 987 79.4 20.6 19.3 25.0 -5.7 Estonia EE 1 366 59.0 38.6 2.5 18.4 20.5 -2.1 Finland FI 5 564 87.9 10.9 1.2 15.3 17.0 -1.7 France (metropolitan) FR 66 017 13.6 79.8 6.5 0.0 24.4 26.9 -2.5 Germany DE 83 119 39.8 60.2 20.7 25.6 -5.0 Greece GR 10 414 0.0 6.0 29.5 41.1 23.3 43.7 43.1 0.6 Hungary HU 9 600 0.0 39.6 59.8 0.7 31.1 39.7 -8.6 Iceland IS 388 72.7 27.3 18.0 16.7 1.3 Ireland IE 5 271 81.4 18.6 18.0 20.6 -2.6 Italy IT 58 997 0.8 11.7 47.0 27.8 12.7 39.6 42.1 -2.5 Latvia LV 1 883 27.8 45.0 27.2 24.6 30.2 -5.6 Liechtenstein LI 40 66.9 33.1 20.0 21.7 -1.6 Lithuania LT 2 857 22.1 55.1 22.8 26.1 32.1 -6.0 Luxembourg LU 661 18.0 82.0 21.4 25.4 -4.0 Malta MT 542 2.6 97.4 42.5 42.9 -0.4 Monaco MC 39 80.5 19.5 27.4 30.7 -3.4 Montenegro ME 617 0.6 28.0 37.4 32.3 1.7 36.0 50.7 -14.8 Netherlands NL 17 811 1.9 98.1 0.0 23.6 28.8 -5.1 North Macedonia MK 1 830 0.2 5.0 7.6 22.9 59.4 4.9 55.9 66.0 -10.1 Norway NO 5 489 66.2 30.8 2.9 0.1 17.2 18.6 -1.4 Poland PL 36 754 0.4 34.5 53.9 11.2 32.7 45.3 -12.6 Portugal (excl. Azores, Madeira) PT 10 022 9.3 58.8 30.7 1.2 27.6 29.3 -1.7 Romania RO 19 055 4.0 41.6 41.8 12.5 0.0 31.6 38.5 -6.9 San Marino SM 34 15.8 84.2 32.6 37.1 -4.5 Serbia (incl. Kosovo) RS 8 350 0.2 10.0 33.7 26.1 30.0 0.0 43.1 57.3 -14.2 Slovakia SK 5 429 1.6 69.1 29.3 28.7 37.7 -9.0 Slovenia SI 2 117 3.0 56.7 39.4 0.9 28.8 33.8 -5.0 Spain (excl. Canarias) ES 45 872 4.8 40.2 53.5 1.5 29.9 31.8 -1.8 Sweden SE 10 522 83.3 14.4 2.2 0.1 0.0 16.6 19.2 -2.6 Switzerland CH 8 815 27.6 70.2 1.9 0.3 21.8 24.2 -2.3 Türkiye TR 85 280 0.0 0.9 4.8 7.9 60.0 26.4 68.6 69.1 -0.5 Total 560 170 14.3 40.7 21.1 7.9 12.1 4.0 34.5 38.4 -3.9 55.0 16.1 EU-27 443 198 16.6 48.8 24.5 7.5 2.6 0.0 28.1 32.9 -4.7 65.4 2.6 Northern Europe 34 055 71.1 23.9 5.0 0.0 0.0 18.3 21.7 -3.4 Western Europe 86 334 13.6 82.8 3.6 24.0 27.8 -3.8 Central Europe 165 805 23.0 55.9 18.4 2.7 24.9 32.2 -7.3 Southern Europe 142 589 3.9 29.6 42.8 16.0 7.8 0.0 34.7 36.4 -1.7 South-Eastern Europe 131 386 0.7 10.4 16.1 12.7 43.1 17.0 57.2 60.2 -3.0 Kosovo XK 1 709 0.1 5.3 56.2 34.6 3.7 0.0 39.0 54.8 -15.8 Serbia (excl. Kosovo) RS 6 641 0.2 11.2 27.9 23.9 36.7 44.2 57.9 -13.7 Note: The percentage value "0.0" indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. 5-year mean, i.e. 5-year mean 2018-2022. Diff., i.e. difference concentrations between 2023 and 5-year mean 2018-2022. ETC HE Report 2025/5 21 No population has been exposed to concentrations above the DLV in 27 out of 41 mapped countries. A limited fraction of the population (up to 4%) was exposed to concentrations above the DLV in Sweden, Romania, Croatia, Montenegro, Bulgaria and Kosovo (in ascending order). More than 10% but less than 50% of the population has been exposed to concentrations above the DLV in Italy, Greece and Serbia. More than half of the population has been estimated to be exposed to concentrations above the DLV in Bosnia and Herzegovina (50%), Albania (58%), North Macedonia (64%), Cyprus (83%) and Türkiye (86%). Figure 2.4 Percentage of the population (%) exposed to different values of PM10 indicator 90.4 percentile of daily means, 2023 The European-wide population-weighted concentration of the 90.4 percentile of PM10 daily means is estimated for 2023 at 34.5 µg/m3 for the total mapped area and 28.1 µg/m3 for the EU-27. The population-weighted concentration of this PM10 indicator decreased by 3.9 µg/m3 for the considered European population and by 4.7 µg/m3 for EU-27 compared to the previous 5-year mean 2018-2022. When assessing the absolute change in individual countries, the steepest decrease was found in Kosovo (15.8 µg/m3), the highest increase was estimated in Cyprus (2.8 µg/m3). Figure 2.5 shows, for the whole mapped area, the population frequency distribution for exposure classes of 2 µg/m3. One can see the highest population frequency for classes between 18 and 50 µg/m3, then a continuous mild decline of population frequency for classes above 50 µg/m3. ETC HE Report 2025/5 22 Figure 2.5 Population exposure frequency distribution, PM10 indicator 90.4 percentile of daily means, 2023. The EU daily limit value (50 µg/m3) is marked by the red line. Note: Apart from the population distribution shown in the graph, it was estimated that 0.13% of population lived in areas with values of PM10 indicator 90.4 percentile of daily means between 135 and 690 µg/m3. Figure 2.6 shows for individual countries the P90.4 of the PM10 daily concentrations to which the population was exposed in 2023. It can be seen that the countries with the highest values of PM10 indicator 90.4 percentile of daily means are located mainly in south-eastern parts of Europe. Figure 2.6 PM10 expressed as indicator 90.4 percentile of daily means to which the population was exposed in 2023, per country. The EU daily limit value (50 µg/m3) is marked by red line Note: For each country, the box plot shows the concentration to which a percentage of the population was exposed: 50% in the case of the black marker, 25% and 75% in the cases of the box’s edges, 2% and 98% in the cases of the whiskers’ edges. As in previous years, the daily LV was more widely exceeded than the annual LV in 2023. ETC HE Report 2025/5 23 2.3 PM2.5 annual average 2.3.1 Concentration map Map 2.5 presents the final combined map for the 2023 PM2.5 annual average. Light green areas show concentrations above WHO AQG of 5 µg/m3, while yellow areas show concentrations above the LV2030 of 10 µg/m3. Red areas show concentrations above the indicative LV of 20 µg/m3 defined as Stage 2 (ILV). The dark red areas show concentrations above the annual limit value (ALV) of 25 µg/m3. Due to the lack of enough rural PM2.5 stations in Türkiye, no proper interpolation results could be estimated for this country in a rural map. Therefore, the estimated PM2.5 values for Türkiye are not presented in the final map. According to Map 2.5, the areas with the highest PM2.5 concentrations appear to be the Po Valley in northern Italy, in the Krakow – Katowice (Poland) – Ostrava (Czechia) industrial region and areas of Bosnia and Herzegovina, Serbia, North Macedonia, Albania, Romania, Bulgaria and Greece. Concentrations above the ALV appear in urban areas in Bosnia and Herzegovina and in North Macedonia. Several other cities and areas in south-eastern and central Europe also show elevated PM2.5 annual average concentrations. Like in the case of PM10, the central and the south and southeastern parts of Europe show higher concentrations than the western and the northern parts. Similarly to the PM10, the final map in 1 km resolution is representative for the rural and the urban background areas, but not for the urban traffic areas (which are smoothed in the 1 km resolution). In order to provide more complete information of the air quality across Europe, the final combined map including the measurement data at stations is presented in Map A4.3 of Annex 4. Map 2.5 Concentration map of PM2.5 annual average, 2023 The relative mean uncertainty of the 2023 map of PM2.5 annual average is 27% for rural and 26% for urban background areas and it is determined exclusively on the actual PM2.5 measurement data points, i.e. not on the pseudo stations (Annex 3, Section A3.2). Similarly as in the case of PM10, this uncertainty ETC HE Report 2025/5 24 is satisfactory, compared to the data quality objective for models of PM2.5 annual average (i.e. 50%) as set in the Air Quality Directive (EC, 2008). Map 2.6 presents the difference between 2023 and the 5-year mean 2018-2022 for annual average for PM2.5. Orange to red areas show an increase of PM2.5 concentration in 2023, while blue areas show a decrease. At the annual average PM2.5 difference map the highest increases are in parts of southern Europe with the highest increases in Greece, Bulgaria and parts of Bosnia and Herzegovina, as well as in the north of Italy and parts of northern Europe (Norway and Sweden). Moderate increases are also observed in southern Italy and parts of southern Spain. On the other hand, there are decreases in most of the remaining countries, mainly in central Europe (especially in Poland, Hungary, Slovakia), extending into Czechia, Germany, the Benelux countries and Austria, and in parts of some countries in south-eastern and southern Europe (northern Spain, northern Italy, Serbia and Romania). Map 2.6 Difference in concentrations between 2023 and the 5-year mean 2018-2022 for PM2.5 annual average 2.3.2 Population exposure Table 2.3 and Figure 2.7 give the population frequency distribution for a limited number of exposure classes to PM2.5 concentrations calculated on a grid of 1 km resolution. Table 2.3 also presents the population-weighted concentration for individual countries, large regions, EU-27 and for the total mapping area. About 96% and 42% of the considered European population (excluding Türkiye), has been exposed to annual average concentrations above the WHO Air Quality Guideline level of 5 µg/m3 (WHO, 2021a) and the new 2030 EU limit value of 10 µg/m3 (EU, 2024), respectively. The same is true for 97% and 41% of the EU-27 population. The total considered and the EU-27 population exposed to concentrations above the EU ALV of 25 µg/m3 has been 0.2% and none, respectively. ETC HE Report 2025/5 25 Table 2.3 Population exposure and population-weighted concentration, PM2.5 annual average 2023 Country ISO Population (inhbs·1000) PM2.5 – annual average, exposed population (%) Population-weighted concentration < 5 5-10 10-15 15-20 20-25 > 25 2023 5-year mean Diff. Albania AL 2 762 0.0 4.1 27.7 45.5 22.7 16.8 17.4 -0.6 Andorra AD 80 23.9 76.1 5.8 8.7 -2.9 Austria AT 9 105 1.3 78.9 19.8 9.1 10.9 -1.8 Belgium BE 11 743 0.3 85.8 13.9 8.8 10.8 -2.0 Bosnia and Herzegovina BA 3 194 4.6 21.2 25.3 33.4 15.5 19.3 23.6 -4.3 Bulgaria BG 6 448 0.0 3.9 54.1 41.7 0.3 14.6 17.2 -2.7 Croatia HR 3 851 0.0 22.2 39.5 36.6 1.7 13.2 15.4 -2.2 Cyprus CY 1 419 2.4 48.0 47.1 2.5 14.7 14.4 0.3 Czechia CZ 10 828 0.0 31.4 63.2 5.4 10.9 14.3 -3.3 Denmark (incl. Faroe Islands) DK 5 987 2.4 97.6 6.6 8.6 -2.0 Estonia EE 1 366 42.9 57.1 5.2 5.9 -0.6 Finland FI 5 564 80.0 20.0 4.3 5.0 -0.7 France (metropolitan) FR 66 017 0.7 89.5 9.8 8.5 9.5 -1.1 Germany DE 83 119 0.1 98.9 1.0 7.9 10.0 -2.1 Greece GR 10 414 2.5 33.0 52.9 11.5 16.5 16.0 0.5 Hungary HU 9 600 2.4 94.6 3.0 12.4 15.1 -2.7 Iceland IS 388 100.0 4.1 4.1 -0.1 Ireland IE 5 271 20.5 79.5 6.2 7.3 -1.1 Italy IT 58 997 0.2 15.4 49.7 26.2 8.5 14.0 14.7 -0.8 Latvia LV 1 883 1.7 82.0 16.3 8.0 10.2 -2.1 Liechtenstein LI 40 0.6 99.4 7.8 8.3 -0.5 Lithuania LT 2 857 91.1 8.9 8.3 11.2 -2.9 Luxembourg LU 661 0.5 99.5 7.2 8.0 -0.9 Malta MT 542 3.5 96.5 10.9 11.4 -0.5 Monaco MC 39 100.0 8.7 11.1 -2.4 Montenegro ME 617 0.0 9.7 60.3 30.0 13.4 17.9 -4.5 Netherlands NL 17 811 99.9 0.1 8.2 10.2 -1.9 North Macedonia MK 1 830 2.2 12.9 41.9 26.3 16.7 19.5 23.2 -3.7 Norway NO 5 489 46.8 53.2 5.0 5.5 -0.5 Poland PL 36 754 6.8 59.5 33.3 0.4 14.0 17.9 -3.9 Portugal (excl. Azores, Madeira) PT 10 022 7.2 79.8 13.0 7.7 8.3 -0.6 Romania RO 19 055 0.0 8.2 68.5 23.2 0.0 13.2 15.3 -2.1 San Marino SM 34 12.1 87.9 12.5 12.9 -0.5 Serbia (incl. Kosovo) RS 8 350 0.8 28.3 42.8 28.0 0.0 17.5 21.5 -4.1 Slovakia SK 5 429 0.0 6.5 86.7 6.8 12.3 15.3 -2.9 Slovenia SI 2 117 0.0 20.5 63.9 15.6 12.1 13.4 -1.4 Spain (excl. Canarias) ES 45 872 2.6 54.5 42.9 9.5 10.2 -0.7 Sweden SE 10 522 60.3 39.7 4.7 5.5 -0.7 Switzerland CH 8 815 3.3 93.9 2.3 0.5 7.7 8.7 -1.0 Total (without Türkiye) 474 890 3.8 54.1 28.4 11.1 2.4 0.2 10.4 12.2 -1.7 57.9 2.6 EU-27 443 198 3.4 55.5 29.4 10.3 1.5 10.3 12.0 -1.7 58.9 1.5 Northern Europe 34 055 42.8 55.5 1.7 5.5 6.8 -1.2 Western Europe 86 334 1.5 90.5 8.0 8.3 9.8 -1.4 Central Europe 165 805 0.3 62.4 28.6 8.6 0.1 10.0 12.7 -2.6 Southern Europe 142 589 1.5 38.8 39.6 15.5 4.5 11.7 12.3 -0.6 South-Eastern Europe without Türkiye 46 106 0.0 6.7 48.4 32.9 10.1 1.8 15.1 17.8 -2.7 Kosovo XK 1 709 0.7 27.3 68.6 3.3 0.1 16.4 20.1 -3.8 Serbia (excl. Kosovo) RS 6 641 0.9 28.6 36.2 34.4 17.8 21.9 -4.1 Note: The percentage value "0.0" indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. 5-year mean, i.e. 5-year mean 2018-2022. Diff., i.e. difference concentrations between 2023 and 5-year mean 2018-2022. ETC HE Report 2025/5 32 Table 3.2 Population exposure and population-weighted concentrations, ozone indicator 93.2 percentile of maximum daily 8-hour means, 2021-2023 (3-year mean) Country ISO Population (inhbs·1000) Ozone — perc93.2, exposed population (%) Population-weighted concentration < 90 90-100 100-110 110-120 120-140 > 140 2021-2023 Albania AL 2 762 4.4 26.9 67.5 1.2 111.8 Andorra AD 80 72.3 12.2 13.6 1.9 98.5 Austria AT 9 105 0.0 10.5 88.4 1.1 114.4 Belgium BE 11 743 6.2 89.2 4.6 105.3 Bosnia and Herzegovina BA 3 194 1.6 75.0 23.4 117.6 Bulgaria BG 6 448 36.2 47.1 13.1 3.6 0.0 93.9 Croatia HR 3 851 10.4 75.0 14.6 114.6 Cyprus CY 1 419 4.4 45.3 40.5 9.8 109.9 Czechia CZ 10 828 18.9 81.0 0.0 112.9 Denmark (incl. Faroe Islands) DK 5 987 16.6 79.1 4.2 94.1 Estonia EE 1 366 28.4 71.6 0.0 92.1 Finland FI 5 564 39.1 60.9 90.6 France (metropolitan) FR 66 017 0.1 5.3 65.6 21.4 7.7 108.7 Germany DE 83 119 3.0 46.0 50.4 0.6 110.2 Greece GR 10 414 0.2 5.2 22.6 28.5 43.5 0.0 116.7 Hungary HU 9 600 21.9 56.3 21.8 114.3 Iceland IS 388 74.8 25.0 0.2 89.0 Ireland IE 5 271 51.4 48.6 0.0 89.5 Italy IT 58 997 0.6 2.2 26.6 26.6 26.4 17.6 121.4 Latvia LV 1 883 2.2 97.6 0.2 93.4 Liechtenstein LI 40 99.9 0.1 116.5 Lithuania LT 2 857 94.2 5.8 98.2 Luxembourg LU 661 87.6 12.4 107.0 Malta MT 542 72.1 27.2 0.7 109.2 Monaco MC 39 100.0 121.9 Montenegro ME 617 3.6 63.6 32.8 0.0 108.4 Netherlands NL 17 811 37.4 61.0 1.6 102.0 North Macedonia MK 1 830 4.0 15.9 55.0 24.8 0.3 104.2 Norway NO 5 489 26.0 73.6 0.4 0.0 91.2 Poland PL 36 754 4.9 59.1 35.8 0.2 107.9 Portugal (excl. Azores, Madeira) PT 10 022 1.8 27.0 60.2 10.6 0.4 102.6 Romania RO 19 055 2.4 40.0 53.8 3.8 100.6 San Marino SM 34 100.0 114.9 Serbia (incl. Kosovo) RS 8 350 0.1 5.9 37.7 56.1 0.2 110.5 Slovakia SK 5 429 54.7 45.3 109.3 Slovenia SI 2 117 3.3 81.8 14.9 116.2 Spain (excl. Canarias) ES 45 872 1.5 14.9 40.0 37.8 5.8 108.2 Sweden SE 10 522 24.0 75.2 0.8 92.4 Switzerland CH 8 815 0.5 71.3 27.5 0.7 118.6 Türkiye TR 85 280 37.6 13.6 15.0 22.6 11.0 0.2 99.0 Total 560 167 8.1 13.8 37.1 31.1 8.0 2.0 107.5 21.9 10 EU-27 443 198 2.9 13.7 42.5 31.2 7.3 2.4 109.0 16.6 9.7 Northern Europe 34 055 23.0 75.5 1.5 0.0 92.7 Western Europe 86 334 3.0 14.9 68.0 13.7 0.4 105.0 Central Europe 165 805 2.6 41.4 52.7 3.3 0.0 110.8 Southern Europe 142 589 0.9 8.2 34.9 28.8 19.7 7.5 114.6 South-Eastern Europe 131 386 26.0 17.8 22.9 25.1 8.0 0.1 101.1 Kosovo XK 1 709 5.1 76.3 17.9 0.7 106.4 Serbia (excl. Kosovo) RS 6 641 0.1 6.2 27.8 65.8 0.1 111.5 Note: The percentage value "0.0" indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. ETC HE Report 2025/5 33 No population has been exposed to concentrations above the TV threshold in 15 countries. In Slovakia, Bulgaria, Kosovo, Montenegro, Liechtenstein, Malta, North Macedonia, Poland, Czechia, Albania, Cyprus, Andorra, Slovenia, Türkiye, Germany (in ascending order) up to 10% of population has been exposed to concentrations above the TV threshold. In Austria, France, Spain, Hungary, Croatia, Italy, this was the case for between 11% and 50% of the population. More than half of population has been exposed to concentrations above the TV threshold in Greece, Bosnia and Herzegovina, Switzerland and Monaco. Figure 3.1 Percentage of the population (%) exposed to different values of the ozone indicator 93.2 percentile of maximum daily 8-hour means (µg/m3), 2023 Figure 3.2 Percentage of the population (%) exposed to different values of the ozone indicator 93.2 percentile of maximum daily 8-hour means (µg/m3), 3-year mean (2021-2023) Based on the 3-year mean (2021-2023), it is estimated that 10% of the considered European population (including Türkiye) and 9.7% of the EU-27 population lived in areas where ozone concentrations exceeded the health-related target value of 120 µg/m3. No population has been exposed to concentrations above the TV in 2021-2023 in 15 countries. In Montenegro, Bulgaria, Czechia, Liechtenstein, Poland, Serbia, North Macedonia, Portugal, Germany, Malta, Kosovo, Austria, Albania, Andorra, Spain, France and Cyprus up to 10% of population has been exposed to concentrations above the TV. In Türkiye, Croatia, Slovenia, Hungary, Bosnia and Herzegovina, Switzerland, Greece and Italy, this was the case for between 11% and 50% of the population. More than half of population has been exposed to concentrations above the TV in Monaco. ETC HE Report 2025/5 34 As the current mapping methodology tends to underestimate high values due to interpolation smoothing (Annex 3, Section A3.3), the percentage of population exposed to values above the TV threshold is most likely somewhat underestimated; additional population exposure above the TV threshold in 2023 might be expected in some other countries: Albania, Austria, Croatia, Czechia, Germany, Luxembourg, Liechtenstein, Montenegro, Slovenia and San Marino, and above the TV in Albania, Austria, Bosnia, Croatia, Czechia, Germany, Hungary, Liechtenstein, San Marino, Serbia, Slovenia and Switzerland (in the case of 3-year mean). The reason is that in these countries the estimated percentage population exposed to the concentrations between 110 µg/m3 and 120 µg/m3 is considerable (more than 50%). The overall population-weighted ozone concentrations in 2023 in terms of the 93.2 percentile of maximum daily 8-hour means has been estimated to be 108.1 µg/m3 and 110.0 µg/m3 for the considered European area and for the EU-27, respectively, i.e. of about 1.9 µg/m3 and 1.1 µg/m3 lower than the 5-year 2018-2022 mean concentration, respectively. When assessing the change in individual countries, the steepest absolute decrease was found in Türkiye (7 µg/m3), the highest increase was estimated in Bosnia and Herzegovina (9 µg/m3). Based on the 3-year mean (2021-2023), the overall population-weighted ozone concentrations in terms of the 93.2 percentile of maximum daily 8-hour means has been estimated to be 107.5 µg/m3 and 109.0 µg/m3 for the considered European area and for the EU-27. One can see that more population was exposed to concentrations above the 120 µg/m3 in 2023 compared to the 3-year mean 2021-2023, indicating that 2023 was a year with higher O3 levels than the previous two. Figure 3.3 (left) shows, for the whole mapped area, the population frequency distribution for exposure classes of 2 µg/m3. The highest population frequency is found for classes between 100 and 120 µg/m3. For classes above 120 µg/m3, a sharp decline of population frequency can be seen. Figure 3.3 (right), based on 3-year mean (2021-2023), shows that the highest frequency is concentrated in the range of approximately 104 to 114 µg/m³. For classes above 114 µg/m3, a sharp decline in population frequency is observed. Figure 3.3 Population exposure frequency distribution, O3 indicator 93.2 percentile of maximum daily 8-hour means, 2023 (left) and 3-year mean 2021-2023 (right). The EU target value threshold (120 µg/m3) is marked by the red line The boxplot showing for individual countries the ozone concentrations expressed as the indicator 93.2 percentile of maximum daily 8-hour means for year 2023 and 3-year mean (2021-2023) to which the population per country was exposed is presented in Figure 3.4. ETC HE Report 2025/5 35 Figure 3.4 Concentrations of O3 indicator 93.2 percentile of maximum daily 8-hour means to which the population per country was exposed in 2023 (top) and 3-year mean 2021-2023 (bottom) Note: For each country, the box plot shows the concentration to which a percentage of the population was exposed: 50% in the case of the black marker, 25% and 75% in the cases of the box’s edges, 2% and 98% in the cases of the whiskers’ edges. 3.2 Ozone – peak season average of maximum daily 8-hour means In September 2021, the WHO introduced new AQG for O3. The new long-term Air Quality Guideline (AQG) level ( 5 ) for O₃ is set to 60 µg/m³ and it is related to the average of daily maximum 8-hour mean ( 5 ) Besides the short-term AQG of 100 µg/m3 defined as 99th percentile of the annual distribution of daily maximum 8-hour mean O3 concentration (WHO, 2021a). ETC HE Report 2025/5 36 O₃ concentration of the so-called peak season. The peak season is defined as the six consecutive months of the year with the highest six-month running-average ozone concentration (WHO, 2021a). For details of the calculations of this indicator, see Annex 2 Section A2.1. The new long-term WHO AQG level for O₃ (i.e. 60 µg/ m³, not to be exceeded by the average of the daily maximum 8-hour mean O₃ concentration during the peak season) was determined based on average concentrations observed in studies that examined the health effects of long-term O3 exposure. By setting this guideline, WHO aims to minimize the health risks associated with long-term ozone exposure and non-accidental mortality (WHO, 2021a). This peak season indicator is not regulated in any of the EU air quality directives, and there are no associated limit or target values defined. 