Interim air quality maps of EEA member and cooperating countries for 2024. PM, O3 and NO2 spatial estimates. ETC HE Report 2025/4.

ETC HE Report 2025/4 Interim air quality maps of EEA member and cooperating countries for 2024 PM, O3, and NO2 spatial estimates Authors: Jan Horálek (CHMI), Leona Vlasáková (CHMI), Markéta Schreiberová (CHMI), Philipp Schneider (NILU), Nina Benešová (CHMI), Ondřej Vlček (CHMI) ETC HE Report 2024/X Cover design: EEA Cover image: Maps showing PM2.5 annual average (top), O3 indicator peak season average of maximum daily 8-hour means (bottom left) and NO2 annual average (bottom right) for 2024. (This report´s Maps 2.3, 3.1 and 4.1.) Layout: EEA / ETC HE (CHMI) Publication Date: 21 November 2025 DOI: 10.5281/zenodo.17601140 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, J., Schreiberová, M., Schneider, P., Benešová, N., Vlček, O. (2025). Interim air quality maps of EEA member and cooperating countries for 2024. PM, O3 and NO2 spatial estimates (Eionet Report – ETC HE 2025/4). European Topic Centre on Human Health and the Environment. (https://doi.org/10.5281/zenodo.17601140) 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., klarFAKT e.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 ....................................................................................................................... 4 Data availability ............................................................................................................................. 4 Summary ....................................................................................................................................... 5 1 Introduction ........................................................................................................................... 6 2 Particulate matter .................................................................................................................. 8 2.1 PM10 annual average ...................................................................................................... 8 2.2 PM2.5 annual average ................................................................................................... 11 3 Ozone ................................................................................................................................... 16 3.1 O3 – peak season average of maximum daily 8-hour means ....................................... 16 4 Nitrogen dioxide .................................................................................................................. 19 4.1 NO2 annual average ..................................................................................................... 19 5 Conclusions .......................................................................................................................... 23 List of abbreviations .................................................................................................................... 24 References ................................................................................................................................... 25 Annex 1 Methodology ................................................................................................... 27 A1.1 Spatial mapping methodology ..................................................................................... 27 A1.2 Pseudo station data estimation ................................................................................... 28 A1.3 Uncertainty analysis ..................................................................................................... 29 A1.4 Validation ..................................................................................................................... 29 A1.5 Population exposure calculation and estimation of trends ......................................... 30 A1.6 Geographical division of the considered European area used in the assessment ...... 30 Annex 2 Data used ........................................................................................................ 31 A2.1 Air quality monitoring data .......................................................................................... 31 A2.2 Chemical transport modelling (CTM) data ................................................................... 32 A2.3 Satellite data ................................................................................................................ 33 A2.4 Other supplementary data........................................................................................... 33 Annex 3 Technical details and uncertainties of interim maps ...................................... 34 A3.1 Particulate matter PM10 ............................................................................................... 34 A3.2 Particulate matter PM2.5 .............................................................................................. 35 A3.3 Ozone ........................................................................................................................... 36 A3.4 Nitrogen dioxide .......................................................................................................... 37 Annex 4 Validation of 2023 interim maps and exposure estimates ............................. 39 A4.1 Concentration maps .................................................................................................... 39 A4.2 Population exposure .................................................................................................... 