3.2.1 Concentration map Map 3.3 presents the final combined map for ozone peak season average of maximum daily 8-hour means. In the map, all coloured area except the dark green ones show values of peak season average of maximum daily 8-hour means above 60 µg/m3. The map shows that in 2023, areas with peak season average of maximum daily 8-hour means exceeding 60 µg/m3 cover the entire mapped region with the exception of a few urban areas in Türkiye and Bulgaria. Similar to other O3 characteristics, lower values (< 80 µg/m3) were observed in northern and western Europe (Finland, northern Sweden and Ireland), while higher values (> 100 µg/m3) are observed in central Europe (Switzerland, Slovenia, parts of Austria), as well as in southern and southeastern Europe (parts of Italy, parts of Spain and Portugal, southern France, parts of Balkan countries, Cyprus, and large areas of Türkiye). In order to provide more complete information of the air quality across Europe, the final combined map including the measurement data at stations is presented in Map A4.5 of Annex 4. Map 3.3 Ozone indicator peak season average of maximum daily 8-hour means, 2023 ETC HE Report 2025/5 37 Since the concentration map for ozone peak season average of maximum daily 8-hour means was presented for the first time for 2022, it was not possible to calculate a 5-year mean (2018-2022) and subsequently create a differences map, as done for other pollutants and indicators. The relative mean uncertainty of the 2023 map of the peak season average of maximum daily 8-hour means is about 8% for rural and 10% for urban areas (Annex 3, Section A3.2). The low uncertainty values are influenced by the character of this ozone indicator. Note that the Air Quality Directive (EC, 2008) sets no modelling uncertainty for ozone long-term indicators. 3.2.2 Population exposure Figure 3.5 and Table 3.3 give, for the peak season average of maximum daily 8-hour means, the population frequency distribution for a limited number of exposure classes for individual countries, large regions, EU-27 and for the total mapped area. Table 3.3 also presents the population-weighted concentration. In 2023, it has been estimated that 98% of the considered European population (including Türkiye) and almost 100% of the EU-27 population lived in areas where the ozone concentration was above the peak season average of maximum daily 8-hour means of 60 µg/m3. Figure 3.5 Percentage of the population (%) exposed to different values of ozone peak season average of maximum daily 8-hour means, 2023 The overall population-weighted ozone concentrations in terms of peak season average of maximum daily 8-hour means has been estimated to be 87.7 µg/m3 in 2023 for the considered European area (including Türkiye) and 89.8 µg/m3 for the EU-27. When considering individual countries, the highest overall population-weighted ozone concentrations in terms of peak season average has been estimated to be 102.3 µg/m3 in Monaco and 99.9 µg/m3 in Bosnia and Herzegovina, respectively, while the lowest was estimated to be 74.4 µg/m3 in Ireland. ETC HE Report 2025/5 38 Table 3.3 Population exposure and population-weighted concentrations, ozone peak season average of maximum daily 8-hour means, 2023 Country ISO Population (inhbs·1000) Ozone — peak season average of maximum daily 8-hour means (%) Population-weighted concentration < 60 60 - 80 80 - 90 90 - 100 100 - 120 > 120 2023 Albania AL 2 762 20.5 78.6 0.9 92.1 Andorra AD 80 73.9 14.9 11.2 87.9 Austria AT 9 105 21.0 78.4 0.6 93.0 Belgium BE 11 743 2.7 97.0 0.4 84.5 Bosnia and Herzegovina BA 3 194 1.0 52.8 45.3 0.9 99.9 Bulgaria BG 6 448 0.7 65.0 28.7 5.7 0.0 78.8 Croatia HR 3 851 30.4 45.8 23.7 0.0 95.7 Cyprus CY 1 419 47.6 37.8 12.7 1.9 81.5 Czechia CZ 10 828 9.1 90.9 0.0 93.0 Denmark (incl. Faroe Islands) DK 5 987 23.3 76.7 0.0 81.6 Estonia EE 1 366 19.0 81.0 81.8 Finland FI 5 564 76.9 23.1 78.7 France (metropolitan) FR 66 017 1.1 69.1 26.7 3.1 88.1 Germany DE 83 119 0.0 59.7 40.3 0.0 89.4 Greece GR 10 414 2.8 30.0 33.3 33.9 94.3 Hungary HU 9 600 29.6 70.4 92.4 Iceland IS 388 93.1 6.9 77.8 Ireland IE 5 271 97.4 2.6 74.4 Italy IT 58 997 2.6 17.0 40.3 40.1 97.7 Latvia LV 1 883 100.0 0.0 85.2 Liechtenstein LI 40 99.9 0.1 93.3 Lithuania LT 2 857 65.6 34.4 89.1 Luxembourg LU 661 94.6 5.4 87.2 Malta MT 542 98.2 1.8 97.0 Monaco MC 39 100.0 102.3 Montenegro ME 617 4.3 94.0 1.7 93.8 Netherlands NL 17 811 0.6 99.4 0.0 85.1 North Macedonia MK 1 830 22.2 67.9 9.4 0.5 83.3 Norway NO 5 489 53.4 46.4 0.2 80.2 Poland PL 36 754 53.1 46.9 0.0 89.3 Portugal (excl. Azores, Madeira) PT 10 022 14.6 64.7 20.8 0.0 86.3 Romania RO 19 055 20.4 70.9 8.6 0.0 84.3 San Marino SM 34 100.0 97.6 Serbia (incl. Kosovo) RS 8 350 8.7 58.7 32.5 0.2 87.5 Slovakia SK 5 429 54.2 45.8 0.0 89.4 Slovenia SI 2 117 18.5 76.9 4.6 93.8 Spain (excl. Canarias) ES 45 872 10.8 15.5 46.0 27.6 93.8 Sweden SE 10 522 51.1 48.9 0.0 80.0 Switzerland CH 8 815 3.4 93.4 3.1 95.0 Türkiye TR 85 280 11.6 48.2 21.3 16.8 2.0 0.0 75.6 Total 555 254 1.7 14.0 43.1 32.7 8.5 0.0 87.7 15.7 8.5 EU-27 443 198 0.0 7.7 48.0 34.5 9.8 0.0 89.8 7.7 9.8 Northern Europe 34 055 42.8 54.2 3.0 81.2 Western Europe 86 334 6.8 80.3 12.8 0.0 85.4 Central Europe 165 805 0.0 47.5 52.2 0.3 90.3 Southern Europe 142 589 6.1 23.6 40.6 29.7 94.8 South-Eastern Europe 131 386 7.3 37.8 32.1 19.5 3.3 0.0 79.7 Kosovo XK 1 709 6.1 73.2 19.8 0.9 86.2 Serbia (excl. Kosovo) RS 6 641 9.3 55.0 35.7 0.0 87.8 Note: The percentage value "0.0" indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. ETC HE Report 2025/5 39 Figure 3.6 shows, for the whole mapped area, the population frequency distribution for exposure classes of 1 µg/m3. The highest population frequency is found for classes between 84 and 96 µg/m3. For classes above 96 µg/m3, a sharp decline of population frequency can be seen. Figure 3.6 Population exposure frequency distribution, ozone peak season average of maximum daily 8-hour means, 2023. The WHO AQG level (60 µg/m3) is marked by the green line The boxplot in Figure 3.7 shows for individual countries the ozone concentrations expressed as peak season average of maximum daily 8-hour means to which the population per country was exposed in 2023. It can be seen that the countries with the highest ozone concentrations are located in the southern, south-eastern and central parts of Europe. Figure 3.7 Population frequency distribution, O3 peak season average of maximum daily 8-hour means, 2023. The WHO AQG level (60 µg/m3) is marked by the green line. ETC HE Report 2025/5 40 3.3 Ozone – SOMO35 and SOMO10 SOMO35 is the annually accumulated ozone maximum daily 8-hourly means in excess of 35 ppb (i.e. 70 µg/m3). It is not regulated in the EU air quality directives and there are no associated limit or target values defined. Nevertheless, it is considered by the WHO as a good indicator of human exposure to ozone (WHO, 2013). Comparing the 93.2 percentile of maximum daily 8-hour means versus the SOMO35 for all background stations shows no simple relationship between the two indicators. However, it seems that the TV threshold of the 93.2 percentile of maximum daily 8-hour means (120 µg/m3) is related approximately with a SOMO35 value in the range of 6 000-8 000 µg/m3·d. This comparison motivates a somewhat arbitrarily chosen threshold of 6 000 µg/m3·d, in order to facilitate the discussion of the estimated SOMO35 levels in their spatial and temporal context. This threshold is used in this and previous papers (Horálek et al., 2023, and the references cited therein) when dealing with the population exposure estimates. SOMO10 is the annually accumulated ozone maximum daily 8-hourly means in excess of 10 ppb (i.e. 20 µg/m3). This indicator was introduced due to its link to the health impact assessment, since the WHO recommended using the SOMO10 as an alternative to the SOMO35 when estimating the health impact of ozone (WHO, 2013). 3.3.1 Concentration maps Maps 3.4 and 3.5 present the final combined map for SOMO35 and SOMO10 in 2023. In the final combined map of SOMO35, the orange areas show values above 6 000 µg/m3·d, while the red and dark red areas show values above 8 000 µg/m3·d. In the case of SOMO10, the boundaries of concentration classes have been chosen quite arbitrary, in order to reflect the concentration distribution of this indicator. In the final combined map of SOMO10, the red and dark red areas show values above 24 000 µg/m3·d. Map 3.4 Concentration map of ozone indicator SOMO35, 2023 Like in the case of the 93.2 percentile of the maximum daily 8-hour means, generally the southern and south-eastern parts of Europe show higher ozone SOMO35 and SOMO10 concentrations than the northern parts. Higher levels of ozone also occur more frequently in mountainous areas south of 50 ETC HE Report 2025/5 41 degrees latitude than in lowlands. In 2023, SOMO35 levels above 6 000 µg/m3·d were estimated in large areas of Türkiye, Italy, Spain, Austria, in parts of many Balkan countries and in southern France. SOMO10 levels above 24 000 µg/m3·d were mainly found in the same areas. The relative mean uncertainty of the 2023 maps of the SOMO35 and SOMO10 is about 33% and 12%, respectively, for rural areas and 39% and 14%, respectively, for urban areas (see Annex 3). Map 3.5 Concentration map of ozone indicator SOMO10, 2023 Map 3.6 presents the difference between 2023 and the 5-year mean 2018-2022 for the O3 indicators SOMO35 and SOMO10. Orange to red areas show an increase of those O3 indicators in 2023, while blue areas show a decrease. Quite similar patterns as in Map 3.2 can be seen. The largest increases in SOMO35 and SOMO10 levels were recorded in some areas of several Balkan countries, in Poland and in the Baltic states. Increases were also observed in smaller areas of Italy, Spain, and Portugal; in the case of SOMO10, also in Germany, Austria, and the Benelux countries. Areas with the highest decreases were found in Türkiye and some Balkan states, plus, parts of Italy, France, Germany, Austria, Slovakia and Hungary in the case of SOMO35. ETC HE Report 2025/5 48 3.4 Ozone – AOT40 vegetation and AOT40 forests In the Ambient Air Quality Directive (EC, 2008) a target value (TV) and a long-term objective (LTO) for the protection of vegetation from high ozone concentrations accumulated during the growing season have been defined. TV and LTO are specified using “accumulated ozone exposure over a threshold of 40 parts per billion” (AOT40). This is calculated as a sum of the difference between hourly concentrations greater than 40 ppb (i.e. 80 µg/m3) and 40 ppb, using only observations between 08:00 and 20:00 Central European Time (CET) each day, calculated over three months from 1 May to 31 July. The TV is 18 000 µg/m3·h (averaged over five years) and the LTO is 6 000 µg /m3·h. Note that the term “vegetation” as used in the Ambient Air Quality Directive (EC, 2008) is not further defined. Nevertheless, the TV used in the directive is quite similar as the critical level used in the Mapping Manual (CLRTAP, 2024) for “agricultural crops” (although the definitions of AOT40 by the EU and the CLRTAP are slightly different), so the term vegetation in the Air Quality Directive has been interpreted as primarily agricultural crops. Therefore, the exposure of agricultural crops has been evaluated here based on the AOT40 for vegetation as defined in the Air Quality Directive and the agricultural areas, defined as the CORINE Land Cover level-1 class 2 Agricultural areas (encompassing the level-2 classes 2.1 Arable land, 2.2 Permanent crops, 2.3 Pastures and 2.4 Heterogeneous agricultural areas), see Annex 2 Section A2.3. Note that in addition to these agricultural areas there are several other CLC classes that could be considered “vegetation”, namely level-2 classes 1.4 Artificial, non-agricultural vegetated areas (encompassing the level-3 classes 1.4.1 Green urban areas and 1.4.2 Sport and leisure facilities), 3.1 Forests (see below) and 3.2 Scrub and/or herbaceous vegetation associations. These other CLC classes have not been considered (apart from forests, see below). Next to the AOT40 for vegetation protection, the Ambient Air Quality Directive (EC, 2008) defines also the AOT40 for forest protection, which is calculated similarly as the AOT40 for vegetation, but is summed over six months from 1 April to 30 September. For AOT40 for forests there is no TV defined in the Air Quality Directive. However, there is a critical level (CL) established by the CLRTAP, see CLRTAP (2024). This critical level is set at 10 000 µg/m3·h. Although CLRTAP (2024) calculates the AOT40 indicators somewhat differently (e.g. it uses the ozone concentration corrected at canopy height), this CL level is further used for the AOT40 for forests calculated according to the EC (2008). For the exposure of forests evaluation, the CLC level-2 class 3.1 Forests has been used. The ecosystem based accumulative ozone indicators described in this section are specifically prepared for calculation of the EEA indicator on ecosystem exposure to ozone (EEA, 2025b). For the estimation of the vegetation and forested area exposure to accumulated ozone, the maps in this section are created on a grid of 2 km resolution. The exposure frequency distribution outcomes are based on the overlay with the 100 m grid resolution of the CLC2018 land cover classes. 3.4.1 Concentration maps The interpolated maps of AOT40 for vegetation and AOT40 for forests are applicable for rural areas only. Map 3.7 presents the final map of AOT40 for vegetation in 2023. Note that in the Ambient Air Quality Directive (EC, 2008) the TV is actually defined as 18 000 µg/m3·h averaged over 5 years. Here both 2023 data and 5-year mean (2019-2023) are presented (5-year mean for AOT40 for vegetation is presented for the first time in this report). Data for the current calendar year (i.e. 2023) are evaluated against what it is called the target value threshold, an AOT40 of 18 000 µg/m3·h for just one year. ETC HE Report 2025/5 49 Map 3.7 Concentration map of ozone indicator AOT40 for vegetation, rural map, 2023 (left) and 5-year mean 2019-2023 (right) The areas in the map with AOT40 for vegetation in 2023 above the TV threshold of 18 000 µg/m3·h are marked in red and dark red. The areas below the long-term objective (LTO) are marked in green. AOT40 levels above the TV threshold for vegetation occur specifically in southern and south-eastern of Europe (Italy, parts of Spain, France and parts of the Balkan countries and Türkiye) and also in central Europe (parts of Switzerland, Germany, Austria, Slovenia, Czechia, Slovakia, Poland and Hungary). The highest levels (dark red) were estimated, similarly to previous years, in the north of Italy, in the south-west of Türkiye and in Cyprus. Based on the 5-year mean of AOT40 for vegetation, the situation is quite similar in terms of the spatial distribution of AOT40 values, a bit worse in Türkiye and the Balkans. The relative mean uncertainty of the 2023 map of the AOT40 for vegetation is about 36% (Annex 3, Section A3.3). Map 3.8 presents the final map of AOT40 for forests in 2023. The areas in the map with concentrations below the critical level (CL) defined by CLRTAP (2024) are marked in green. One can see large forested areas exceeding this level in most of Europe, with the exception of parts of some northern countries and Ireland. The highest values of the AOT40 for forests in 2023 were found mainly in southern Europe (Spain, south of France, parts of the Balkan states, the Po Valley and Cyprus) and Türkiye. The relative mean uncertainty of the 2023 map of the AOT40 for forests is about 30% (Annex 3, Section A3.3). In order to provide more complete information of the air quality across Europe, the AOT40 maps including the AOT40 values based on the actual rural background measurement data at stations are presented in Maps A4.8 and A4.9 of Annex 4. ETC HE Report 2025/5 50 Map 3.8 Concentration map of ozone indicator AOT40 for forests, rural map, 2023 Map 3.9 presents the difference between 2023 and the 5-year mean 2018-2022 for AOT40 for vegetation and for AOT40 for forests. Orange to red areas show an increase of AOT40 in 2023, while blue areas show a decrease. Map 3.9 Difference concentrations between 2023 and the 5-year mean 2018-2022 for ozone indicators, AOT40 for vegetation (left) and AOT40 for forests (right) ETC HE Report 2025/5 51 The most notable increases in AOT40 for vegetation were recorded in southern Spain, eastern France, parts of Germany and northern Europe. In northern Europe, specifically in the Baltic states, an increase also in AOT40 for forests was also observed. Elsewhere, a decrease in AOT40 has been observed across various parts throughout the mapped area, with the most notable decrease in Türkiye, the Balkan states, Italy and parts of France and Spain. Regarding AOT40 levels for forests, a noticeable decline is observed in similar areas, i.e. in Türkiye, the Balkan states, Italy and parts of Spain and most of France and Germany. 3.4.2 Vegetation exposure Agricultural crops The rural map with the ozone indicator AOT40 for vegetation has been combined with the land cover CLC2018 map. Following a similar procedure as described in Horálek et al. (2007), the exposure of agricultural areas (as defined above) has been calculated at the country-level. Tables 3.6 and 3.7 give the absolute and relative agricultural area in 2023 and for the 5-year mean (2013-2023) for each country and for five European regions where ozone concentrations are above the target value (TV) threshold and the long-term objective (LTO) for the protection of vegetation as defined in the Ambient Air Quality Directive (EC, 2008). The frequency distribution of the agricultural area over some exposure classes per country is presented as well. Table 3.6 illustrates that in 2023, 12% of all European agricultural land including Türkiye was exposed to ozone concentrations above the TV threshold of 18 000 µg/m3·h. For the area for the EU-27, it was also 12%. None of the agricultural area presents ozone levels in excess of the TV in 17 countries. Agricultural areas with ozone concentrations above TV threshold covered less than 25% in Portugal, Slovakia, Poland, Hungary North Macedonia, Montenegro, Kosovo, France, Albania, Liechtenstein, Czechia, Croatia, San Marino, Slovenia, Türkiye, Austria, Greece, Germany (in ascending order). In Spain, Italy and Switzerland, between 26% and 50% of agricultural area has been exposed to ozone concentrations above the TV threshold. The largest proportion of the agricultural area exposed to ozone concentrations above the TV threshold is Cyprus, Andorra and Malta (from 54% up to 100%). Considering the LTO of 6 000 µg/m3·h, the total European area (including Türkiye) in excess has been 91%. For the area for the EU-27, it has been 94%. In 2023, values of the AOT40 for vegetation above the LTO have occurred in all countries with the exception of Iceland and Ireland. Fewer than 50% of the areas with values above the LTO have occurred in Finland (17%) and Norway (27%). In the remaining countries, the agricultural area exposed above the LTO in 2023 has been between 75% and 100%. In Albania, Andorra, Austria, Belgium, Croatia, Cyprus, Czechia, Germany, Hungary, Kosovo, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Netherlands, North Macedonia, Poland, San Marino, Serbia, Slovenia, Switzerland, the entire agriculture area was exposed to AOT40 levels above the LTO for protection of vegetation. In 2023, the agricultural-weighted ozone concentration of vegetation-related AOT40 for the total mapped area was estimated to be 12 103 µg/m3·h, i.e. 2 049 µg/m3·h lower than the 5-year 2018-2022 mean. For the EU-27 area, it was estimated to be 12 407 µg/m3·h, i.e. 538 µg/m3·h lower than the 5year 2019-2023 mean. When assessing the change in individual countries, the steepest decrease was found in Türkiye (10 508 µg/m3·h), while the highest increase was estimated in Latvia (3 838 µg/m3·h, respectively). ETC HE Report 2025/5 52 Table 3.6 Agricultural area exposure and agricultural-weighted concentrations, ozone indicator AOT40 for vegetation, 2023 Country Agricultural area, 2023 Percentage of agricultural area (%) Agricultural-weighted conc. Total area > LTO (6 000 µg/m3·h) > TV threshold (18 000 µg/m3·h) < 6 000 6 00012 000 12 00018 000 18 00027 000 > 27 000 (µg/m3·h) (km2) (km2) (%) (km2) (%) (µg/m3·h) 2023 5-year mean Diff. Albania 8 017 8 017 100.0 416 5.2 21.4 73.5 5.2 14 254 16 602 -2 347 Andorra 13 13 100.0 9 64.8 35.2 64.8 19 607 Austria 26 827 26 827 100.0 4 731 17.6 0.4 81.9 17.6 0.0 16 601 17 413 -812 Belgium 17 473 17 473 100.0 21.9 78.1 13 092 11 781 1 312 Bosnia-Herzegovina 17 023 17 011 99.9 0.1 89.3 10.7 10 288 10 857 -570 Bulgaria 57 390 51 940 90.5 9.5 87.8 2.7 8 330 11 538 -3 209 Croatia 22 168 22 168 100.0 1 346 6.1 44.5 49.5 6.1 12 543 13 450 -907 Cyprus 4 291 4 291 100.0 2 322 54.1 45.9 54.1 18 368 24 179 -5 811 Czechia 44 784 44 784 100.0 2 524 5.6 94.4 5.6 16 229 16 883 -654 Denmark (incl. Faroes) 31 236 31 177 99.8 0.2 99.8 7 889 6 977 913 Estonia 14 252 14 182 99.5 0.5 99.5 6 893 4 101 2 792 Finland 27 505 4 791 17.4 82.6 17.4 4 208 3 050 1 158 France 323 377 321 366 99.4 5 294 1.6 0.6 61.3 36.4 1.6 0.0 11 330 11 612 -282 Germany 204 463 204 463 100.0 51 049 25.0 32.6 42.4 25.0 14 385 14 096 289 Greece 50 052 50 019 99.9 9 994 20.0 0.1 23.6 56.4 17.1 2.9 15 028 19 830 -4 803 Hungary 60 390 60 390 100.0 23 0.0 25.1 74.9 0.0 13 201 15 843 -2 642 Iceland 2 518 0 0.0 100.0 0.0 943 1 102 -158 Ireland 46 756 0 0.0 100.0 0.0 2 116 2 932 -816 Italy 155 718 155 673 100.0 59 443 38.2 0.0 23.7 38.1 24.0 14.2 17 674 20 868 -3 194 Latvia 25 532 25 532 100.0 100.0 8 039 4 201 3 838 Liechtenstein 37 37 100.0 2 5.6 94.4 5.6 16 210 17 648 -1 438 Lithuania 38 155 38 155 100.0 100.0 9 203 5 465 3 738 Luxembourg 1 351 1 351 100.0 100.0 16 134 13 843 2 291 Malta 125 125 100.0 125 100.0 100.0 20 597 19 983 614 Monaco Montenegro 2 243 2 236 99.7 5 0.2 0.3 56.9 42.6 0.2 11 762 13 547 -1 784 Netherlands 23 644 23 644 100.0 69.8 30.2 11 021 9 399 1 622 North Macedonia 9 146 9 146 100.0 15 0.2 58.2 41.6 0.2 11 731 15 892 -4 161 Norway 15 637 4 237 27.1 72.9 27.1 0.0 4 622 3 590 1 032 Poland 183 268 183 268 100.0 50 0.0 42.6 57.4 0.0 12 482 11 953 529 Portugal 42 566 41 800 98.2 0 0.0 1.8 60.7 37.5 0.0 11 206 11 109 97 Romania 135 279 122 047 90.2 9.8 85.3 4.9 8 380 10 182 -1 802 San Marino 42 42 100.0 5 11.9 88.1 11.9 17 151 19 108 -1 957 Serbia (incl. Kosovo) 46 768 46 768 100.0 18 0.0 39.3 60.6 0.0 12 249 14 078 -1 829 Slovakia 23 100 23 090 100.0 3 0.0 0.0 31.1 68.9 0.0 13 015 14 890 -1 875 Slovenia 6 986 6 986 100.0 1 010 14.4 2.4 83.2 14.4 0.0 15 911 16 878 -967 Spain 241 014 231 836 96.2 78 106 32.4 3.8 12.1 51.7 32.4 0.0 15 676 15 744 -67 Sweden 39 035 34 395 88.1 11.9 88.1 7 185 5 679 1 506 Switzerland 11 359 11 359 100.0 5 466 48.1 0.6 51.2 48.0 0.1 17 937 18 575 -638 Türkiye 339 984 255 707 75.2 54 998 16.2 24.8 39.6 19.4 13.6 2.5 10 717 21 225 -10 508 Total 2 299 526 2 096 348 91.2 276 952 12.0 8.8 43.2 35.9 10.6 1.4 12 103 14 153 -2 049 EU-27 1 846 681 1 741 774 94.3 216 019 11.7 5.7 44.0 38.6 10.4 1.3 12 407 12 945 -538 Northern Europe 193 869 152 469 78.6 21.4 78.6 0.0 7 076 4 934 2 142 Western Europe 346 009 298 273 86.2 1 753 0.5 13.8 50.5 35.1 0.5 10 226 10 059 167 Central Europe 561 215 561 205 100.0 64 858 11.6 0.0 29.8 58.6 11.6 0.0 13 924 14 123 -199 Southern Europe 560 414 549 360 98.0 153 544 27.4 2.0 26.3 44.3 23.2 4.2 15 306 16 907 -1 602 South-Eastern Europe 638 018 535 040 83.9 56 797 8.9 16.1 55.2 19.8 7.5 1.4 10 234 16 780 -6 546 Kosovo 4 167 4 167 100.0 18 0.4 18.3 81.2 0.4 13 506 16 119 -2 613 Serbia (without Kosovo) 42 601 42 601 100.0 41.4 58.6 12 126 13 878 -1 752 Note: The percentage value "0.0" indicates that an exposed agricultural area exists, but it is small and estimated to be less than 0.05%. Empty cells mean no agricultural area in exposure. 5-year mean, i.e. 5-year mean 2018-2022. Diff., i.e. difference concentrations between 2023 and the 5-year mean 2018-2022. Relating to 5-year mean 2019-2023 (Table 3.7), 16% of all European agricultural land including Türkiye has been exposed to ozone concentrations above the TV of 18 000 µg/m3·h. For the area for the EU27, it was 10%. None of the agricultural area presents ozone levels in excess of the TV in 17 countries. Agricultural areas with ozone concentrations above the TV covered less than 25% in Serbia, Portugal, Slovakia, Czechia, Montenegro, Germany, Hungary, France, Austria, Liechtenstein, Croatia, North Macedonia, Kosovo. Albania, Slovenia and Spain (in ascending order). In Switzerland, Andorra, Greece and Italy, between 27% and 50% of agricultural area has been exposed to ozone concentrations above the TV. The largest proportion of the agricultural area exposed to ozone concentrations above the TV is Türkiye, Malta, San Marino and Cyprus (from 54% up to 100%). ETC HE Report 2025/5 53 Table 3.7 Agricultural area exposure and agricultural-weighted concentrations, ozone indicator AOT40 for vegetation, 5-year mean (2019-2023) Country Agricultural area, 2019-2023 Percentage of agricultural area (%) Agriculturalweighted conc. Total area > TV (18 000 µg/m3·h) < 6 000 6 00012 000 12 00018 000 18 00027 000 > 27 000 (µg/m3·h) (km2) (km2) (%) (µg/m3·h) 2019-2023 Albania 8 017 1200 15.0 2.8 82.2 15.0 15 839 Andorra 13 6 44.7 55.3 44.7 17 290 Austria 26 827 1074 4.0 0.6 95.4 4.0 0.0 16 087 Belgium 17 473 91.6 8.4 10 464 Bosnia-Herzegovina 17 023 0.5 89.3 10.3 9 975 Bulgaria 57 390 75.5 24.5 10 609 Croatia 22 168 1545 7.0 58.2 34.8 7.0 0.0 12 313 Cyprus 4 291 4291 100.0 91.9 8.1 23 500 Czechia 44 784 52 0.1 0.2 99.7 0.1 15 002 Denmark (incl. Faroes) 31 236 66.4 33.6 5 749 Estonia 14 251 98.3 1.7 4 435 Finland 27 504 100.0 2 771 France 323 377 10839 3.4 0.4 78.1 18.2 3.3 0.0 10 579 Germany 204 463 978 0.5 0.2 44.0 55.3 0.5 12 551 Greece 50 051 22801 45.6 5.9 48.5 42.5 3.0 17 870 Hungary 60 387 1408 2.3 12.2 85.5 2.3 14 290 Iceland 2 517 100.0 681 Ireland 46 756 100.0 2 104 Italy 155 718 77144 49.5 2.6 47.9 30.7 18.8 20 284 Latvia 25 530 99.4 0.6 4 545 Liechtenstein 37 2 5.5 94.5 5.5 15 711 Lithuania 38 148 76.3 23.7 5 579 Luxembourg 1 351 12.3 87.7 12 622 Malta 125 70 56.3 43.7 56.3 19 842 Monaco Montenegro 2 243 10 0.5 26.7 72.9 0.5 13 018 Netherlands 23 644 0.1 99.6 0.3 8 698 North Macedonia 9 146 805 8.8 17.4 73.8 8.8 14 647 Norway 15 636 96.3 3.7 2 970 Poland 183 258 64.3 35.7 10 998 Portugal 42 566 1 0.0 2.4 59.7 37.9 0.0 11 374 Romania 135 270 1.7 85.9 12.4 9 495 San Marino 42 30 71.1 28.9 71.1 18 401 Serbia (incl. Kosovo) 46 768 477 1.0 21.1 77.9 1.0 13 325 Slovakia 23 100 3 0.0 30.9 69.1 0.0 13 385 Slovenia 6 986 1189 17.0 1.7 81.3 17.0 0.0 16 131 Spain 241 014 55388 23.0 4.4 8.7 63.9 23.0 0.0 15 344 Sweden 39 035 73.7 26.3 4 909 Switzerland 11 359 3065 27.0 0.1 72.9 26.6 0.4 17 367 Türkiye 339 966 183555 54.0 1.8 9.8 34.4 44.7 9.3 18 616 Total 2 299 474 365935 15.9 10.1 36.2 37.8 13.2 2.7 13 002 EU-27 1 846 648 176785 9.6 11.3 41.8 37.4 7.9 1.7 12 043 Northern Europe 193 857 84.1 15.9 4 579 Western Europe 346 009 1728 0.5 13.7 71.4 14.4 0.5 9 046 Central Europe 561 202 7772 1.4 0.1 39.7 58.8 1.4 0.0 12 772 Southern Europe 560 414 168844 30.1 2.2 17.6 50.1 24.6 5.6 16 300 South-Eastern Europe 637 992 187592 29.4 1.3 36.6 32.7 24.5 4.9 15 013 Kosovo 4 167 477 11.4 4.2 84.3 11.4 15 754 Serbia (without Kosovo) 42 601 0 0.0 22.8 77.2 0.0 13 087 Note: The percentage value "0.0" indicates that an exposed agricultural area exists, but it is small and estimated to be less than 0.05%. Empty cells mean no agricultural area in exposure. In 2023, the agricultural-weighted ozone concentration of vegetation-related AOT40 (5-year mean) for the total mapped area was estimated to be 13 002 µg/m3·h. For the EU-27 area, it was estimated to be 12 043 µg/m3·h. ETC HE Report 2025/5 54 Forests The rural map with ozone indicator AOT40 for forests was combined with the land cover CLC2018 map. Following a similar procedure as described in Horálek et al. (2007), the exposure of forest areas (as defined above) has been calculated for each country, for the same five European regions as for crops and for Europe as a whole. Table 3.8 gives the absolute and relative forest area where the critical level (CL) set at 10 000 µg/m3·h, the same level as defined in CLRTAP (2024), and the value 20 000 µg/m3·h (which is equal to the earlier used reporting value, RV, as was defined in the repealed ozone directive 2002/3/EC, EC, 2002) are exceeded. Next to the forest area in exceedance, the table presents the frequency distribution of the forest area over some exposure classes. The CL was exceeded in 2023 at 73% of all European forested area including Türkiye. For the area of the EU-27 it was exceeded at about 74%. As in previous years, most countries continue to have in 2023 the whole or considerable forest areas in excess of the CL. In 2023, there were only two countries with almost no CL exceedance (Iceland and Ireland). Up to 50% of the areas with values above the CL have occurred in Finland, Norway and Sweden. In the remaining countries, the CL for AOT40 was exceeded in more than 90% of the area of the respective country. In 27 countries, the entire forested area was exposed to AOT40 levels above the CL for forests. In 2023, the forest-weighted ozone concentration of forest-related AOT40 for the total mapped area was estimated to be 18 143 µg/m3·h, i.e. 2 176 µg/m3·h less than the 5-year 2018-2022 mean. For the EU-27 area, it was estimated to be 17 993 µg/m3·h, i.e. 1 445 µg/m3·h less than the 5-year 2018-2022 mean. When assessing the change in individual countries, the steepest absolute decrease was found in Türkiye (12 395 µg/m3·h), while the highest increase was estimated in Lithuania (4 553 µg/m3·h). In this context, it should be mentioned that the AOT40 indicator is not the best proxy for vegetation damage assessment. AOT40 does not take into account plant physiological control of ozone absorbed doses, which is taken into account in the POD (i.e. Phytotoxic Ozone Dose) indicators, as discussed in Section 3.5. POD indicators are known to be more related with ozone effects on plant growth than ambient air ozone concentrations alone. The AOT40 does not take into account the influence of meteorological conditions on growing season timing. Growing season´s start and end dates can change across Europe, and between years for a given site, depending on factors such as air temperature, solar radiation, photoperiod or rainfall. High temperature and dry weather favouring ozone pollution cause a reduction of ozone absorbed doses by plants due to plant physiological response to drought (i.e. the vegetation closes its stomata protecting itself from the exposure to ozone). However, plants may still be sensitive to ozone in such weather conditions, as illustrated by foliar injury records in Aleppo pine stands growing in southern France (CLRTAP, 2016) or controlled experimental results (e.g. Alonso et al., 2014). ETC HE Report 2025/5 55 Table 3.8 Forested area exposure and forest-weighted concentrations, ozone indicator AOT40 for forests, 2023 Country Forested area Percentage of forested area [%] Forest-weighted conc. Total area > CL (10 000 µg/m3·h) > RV (20 000 µg/m3·h) < 10 000 10 00020 000 20 00030 000 30 00050 000 > 50 000 (µg/m3·h) (km2) (km2) (%) (km2) (%) (%) (µg/m3·h) (µg/m3·h) (µg/m3·h) (µg/m3·h) 2023 5-year mean Diff. Albania 7 104 7 104 100.0 7 031 99.0 1.0 50.5 48.4 0.0 29 921 34 999 -5 078 Andorra 128 128 100.0 128 100.0 0.0 0.0 14.1 85.9 34 970 Austria 36 667 36 667 100.0 36 135 98.5 1.5 77.3 21.2 27 608 30 799 -3 190 Belgium 6 089 6 089 100.0 3 063 50.3 49.7 50.3 20 196 22 794 -2 598 Bosnia-Herzegovina 23 911 23 911 100.0 19 205 80.3 19.7 77.3 3.0 23 160 23 062 97 Bulgaria 34 675 34 636 99.9 23 084 66.6 0.1 33.3 59.8 6.8 22 195 27 419 -5 224 Croatia 19 734 19 734 100.0 17 995 91.2 8.8 75.1 16.1 25 163 27 103 -1 940 Cyprus 1 458 1 458 100.0 1 458 100.0 99.3 0.7 41 404 48 212 -6 808 Czechia 25 867 25 867 100.0 25 867 100.0 94.0 6.0 27 189 30 280 -3 091 Denmark (incl. Faroes) 3 747 3 707 98.9 1.1 98.9 13 280 12 880 400 Estonia 21 080 20 143 95.6 4.4 95.6 10 962 7 871 3 092 Finland 211 668 3 040 1.4 98.6 1.4 6 079 5 742 337 France 143 376 142 897 99.7 74 278 51.8 0.3 47.9 37.7 14.0 0.1 21 897 25 651 -3 754 Germany 108 031 108 030 100.0 84 900 78.6 0.0 21.4 71.5 7.1 24 096 27 670 -3 574 Greece 26 122 26 122 100.0 25 070 96.0 4.0 53.0 41.6 1.4 29 647 37 853 -8 206 Hungary 17 407 17 407 100.0 15 474 88.9 11.1 86.9 2.0 24 260 29 099 -4 839 Iceland 537 3 0.5 99.5 0.5 4 215 4 778 -563 Ireland 4 510 11 0.2 99.8 0.2 4 857 6 611 -1 754 Italy 79 052 79 052 100.0 77 836 98.5 1.5 41.2 51.9 5.4 33 204 37 591 -4 388 Latvia 24 261 24 261 100.0 100.0 12 772 8 558 4 214 Liechtenstein 79 79 100.0 79 99.4 0.6 34.5 65.0 31 505 35 937 -4 432 Lithuania 19 455 19 455 100.0 100.0 15 546 10 993 4 553 Luxembourg 937 937 100.0 937 100.0 100.0 22 415 25 465 -3 050 Malta 2 2 100.0 2 100.0 100.0 41 655 43 548 -1 893 Monaco 1 1 100.0 1 100.0 100.0 52 516 45 236 7 280 Montenegro 5 777 5 780 100.0 5 513 95.4 4.6 75.1 20.3 26 526 30 080 -3 555 Netherlands 3 118 3 115 99.9 137 4.4 0.1 95.5 4.4 17 660 17 794 -134 North Macedonia 8 144 8 144 100.0 8 083 99.3 0.7 68.9 30.4 28 147 34 703 -6 557 Norway 103 494 43 042 41.6 102 0.1 58.4 41.5 0.1 9 742 9 498 244 Poland 96 966 96 966 100.0 54 203 55.9 44.1 55.4 0.5 20 865 21 583 -718 Portugal 16 512 16 368 99.1 8 716 52.8 0.9 46.3 52.7 0.1 19 657 21 956 -2 299 Romania 71 273 71 033 99.7 23 867 33.5 0.3 66.2 33.2 0.3 18 453 20 766 -2 313 San Marino 6 6 100.0 6 100.0 100.0 35 064 35 121 -57 Serbia (incl. Kosovo) 27 211 27 211 100.0 26 286 96.6 3.4 86.4 10.2 25 543 28 656 -3 113 Slovakia 20 484 20 484 100.0 16 016 78.2 21.8 76.0 2.2 23 067 26 719 -3 652 Slovenia 11 441 11 441 100.0 11 331 99.0 1.0 61.3 37.7 28 918 32 708 -3 790 Spain 107 927 101 906 94.4 85 131 78.9 5.6 15.5 44.0 34.9 0.0 25 826 26 200 -374 Sweden 261 757 118 467 45.3 0 0.0 54.7 45.3 0.0 9 382 8 845 537 Switzerland 11 850 11 850 100.0 11 643 98.3 1.7 58.4 39.2 0.7 29 716 35 917 -6 202 Türkiye 114 886 91 315 79.5 60 638 52.8 20.5 26.7 27.8 23.1 1.9 21 685 34 080 -12 395 Total 1 676 742 1 227 868 73.2 724 217 43.2 26.8 30.0 32.0 10.8 0.4 18 143 20 319 -2 176 EU-27 1 373 615 1 009 295 73.5 585 501 42.6 26.5 30.9 32.1 10.1 0.3 17 993 19 439 -1 445 Northern Europe 645 997 232 117 35.9 102 0.0 64.1 35.9 0.0 8 740 7 975 765 Western Europe 104 660 99 683 95.2 43 083 41.2 4.8 54.1 36.6 4.6 19 394 23 109 -3 715 Central Europe 328 793 328 792 100.0 255 649 77.8 0.0 22.2 69.4 8.3 0.0 24 095 26 920 -2 825 Southern Europe 284 577 278 409 97.8 233 680 82.1 2.2 15.7 43.1 37.4 1.7 27 785 30 681 -2 896 South-Eastern Europe 312 715 288 867 92.4 191 703 61.3 7.6 31.1 46.9 13.7 0.7 22 115 28 515 -6 400 Kosovo 4 316 4 316 100.0 4 312 99.9 0.1 59.2 40.7 29 519 33 871 -4 352 Serbia (without Kosovo) 22 894 22 894 100.0 21 974 96.0 4.0 91.5 4.5 24 793 27 673 -2 880 Note: The percentage value “0.0” indicates that an exposed forested area exists, but it is small and estimated to be less than 0.05%. Empty cells mean no forested area in exposure. 5-year mean, i.e. 5-year mean 2018-2022. Diff., i.e. difference concentrations between 2023 and the 5-year mean 2018-2022. 3.5 Ozone – Phytotoxic Ozone Dose (PODY) for crops and forest trees Ozone is generally recognized to be the most relevant pollutant for plants. Visible injury, reduction in growth, changes in biomass partitioning, or a higher susceptibility to pathogen attack can be the effect of ozone influence (Krupa et al., 2000). As mentioned above, scientific evidence suggests that observed effects of ozone on vegetation are more strongly related to the uptake of ozone through the stomatal leaf pores into the leaf interior (stomatal flux) than to the concentration in the atmosphere around the plants (Reich, 1987; Ashmore et al., 2004; Mills et al., 2011). The cumulative stomatal ozone fluxes (Fsto) through the stomata of leaves found at the top of the canopy are calculated over the course of the growing season based on ambient ozone concentration and stomatal conductance (gsto) to ozone. The stomatal conductance has been calculated using a multiplicative stomatal conductance model (Emberson et al., 2000a) based on Jarvis (1976) as a ETC HE Report 2025/5 56 function of species-specific maximum gsto (expressed on a single leaf-area basis), phenology, and prevailing environmental conditions (photosynthetic photon flux density, PPFD), air temperature, vapour pressure deficit (VPD), and soil moisture. PODY (Phytotoxic Ozone Dose) is the accumulated plant uptake (flux) of ozone above a threshold of Y during a specified time or growth period. The flux-based PODY metrics are preferred in risk assessment over the concentration-based AOT40 exposure index. AOT40 accounts for the ambient ozone concentration and is therefore biologically less relevant for ozone impact assessment than PODY as it does not take into account how ozone uptake is affected by climate, soil and plant factors. Several PODY indicators are described in CLRTAP (2024). PODYSPEC is a species or group of speciesspecific PODY that requires comprehensive input data and is suitable for detailed risk assessment. PODYIAM is a vegetation-type specific PODY that requires less input data and is suitable for large-scale modelling, including integrated assessment modelling. PODYSPEC is further used in this report. For crops (wheat, potato and tomato), the Y value is taken equal to 6 nmol/m2 PLA/s (i.e. per unit projected leaf area) as recommended in CLRTAP (2024). For the details of PODY (and specifically POD6SPEC as used in this report) calculation, see Annex 1, Section A1.3. The species-specific flux models and associated response functions and critical levels (CL) for ozonesensitive crops and cultivars can be used to quantify the potential negative impacts of O3 on the security of food supplies at the local and regional scale. They can be used to estimate yield losses, including economic losses. A flux-threshold Y of 6 (POD6SPEC) provides the strongest flux-effect relationships for crops (Pleijel et al., 2007). O3 effects proved to be significant at a 5% reduction of the effect parameter (Mills et al., 2011), hence CLs were determined for this 5% reduction of the effect parameter (i.e. yield, weight or quality of grain, tuber or fruit), based on the slope of the function. The POD6SPEC CLs for crops were determined based on this reduction of relevant yield or weight, as shown in Table 3.9. Wheat, potato and tomato are considered as representative species of crops in Europe (tomato can be regarded as representative horticultural crop for the Mediterranean and Black Sea regions, which is the case of potato for other regions). Therefore, POD6SPEC for these crops (labelled further simply as POD6 for wheat, potato and tomato, respectively) are recommended for regular map construction. This report presents maps of POD6 for wheat (Triticum aestivum), potato (Solanum tuberosum) and tomato (Solanum lycopersicum). Table 3.9 POD6SPEC critical levels for crops as determined by CLRTAP Crop Effect parameter POD6SPEC critical level Wheat grain yield 1.3 mmol/m2 PLA Wheat 1000-grain weight 1.5 mmol/m2 PLA Wheat protein yield 2.0 mmol/m2 PLA Potato tuber yield 3.8 mmol/m2 PLA Tomato fruit yield 2 mmol/m2 PLA Tomato fruit quality 3.8 mmol/m2 PLA Source: CLRTAP, 2024 ETC HE Report 2025/5 57 Regarding trees, beech (Fagus sylvatica) and Norway spruce (Picea abise) were selected as the tree species for which the most comprehensive parameterization for POD is available. For them, the Y value is taken equal to 1 nmol/m2 PLA/s (i.e. per unit projected leaf area). For the details of PODY (and specifically POD1SPEC as used in this report) calculation, see Vlasáková et al. (2023). A uniform O3 flux threshold of Y = 1 nmol/m2 PLA/s was adopted for use in species-specific phytotoxic O3 doses (PODYSPEC) for all tree species at the O3 CLs workshop in Madrid, November 2016 (CLRTAP, 2024), based on data and analyses presented in Büker et al. (2015). Anav et al. (2022) illustrated that POD1 is the most reliable simple estimate of O3 risk and recommended the use of this metric by policy makers as an air quality standard to protect vulnerable forest ecosystems in the future. The POD1SPEC CLs for forest trees were set to values for an acceptable biomass loss, as shown in Table 3.10. Table 3.10 POD1SPEC critical levels for trees as determined by CLRTAP Tree Effect parameter POD1SPEC critical level Beech 4% annual reduction of the whole tree biomass 5.2 mmol/m2 PLA Spruce 2% annual reduction of the whole tree biomass 9.2 mmol/m2 PLA Source: CLRTAP, 2024 The POD maps have been calculated based on the hourly ozone concentration maps, together with the meteorological and soil hydraulic properties data, based on the methodology described in Annex 1, Section A1.3. The ozone maps for each hour of the year 2023 have been constructed using the same methodology as the annual maps, i.e. the multiple linear regression followed by the kriging of its residuals (see Annex 1, Section A1.1) based on the measurement data, chemical transport model (CAMS-Ensemble forecast) output, altitude and the surface solar radiation. For details, see Annex 3, Section A3.3. The ozone maps have been calculated at the 2 km resolution, while the complete POD calculation has been executed in 0.1° resolution. The hourly ozone maps are created for rural areas only, based on rural background stations. The POD maps are therefore applicable to rural areas only. Next to this, it should be noted that in the POD calculations for wheat and potato, all growing areas are considered rain-fed (i.e. without irrigation), see Colette et al. (2018). Thus, the maps are directly applicable only for areas without irrigation. If applied for irrigated areas, the POD values for wheat and potato might be somewhat underestimated. On the other hand, no limitation of stomatal conductance due to soil moisture can be assumed for tomato, since it is an irrigated horticultural crop (see Annex 1, Section A1.3). 3.5.1 Phytotoxic Ozone Dose maps Maps 3.10 to 3.12 present the maps of Phytotoxic Ozone Dose (POD6) for wheat, potato and tomato and maps 3.13 and 3.14 present the maps of POD1 for beech and spruce in 2023. Generally, high values of the POD can be found in different parts of Europe since the POD is dependent not only on ozone levels but also on the environmental conditions and plant phenology. The lowest levels of the POD usually occur in areas with lower ozone concentrations (e.g. northern European regions) and/or in areas where environmental conditions limit the ozone stomatal conductance (dry and warm areas, including parts of the southern, south-western and south-eastern Europe). On the other hand, higher POD values can occur in areas with lower ozone concentrations, but favourable conditions for the stomatal conductance. Crops The areas in the Map 3.10 with POD6 values below the lowest CL for wheat (i.e. 1.3 mmol/m2 PLA for grain yield) are marked in dark green and green. The areas with POD6 values in between CLs for grain ETC HE Report 2025/5 64 Map 4.2 presents the difference between 2023 and the 5-year mean 2018-2022 for NO2 annual average. Orange to red areas show an increase of NO2 concentration in 2023, while blue areas show a decrease. At the annual average NO2 difference map the highest increases are observed in Türkiye and Greece. There are increases also in parts of southern and south-eastern Europe. A smaller increase is also observed in northern Europe, specifically in disconnected areas of Scandinavia and in the Baltic states. On the other hand, there are decreases in several countries, including the Benelux region, Germany, Switzerland, northern Italy, the Île-de-France region, Romania and eastern Türkiye. Map 4.2 Difference concentrations between 2023 and the 5-year mean 2018-2022 for NO2 annual average 4.1.2 Population exposure Table 4.1 and Figure 4.1 give the population frequency distribution for a limited number of exposure classes calculated on a grid of 1 km resolution for individual countries, large regions, EU-27 and for the whole mapping area. Table 4.1 also presents the population-weighted concentration. It has been estimated that in 2023, 2.6% of the considered European population including Türkiye and 0.2% of the EU-27 population lived in areas with NO2 annual average concentrations above the annual EU limit value of 40 µg/m3. About 67% of the considered European population including Türkiye and 63% of the EU-27 population has been exposed to annual average concentrations above the WHO AQG level of 10 µg/m3 (WHO, 2021a). No population has been exposed to concentrations above the ALV in 35 countries. In France, Italy, Romania, Greece up to 4% of population has been exposed to concentrations above the limit value. In Cyprus and Türkiye, this is the case for 8% and 19% of the population, respectively. ETC HE Report 2025/5 65 Table 4.1 Population exposure and population-weighted concentration, NO2 annual average, 2023 Country ISO Population (inhbs·1000) NO2 — annual average, exposed population (%) Population-weighted concentration < 10 10-20 20-30 30-40 40-45 > 45 2023 5-year mean Diff. Albania AL 2 762 52.9 33.0 13.5 0.6 11.7 13.6 -1.9 Andorra AD 80 27.5 72.5 12.5 17.0 -4.5 Austria AT 9 105 32.9 52.7 14.3 0.2 13.1 15.4 -2.2 Belgium BE 11 743 25.9 66.4 7.7 0.0 12.7 16.9 -4.2 Bosnia and Herzegovina BA 3 194 25.3 61.3 13.5 14.1 13.8 0.3 Bulgaria BG 6 448 24.5 57.4 16.3 1.8 14.5 17.5 -3.0 Croatia HR 3 851 36.5 51.0 12.0 0.6 12.4 13.4 -1.0 Cyprus CY 1 419 10.7 8.7 72.5 0.8 7.3 24.6 22.4 2.2 Czechia CZ 10 828 40.8 55.9 3.3 11.5 13.6 -2.1 Denmark (incl. Faroe Islands) DK 5 987 88.9 11.0 0.1 6.4 8.2 -1.8 Estonia EE 1 366 80.6 19.4 6.6 6.9 -0.3 Finland FI 5 564 84.2 15.8 6.9 7.6 -0.7 France (metropolitan) FR 66 017 53.2 36.6 8.0 2.2 0.0 11.1 13.7 -2.7 Germany DE 83 119 34.0 58.4 7.2 0.4 12.4 16.2 -3.8 Greece GR 10 414 35.0 27.8 21.0 12.3 0.9 3.0 17.4 18.4 -1.0 Hungary HU 9 600 31.8 55.4 11.2 1.6 12.9 15.7 -2.8 Iceland IS 388 86.4 13.3 0.3 7.2 8.7 -1.5 Ireland IE 5 271 67.5 30.2 2.2 0.1 8.0 9.3 -1.3 Italy IT 58 997 17.2 54.3 23.5 4.8 0.2 0.0 16.7 18.7 -2.0 Latvia LV 1 883 62.1 36.5 1.4 9.0 10.4 -1.4 Liechtenstein LI 40 6.9 91.9 1.2 12.2 15.3 -3.2 Lithuania LT 2 857 55.5 41.1 3.5 9.9 11.0 -1.1 Luxembourg LU 661 44.8 53.8 1.4 10.8 16.4 -5.6 Malta MT 542 45.7 47.3 7.0 10.8 10.9 -0.1 Monaco MC 39 80.5 19.5 16.5 20.5 -3.9 Montenegro ME 617 20.7 75.3 4.0 12.9 13.4 -0.5 Netherlands NL 17 811 16.7 78.4 4.9 13.5 17.4 -3.8 North Macedonia MK 1 830 50.5 44.6 3.4 1.5 10.8 16.2 -5.4 Norway NO 5 489 72.9 25.4 1.7 7.4 8.6 -1.2 Poland PL 36 754 43.4 49.4 6.7 0.4 11.8 13.9 -2.1 Portugal (excl. Azores, Madeira) PT 10 022 42.1 49.1 8.0 0.8 11.7 13.1 -1.4 Romania RO 19 055 27.5 53.0 14.3 4.4 0.5 0.3 14.9 17.9 -3.0 San Marino SM 34 30.1 65.2 4.7 11.0 14.0 -3.0 Serbia (incl. Kosovo) RS 8 350 22.2 63.1 14.2 0.4 14.3 16.2 -2.0 Slovakia SK 5 429 40.8 56.9 2.2 11.1 12.7 -1.6 Slovenia SI 2 117 40.5 56.3 3.1 11.4 13.4 -2.1 Spain (excl. Canarias) ES 45 872 30.9 45.1 18.9 5.1 15.0 16.6 -1.7 Sweden SE 10 522 88.4 11.6 6.3 7.3 -1.0 Switzerland CH 8 815 22.4 69.3 8.3 12.8 15.4 -2.6 Türkiye TR 85 280 8.7 11.6 30.4 30.0 6.3 13.0 30.7 26.4 4.3 Total 560 170 33.1 43.8 13.9 6.2 1.0 2.0 15.5 16.9 -1.4 73.1 2.6 EU-27 443 198 37.4 49.0 11.2 2.2 0.1 0.1 12.9 15.4 -2.5 81.4 0.2 Northern Europe 34 055 80.6 18.7 0.7 7.1 8.3 -1.2 Western Europe 86 334 43.3 48.1 7.0 1.6 0.0 11.7 15.0 -3.3 Central Europe 165 805 36.1 56.2 7.3 0.4 12.2 15.2 -3.0 Southern Europe 142 589 28.1 47.4 19.5 4.6 0.1 0.3 15.3 17.0 -1.7 South-Eastern Europe 131 386 16.3 27.6 24.2 19.7 4.1 8.2 24.5 22.7 1.8 Kosovo XK 1 709 30.5 69.1 0.4 11.6 15.4 -3.9 Serbia (excl. Kosovo) RS 6 641 20.0 61.6 17.8 0.6 14.9 16.4 -1.5 Note: The percentage value "0.0" indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. 5-year mean, i.e. 5-year mean 2018-2022. Diff., i.e. difference concentrations between 2023 and the 5-year mean 2018-2022. ETC HE Report 2025/5 66 The population-weighted concentration of the NO2 annual mean has been estimated for 2023 at 15.5 µg/m3 for total mapped area and 12.9 µg/m3 for the EU-27, which means a decrease of 1.4 µg/m3 and 2.5 µg/m3 compared to the previous 5-year mean, respectively. When assessing the change in individual countries, the steepest absolute decrease was found in Luxembourg (5.6 µg/m3), and the highest increase was estimated in Türkiye (4.3 µg/m3). Figure 4.1 Percentage of the population (%) exposed to different values of NO2 annual average (µg/m3), 2023 Figure 4.2 shows, for the whole mapped area, the population frequency distribution for exposure classes of 1 µg/m3. One can see the highest population frequency for classes between 4 and 22 µg/m3, a continuous decline of population frequency for classes between 22 and 40 µg/m3 and continuous mild decline of population frequency for classes between 41 and 75 µg/m3. Figure 4.2 Population exposure frequency distribution, NO2 annual average, 2023. The WHO AQG level (10 µg/m3) is marked by the green line, the 2030 EU limit value (20 µg/m3) is marked by the yellow line, the EU annual limit value (40 µg/m3) is marked by the red line Note: Apart from the population distribution shown in the graph, it was estimated that 0.13% of population lived in areas with PM2.5 annual average concentrations between 75 and 95 µg/m3. ETC HE Report 2025/5 67 The boxplot showing for individual countries the NO2 annual average concentrations to which the population per country was exposed in 2023 is presented in Figure 4.3. Figure 4.3 NO2 annual average concentrations to which the population per country was exposed in 2023. The WHO AQG level (10 µg/m3) is marked by the green line, the 2030 EU limit value (20 µg/m3) is marked by the yellow line, the 2008 EU limit value (40 µg/m3) is marked by the red line Note: For each country, the box plot shows the concentration to which a percentage of the population was exposed: 50% in the case of the black marker, 25% and 75% in the cases of the box’s edges, 2% and 98% in the cases of the whiskers’ edges. 4.2 NOx – Annual mean 4.2.1 Concentration map The Ambient Air Quality Directive (EC, 2008) sets a critical level (CL) for the protection of vegetation for the NOx annual mean at 30 µg/m3 (and this standard remains unchanged under EU, 2024). According to these directives, the sampling points targeted at the protection of vegetation and natural ecosystems shall be in general sited more than 20 km away from agglomerations or more than 5 km away from other built-up areas. Thus, only the observations at rural background stations are used for the NOx mapping and the resulting map is representative for rural areas only. The number of NOx measurement stations is limited. The mapping of the NOx annual average has been therefore performed on the basis of an approach presented in Horálek et al. (2007). This approach derives additional pseudo NOx annual mean concentrations from NO2 annual mean measurement concentrations and increases as such the number and spatial coverage of NOx ‘data points’, and applies these data to the NOx mapping. Section A1.1 of Annex 1 provides some details. Map 4.3 presents the concentration map of NOx annual average. It concerns rural areas only, representing an indicator for vegetation exposure to NOx. Most of the European area shows NOx levels below 20 µg/m3. However, in the Po Valley, parts of Türkiye and around some larger European cities (Athens and Barcelona) NOx concentrations above the ETC HE Report 2025/5 68 CL are observed. These concentrations are expected to be the result of large emissions from transport in and around the cities, as well as energy production and industrial facilities taking place at these areas. These values above the CL would be relevant only for the so-called peri-urban vegetation where patches of agricultural land and of natural or planted vegetation can be found. On the contrary, low concentrations (below 10 µg/m³) are observed across large areas in most of the mapped countries, with the exception of more extensive regions in Greece, Bulgaria, Romania, the Po Valley, the Benelux countries, Türkiye, and several smaller, discontinuous areas in Central Europe and on the Iberian Peninsula. Map 4.3 Concentration map of NOx annual average, rural map, 2023 The relative mean uncertainty of this NOx rural map is 43%. This means higher mapping uncertainty compared to the quality objective for models of NOx annual average (i.e. 30%) as set in the Air Quality Directive (EC, 2008). This higher relative uncertainty is strongly influenced by the low concentration values of NOx in most of the areas. The NOx annual average rural map including the data measured at rural background stations is presented in Map A4.11 of Annex 4. The map illustrates the lack of the NOx rural stations in the Balkan area. Map 4.4 presents the difference between 2023 and the 5-year mean 2018-2022 for annual average for NOX. Orange to red areas show an increase of NOX concentration in 2023, while blue areas show a decrease. The highest increases are observed in Türkiye, with increases also in the eastern regions of Bulgaria and Romania. Notable decreases are primarily observed in the Île-de-France region and the Benelux countries, the Po Valley, parts of Germany, and southern Türkiye. Additional areas showing a widespread decline in NOX concentrations include Hungary and large parts of the Balkan states. A continuous area of decrease is also observed in Germany and in several smaller regions across Central and Western Europe. Vegetation exposure has not been calculated for NOx, since values above the CL for protection of the vegetation would occur in limited vegetation areas only and, as such, is considered not to provide essential information from the European scale perspective. Furthermore, contrary to vegetation exposure to high ozone concentrations in Europe that leads to considerable damage, vegetation ETC HE Report 2025/5 69 exposure to NOx pollution is of minor importance in terms of actual impacts. On the other hand, NOx concentrations contribute in part to the total N-deposition, which leads to acidifying and eutrophying effects on vegetation. These effects, especially eutrophication, are still very important in Europe (e.g. EMEP, 2020). However, these effects on vegetation cannot be expressed by an exposure to NOx as many oxidized and reduced nitrogen compounds contribute to total atmospheric nitrogen deposition. Map 4.4 Difference concentrations between 2023 and the 5-year mean 2018-2022 for NOx annual average Concerning the potential exposure estimate of vegetation and natural ecosystems to NOx there is an additional dilemma: which receptor types should be selected to estimate the exposure and CL exceedance of vegetation and natural ecosystems? An option would be the use of CLC classes (e.g. like in Horálek et al., 2008, i.e. natural areas); nevertheless, this classification is too general. Another option would be the NATURA 2000 database. However, that data source contains a wide series of receptor types, species and classes. Serious additional efforts would be needed to conclude on the most relevant set of receptors from the NATURA 2000 geographical database. ETC HE Report 2025/5 70 5 Benzo(a)pyrene An annual average map for benzo(a)pyrene (BaP) has been produced and is presented in the regular mapping report for the fourth time. In agreement with the conclusions of Horálek et al. (2022b), it is labelled as an experimental map to indicate that it does not yet meet the same accuracy standards as the regularly produced maps of other pollutants. The map of BaP is based on the mapping methodology developed and tested in Horálek et al. (2022b). The methodology for creating the concentration maps follows the same principle as for the rest of pollutants: a linear regression model based on European wide station measurement data, followed by kriging of the residuals produced from that regression model (residual kriging). The map layers are created for the rural and urban background areas separately on a grid at 1 km resolution. For details, see Annex 1, Section A1.1. Supplementary data used in the linear regression for rural areas consist of chemical transport model (CTM) output, altitude, temperature, wind speed and land cover; for urban background areas they are CTM output and temperature (Annex 3, Section A3.5). The final concentration map is presented at a 1 km grid resolution. Due to the poor spatial coverage of the BaP measurement stations, so-called pseudo-BaP stations have been used in addition. Pseudo-BaP data in locations with PM2.5 measurements (or with pseudo-PM2.5 data based on PM10 measurements) and with no BaP measurements have been estimated based on the exponential regression of the observed BaP concentrations with the PM2.5 data, geographical coordinates and the land cover. Due to quite high uncertainty of the pseudo data, they are only used in areas with a lack of BaP measurements. Due to the serious lack of Turkish data, Türkiye is not included in the mapping area. Annex 3, Section A3.5 provides details on the regression and kriging parameters applied for deriving the BaP map, as well as the uncertainty analysis of this map. The 2004 Ambient Air Quality Directive (EC, 2004) sets a target value (TV) for ambient air concentration of BaP, as a marker for the carcinogenic risk of polycyclic aromatic hydrocarbons (PAHs) in ambient air. The TV for BaP (measured in PM10) is set to 1 ng/m3 as an annual mean. The revised Ambient Air Quality Directive (EU, 2024) introduces the EU limit value to be reached by 2030 (LV2030), which is set to 1.0 ng/m3 for the annual mean. Both the 2030 EU LV (1.0 ng/m3) and the estimated WHO reference level (RL, 0.12 ng/m3) are based on the WHO lung cancer unit risk for PAH mixtures and correspond to an additional lifetime cancer risk of approximately 9 cases and 1 case in 100 000 exposed individuals, respectively (WHO, 2021b). 5.1 Benzo(a)pyrene – Annual mean 5.1.1 Concentration map Map 5.1 presents the final combined 1 km resolution concentration map for the 2023 annual average of BaP. Red and purple areas indicate concentrations above 1.0 ng/m3 (i.e. above the 2030 LV). BaP concentrations above 1.0 ng/m3 are found in Poland, the north-eastern part of Czechia, west Balkan, the eastern Po Valley in northern Italy, and some populated locations in central and southeastern Europe, Latvia and Finland. By contrast, western and southern Europe (except Italy) have low BaP values. Generally, lower levels of BaP concentrations in natural areas can be seen, compared with the other land cover types. The relative mean uncertainty of the 2023 map of BaP annual average is 137% for rural and 67% for urban areas and determined exclusively on the actual BaP measurement data points, i.e. not the pseudo stations (Annex 3, Section A3.5). This uncertainty is considerably high (especially in the rural areas) with respect to the quality objective for models for BaP annual average (i.e. 60%) as set in the ETC HE Report 2025/5 71 European directive (EC, 2004). The high uncertainty in the rural areas is probably strongly affected (besides the low density of the rural stations) by the fact that stations classified as “rural background” comprise both regional stations with low BaP values and stations located in villages, which are often highly influenced by local heating leading to high BaP concentrations. Map 5.1 Concentration map of benzo(a)pyrene annual average, 2023, experimental map Map 5.2 Difference concentrations between 2023 and the five-year mean 2018-2022 benzo(a)pyrene annual average Map 5.2 presents the difference between 2023 and the five-year mean 2018-2022 for BaP annual average. Orange to red areas show an increase of BaP concentration in 2023, while blue areas show a decrease. Areas with the highest decreases have been found in Poland. Areas with the highest increase ETC HE Report 2025/5 72 are shown in the western Balkans, however, this estimated increase can be influenced by a high uncertainty of the map and a low density of BaP stations in this area. 5.1.2 Population exposure Table 5.1 gives the population frequency distribution for a limited number of exposure classes to BaP concentrations, as well as the population-weighted concentration. Due to the experimental character of the BaP map and its high uncertainty, the population exposure is presented only for EU-27, for five European regions and for the total mapping area, not for individual countries. Table 5.1 Population exposure and population-weighted concentration, benzo(a)pyrene annual average, 2023, based on the experimental map Area Population [inhbs·1000] BaP – annual average, exposed population, 2023 [%] Population-weighted concentration < 0.12 0.12-0.4 0.4-0.6 0.6-1.0 1.0-1.5 > 1.5 2023 5-year mean Diff. Northern Europe 34 055 9.6 62.5 16.5 10.6 0.8 0.37 0.53 -0.16 Western Europe 86 334 63.7 36.2 0.0 0.0 0.12 0.15 -0.03 Central Europe 165 805 27.2 35.5 8.0 14.0 8.2 7.2 0.53 0.96 -0.44 Southern Europe 142 589 21.3 63.8 8.9 4.0 1.9 0.1 0.29 0.27 0.02 South-Eastern Europe without Türkiye 46 106 1.8 18.6 15.5 39.9 17.0 7.2 0.84 1.29 -0.45 Total 474 890 28.0 44.3 8.2 10.9 5.2 3.3 0.40 0.62 -0.21 EU-27 443 198 29.7 44.3 8.3 10.7 4.2 2.8 0.38 Note: The percentage value “0.0” indicates that an exposed population exists, but it is small and estimated to be less than 0.05%. Empty cells mean no population in exposure. Based on the experimental map, it is estimated that 8.5% of the population living in the considered European area was exposed to concentrations above 1.0 ng/m3 in 2023 (7% for EU27). Further, it is estimated that about 72% of the population living in the considered European area was exposed to concentrations above the WHO RL of 0.12 ng/m3 (70.3% for EU27). The population-weighted concentration of the BaP annual average for 2023 for the considered European countries is estimated to be 0.4 ng/m3 (0.38 ng/m3 for EU27). Figure 5.1 shows, for the whole mapped area, the population frequency distribution for exposure classes of 0.05 ng/m3. The highest population frequency is found for classes between 0.05 and 0.20 ng/m3. A quite continuous decline of population frequency is visible for classes above 0.30 ng/m3. Figure 5.1 Population frequency distribution, benzo(a)pyrene annual average, 2023. The value 1.0 ng/m3 is marked by the red line Note: Apart from the population distribution shown in the graph, it was estimated that 0.15% of population lived in areas with BaP annual average concentrations between 3.0 and 4.3 ng/m3. ETC HE Report 2025/5 73 6 Accumulated risks Although the spatial distributions of PM, NO2 and ozone concentrations differ widely, the possibility of an accumulation of risk resulting from high exposures to all three pollutants cannot be excluded. The maps for the three most frequently exceeded EU standards (PM10 daily limit value as presented in Map 2.3, O3 target value as shown in Map 3.1 left and NO2 annual limit value as presented in Map 4.1) have been combined, see Map 6.1. Map 6.1 Exceedance of Health-Related Air Quality Standards, 2023 The combined population exposure shows the following results: out of the total population of 560 million in the considered area, 5% (30.2 million) people live in areas where two or three of these air quality standards are exceeded; 0.1% (0.5 million) people live in areas where all three standards are exceeded (in some urban areas affected by traffics, not visible in the 1 km resolution map). The worst situation in 2023 was observed in Greece and Türkiye, where 4% and 0.2% lived in areas where all three standards are exceeded, respectively. ETC HE Report 2025/5 80 The trend analysis of the forest-weighted concentrations for the AOT40 for forests across the period 2005-2023 for the total considered mapping area shows no trend. The evolution of the population-weighted and the agricultural-weighted concentrations for ozone indicators SOMO35 and AOT40 for vegetation is shown in Figure 7.5. Figure 7.5 Population-weighted concentration of ozone indicator SOMO35 (left) and agriculturalweighted concentration of ozone indicator AOT40 for vegetation (right) in 2005-2023 7.3 Human health NO2 indicator Table 7.5 summarises the development in exposure levels of the considered European population for the human health NO2 indicator (annual average), in terms of population-weighted concentrations and percentage of population exposed to concentrations above the 2008 annual LV (40 µg/m3) and the 2030 annual LV (20 µg/m3), for the years 2005, 2009, 2010 and 2013 to 2023, for which the maps based on the current methodology are available. The population-weighted concentration is presented additionally also for 2007, although based on different mapping methodology than the other years. This 2007 value is probably slightly underestimated; based on Horálek et al. (2017b), one can suppose the true value would be of about 1 percentage point higher (i.e. it would be about 23.5 µg/m3). Table 7.5 Population-weighted concentration and percentage of the considered European population (including United Kingdom, without Türkiye) exposed to concentrations above the 2008 NO2 limit value (LV) of 40 µg/m3 and the 2030 NO2 limit value (LV2030) of 20 µg/m3 for the protection of health for 2005 to 2023 NO2#### 2006 #### 2008 #### #### #### #### #### #### #### #### #### #### #### #### #### #### #### Pop.-weighted conc. [μg/m3]23.3 23.3 22.1 22.1 19.4 18.6 18.8 18.6 18.4 17.6 16.8 14.0 14.4 14.4 12.9 Pop. exp. > LV (40 μg/m3) [%] 7.9 5.6 4.9 3.2 2.8 3.2 2.8 3.0 1.8 1.3 0.2 0.2 0.2 0.1 Pop. exp. > LV2030 (20 μg/m3) [%] 57.1 53.8 55.0 41.5 39.4 38.1 38.5 37.2 33.1 29.5 16.0 17.5 17.8 13.0 Annual average not mappe d not mappe d not mapped In 2023 the fractions of the population exposed to NO2 annual mean concentrations above the 2008 limit value of 40 µg/m3 and above the 2030 limit value of 20 µg/m3 have been 0.1% and 13% of the total population, respectively, which are the lowest values in the whole series. Furthermore, it is estimated that the considered European inhabitants have been exposed on average to an annual mean NO2 concentration of 13 µg/m3, again the lowest in the whole series. Trend analysis on the population-weighted concentration for the total mapping area shows a downward trend of about -0.7 µg/m3 per year, for the period 2005-2023, which is statistically significant (at the strongest level 0.001). Figure 7.6 presents the NO2 population exposure for six concentration classes, for 2012-2023. ETC HE Report 2025/5 81 Figure 7.6 Population exposure, NO2 annual average, 2012-2023 The evolution of the population-weighted concentrations for NO2 is shown in Figure 7.7. As the regular NO2 maps are not available for all years, an alternative mapping results prepared based on a subset of stations for the purpose of trend analysis 2005-2019 (Horálek et al., 2022a) are also presented. Figure 7.7 Population-weighted concentration of NO2 annual mean. Results based on both regular maps (red) and maps for trend analysis (blue) are presented, where available 7.4 Human health BaP indicator Table 7.6 summarises for the human health BaP indicator (annual average) the population-weighted concentration and the percentage of the considered European population exposed to BaP concentrations above the 2030 EU limit value (LV2030) of 1.0 ng/m3, for years 2012-2023. Table 7.6 Population-weighted concentration and percentage of the considered European population (including United Kingdom, without Türkiye) exposed to concentrations above the 2030 BaP limit value of 1.0 ng/m3 for the protection of health in 2012-2023 BaP 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Population-weighted concentration [ng/m3]0.84 0.62 0.56 0.54 0.62 0.47 0.38 Population exposed > LV2030 (1.0 ng/m3) [%] 20.5 16.4 15.1 13.9 15.5 11.9 7.5 Annual average not mapped In 2023 the population exposed to BaP annual mean concentrations above the level of 1.0 ng/m3 has been 7.5% of the total population, which is the lowest value in the whole series. Trend analysis on the population-weighted concentration for the period 2012-2023 shows a downward trend of cc. -0.04 ng/m3 per year, which is statistically significant (at a level 0.05). However, one should have in mind both short period of the time series and the experimental character of the BaP maps. ETC HE Report 2025/5 82 List of abbreviations Abbreviation Name Reference ALV Annual Limit Value AOT40 Accumulated Ozone exposure over a Threshold of 40 ppb (i.e. 80 µg/m³) in a specific period http://eurlex.europa.eu/LexUriServ/L exUriServ.do?uri=OJ:L:200 8:152:0001:0044:EN:PDF AQ Air Quality BaP Benzo(a)pyrene CL Critical Level https://icpvegetation.ceh.a c.uk/chapter-3-mappingcritical-levels-vegetation CLC CORINE Land Cover https://land.copernicus.eu /pan-european/corineland-cover CLRTAP Convention on Long-range Transboundary Air Pollution (Air Convention) https://unece.org/environ ment-policy/air CORINE Co-ORdinated INformation on the Environment https://land.copernicus.eu /pan-european/corineland-cover CTM Chemical Transport Model Defra United Kingdom Department for Environment Food & Rural Affairs DLV Daily Limit Value ECMWF European Centre for Medium-Range Weather Forecasts https://www.ecmwf.int/ EBAS EMEP dataBASe https://ebas.nilu.no/ EEA European Environment Agency www.eea.europa.eu EMEP European Monitoring and Evaluation Programme https://www.emep.int/ ETC/ACM European Topic Centre on Air pollution and Climate change Mitigation https://www.eionet.europ a.eu/etcs ETC/ATNI European Topic Centre on Air pollution, Noise, Transport and Industrial pollution https://www.eionet.europ a.eu/etcs ETC HE European Topic Centre on Human Health and the Environment https://www.eionet.europ a.eu/etcs EU European Union https://europeanunion.europa.eu GMTED Global Multi-resolution Terrain Elevation Data GRIP Global Roads Inventory Dataset HLV Hourly Limit Value ICP International scientific Cooperative Programme https://icpvegetation.ceh.a c.uk/ ILV Indicative Limit Value JRC Joint Research Centre https://ec.europa.eu/info/ departments/jointresearch-centre_en ETC HE Report 2025/5 83 Abbreviation Name Reference LV Limit Value http://eurlex.europa.eu/LexUriServ/L exUriServ.do?uri=OJ:L:200 8:152:0001:0044:EN:PDF LV2030 New or revised 2030 EU Limit Value NILU Norwegian Institute for Air Research https://www.nilu.no/ NO2 Nitrogen dioxide NOx Nitrogen oxides O3 Ozone ORNL Oak Ridge National Laboratory https://www.ornl.gov/ PLA Projected Leaf Area https://icpvegetation.ceh.a c.uk/chapter-3-mappingcritical-levels-vegetation PM10 Particulate matter with a diameter of 10 micrometres or less PM2.5 Particulate matter with a diameter of 2.5 micrometres or less POD6 Phytotoxic Ozone Doze above a threshold of 6 nmol/m2 PLA/s https://icpvegetation.ceh.a c.uk/chapter-3-mappingcritical-levels-vegetation POD1 Phytotoxic Ozone Doze above a threshold of 1 nmol/m2 PLA/s https://icpvegetation.ceh.a c.uk/chapter-3-mappingcritical-levels-vegetation R2 Coefficient of determination RIMM Regression – Interpolation – Merging Mapping RMSE Root Mean Square Error SOMO10 Sum of Ozone Maximum daily 8-hour means Over 10 ppb (i.e. 20 µg/m3) SOMO35 Sum of Ozone Maximum daily 8-hour means Over 35 ppb (i.e. 70 µg/m3) TV Target Value http://eurlex.europa.eu/LexUriServ/L exUriServ.do?uri=OJ:L:200 8:152:0001:0044:EN:PDF UN United Nations https://www.un.org UNECE United Nations Economic Commission for Europe https://unece.org/ UTC Coordinated Universal Time WHO World Health Organization https://www.who.int/ ETC HE Report 2025/5 84 References Alonso, R., et al., 2014, ‘Drought stress does not protect Quercus ilex L. from ozone effects: results from a comparative study of two subspecies differing in ozone sensitivity’, Plant Biology 16, pp. 375384 (https://onlinelibrary.wiley.com/doi/10.1111/plb.12073) accessed 21 January 2021. Anav, A., et al., 2022, 'Legislative and functional aspects of different metrics used for ozone risk assessment to forests', Environmental Pollution 295(118690) (https://doi.org/10.1016/j.envpol.2021.118690) accessed 11 July 2022. Ashmore, M., et al., H., 2004, ‘New directions: a new generation of ozone critical levels for the protection of vegetation in Europe’, Atmospheric Environment 38, pp. 2213-2214 (https://doi.org/10.1016/j.atmosenv.2004.02.029) accessed 20 November 2020. Büker, P., et al. , 2015, 'New flux based dose-response relationships for ozone for European forest tree species', Environmental Pollution 206, pp. 163-174 (https://doi.org/10.1016/j.envpol.2015.06.033) accessed 27 April 2022. CAMS, 2024, CAMS European air quality forecasts, ENSEMBLE data. Copernicus Atmosphere Monitoring Service (CAMS) Atmosphere Data Store (ADS) (https://ads.atmosphere.copernicus.eu/datasets/cams-europe-air-quality-forecasts?tab=overview) accessed on 21 October 2024. CEIP, 2025, Inventory Review 2025. Review of emission data reported under the LRTAP Convention and NEC Directive, Stage 1, 2 and 3 review, Status of gridded and LPS data, Technical Report CEIP 03/2025. (In preparation) CLRTAP, 2016, Forest condition in Europe, 2016 Technical Report of ICP Forests, UNECE Convention on Long-range Transboundary Air Pollution (https://www.icp-forests.org/pdf/TR2016.pdf) accessed 19 November 2020. CLRTAP, 2017, Scientific Background Document A of Chapter 3 of "Manual on methodologies and criteria for modelling and mapping critical loads and levels of air pollution effects, risks and trends" (https://icpvegetation.ceh.ac.uk/sites/default/files/ScientificBackgroundDocumentAOct2018.pdf) accessed 11 December 2020. CLRTAP, 2020, Scientific Background Document B of Chapter 3 of “Manual on methodologies and criteria for modelling and mapping critical loads and levels of air pollution effects, risks and trends” (https://icpvegetation.ceh.ac.uk/sites/default/files/Scientific%20Background%20document%20B%20 June%202020.pdf) accessed 11 December 2020. CLRTAP, 2024, Manual on Methodologies and Criteria for Modelling and Mapping Critical Loads and Levels and Air Pollution Effects, Risks, and Trends. Update 2024. Chapter III: "Mapping Critical levels for Vegetation and Lichens", UNECE Convention on Long-range Transboundary Air Pollution (https://www.umweltbundesamt.de/sites/default/files/medien/11850/publikationen/123_2024_tex te_manual_on_methodologies_and_criteria.pdf) accessed 18 July 2025). Colette, A., et al., 2018, Long term evolution of the impacts of ozone air pollution on agricultural yields in Europe. A modelling analysis for the 1990-2010 period, Eionet Report ETC/ACM 2018/15 (https://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/eionet_rep_etcacm_2018_15_o3impacttrends) accessed 26 August 2020. ETC HE Report 2025/5 85 Cressie, N., 1993, Statistics for spatial data, Wiley series, New York. Danielson, J. J. and Gesch, D. B., 2011, Global multi-resolution terrain elevation data 2010 (GMTED2010), U.S. Geological Survey Open-File Report, pp. 2011-1073 (https://pubs.er.usgs.gov/publication/ofr20111073) accessed 19 November 2020. Defra, 2025, UK Air information resource, Data archive, UK Department for Environment Food & Rural Affairs (https://uk-air.defra.gov.uk/data/). Data extracted in March 2025. De Leeuw, F., 2012, AirBase: a valuable tool in air quality assessments at a European and local level, ETC/ACM Technical Paper 2012/4 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacm_tp_2012_4_airbase_aqassessment) accessed 26 August 2020. Denby, B., et al., 2008, ‘Comparison of two data assimilation methods for assessing PM10 exceedances on the European scale’, Atmospheric Environment 42, pp. 7122-7134 (https://doi.org/10.1016/j.atmosenv.2008.05.058) accessed 26 August 2020. Denby, B., et al., 2011, Mapping annual mean PM2.5 concentrations in Europe: application of pseudo PM2.5 station data, ETC/ACM Technical Paper 2011/5 (http://www.eionet.europa.eu/etcs/etcatni/products/etc-atni-reports/etcacm_tp_2011_5_spatialpm2-5mapping) accessed 26 August 2020. De Smet, P., et al., 2011, European air quality maps of ozone and PM10 for 2008 and their uncertainty analysis, ETC/ACC Technical Paper 2010/10 (http://www.eionet.europa.eu/etcs/etcatni/products/etc-atni-reports/etcacc_tp_2010_10_spataqmaps_2008) accessed 26 August 2020. Deumier, J. M. and Hannon, C., 2010, ‘La période d’ initiation de la tubérisation: comment la repérer?’ (in French), CNIPT (http://www.cnipt.fr/wp-content/uploads/2013/10/Irrigation-juin2010.pdf) accessed 8 February 2021. EC, 2002, Directive 2002/3/EC of the European Parliament and of the Council of 12 February 2002 relating to ozone in ambient air, OJ L 67, 9.3.2002, p. 14-30 (https://eur-lex.europa.eu/legalcontent/EN/TXT/PDF/?uri=CELEX:32002L0003) accessed 6 November 2024. EC, 2004, Directive 2004/107/EC of the European Parliament and of the Council of 15 December 2004 relating to arsenic, cadmium, mercury, nickel and polycyclic aromatic hydrocarbons in ambient air, OJ L 23, 26.1.2005, p. 3-16 (https://eur-lex.europa.eu/legalcontent/EN/TXT/PDF/?uri=CELEX:32004L0107&from=EN) accessed 10 June 2022. EC, 2008, Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe, OJ L 152, 11.06.2008, p. 1-44 (http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:EN:PDF) accessed 26 May 2021. EC, 2024. ‘Fruit and Vegetable Production Dashboard‘, (https://agridata.ec.europa.eu/extensions/DashboardFruitAndVeg/FruitandVegetableProduction.ht ml) accessed 10 June 2025. ECMWF, 2024, CAMS Regional: European air quality analysis and forecast data documentation. WWW: (https://confluence.ecmwf.int/display/CKB/CAMS+Regional%3A+European+air+quality+analysis+and +forecast+data+documentation) accessed 1 October 2024. ETC HE Report 2025/5 86 ECMWF, 2025, Operational archive (https://www.ecmwf.int/en/forecasts/dataset/operationalarchive) accessed 5 September 2025. EEA, 2018, Guide for EEA map layout. EEA operational guidelines, January 2015, version 5 (https://www.eionet.europa.eu/gis/docs/GISguide_v5_EEA_Layout_for_map_production.pdf) accessed 26 August 2020. EEA, 2024, Harm to human health from air pollution in Europe: burden of disease status, EEA Briefing Published 10 December 2024 (https://www.eea.europa.eu/en/analysis/publications/harm-to-humanhealth-from-air-pollution-2024) accessed 19 September 2025. EEA, 2025a, Air Quality e-Reporting. Air quality database (https://www.eea.europa.eu/data-andmaps/data/aqereporting-8). Data extracted in February 2025. EEA, 2025b, Exposure of Europe’s ecosystems to ozone (https://www.eea.europa.eu/en/analysis/indicators/exposure-of-europes-ecosystems-to-ozone) accessed 27 August 2025. Emberson, L. D., et al., 2000a, ‘Modelling stomatal ozone flux across Europe’, Environmental Pollution 109, pp. 403-413 (https://doi.org/10.1016/S0269-7491(00)00043-9) accessed 26 May 2021. Emberson, L. D., et al., 2000b, Towards a model of ozone deposition and stomatal uptake over Europe, Norwegian Meteorological Institute, Research Note No. 42 (https://emep.int/publ/reports/2000/dnmi_note_6_2000.pdf) accessed 24 April 2025. EMEP, 2020, Transboundary particular matter, photo-oxidants, acidifying and eutrophying components, EMEP Report 1/2020 (https://emep.int/publ/reports/2020/EMEP_Status_Report_1_2020.pdf) accessed 22 January 2021. ESA, 2019, Land cover classification gridded maps from 1992 to present derived from satellite observations (https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-land-cover) accessed 17 February 2021. EU, 2020, Corine land cover 2018 (CLC2018) raster data, 100x100m2 gridded version 2020_20 (https://land.copernicus.eu/pan-european/corine-land-cover/clc2018) accessed 19 November 2020. EU, 2024, Directive (EU) 2024/2881 of the European Parliament and of the Council of 23 October 2024 on ambient air quality and cleaner air for Europe (recast), OJ L, 20.11.2024, p. 1-70 (https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L_202402881) accessed 2 September 2025. Eurostat, 2020, JRC-GEOSTAT 2018 grid dataset, Population distribution dataset (https://ec.europa.eu/eurostat/web/gisco/geodata/reference-data/population-distributiondemography/geostat) accessed 12 April 2024. Eurostat, 2025, Population on 1 January by age and sex (https://ec.europa.eu/eurostat/databrowser/view/DEMO_PJAN/default/table) accessed 21 March 2025. ETC HE Report 2025/5 87 Gilbert, R. O., 1987, Statistical Methods for Environmental Pollution Monitoring, Van Nostrand Reinhold, New York. Haberle, J. and Svoboda, P., 2015, ‘Calculation of available water supply in crop root zone and the water balance of crop’s, Contributions to Geophysics and Geodesy 45, pp. 285-298 (https://www.researchgate.net/publication/293194127_Calculation_of_available_water_supply_in_ crop_root_zone_and_the_water_balance_of_crops) accessed 21 January 2021. Horálek, J., et al., 2007, Spatial mapping of air quality for European scale assessment, ETC/ACC Technical paper 2006/6 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacc_technpaper_2006_6_spat_aq) accessed 26 August 2020. Horálek, J., et al., 2008, European air quality maps for 2005 including uncertainty analysis, ETC/ACC Technical paper 2007/7 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacc_tp_2007_7_spataqmaps_ann_interpol) accessed 26 August 2020. Horálek, J., et al., 2010, Methodological improvements on interpolating European air quality maps, ETC/ACC Technical Paper 2009/16 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacc_tp_2009_16_improv_spataqmapping) accessed 26 August 2020. Horálek, J., et al., 2016, Application of FAIRMODE Delta tool to evaluate interpolated European air quality maps for 2012, ETC/ACM Technical Paper 2015/2 (http://www.eionet.europa.eu/etcs/etcatni/products/etc-atni-reports/etcacm_tp_2015_2_delta_evaluation_aqmaps2012) accessed 26 August 2020. Horálek, J., et al., 2017a, European air quality maps for 2014, ETC/ACM Technical Paper 2016/6 (https://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacm_tp_2016_6_aqmaps2014) accessed 30 September 2022. Horálek, J., et al., 2017b, Inclusion of land cover and traffic data in NO2 mapping methodology, ETC/ACM Technical Paper 2016/12 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atnireports/etcacm_tp_2016_12_lc_and_traffic_data_in_no2_mapping) accessed 27 May 2021. Horálek, J., et al., 2018, Satellite data inclusion and kernel based potential improvements in NO2 mapping, ETC/ACM Technical Paper 2017/14 (www.eionet.europa.eu/etcs/etc-atni/products/etcatni-reports/etcacm_tp_2017_14_improved_aq_no2mapping) accessed 27 May 2021. Horálek, J., et al., 2019, Land cover and traffic data inclusion in PM mapping, Eionet Report ETC/ACM 2018/18 (http://www.eionet.europa.eu/etcs/etc-atni/products/etc-atni-reports/etc-acm-report-182018-land-cover-and-traffic-data-inclusion-in-pm-mapping) accessed 27 May 2021. Horálek, J., et al., 2022a, Air quality evolution and trends in Europe in 2005-2019 based on spatial maps, Eionet Report ETC/ATNI 2021/11 (https://doi.org/10.5281/zenodo.6586861) accessed 15 October 2024. Horálek, J., et al., 2022b, Benzo(a)pyrene (BaP) annual mapping, Eionet Report ETC/ATNI 2021/18 (https://doi.org/10.5281/zenodo.5898376) accessed 26 August 2022. Horálek, J., et al., 2024, Air quality maps of EEA member and cooperating countries for 2022, Eionet Report ETC HE 2023/3 (https://doi.org/10.5281/zenodo.14639622) accessed 23 June 2025. ETC HE Report 2025/5 88 ICP Vegetation, 2022. ‘Review of NOx Critical Levels – Minutes of the First Workshop’, (https://icpvegetation.ceh.ac.uk/sites/default/files/NOx_critical_levels_workshop_MINUTES_FINAL.p df) accessed 23 June 2025. Jarvis, P. G., 1976, ‘The interpretation of the variation in leaf water potential and stomatal conductance found in canopies in the field’, Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 273, pp. 593-610 (https://doi.org/10.1098/rstb.1976.0035) accessed 26 May 2021. JRC, 2016, Maps of indicators of soil hydraulic properties for Europe, dataset/maps downloaded from the European Soil Data Centre (http://esdac.jrc.ec.europa.eu/content/maps-indicators-soilhydraulic-properties-europe) accessed 8 December 2020. JRC, 2023, GHS-POP R2023A - GHS population grid multitemporal (1975-2030) (https://doi.org/10.2905/2FF68A52-5B5B-4A22-8F40-C41DA8332CFE) accessed 12 April 2024. Krupa, S., et al., 2000, ‘Ambient ozone and plant health’, Plant Disease 85, pp. 4-12 (https://apsjournals.apsnet.org/doi/pdf/10.1094/pdis.2001.85.1.4) accessed 8 December 2020. Kuenen, J., et al., 2024, Copernicus Atmosphere Monitoring Service regional emissions (CAMS-REGANT) Copernicus Atmosphere Monitoring Service [publisher] ECCAD [distributor], (https://doi.org/10.24380/0vzb-a387) accessed 21 October 2024. LMD, INERIS, LISA, 2020, CHIMERE Chemistry-Transport Model (version 2020r1) documentation, pp. 161-162 (https://www.lmd.polytechnique.fr/chimere/docs/CHIMEREdoc_v2020r1.pdf) accessed 17 September 2024. MDA, 2015, World Land Cover at 30m resolution from MDAUS BaseVue 2013 (https://www.arcgis.com/home/item.html?id=1770449f11df418db482a14df4ac26eb ) accessed 17 February 2021. Meijer, J. R., et al., 2018, ‘Global patterns of current and future road infrastructure’, Environmental Research Letters 13, 0640 (https://doi.org/10.1088/1748-9326/aabd42) accesed 10 June 2019. Mills, G., et al., 2011, ‘New stomatal flux-based critical levels for ozone effects on vegetation’, Atmospheric Environment 45, pp. 5064-5068 (https://doi.org/10.1016/j.atmosenv.2011.06.009) accessed 19 November 2020. NILU, 2025, EBAS, database of atmospheric chemical composition and physical properties (http://ebas-data.nilu.no). Data extracted in March 2025. Pedersen, S. M., et al., 2005, Potato production in Europe - a gross margin analysis, University of Copenhagen, FOI Working Paper Vol. 2005 No. 5, pp. 1-39 (https://curis.ku.dk/ws/files/135440168/5.pdf) accessed 26 May 2021. Pleijel, H., et al., 2007, ‘Ozone risk assessment for agricultural crops in Europe: Further development of stomatal flux and flux-response relationships for European wheat and potato’, Atmospheric Environment 41, pp.3022-3040 (https://doi.org/10.1016/j.atmosenv.2006.12.002) accessed 8 December 2020. ETC HE Report 2025/5 89 Reich, P. B., 1987, ‘Quantifying plant response to ozone: a unifying theory’, Tree Physiology 3, pp. 6391 (https://doi.org/10.1093/treephys/3.1.63) accessed 19 November 2020. Simpson, D., et al., 2012, ‘The EMEP MSC-W chemical transport model – technical description’, Atmospheric Chemistry and Physics 12, pp. 7825-7865 (https://doi.org/10.5194/acp-12-7825-2012) accessed 26 August 2020. Targa, J., et al., 2025, Status report of air quality in Europe for year 2023, using validated data, Eionet Report ETC HE 2025/2 (https://www.eionet.europa.eu/etcs/etc-he/products/etc-he-products/etche-reports/etc-he-report-2025-2-status-report-of-air-quality-in-europe-for-year-2023-usingvalidated-data) accessed 27 August 2025. UN, 2025, World Population Prospects 2024, United Nations, Department of Economic and Social Affairs, Population Division (https://population.un.org/wpp/downloads?folder=Standard%20Projections&group=Most%20used) accessed 21 March 2025. van Geffen, J., et., 2019, ‘TROPOMI ATBD of the total and tropospheric NO2 data products’, KNMI (https://sentinel.esa.int/documents/247904/2476257/Sentinel-5P-TROPOMI-ATBD-NO2-dataproducts) accessed 30 August 2021. van Geffen, J., et al., 2020, ‘S5P TROPOMI NO2 slant column retrieval: Method, stability, uncertainties and comparisons with OMI’, Atmospheric Measurement Techniques 13, pp. 1315-1335 (https://doi.org/10.5194/amt-13-1315-2020) accessed 30 August 2021. Veefkind, J. P., et al., 2012. ‘TROPOMI on the ESA Sentinel-5 Precursor: A GMES mission for global observations of the atmospheric composition for climate, air quality and ozone layer applications’, Remote Sensing of Environment 120, pp. 70-83 (https://doi.org/10.1016/j.rse.2011.09.027) accessed 30 August 2021. Vlasáková, L., et al., 2023. Evaluation of European-wide map creation of flux-based ozone indicator POD for selected tree species, Eionet Report ETC HE 2022/23 (https://doi.org/10.5281/zenodo.10688319) accessed 17 September 2024. WHO, 2013, Review of evidence on health aspects of air pollution: REVIHAAP project: technical report, World Health Organization (https://apps.who.int/iris/handle/10665/341712) accessed 16 November 2022. WHO, 2021a, WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide, World Health Organization (https://apps.who.int/iris/handle/10665/345329) accessed 10 December 2021. WHO, 2021b, Human health effects of polycyclic aromatic hydrocarbons as ambient air pollutants (https://www.who.int/europe/publications/i/item/9789289056533) accessed 22 September 2022. WRF, 2024, WRF Source Codes and Graphics Software Download Page. (https://www2.mmm.ucar.edu/wrf/users/download/get_sources.html#WRF-ARW) accessed 5 September 2025. ETC HE Report 2025/5 96 while 𝑅𝑎(𝑧𝑡𝑔𝑡, 𝑧𝑚,𝑂3)= 1 𝑘.𝑢∗[𝑙𝑛(𝑧𝑚,𝑂3−𝑑 𝑧𝑡𝑔𝑡−𝑑 ) − 𝛹𝐻(𝑧𝑚,𝑂3−𝑑 𝐿) + 𝛹𝐻(𝑧𝑡𝑔𝑡−𝑑 𝐿)] (A1.8a) 𝑅𝑎(𝑑+𝑧0, 𝑧𝑚,𝑂3)= 1 𝑘.𝑢∗[𝑙𝑛(𝑧𝑚,𝑂3−𝑑 𝑧0) − 𝛹𝐻(𝑧𝑚,𝑂3−𝑑 𝐿) + 𝛹𝐻(𝑧0 𝐿)] (A1.8b) 𝑅𝑏= 2 𝑘· 𝑢∗ (𝑆𝑐 𝑃𝑟)2/3 (A1.8c) 𝑅𝑠𝑢𝑟𝑓 =1 𝐿𝐴𝐼 𝑅𝑠𝑡𝑜+ 𝑆𝐴𝐼 𝑅𝑒𝑥𝑡+ 1 𝑅𝑖𝑛𝑐+ 𝑅𝑠𝑜𝑖𝑙 (A1.8d) while 𝐿= − 𝑢∗3 𝑘·𝑔·𝐻·𝑅 𝑃·𝑐𝑝 (A1.8e) where k is the von Kármán constant (equal to 0.41), ztgt is the top canopy height (the target height), zm, O3 is the height of the available ozone measurement above the canopy, z0 is the roughness length, usually assumed as 1/10 of the canopy height, L is the Obukhov length, H is the sensible heat flux in [Wm2], d is the displacement height, usually assumed as 2/3 of the canopy height, u* is the friction velocity in [m/s], Sc is the Schmidt number for ozone (equal to 0.41), Pr is the Prandtl number of air (equal to 0.71), P is the atmospheric pressure in [Pa], g is the gravity acceleration (equal to 9.8 m/s2), cp is the specific heat at constant pressure (equal to 1048 J/kg/K), R is the specific gas constant for dry air (equal to 287.05 J/kg/K), LAI is the projected leaf area in [m2/m2] ( 8 ), SAI is the surface area of the canopy in [m2/m2], ΨH(..) = ΨH(ζ) is the similarity function for heat with ζ as the argument ( 9 ), according to (𝜁)=2 when 𝜁<0 =−5𝜁 when 𝜁≥0 (A1.8f) with x = (1 – 16 * ζ)1/4 (A1.8g) and Rext is the resistance to cuticular deposition of ozone (equal to 2 500 s/m); Rsoil is the soil resistance (equal to 200 s/m1), while Rsto = 1/gsto (A1.8h) Rinc = b.SAI.h / u* (A1.8i) where gsto is the actual stomatal conductance, b is the empirical constant (equal to 14 m-1), h is the height of the canopy. The calculations of LAI and SAI are based on Simpson et al. (2012) and Emberson et al. (2000b), as well as the schematics regarding LAI presented therein (see Eq. 2.8i to 2.8n). ( 8 ) For more details see LMD et al. (2020). ( 9 ) For more details see CLRTAP (2017). ETC HE Report 2025/5 97 LAI = LAImin + (d – SGS) * (LAImax − LAImin) / SGSlength, (A1.8j) SAI = 5 * LAI / 3.5 (A1.8k) when SGS ≤ d < SGS + SGSlength LAI = LAImax, (A1.8l) SAI = LAI + 1.5, (A1.8m) when SGS + SGSlength ≤ d < EGS −EGSlength LAI = LAImax − (d – (EGS –EGSlenght)) * (LAImax − LAImin) / EGSlength, (A1.8n) SAI = LAI + 1.5, (A1.8o) when EGS − EGSlength ≤ d ≤ EGS where LAImin is the minimum (within the growing season) LAI values [m2/m2] LAImax is the maximum (within the growing season) LAI values [m2/m2] SGS is the start of the growing season [day of year] SGSlenght is the duration of the start of the growing season [days] EGSlenght is the duration of the end of the growing season [days] EGS is the end of the growing season [day of year] [day of year] Calculation of the hourly stomatal conductance of ozone (gsto) The basis of the approach used for calculating phytotoxic ozone doses is the calculation of an instantaneous stomatal conductance gsto in the given hour H, according to the equation gsto = gmax * [min(fphen, fO3)] * flight * max[fmin, (ftemp * fVPD * fSW)] (A1.9) where gsto is the actual stomatal conductance in [mmol O3 /m2 PLA per second], gmax is the species-specific maximum stomatal conductance in [mmol O3 /m2 PLA per second], see Table A1.2, fphen is the relative proportion function for the phenology for the different stage of growing, fO3 is the relative proportion function for the influence of ozone on stomatal flux by promoting premature senescence, fmin is the species-specific relative minimum stomatal conductance that occurs during daylight hours, see Table A1.2, ftemp, fVPD, fSW, flight are relative proportion functions for leaf stomata respond to temperature, air humidity, soil moisture and light. Parameters fphen, fO3, flight, ftemp, fVPD, fSW and fmin are expressed as relative proportion functions, taking values between 0 and 1 as a proportion of gmax. These functions allow taking into account irradiance (flight), temperature (ftemp), water vapour deficit at leaves level (fvpd), soil moisture (fsw), phenology for the different stage of growing (fphen) and the influence of ozone on stomatal flux by promoting premature senescence (fO3). fmin is the minimum relative value of stomatal conductance during the daylight. The parameter fphen is calculated based on the accumulation of thermal time over the growing season of the crop being considered (Colette et al., 2018), according to CLRTAP (2024). For wheat and potato, the accumulation period is defined for each year using the effective temperature sum (ETS) in ˚C for days in excess of 0 ˚C, while for tomato for days in excess of 10 ˚C. For wheat, the total accumulation period during which wheat is sensitive to ozone exposure is 200 °C days and 300 °C days before mid-anthesis (mid-point in flowering) to 700 °C days to 550 °C days after ETC HE Report 2025/5 98 mid-anthesis for Atlantic, Boreal and Continental regions and Mediterranean region, respectively. The timing of mid-anthesis is estimated by starting at the first date after 1 January (or just 1 January) when the temperature exceeds 0 °C. The mean daily temperature is then accumulated (temperature sum), and mid-anthesis is estimated to be a temperature sum of 1075 °C days for Atlantic, Boreal and Continental regions and 1250 °C days for Mediterranean region, which in general corresponds to bread wheat. For potato, the accumulation period stands between 330 °C days before the tuber initiation date and 800 °C days after this date. The tuber initiation date is considered to be homogeneous throughout Europe. The reasons for its simplification are a) heterogeneous climatic conditions in the European countries naturally lead to different time of potato planting (Pedersen et al., 2005) followed by different time of the tuber initiation and b) lack of detailed local data availability for modelling and mapping. As discussed ( 10 ) with the French national Chamber of agriculture (APCA, http://chambresagriculture.fr), the tuber initiation starts 15 days after the transplantation in the field, which occurs in May. Therefore, the fixed date for the tuber initiation was set to June 1st. For tomato, the accumulation period is from 250 ˚C days to 1500 ˚C days after transplantation in the field over a base temperature of 10 °C. The timing of the transplantation is set on the date June 1st. The parameter fphen is calculated according to following equations: in the case of wheat: 𝑓𝑝ℎ𝑒𝑛 =1 when (fphen_2_ETS + fphen_1_ETS) ≤ ETS ≤ (fphen_2_ETS + fphen_3_ETS) =1−( 𝑓𝑝ℎ𝑒𝑛_𝑎 𝑓𝑝ℎ𝑒𝑛_4_𝐸𝑇𝑆−𝑓𝑝ℎ𝑒𝑛_3_𝐸𝑇𝑆) * (ETS – fphen_3_ETS) when (fphen_2_ETS + fphen_3_ETS) < ETS ≤ (fphen_2_ETS + fphen_4_ETS) =𝑓𝑝ℎ𝑒𝑛_𝑒 −( 𝑓𝑝ℎ𝑒𝑛_𝑒 𝑓𝑝ℎ𝑒𝑛_5_𝐸𝑇𝑆−𝑓𝑝ℎ𝑒𝑛_4_𝐸𝑇𝑆) * (ETS – fphen_4_ETS) when (fphen_2_ETS + fphen_4_ETS) < ETS ≤ fphen_5_ETS (A1.9a) in the case of potato (formulated based on CLRTAP, 2017): 𝑓𝑝ℎ𝑒𝑛 =1−(1 − 𝑓𝑝ℎ𝑒𝑛_𝑎 𝑓𝑝ℎ𝑒𝑛_1_𝐸𝑇𝑆)∗𝐸𝑇𝑆 when fphen_1_ETS ≤ ETS < 0 =1−(1 − 𝑓𝑝ℎ𝑒𝑛_𝑒 𝑓𝑝ℎ𝑒𝑛_2_𝐸𝑇𝑆)∗𝐸𝑇𝑆 when 0 < ETS ≤ fphen_2_ETS (A1.9b) in the case of tomato (formulated based on CLRTAP, 2017): 𝑓𝑝ℎ𝑒𝑛 =𝐸𝑇𝑆−𝑓𝑝ℎ𝑒𝑛_2_𝐸𝑇𝑆 𝐴𝑠𝑡𝑎𝑟𝑡_𝐸𝑇𝑆−𝑓𝑝ℎ𝑒𝑛_2_𝐸𝑇𝑆 when Astart_ETS ≤ ETS < Aend_ETS (A1.9c) ( 10 ) There is a lack of information on a date of potato tuber initiation in Europe. It should ideally rely on existing models based on agricultural practices, local climatology, ground properties, and location. INERIS, while developing the POD script, relied on contents of discussions with the French National Chamber of Agriculture (consultation with APCA, March 2018; Deumier and Hannon, 2010). Based on the information given that the tuber initiation starts 15 days after the transplantation in the field, which occurs in May in France, a fixed date of June 1st has been chosen for France and applied also for the rest of Europe. This date should be revised according to the availability of more accurate information on potato plantations in Europe. ETC HE Report 2025/5 99 where ETS is the effective temperature sum in ˚C days using a base temperature of 0 ˚C for wheat and potato and a base temperature of 10 °C for tomato (see Table A1.2); for wheat, ETS is set to 0 °C days at mid-anthesis day. Then Astart_ETS will be at 200 ˚C days before mid-anthesis, and Aend_ETS will be at 700 ˚C days after mid-anthesis over a base temperature of 0 °C; for potato, ETS is set to 0 °C days at tuber initiation day. Then Astart_ETS will be at 330 ˚C days before tuber initiation and Aend_ETS at 800 ˚C days after tuber initiation over a base temperature of 0 °C; for tomato, ETS is set to 0 °C days at transplantation day in the field. Then Astart_ETS will be at 250 ˚C days after transplantation in the field and Aend_ETS at 1500 ˚C days after transplantation in the field over a base temperature of 10 °C, fphen_a, fphen_e is the phenology function, which consists of terms describing rate changes of gmax expressed as fractions (see Table A1.2), fphen_1_ETS, fphen_2_ETS , fphen_3_ETS , fphen_4_ETS, fphen_5_ETS are °C days (see Table A1.2; fphen_1_ETS and fphen_5_ETS define period crops to be sensitive to ozone exposure), Astart_ETS and Aend_ETS are the effective temperature sums (counted from the day of the midanthesis for wheat, from the day of the tuber initiation for potato and from the day of the transplantation in the field for tomato) above a base temperature of 0 ˚C for wheat and potato and 10 ˚C for tomato at the start and end of the O3 accumulation period respectively; see Table A1.2. The parameter fO3 in the case of wheat is calculated according to equation fO3 = [(1+(POD0/14)8]-1 (A1.9d) while 𝑃𝑂𝐷0=∑𝐹𝑠𝑡𝑜 𝐻−1 𝑛=𝐴𝑠𝑡𝑎𝑟𝑡 (𝑛)·3600 106 (A1.9e) where POD0 is the ozone flux already accumulated since the beginning of the vegetation period Astart up to the last hour H-1, Fsto(n) is the hourly ozone flux in the hour n, calculated in the previous steps based on Equation A1.10, while Fsto(Astart) is equal to 0. The parameter (ozone function) fO3 in the case of potato is calculated according to equation fO3 = [(1+(AOT0/40)5]-1 (A1.9f) where AOT0 is accumulated ozone concentration from the start of the vegetation period Astart up to the last hour H-1. The parameter (ozone function) fO3 in the case of tomato is not determined. The parameter flight is calculated according to flight = 1 – EXP[(–light_a)*PPFD] (A1.9g) while PPFD = SSRD * 0.5 * 4.5 (A1.9h) where PPFD represents the photosynthetic photon flux density [μmol/m2 per second], light_a is a light parameter (see Table A1.2), SSRD represents the surface net solar radiation in [W/m2]. The parameter ftemp is calculated according to: ETC HE Report 2025/5 100 ftemp = max {fmin, [(T – Tmin) / (Topt – Tmin)] * [(Tmax – T) / (Tmax – Topt)]bt} when Tmin < T < Tmax = fmin when Tmin > T > Tmax (A1.9i) while bt = (Tmax – Topt) / (Topt – Tmin) (A1.9j) where Tmin, Tmax and Topt are minimum, maximum and optimum temperatures determining leaf stomata opening (see Table A1.2) Table A1.2 Parametrisation for POD6SPEC for wheat flag leaves and the upper-canopy sunlit leaves of potato and tomato, for different biogeographical regions Parameter Units (Bread) Wheat Potato Tomato Atlantic, Boreal, Continental (Pannonia, Steppic) Mediterranean Atlantic, Boreal, Continental (Mediterranean Pannonia, Steppic) Mediterranean gmax mmol O3 /m2 PLA per second 500 430 750 330 fmin fraction 0.01 0.01 0.01 0.06 light_a - 0.0105 0.0105 0.005 0.0125 Tmin °C 12 12 13 18 Topt °C 26 28 28 28 Tmax °C 40 39 39 37 VPDmax kPa 1.2 3.2 2.1 1 VPDmin kPa 3.2 4.6 3.5 4 ΣVPD_crit kPa 8 16 10 - fO3 POD0 mmol O3/m2 PLA per second 14 - - - fO3 AOT0, ppmh - - 40 - fO3 exponent 8 - 5 - Astart_ETS ºC day - - - 250 Aend_ETS ºC day - - - 1500 SGS ºC day Astart_ETS Astart_ETS Astart_ETS Astart_ETS EGS ºC day Aend_ETS Aend_ETS Aend_ETS Aend_ETS SGSlength days 70 70 35 70 EGSlength days 22 44 65 44 LAImin m2/m2 0 0 0 0 LAImax m2/m2 5 3.5 6 3.5 Leaf dimension cm 2 2 4 3 Canopy height m 1 0.75 1 2 fphen_a fraction 0.3 0.5 0.4 1 fphen_b fraction - - - - fphen_c fraction - - - - fphen_d fraction - - - - fphen_e fraction 0.7 0.5 0.2 0.0 fphen_1_ETS °C day -200 -300 -330 0 fphen_2_ETS °C day 0 0 800 2770 fphen_3_ETS °C day 100 70 - - fphen_4_ETS °C day 525 312 - - fphen_5_ETS °C day 700 550 - - mid-anthesis °C day 1075 1250 - - Source: CLRTAP, 2024; Emberson et al., 2000b; González-Fernández et al., 2013; González-Fernández (personal communication, May 2021). The parameter fVPD is calculated according to: fVPD = min{1,max {fmin, [(1–fmin)*(VPDmin – VPD) / (VPDmin – VPDmax)] + fmin}} (A1.9k) ETC HE Report 2025/5 101 while VPD = es(T) * (1− RH) (A1.9l) es(T) = 0.61 exp[17.502T/(240.97+T)] (A1.9m) where VPDmin is the minimum vapour pressure deficit determining leaf stomata opening, VPDmax is the maximum vapour pressure deficit determining leaf stomata opening, T is the air temperature [°C], RH is the relative humidity [%]/100 (i.e. in the range 0-1), es(T) is the potential (saturation) water vapour pressure, The ΣVPD (i.e. the function describing stomatal re-opening in the afternoon) is taken into account for maps PODYSPEC for wheat and potato. ΣVPD (kPa) should be calculated for daylight hours until dawn of the next day. If ΣVPD ≥ ΣVPD_crit, gsto calculated using Equation A1.9 is valid if smaller or equal to gsto of the preceding hour. If gsto is larger than gsto of the preceding hour, given that ΣVPD is larger than or equal to ΣVPD_crit, it is replaced by the gsto of the preceding hour. The parameter fSW is replaced by fSMI (where SMI represents Soil Moisture Index with maximum at field capacity), taking values between 0 and 1 as a proportion of gmax (with 0 for soil moisture at and below wilting point), following the parameterization given in Simpson et al. (2012), similar to the plant available water (PAW) parameterization fPAW as defined for wheat in CLRTAP (2024). The basic equation used for fSW resp. fSMI is: 𝑓𝑆𝑀𝐼 = 0 for SMI ≤ 0 =SMI PAW𝑡 for 0 <SMI ≤ PAWt =1 for SMI > PAWt (A1.9n) while SMI=SWLL − PWP FC − PWP (A1.9o) where PAWt is the threshold amount of water in the soil available to the plants, above which stomatal conductance is at a maximum, set to 0.5, SWLL is the soil moisture in [m3/m3], PWP is the permanent wilting point in [cm3/cm3], FC is the field capacity in [cm3/cm3]. Parameter fSW represents below-ground soil water, which is extremely difficult to verify due to the limited availability of measured data across Europe. Moreover, the relationship between precipitation, soil water content and soil water pressure is highly non-linear and sensitive to soil characteristics, making reliable estimation problematic. Soils and plant responses are highly heterogeneous, and simple models often fail to reflect this complexity. Therefore, using fSMI provides a more practical and robust alternative for capturing the effects of soil moisture stress in large-scale modelling (CLRTAP, 2020). The Soil Moisture Index using the EMEP methodology as described in Simpson et al. (2012) and CLRTAP (2020) is used. It is computed using the soil moisture variable available from a meteorological model, which represents the water content in m3 of water per m3 of ground [m3/m3] in a specific ground level, in dependence on the available dataset. For soil moisture, the ECWMF’s ERA5-Land variable Volume of water in soil layer 3 (i.e. 28-100 cm) has been used, see Section 3.3. The level of soil layer was chosen based on recommendation of Haberle and Svoboda (2015). The soil moisture is quite a sensitive parameter in the calculation of the POD. Next to the soil moisture, the soil moisture index also takes into account the permanent wilting point and the field capacity; they are taken from JRC soil database (JRC, 2016), see Annex 2, Section A2.3. ETC HE Report 2025/5 102 No limitation of stomatal conductance due to soil moisture can be assumed for tomato, since it is an irrigated horticultural crop. Thus, fSMI for this crop could be established to fSMI = 1 over the whole range of SMI values to remove limitation due to soil moisture deficit. Modelling the hourly stomatal flux of ozone (Fsto) Once the hourly stomatal conductance of ozone (gsto) and all relevant variables are computed, the stomatal flux of ozone (Fsto) can be calculated, based on the assumption that the concentration of ozone at the top of the canopy represents a reasonable estimate of the concentration at the upper surface of the laminar layer for a sunlit upper canopy leaf. Fsto is calculated according to the CLRTAP (ICP Vegetation) methodology, thus the fraction of the ozone taken up by the stomata is given using a combination of the stomatal conductance, the external leaf, or cuticular, resistance and the leaf surface resistance. The hourly stomatal flux in the given hour H is calculated according to 𝐹𝑠𝑡𝑜 =𝑐(𝑧1)∗𝑔𝑠𝑡𝑜 ∗𝑟𝑐 𝑟𝑏+𝑟𝑐 (A1.10) where Fsto is the hourly stomatal flux of ozone in [nmol/m2 PLA per second] c(z1) is the concentration of ozone at canopy top in [nmol/m3] rb is the quasi-laminar resistance in [s/m] rc is the leaf surface resistance in [s/m] gsto is the actual stomatal conductance in [m/s], while rc = 1/(gsto + gext) (A1.10a) 𝑟𝑏=1.3∗150∗√𝐿 𝑢(𝑧1) (A1.10b) where gext is the external leaf, or cuticular, resistance in [m/s], equal to 1/2500 m/s u(z1) is the wind speed at height z1 (z1 is the canopy top) L is the cross-wind leaf dimension (2 cm, see Table A1.2) while 𝑢(𝑧1) = 𝑢∗ 𝑘∗𝑙𝑛 (𝑧1−𝑑 𝑧0) for wheat (A1.10c) 𝑢(𝑧1) = 𝑢∗ 𝑘∗𝑙𝑛 (𝑧1 𝑧0) for potato and tomato (A1.10d) where k is the von Kármán constant (equal to 0.41) d is the displacement height usually assumed as 2/3 of the canopy height, z1 is the top of the canopy z0 is the roughness length usually assumed as 1/10 of the canopy height u* is the friction velocity. Box A1.1 shows the conversion of stomatal conductance and ozone concentration to units demanded for PODY calculation. ETC HE Report 2025/5 103 Box A1.1 Conversion of stomatal conductance gsto and ozone concentration to units demanded for PODy calculation Stomatal conductance gsto has to be converted from units mmol/m2 per second to units m/s (since all the resistances are expressed in the unit of s/m). At standard temperature (20 °C) and air pressure (1.013 x 105 Pa), the conversion is made by dividing the conductance in mmol/m2 per second by 41 000 to give conductance in m/s. To convert the ozone concentration (C) at canopy height from µg/m3 resp. ppb to nmol/m, the following equation should be used: C [nmol·m-3] = C [ppb] * P/(R·T) = C [µg/m3] / 2 * P/(R·T) (A1.11) where P is the atmospheric pressure in Pa, R is the universal gas constant of 8.31447 J/mol per Kelvin T is the air temperature in Kelvin. At standard temperature (20 °C) and air pressure (1.013 x 105 Pa), the concentration in ppb should be multiplied by 41.56 to calculate the concentration in nmol/m3. Source: CLRTAP, 2024 Calculation of PODY from Fsto Hourly averaged stomatal ozone fluxes (Fsto) in excess of a Y threshold are accumulated over a species or vegetation-specific accumulation period using the following equation: 𝑃𝑂𝐷𝑌=∑(𝐹𝑠𝑡𝑜(𝑛)−𝑌) 𝑛·3600 106 (A1.12) while Y (for wheat, potato or tomato) = 6 nmol/m2 PLA per second where PODY is the phytotoxic ozone dose related to the threshold Y, in [mmol/m2 PLA], Fsto(n) is the hourly ozone flux in the hour n of the accumulation period. The value Y (in [nmol/m2 PLA/s]) is subtracted from each hourly averaged Fsto (in [nmol/m2 PLA/s]) value and the Fsto (after the subtracting of Y) is accumulated only when Fsto>Y, during daylight hours (when global radiation is more than 50 W/m2). The value is then converted to hourly fluxes by multiplying by 3 600 and to mmol by dividing by 106 to get the stomatal ozone flux in mmol/m2 PLA. Trees The POD maps for selected trees, i.e. beech (F. sylvatica) and spruce (P. abies), are created with calculated hourly POD values which are based on hourly O3 concentrations, hourly meteorological parameters such as temperature, vapour pressure deficit, solar radiation and soil hydraulic property data. The hourly O3 concentrations are calculated by combining the monitoring data from rural background stations, chemical transport modelling data and other supplementary data (Horálek et al., 2023). The calculation of the phytotoxic O3 dose above a threshold Y (POD1) as described in Vlasáková et al. (2023) follows precisely the methodology described in the Manual for modelling and mapping critical loads & levels of the CLRTAP in its most recent available revision (CLRTAP, 2024), including some specifications presented in the Scientific Background Documents of this manual (CLRTAP, 2017, 2020), ETC HE Report 2025/5 104 as prepared by the International scientific Cooperative Programme on effects of air pollution on natural vegetation and crops of the Working Group on Effects of the CLRTAP (ICP Vegetation). A1.4 Methods for uncertainty analysis The uncertainty estimation of the European map is based on leave-one-out cross-validation. This crossvalidation method computes the quality of the spatial interpolation for each point of measurement (i.e. monitoring station) from all available information except from the point in question, i.e. it withholds one data point and then makes a prediction at the spatial location of that point. This procedure is repeated for all measurement points in the available set. The predicted and measurement values at these points are plotted in the form of a scatter plot. With help of statistical indicators (see below), the quality of the predictions is demonstrated objectively. The advantage of the nature of this cross-validation technique is that it enables evaluation of the quality of the predicted values at locations without measurements, as long as they are within the area covered by the measurements. In addition, a simple comparison is made between the point measurement data and the estimated values of the 1 km x 1 km grid cells (for PM and NO2) or the 10 km x10 km grid cells (for ozone) for the separate rural and urban background (and urban traffic, where relevant) map layers and the 1 km x 1 km grid cells for the final combined maps, for the health-related indicators, and the 2 x 2 km grid cells in the case of AOT40 and NOx. Note that the grid cell value is the mean estimated value of this grid cell area. The estimated value within a grid cell will only approximate the predicted value(s) at the station(s) lying within that cell. This additional analysis has not been performed for BaP. Cross-validation The results of cross-validation are described by the statistical indicators and scatter plots. The main indicator used is root mean squared error (RMSE) and the additional ones are relative RMSE (RRMSE), which is expressed in relative terms (by relating the RMSE to the mean of the air pollution indicator value for all stations), and bias (mean prediction error, MPE): 𝑅𝑀𝑆𝐸=√1 𝑁∑(𝑍 󰆹(𝑠𝑖)−𝑍(𝑠𝑖))2 𝑁 𝑖=1 (A1.13) 𝑅𝑅𝑀𝑆𝐸=𝑅𝑀𝑆𝐸 𝑍 .100 (A1.14) 𝑏𝑖𝑎𝑠(𝑀𝑃𝐸)= 1 𝑁∑(𝑍 󰆹(𝑠𝑖)−𝑍(𝑠𝑖)) 𝑁 𝑖=1 (A1.15) where 𝑍 󰆹(𝑠𝑖) is the air quality indicator value derived from the measured concentration at the ith point, i = 1, …, N, 𝑍(𝑠𝑖) is the air quality estimated indicator value at the ith point using other information, without the indicator value derived from the measured concentration at the ith point, RRMSE is the relative RMSE, expressed in percent, 𝑍 is the arithmetic average of the indicator values Z(s1), …, Z(sN), as derived from measurement concentrations at the stations i = 1, … , N, N is the number of the measuring points. Other indicators are R2 and the regression equation (y = a.x + c) parameters slope (a) and intercept (c), following from the scatter plot between the predicted (using cross-validation) and the observed concentrations. RMSE should be as small as possible, bias (MPE) should be as close to zero as possible, R2 should be as close to 1 as possible, slope a should be as close to 1 as possible, and intercept c should be as close to zero as possible (in the regression equation y = a.x + c). ETC HE Report 2025/5 105 In the cross-validation of PM2.5, NOx and BaP, only stations with PM2.5, NOx and BaP measurement data, respectively, are used (not the pseudo PM2.5, NOx and BaP stations, see Annex 1 Section A1.1). Comparison of the point measurement and interpolated grid values The comparison of point measurement and predicted grid values is described by the linear regression equation and its parameters and statistical values. The comparison is executed separately for rural and urban background (and urban traffic, where relevant) map layers and for the final combined map. In the case of PM2.5 and NOx, only the stations with actual PM2.5 and NOx measurement data are used (not the pseudo PM2.5 and NOx stations). This analysis is done for PM, ozone, NO2 and NOx, not for BaP. The point observation – point cross-validation prediction analysis (Annex 3, sections “Uncertainty estimated by cross-validation”) describes interpolation performance at point locations when there is no observation (as it follows the leave-one-out approach). In this case, the smoothing effect of the interpolation is most prevalent. The point observation – grid prediction approach indicates performance of the value for the grid cell (either in 1 km, 2 km or 10 km resolution) with respect to the observations that are located within that cell. As such, some variability is due to smoothing but it also includes smoothing due to spatial averaging into the grid cells. As such, the point-grid validation approach tells us how well our interpolated and aggregated grid values approximate the measurements at the actual station (point) locations. Whereas the point-point approach tells us how well our interpolated values estimate the indicator at a point where there is no actual measurement at that location, under the constraint that the point lies within the area covered by measurements. ETC HE Report 2025/5 112 Table A2.2 General land cover classes, based on CLC2018 classes, used in mapping Label General class description CLC classes grid codes CLC classes codes CLC classes description HDR High density residential areas 1 111 Continuous urban fabric LDR Low density residential areas 2 112 Discontinuous urban fabric AGR Agricultural areas 12-22 211-244 Agricultural areas NAT Natural areas 23-34 311-335 Forest and semi natural areas Two aggregations are used, i.e. into 1 km resolution grid and into the circle with radius of 5 km. For each general CLC class, the high land use resolution is spatially aggregated into the 1 km EEA standard grid resolution. The aggregated grid square value represents for each general class the total area of this class as percentage of the total 1 km x 1 km area. For details, see Horálek et al. (2017b). Road type vector data GRIP (Meijer et al., 2018) vector road type data provided by the Netherlands Environmental Assessment Agency (PBL) are used for the weighting procedure of the urban background and the urban traffic map layers (Annex 1, Section A1.1). The road types are distributed into 5 classes, from highways to local roads and streets. In agreement with Horálek et al. (2017b), road classes No. 1 “Highways”, No. 2 “Primary roads” and No. 3 “Secondary roads” are used. Percentage of the area influenced by traffic is represented by buffers around the roads: for the individual classes 1-3 and for classes 1-3 together, at all 1 km x 1 km grid cells; a buffer of 75 metres distance at each side from each road vector is taken for the roads of classes 1 and 2, while a buffer of 50 metres is taken for the roads of class 3. For details, see Horálek et al. (2017b). Satellite data The annual average NO2 dataset was constructed based on data from the TROPOspheric Monitoring Instrument (TROPOMI) onboard of the Sentinel-5 Precursor satellite (Veefkind et al., 2012). All available swath-based Level-2 data with an irregular pixel geometry was acquired for the year 2023. The spatial resolution of the product is cc. 5.5 km by 3.5 km. The product used is the S5P_OFFL_L2__NO2 product (van Geffen et al., 2019, 2020) and it provides the tropospheric vertical column density of NO2, i.e. a vertically integrated value over the entire troposphere. All overpasses for a specific day were then mosaicked using HARP (https://stcorp.github.io/harp/doc/html/index.html) and retrievals with a quality assurance values greater than 0.75 (indicating high quality and cloud-free conditions) were gridded to a regular projected grid for all area with a 1 km spatial resolution in a ETRS89 / ETRS-LAEA (EPSG 3035) projection. The daily gridded files were subsequently averaged to an annual mean. I.e. the parameter used is NO2 – annual average tropospheric vertical column density (VCD) [number of NO2 molecules per cm2 of earth surface], year 2023 (aggregated from cloud-free high-quality daily data). Soil hydraulic properties data JRC data called "Maps of indicators of soil hydraulic properties for Europe" in 1 km resolution are used for POD calculations, JRC (2016). Namely the following indicators are used: Wilting Point – water content at wilting point [cm3/cm3], Field Capacity – water content at field capacity [cm3/cm3]. ETC HE Report 2025/5 113 Annex 3 Technical details and mapping uncertainties This annex contains technical details on the linear regression models and the residual kriging as used in the mapping. Furthermore, uncertainty estimates for the maps of the indicators are given. A3.1 PM10 Technical details on the mapping and uncertainty estimates for both PM10 indicators maps annual average (Map 2.1) and 90.4 percentile of daily means (Map 2.3) are presented in this section. Technical details on the mapping Table A3.1 presents the estimated parameters of the linear regression models (c, a1, a2, …) and of the residual kriging (nugget, sill, range) and includes the statistical indicators of both the regression and the kriging, for both PM10 indicators. The linear regression and ordinary kriging of its residuals are applied on the logarithmically transformed data of both measurement and modelled PM10 values. In Table A3.1 the standard error and variogram parameters (nugget, sill and range) refer to these transformed data, whereas RMSE and bias refer to the interpolation after a back-transformation. Since 2017 maps, an updated methodology as developed and tested under Horálek et al. (2019) has been used, i.e. including land cover among the supplementary data and using the traffic urban map layer. The adjusted R2 and standard error are indicators for the fit of the regression relationship, where the adjusted R2 should be as close to 1 as possible and the standard error should be as small as possible. The adjusted R2 for the rural areas was 0.67 at the annual average and 0.66 at the 90.4 percentile of daily means (P90.4); for the urban background areas 0.38 at the annual average and 0.37 at the P90.4; for the urban traffic areas 0.38 at the annual average and 0.28 at the P90.4. Table A3.1 Parameters and statistics of linear regression model and ordinary kriging of PM10 indicators annual average and 90.4 percentile of daily means for 2023 in rural, urban background and urban traffic areas for the final combined map Rural areas Urb. b. ar. Urb. tr. ar. Rur. ar. Urb. b. ar. Urb. tr. ar. c (constant) 1.40 0.74 1.68 1.44 1.06 2.44 a1 (log. CAMS model) 0.862 0.842 0.56 0.788 0.764 0.41 a2 (altitude GMTED) -0.00015 non signif. a3 (wind speed) -0.03171 -0.053 non signif. -0.075 a4 (relative humidity) -0.012 -0.010 a5 (land cover NAT) -0.0009 -0.0011 Adjusted R20.67 0.38 0.38 0.66 0.37 0.28 Stand. Error [µg/m3]0.23 0.32 0.29 0.22 0.33 0.33 Nugget 0.020 0.024 0.026 0.020 0.023 0.031 Sill 0.054 0.051 0.048 0.039 0.058 0.060 Range [km] 1000 210 330 990 220 250 RMSE [µg/m3]3.4 6.4 4.8 6.1 11.4 9.4 Relative RMSE [%] 24.7 31.1 23.5 25.9 32.7 26.9 Bias (MPE) [µg/m3]0.0 0.1 0.0 -0.1 0.0 -0.1 90.4 percentile of daily means Linear regresion model (LRM, Eq. A1.3) Ordinary kriging (OK) of LRM residuals LRM + OK of its residuals Annual average RMSE (the smaller the better) and bias (the closer to zero the better), highlighted by orange, are the cross-validation indicators, showing the quality of the resulting map. The bias indicates to what extent the predictions are underor overestimated on average. Further in this section, more detailed uncertainty analysis is presented. ETC HE Report 2025/5 114 Uncertainty estimated by cross-validation Using RMSE as the most common indicator, the absolute mean uncertainty of the final combined map at areas 'in between' the station measurements (i.e. at locations without measurements, as long as they are within the area covered by the measurements) can be expressed in µg/m3. Table A3.1 shows that the absolute mean uncertainty of the final combined map of PM10 annual average and 90.4 percentile of daily means expressed by RMSE is 3.4 µg/m3 and 6.1 µg/m3 for the rural areas, 6.4 µg/m3 and 11.4 µg/m3 for the urban background areas, and 4.8 µg/m3 and 9.4 µg/m3 for the urban traffic areas, respectively. Alternatively, one can express this uncertainty in relative terms by relating the absolute RMSE uncertainty to the mean air pollution indicator value for all stations. This relative mean uncertainty (Relative RMSE) of the final combined map of PM10 annual average and 90.4 percentile of daily means is 24.7% and 25.9% for rural areas, 31.1% and 32.7% for urban background areas, and 23.5% and 26.9% for urban traffic areas, respectively. These quite high numbers in urban background areas compared to previous years up to 2015 are caused by inclusion of Türkiye since 2016 mapping. For the mapping results without Türkiye, the relative mean uncertainty is 19.8% and 19.6% for rural areas, 19.5% and 22.1% for urban background areas and 18.0% and 20.4% for urban traffic areas, respectively. Nevertheless, the relative uncertainty values including Türkiye fulfil the data quality objectives for models as set in Annex I of the Air Quality Directive (EC, 2008). Figure A3.1 shows the cross-validation scatter plots, obtained according to Annex 1, Section A1.4 for rural, urban background and urban traffic areas, for both PM10 indicators. The R2 indicates that the variability is attributable to the interpolation for about 70% and 67% at the rural areas, for 66% and 65% at the urban background areas, and for about 72% and 68% at the urban traffic areas, for the annual average and the 90.4 percentile of daily means, respectively. Figure A3.1 Correlation between cross-validated predicted (y-axis) and measurement values for PM10 indicators annual average (top) and 90.4 percentile of daily means (bottom) for 2023 for rural (left), urban background (middle) and urban traffic (right) areas ETC HE Report 2025/5 115 The trend line in the scatter-plots deviates at the lowest values somewhat above, and at higher values below the symmetry axis, indicating that the interpolation methods tend to underestimate the high concentrations and overestimate the low concentrations. For example, in urban background areas for annual average an observed value of 40 µg/m3 is estimated in the interpolations to be about 36 µg/m3, about 10% lower. This underestimation at high values is common to all spatial interpolation methods. It could be reduced by either using a higher number of stations with an improved spatial distribution, or by introducing an improved regression that uses either other supplementary data or more advanced chemical transport model (or a model in finer resolution). Comparison of point measurement values with the predicted grid value In addition to the above point observation – point prediction cross-validation, a simple comparison has been made between the point observation values and interpolated prediction values spatially averaged at grid cells. This point observation – grid averaged prediction comparison indicates to what extent the predicted value of a grid cell represents the corresponding measurement values at stations located in that cell. The comparison has been made primarily for the separate rural, urban background and urban traffic map layers at 1 km resolution. (One can directly relate this comparison result to the cross-validation results of Figure A3.1). Apart from this, the comparison has been done also for the final combined maps at the same 1 km resolution. Figure A3.2 shows the scatterplots for these comparisons, for PM10 annual average only as an illustration. The results of the point observation – point prediction cross-validation of Figure A3.1 and those of the point observation – grid averaged prediction validation for separate rural, urban background and urban traffic map layers, and for the final combined maps are summarised in Table A3.2 for both PM10 indicators. Figure A3.2 Correlation between predicted grid values from rural (upper left), urban background (upper middle) and urban traffic (upper right) map layer and final combined map (all bottom) (y-axis) versus measurements from rural (left), urban/suburban background (middle) and urban/suburban traffic stations (right) (x-axis) for PM10 annual average 2023 ETC HE Report 2025/5 116 Table A3.2 Statistical indicators from the scatter plots for the predicted grid values from separate (rural, urban background or urban traffic) map layers and final combined map versus the measurement point values for rural (upper left), urban background (upper right) and urban traffic (bottom left) stations for PM10 indicators annual average (top) and 90.4 percentile of daily means (bottom) for 2023 RMSE Bias R2Lin. r. equation RMSE Bias R2Lin. r. equation cross-val. prediction, separate (r or ub) map layer 3.4 0.0 0.699 y = 0.720x + 3.8 6.4 0.1 0.657 y = 0.782x + 4.6 grid prediction, 1 km res., separ. (r or ub) map layer 2.3 -0.2 0.861 y = 0.796x + 2.6 4.3 0.0 0.838 y = 0.834x + 3.3 grid prediction, 1 km res., final combined map 2.6 0.2 0.823 y = 0.846x + 2.3 7.4 -0.3 0.831 y = 0.835x + 3.4 cross-val. prediction, separate (r or ub) map layer 6.1 -0.1 0.668 y = 0.669x + 7.7 11.4 0.0 0.654 y = 0.724x + 9.6 grid prediction, 1 km res., separ. (r or ub) map layer 4.7 -0.4 0.814 y = 0.739x + 5.7 7.5 -0.5 0.855 y = 0.805x + 6.5 grid prediction, 1 km res., final combined map 5.0 0.2 0.780 y = 0.804x + 4.8 7.5 -0.2 0.847 y = 0.807x + 6.6 RMSE Bias R2Lin. r. equation cross-valid. prediction, urban traffic map layer 4.8 0.0 0.716 y = 0.748x + 5.2 grid prediction, 1 km res., urban traffic map layer 3.8 -0.1 0.830 y = 0.806x + 3.9 grid prediction, 1 km res., final combined map 5.2 -2.4 0.745 y = 0.752x + 2.7 cross-valid. prediction, urban traffic map layer 9.4 -0.1 0.675 y = 0.691x + 10.7 grid prediction, 1 km res., urban traffic map layer 7.0 -0.2 0.822 y = 0.774x + 7.7 grid prediction, 1 km res., final combined map 10.1 -4.2 0.688 y = 0.682x + 6.9 Annual average 90.4 percentile of daily means Rural background stations Urban/suburban background stations PM10 Annual average 90.4 percentile of daily means PM10 Urban/suburban traffic stations By comparing the scatterplots and the statistical indicators for the separate rural, urban background and urban traffic map layers with the final combined map, one can evaluate the level of representation of the rural, urban background and urban traffic areas in the final combined map. Both the rural and the urban air quality are fairly well represented in the 1 km final combined map, while the traffic air quality is underestimated in this spatial resolution. One can conclude that the final combined map in 1 km resolution is representative for rural and urban background areas, but not for urban traffic areas. The Table A3.2 shows a better relation (i.e. lower RMSE, higher R2, smaller intercept and slope closer to 1) between station measurements and the interpolated values of the corresponding grid cells at either rural, urban background or urban traffic areas than it does at the point cross-validation predictions. That is because the simple comparison between point measurements and the gridded interpolated values shows the uncertainty at the actual station locations (points), while the point crossvalidation prediction simulates the behaviour of the interpolation at point positions assuming no actual measurement would exist at that point. The uncertainty at measurement locations is introduced partly by the smoothing effect of the interpolation and partly by the spatial averaging of the values in the 1 km grid cells. The level of the smoothing effect leading to underestimation at areas with high values is there smaller than in situations where no measurement is represented in such areas. For example, in urban background areas the predicted interpolation gridded annual average value in the separate urban background map will be about 37 µg/m3 at the corresponding station with the measurement value of 40 µg/m3. This means an underestimation of about 8%. It is a slightly less than the prediction underestimation of 10% at the same point location, when leaving out this one actual measurement point and the interpolation is done without this station (see the previous subsection). ETC HE Report 2025/5 117 A3.2 PM2.5 Technical details and uncertainty estimates for Map 2.5 with the PM2.5 annual average are presented in this section. Technical details on the mapping Like for PM10, an updated methodology as developed and tested under Horálek et al. (2019) has been used, i.e. including the land cover among supplementary data and using the traffic urban map layer. Table A3.3 presents the regression coefficients determined for pseudo PM2.5 stations data estimation, based on the 1205 rural and urban/suburban background and 458 urban/suburban traffic stations that have both PM2.5 and PM10 measurements available (see Section 2.1.1). Table A3.3 Parameters and statistics of linear regression model for generating pseudo PM2.5 annual average data for 2023 in rural and urban background (left) and urban traffic (right) areas c (constant) 22.0 42.3 b (PM10 measurement data) 0.625 0.444 a1 (surface solar radiation) -0.003 -0.004 a2 (latitude) -0.261 -0.552 a3 (longitude) 0.078 0.106 Adjusted R20.82 0.74 Standard Error [µg.m-3]1.7 2.0 Linear regresion model (LRM, Eq. A1.1) PM2.5, Annual avarage Rural and urban background areas Urban traffic areas Due to sufficient data coverage of PM2.5 measurements in urban background areas in some countries, the pseudo PM2.5 stations have not been used in urban background areas of these countries. This refers to Austria, Benelux, Czechia, Croatia, France, Germany, Ireland, Italy, Poland, Slovenia, Slovakia and Spain. Table A3.4 presents the estimated parameters of the linear regression models (c, a1, a2,…) and of the residual kriging (nugget, sill, range) and includes the statistical indicators of both the regression and the kriging of its residuals. Table A3.4 Parameters and statistics of linear regression model and ordinary kriging of PM2.5 annual average 2023 in rural, urban background and urban traffic areas for final combined map Rural areas Urban b. areas Urban tr.. areas c (constant) 0.55 0.56 0.73 a1 (log. CAMS model) 0.820 0.780 0.706 a2 (altitude GMTED) -0.00030 a3 (wind speed) -0.055 a4 (land cover NAT1) non signif. Adjusted R20.65 0.46 0.59 Standard Error [µg.m-3]0.26 0.29 0.25 nugget 0.028 0.023 0.016 sill 0.076 0.052 0.035 range [km] 1000 220 180 RMSE [µg.m-3]2.2 2.9 2.4 Relative RMSE [%] 27.0 25.7 21.9 Bias (MPE) [µg.m-3]-0.1 0.0 -0.1 Linear regresion model (LRM, Eq. A1.3) Ordinary kriging (OK) of LRM residuals LRM + OK of its residuals PM2.5 Annual average ETC HE Report 2025/5 118 The same supplementary data as in Horálek et al. (2019) has been used. Like in the case of PM10, the linear regression is applied on the logarithmically transformed data of both measurement and modelled PM2.5 values. Thus, the standard error and variogram parameters refer to these transformed data, whereas RMSE and bias refer to the interpolation after the back-transformation. The adjusted R2 and standard error are indicators for the quality of the fit of the regression relation. The adjusted R2 is 0.65 for the rural areas, 0.46 for urban background areas and 0.59 for urban traffic areas. Quite weaker regression relation in the urban background areas causes a higher impact of the interpolation part of the interpolation-regression-merging mapping methodology in these areas. RMSE and bias – highlighted in orange – are the cross-validation indicators, showing the quality of the resulting map; the bias indicates to what extent the predictions are underor overestimated on average. Only stations with PM2.5 measurement data are used for calculating the RMSE and the bias (i.e. the pseudo PM2.5 stations are not used). These statistical indicators are calculated excluding the pseudo stations because they are estimated values only, not actual measurement values. According to Denby et al (2011), the pseudo PM2.5 data does not satisfy the quality objectives for fixed monitoring alone. The pseudo stations are used as they improve the mapping estimate, whereas the actual measurements can be used for evaluating the quality of the map. For the future, it will be considered to quit the application of the PM2.5 pseudo stations as the current number of the actual PM2.5 measurement stations has increased over time such that the use of pseudo PM2.5 stations may not contribute enough any longer to improve the mapping estimates. Due to the lack of rural stations in Türkiye for PM2.5, no proper interpolation results could be presented for this country in a rural map, so the estimated PM2.5 values for Türkiye are not presented in the final map. Thus, the stations located in Türkiye have not been used in the uncertainty estimates (although used in the mapping process), as they lie outside the mapping area. Uncertainty estimated by cross-validation Table A3.4 shows that the absolute mean uncertainty of the final combined map of PM2.5 annual average expressed as RMSE is 2.62 µg/m3 for the rural areas, 2.9 µg/m3 for the urban background areas and 2.4 µg/m3 for the urban traffic areas. On the other hand, the relative mean uncertainty (Relative RMSE) of the final combined map of PM2.5 annual average is 27% for rural areas, 26% for urban background areas and 22% for urban traffic areas. These relative uncertainty values fulfil the data quality objectives for models as set in Annex I of the Air Quality Directive (EC, 2008). Figure A3.3 shows the cross-validation scatter plots, obtained according to Section A1.3, for different area types. The R2 indicates that about 75% of the variability is attributable to the interpolation for the rural areas, 78% for the urban background areas and 74% for the urban traffic areas. The scatter plots indicate that in areas with high concentrations the interpolation methods tend to underestimate the levels. E.g., in urban background areas an observed value of 25 µg/m3 is estimated in the interpolations to be about 22 µg/m3, which is an underestimated prediction of about 11%. ETC HE Report 2025/5 119 Figure A3.3 Correlation between cross-validated predicted and measurement values for PM2.5 annual average 2023 for rural (left), urban background (middle) and urban traffic (right) areas Comparison of point measurement values with the predicted grid value Like for PM10, a simple comparison has been made between the point observation values and interpolated prediction values spatially averaged in grid cells, in addition to the cross-validation. The comparison has been made primarily for the separate rural, urban background and urban traffic map layers at 1 km resolution. Next to this, the comparison has been done also for the final combined maps at the same 1 km resolution. Figure A3.4 shows the scatterplots for these comparisons. Figure A3.4 Correlation between predicted grid values from rural (upper left), urban background (upper middle) and urban traffic (upper right) map layer and final combined map (all bottom) (y-axis) versus measurements from rural (left), urban/suburban background (middle) and urban/suburban traffic stations (right) (x-axis) for PM2.5 annual average 2023 ETC HE Report 2025/5 120 The results of the point observation – point prediction cross-validation of Figure A3.3 and those of the point observation – grid averaged prediction validation Figure A3.4 for separate map layers and for the final combined map are summarised in Table A3.5. By comparing the scatterplots and the statistical indicators for separate rural, urban background and urban traffic map layers with the final combined maps, one can evaluate the level of representation of the rural, urban background and urban traffic areas in the final combined map. Similar results as for PM10 can be observed: the final combined map in 1 km resolution is fairly well representative for rural and urban background areas, but not for urban traffic areas. Like in the case of PM10 and for the same reason, Table A3.5 shows a better correlated relation with the station measurements for the simply interpolated gridded values than for the point cross-validation predictions, at all area types. The uncertainty at measurement locations is introduced partly by the smoothing effect of the interpolation and partly by the spatial averaging of the values in the 1 km grid cells. For example, in urban background areas the predicted interpolation gridded value in the final map will be about 23 µg/m3 at the corresponding station with the measurement value of 25 µg/m3 (calculated based on the linear regression equation), which coincides with an underestimation of about 8%. Table A3.5 Statistical indicators from the scatter plots for the predicted grid values from separate (rural, urban background or urban traffic) map layers and final combined map versus the measurement point values for rural (upper left), urban background (upper right) and urban traffic (bottom left) stations for PM2.5 annual average 2023 RMSE Bias R2Lin. r. equation RMSE Bias R2Lin. r. equation Cross-val. prediction, separate (r or ub) map layer 2.2 -0.1 0.749 y = 0.709x + 2.3 2.9 0.0 0.780 y = 0.794x + 2.3 Grid prediction, 1 km res., separ. (r or ub) map layer 1.6 -0.2 0.861 y = 0.765x + 1.7 2.0 0.0 0.845 y = 0.852x + 1.7 Grid prediction, 1 km res., final merged map 1.6 0.0 0.849 y = 0.787x + 1.7 2.0 0.0 0.831 y = 0.842x + 1.8 RMSE Bias R2Lin. r. equation Cross-val. prediction, urban traffic map layer 2.4 -0.1 0.738 y = 0.725x + 2.8 Grid prediction, 1 km res., urban traffic map layer 1.7 0.2 0.853 y = 0.873x + 1.5 Grid prediction, 1 km res., final merged map 2.3 -0.6 0.787 y = 0.861x + 1.0 Rural background stations Urban/suburban background stations PM2.5, Annual average PM2.5, Annual average Urban/suburban traffic stations A3.3 Ozone In this section, the technical details and the uncertainty estimates are presented for the maps of ozone health-related indicators 93.2 percentile of maximum daily 8-hour means, peak season average of maximum daily 8-hour means, SOMO35 and SOMO10 (Maps 3.1, 3.3-3.5), as well as for the maps of ozone vegetation-related indicators AOT40 for vegetation and AOT40 for forests (Maps 3.7 and 3.8). Next to this, the details of PODY (i.e. POD6 and POD1) maps are presented. Technical details on the mapping Table A3.6 presents the estimated parameters of the linear regression models and of the residual kriging, including the statistical indicators of both the regression and the kriging. The adjusted R2 and standard error show the quality of the fit of the regression relation. For the rural areas, all indicators show the value of the adjusted R2 between 0.37 and 0.53. For the urban areas, the adjusted R2 is in between 0.26 and 0.28 for all indicators apart from SOMO10, for which it is 0.12. ETC HE Report 2025/5 121 For the vegetation-related indicators the urban maps are not constructed. RMSE and bias – highlighted by orange – are the cross-validation indicators, showing the quality of the resulting map. Table A3.6 Parameters and statistics of linear regression model and ordinary kriging for ozone indicators 93.2 percentile of maximum daily 8-hourly means, peak season average of maximum daily 8-hourly means, SOMO35 and SOMO10 in rural and urban areas and for ozone indicators AOT40 for vegetation and for forests in rural areas for 2023 AOT40v AOT40f Rur. ar. Urb. ar. Rur. ar. Urb.ar. Rur. ar. Urb.ar. Rur. ar. Urb.ar. Rur. ar. Rur. ar. c (constant) -6.5 38.8 6.0 33.9 332 2007 4268 6571 3422 3802 a1 (CAMS model) 1.11 0.76 0.93 0.68 0.90 0.58 0.76 0.62 0.91 0.90 a2 (altit. GMTED) 0.0149 0.0119 3.17 3.79 7.81 14.31 a3 (wind speed) -2.98 -2.05 -335.6 n. sign. a4 (s. solar rad.) n.sign. n.sign. n.sign. n.sign. n.sign. 0.22 n.sign. n. sign. n. sign. n. sign. Adjusted R20.48 0.28 0.47 0.26 0.47 0.27 0.37 0.12 0.46 0.53 St. Err. [µg/m3·x]* 9.4 13.7 7.4 11.0 1779 2011 2732 3572 5513 9031 Nugget 51 69 33 35 2.2E+06 1.3E+06 4.9E+06 3.8E+06 1.4E+07 4.1E+07 Sill 59 109 38 61 2.6E+06 4.0E+06 5.3E+06 6.2E+06 2.5E+07 7.0E+07 Range [km] 700 120 390 80 690 1000 700 380 1000 1000 RMSE [µg/m3·x]* 9.1 11.6 7.2 8.9 1764 1759 2648 2799 5125 8747 Rel. RMSE [%] 8.2 10.6 7.8 9.9 33.0 39.0 12.2 14.3 35.5 35.8 Bias [µg/m3·x]* 0.1 -0.1 0.1 -0.2 3 101 14 -19 -13 -22 SOMO10 SOMO35 Linear regresion model (LRM, Eq. A1.3) Ord. krig. (OK) of LRM LRM + OK of its residuals 93.2 perc. Peak seas. avg * Units: 93.2 percentile of daily 8-h maximums and peak season average of daily 8-h maximums: [µg/m3], SOMO35 and SOMO10: [µg/m3·d], AOT40v and AOT40f: [µg/m3·h]. Uncertainty estimated by cross-validation The basic uncertainty analysis is provided by cross-validation. Table A3.6 shows both absolute and relative mean uncertainty, expressed by RMSE and Relative RMSE. The relative mean uncertainty of the 2023 ozone map is around 7-11% for the 93.2 percentile of maximum daily 8-h means and the peak season average of maximum daily 8-h means, around 33-39% for SOMO35, around 12-15% for SOMO10 and around 35-36% at AOT40 indicators. The small levels of the relative uncertainty for the 93.2 percentile of maximum daily 8-h means, the peak season indicator and SOMO10 are highly influenced by the low ratio between the relevant standard error and mean calculated based on all annual station concentration data: for these three indicators the ratio is at the level of about 0.110.19, while for SOMO35 and for both AOT40 indicators it is at the level of about 0.46-0.54. Figure A3.5 shows the cross-validation scatter plots for both the rural and urban areas of the 2023 map for the health-related ozone indicators. Based on the R2, one can see that for the health-related ozone indicators, about 41-52% of the variability is attributable to the interpolation. The scatter plots indicate that the higher values are underestimated and the lower values somewhat overestimated by the interpolation method; a typical smoothing effect inherent to the interpolation method with the linear regression and its residuals kriging. For example, in the case of the 93.2 percentile of daily 8-h maximums, in urban areas (Figure A3.5, upper right panel) an observed value of 150 µg/m3 is estimated in the interpolation as 130 µg/m3, which is 13% lower. Or, in the case of SOMO35, in rural areas (Figure A3.5, lower middle left panel) an observed value of 8 500 µg/m3·d is estimated in the interpolation as about 6 900 µg/m3·d, which is 19% lower. In addition, an overestimation at the lower end of predicted values occurred. ETC HE Report 2025/5 128 Uncertainty estimated by cross-validation Table A3.10 shows both absolute and relative mean uncertainty, expressed by RMSE and Relative RMSE. The absolute mean uncertainty of the final combined map of NO2 annual average expressed as RMSE is 1.9 µg/m3 for the rural areas, 4.4 µg/m3 for the urban background areas and 6.2 µg/m3 for the urban traffic areas. For the NOx rural map it is 3.3 µg/m3. The relative mean uncertainty of the NO2 annual average map is 34% for rural areas, 29% for urban background areas and 26% for urban traffic areas. The NOx annual average rural map has a relative mean uncertainty of 43%. Figure A3.9 shows the point observation – point prediction cross-validation scatter plots for NO2 annual average. The R2 indicates that about 75% of the variability is attributable to the interpolation for the rural areas, while for the urban background areas it is 69% and for the urban traffic 64%. Figure A3.9 Correlation between cross-validated predicted and measurement values for NO2 annual average 2023 for rural (left), urban background (middle) and urban traffic (right) areas Like in the case of other pollutants, the cross-validation scatter plots show the underestimation of predictions at high concentrations at locations with no measurements. For example, in urban background areas an observed value of 40 µg/m3 is estimated in the interpolations to be about 34 µg/m3, which is an underestimated prediction of about 16%. Figure A3.10 shows the cross-validation scatter plot for NOx annual average rural map. The R2 indicates that about 65% of the variability is attributable to the interpolation. Figure A3.10 Correlation between cross-validated predicted and measurement values for NOx annual average 2023 for rural areas ETC HE Report 2025/5 129 Comparison of point measurement values with the predicted grid value Next to the above presented cross-validation, a simple comparison was made between the point observation values and interpolated predicted 1 km and 2 km grid values, respectively. For NO2 annual average, the comparison has been made primarily for the separate map layers at 1 km resolution. Besides, the comparison has been done also for the final combined map. Table A3.11 presents the results of this comparison, together with the results of cross-validation prediction of Figure A3.10. One can conclude that the final combined map in 1 km resolution is representative for rural and urban background areas, but not for urban traffic areas. Table A3.11 Statistical indicators from the scatter plots for the predicted grid values from separate (rural, urban background or urban traffic) map layers and final combined map versus the measurement point values for rural (upper left), urban background (upper right) and urban traffic (bottom left) stations for NO2 annual average 2023 RMSE Bias R2Lin. r. equation RMSE Bias R2Lin. r. equation Cross-val. prediction, separate (r or ub) map layer 1.9 0.0 0.745 y = 0.756x + 1.4 4.4 0.0 0.686 y = 0.738x + 4.0 Grid prediction, 1 km res., separ. (r or ub) map layer 1.3 -0.1 0.877 y = 0.835x + 0.8 2.7 0.2 0.880 y = 0.846x + 2.6 Grid prediction, 1 km res., final combined map 2.2 0.4 0.736 y = 0.930x + 0.8 3.2 0.8 0.846 y = 0.884x + 2.5 RMSE Bias R2Lin. r. equation Cross-valid. prediction, urban traffic map layer 6.2 -0.1 0.639 y = 0.669x + 7.7 Grid prediction, 1 km res., urban traffic map layer 3.5 0.0 0.889 y = 0.816x + 4.4 Grid prediction, 1 km res., final combined map 9.6 -7.0 0.612 y = 0.504x + 4.8 Rural backgr. stations Urban/suburban background stations NO2, Annual average NO2, Annual average Urban/suburban traffic stations Table A3.12 presents the cross-validation results of Figure A3.11 and those of the point observation – grid averaged prediction validation for the rural map of NOx annual average. Table A3.12 Statistical indicators from the scatter plots for predicted point values based on crossvalidation and predicted grid values from rural 2 km resolution map versus measurement point values for rural background stations for NOx annual average 2023 RMSE Bias R2Linear regression equation Cross-valid. prediction, rural map 3.3 0.1 0.647 y = 0.660x + 2.63 Grid prediction, 2 km resolution, rural map 2.7 0.1 0.765 y = 0.720x + 2.16 Rural background stations NOx A3.5 BaP In this section, the technical details and the uncertainty estimates for Map 5.1 of BaP annual average are presented. Technical details on the mapping The methodology as developed and tested in Horálek et al. (2022b) has been used. Table A3.13 presents the regression coefficients determined for pseudo BaP stations data estimates, based on the 341 rural and urban/suburban background that have both BaP and PM2.5 measurements available (see Annex 1 Section A1.1). Looking at the parameters of the regression, one can note that the adjusted R2 of 0.62 is a quite poor correlation. Based on this and in agreement with Horálek et al. (2022b), the pseudo stations have only been used in areas with a significant lack of the BaP measurements. The pseudo stations have been applied for countries and areas, as follows. For the rural areas: All the mapping area, apart from Austria, Benelux, Czechia, Germany, Poland, Slovakia, Spain, and Italy north ETC HE Report 2025/5 130 of 44 degrees latitude. For the urban background areas: Finland, Greece, Iceland, Norway, Portugal, Sweden and west Balkan countries (namely, Albania, Bosnia and Herzegovina, Montenegro, Northern Macedonia, and Serbia including Kosovo). Table A3.13 Parameters and statistics of linear regression model for generating pseudo BaP annual average data for 2023 in rural and urban background areas c (constant) -6.31 a1 (PM2.5 annual average) 0.148 a2 (latitude) 0.061 a3 (longitude) 0.045 a4 (land cover NAT_1km) -0.0071 a5 (land cover NAT_5km_r) 0.0110 Adjusted R20.62 Nonlinear regresion model (NLRM, Eq. A1.3) Rural and urban background areas Table A3.14 presents the estimated parameters of the linear regression models (c, a1, a2,…) and of the residual kriging (nugget, sill, range) and includes the statistical indicators of both the regression and the kriging of its residuals. The same supplementary data as in Horálek et al. (2022b) has been used. Table A3.14 Parameters and statistics of linear regression model and ordinary kriging of BaP annual average 2023 in rural, urban background and urban traffic areas for the final combined map Rural areas Urban b. areas c (constant) 0.74 1.43 a1 (log. EMEP model) 0.370 0.442 a2 (altitude GMTED) -0.00120 a3 (wind speed) -0.354 a4 (temperature) n. sign. -0.12 a5 (land cover NAT_1km) -0.0059 Adjusted R20.37 0.42 Standard Error [ng/m3]1.01 0.89 nugget 0.333 0.187 sill 0.794 0.678 range [km] 300 530 RMSE [ng/m3]0.37 0.55 Relative RMSE [%] 136.5 67.3 Bias (MPE) [ng/m3]0.06 0.01 Linear regresion model (LRM, Eq. A1.3) Ordinary kriging (OK) of LRM residuals LRM + OK of its residuals BaP Annual average The adjusted R2 is 0.37 for the rural areas and 0.42 for urban background areas. Uncertainty estimated by cross-validation Table A3.14 shows that the absolute mean uncertainty of the final combined map of BaP annual average expressed as RMSE is 0.37 ng/m3 for the rural areas and 0.55 ng/m3 for the urban background areas. The RRMSE of this map is 136.5% for rural areas and 67.3% for urban background areas. The cross-validation relative uncertainty RRMSE is still at the considerably higher level (especially in the rural areas) compared to the 60%, being the data quality objective for the modelling uncertainty in the European directive (EC, 2004). ETC HE Report 2025/5 131 Annex 4 Concentration maps including stations Throughout the report, the concentration maps presented do not include the concentration values measured at the stations. The reason is to better visualise the health-related indicators with their distinct concentration levels at the more fragmented and smaller urban areas. As presented in Annex 3, the kriging interpolation methodology somewhat smooths the concentration field. Therefore, it is valuable to present in this Annex 4 the indicator maps including the concentration values resulting from the measurement data at the stations. These points provide important additional visual information on the smoothing effect caused by the interpolation. For instance, maps A4.1 and A4.2 present PM10 indicators annual average and 90.4 percentile of daily means and include the stations points used in the interpolation. They correspond to Maps 2.1 and 2.3 of the main report, which do not have stations. Table A4.1 provides an overview of the maps in the main report and the corresponding maps including stations point values as presented in this annex. Both the rural and the urban/suburban background stations and also urban/ traffic stations for PM and NO2 are included in the maps of the health-related indicators, while the rural stations only are shown in the maps of vegetation related indicators. For PM2.5, NOx and BaP, only the stations with relevant measured data (i.e. not the pseudo stations) are presented. Table A4.1 Overview of maps presented in this Annex 4 and their relationship with the maps presented in the main report Air pollutant Indicator Map including stations Map without stations PM10 Annual average A4.1 2.1 90.4 percentile of daily means A4.2 2.3 PM2.5 Annual average A4.3 2.5 Ozone 93.2 percentile of maximum daily 8-hour means A4.4 3.1 left Peak season average of maximum daily 8-hour means A4.5 3.3 SOMO35 A4.6 3.4 SOMO10 A4.7 3.5 AOT40 for vegetation (a) A4.8 3.7 left AOT40 for forests (a) A4.9 3.8 NO2 Annual average A4.10 4.1 NOx Annual average (a) A4.11 4.3 BaP Annual average A4.12 5.1 (a) Rural map, applicable for rural areas only. ETC HE Report 2025/5 132 Map A4.1 Concentration map of PM10 annual average including station measurement values, 2023 ETC HE Report 2025/5 133 Map A4.2 Concentration map of PM10 indicator 90.4 percentile of daily means including station measurement values, 2023 ETC HE Report 2025/5 134 Map A4.3 Concentration map of PM2.5 annual average including station measurement values, 2023 ETC HE Report 2025/5 135 Map A4.4 Concentration map of ozone indicator 93.2 percentile of maximum daily 8-hour means including station measurement values, 2023 ETC HE Report 2025/5 136 Map A4.5 Concentration map of ozone peak season average of maximum daily 8-hour means, 2023 ETC HE Report 2025/5 137 Map A4.6 Concentration map of ozone indicator SOMO35 including station measurement values, 2023 The European Topic Centre on Human health and the environment (ETC HE) is a consortium of European institutes under contract of the European Environment Agency. European Topic Centre on Human Health and the Environment https://www.eionet.europa.eu/etcs/etc-he