45 ETC HE Report 2025/5 4 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). There were other ETC HE contributions. The air quality monitoring data were extracted from the AQ e-reporting database by María Colina and Jaume Targa (4sfera, Spain). The data preparation and the analysis were assisted by Pavel Kurfürst and Lucie Školoudová (CHMI, Czechia). Data availability The interim maps presented in this report are available in the GeoTIFF format on the EEA geospatial data catalogue, see below. PM2.5, annual average, 2024: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/72a465a6-4af1-4b0c-ade5ddfa23cc19fe (accessed 7 October 2025). O3, peak season average of maximum daily 8-hour means, 2024: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/839407de-bb7c-4cc4-91b9a8282b05203a (accessed 7 October 2025). NO2, annual average, 2024: https://sdi.eea.europa.eu/catalogue/srv/eng/catalog.search#/metadata/9606e270-8ce5-4ce9-94945554290ab888 (accessed 7 October 2025). ETC HE Report 2025/5 5 Summary This report presents the interim air quality maps for the area of the member and cooperating countries of the European Environment Agency (EEA) for the year 2024. These maps are based on the nonvalidated, up-to-date measurement data and CAMS Ensemble Forecast modelling results, together with other supplementary data. The interim maps and further assessment present annual average particulate matter (both PM10 and PM2.5) concentration, annual average nitrogen dioxide (NO2) concentration, and ground-level ozone (O3) concentrations, expressed as the peak season average of maximum daily 8-hour means. The share of population living in the considered (i.e. presented) European area exposed to annual average PM10 concentration above the current (EU, 2008) limit value (LV) of 40 µg/m3 is estimated to be 0.1%; for the EU-27, no population is estimated to be exposed to LV exceedances. In comparison, 29% of the population in both the considered European and the EU-27 was exposed to annual average concentrations above the revised (EU, 2024) limit value of 20 µg/m3 to be attained in 2030 (LV2030). Almost 56% of both the considered European and the EU-27 populations have been exposed to annual average concentrations above the WHO Air Quality Guideline (AQG) level of 15 µg/m3. The populationweighted concentration of the PM10 annual average for 2024 for both the considered European countries and for the EU-27 is estimated to be about 17 µg/m3. The population-weighted concentration of the PM10 annual averages shows quite a steady decrease of about 0.5 µg/m3 per year in the period 2005-2024, with the second lowest concentration in this period being recorded in 2024. Regarding PM2.5, it is estimated that 0.1% of the population living in the considered (i.e. presented) European area has been exposed to concentrations above the EU annual LV of 25 µg/m3. For the EU27, less than 0.05% of the population is estimated to be exposed to those levels above the LV. In comparison, 41% of the considered European and 40% of the EU-27 population were exposed to annual average concentrations above the revised LV2030 of 10 µg/m3. About 97% of the population living in both the considered European area and the EU-27 has been exposed to concentrations above the WHO AQG level of 5 µg/m3. The population-weighted concentration of the PM2.5 annual average for 2024 is estimated to be 10 µg/m3 for both the EEA member and cooperating countries and the EU27. The population-weighted concentration of the PM2.5 annual averages shows a fairly steady decrease of about 0.5 µg/m3 per year in the period 2005-2024, with the lowest concentration in this period being recorded in 2024. Based on the interim map for 2024, it has been estimated that 100% of the population in both the considered European area and the EU-27 lived in regions where the O3 concentration was above the peak season average of maximum daily 8-hour means of 60 µg/m3. The population-weighted concentration of the O3 peak season average of maximum daily 8-hour means for 2024 for both the considered European and the EU-27 population is estimated to be about 89 µg/m3, which is a slightly lower value compared to both 2023 (90 µg/m3) and 2022 (92 µg/m3). The share of the population living in the European area considered in this report and the EU-27 area exposed to annual average NO2 concentration above the LV of 40 µg/m3 is estimated to be less than 0.1%. In comparison, 10% of both the considered European and the EU-27 population were exposed to annual average concentrations above the revised LV2030 of 20 µg/m3. Around 63% of the population living in both the considered European area and the EU-27 has been exposed to concentrations above the WHO AQG level of 10 µg/m3. The population-weighted concentration of the NO2 annual average for 2024 for both areas is estimated to be about 12 µg/m3. The population-weighted concentration for the NO2 annual average shows a steady decrease of about 0.7 µg/m3 per year in the period 2005-2024, with the lowest concentration in this period recorded in 2024. ETC HE Report 2025/5 6 1 Introduction European-wide air quality (AQ) annual maps have been routinely constructed under the ETC HE (and the previous consortia) since 2005 (Horálek, 2024a and references therein). The mapping methodology combines monitoring data, chemical transport model (CTM) results and other supplementary data using a linear regression model followed by kriging of the residuals produced from that model (‘residual kriging’). Separate mapping layers (rural, urban background and urban traffic, where relevant) are created separately and subsequently merged to the final map. In order to reflect the three steps applied, the methodology is called Regression – Interpolation – Merging Mapping (RIMM). The regular maps (i.e. maps presented under the ETC’s regular mapping reports, e.g. Horálek et al., 2024a) are based on the validated air quality monitoring data as stored in the EEA’s AQ e-reporting database (in the so-called E1a data set), the modelling results and other supplementary data. Due to the time schedule of the production and availability of the validated AQ measurement data, the regular RIMM maps of a year Y are typically available in May of year Y+2. Thus, the regular 2024 maps based on the validated data will be available around May 2026. This report presents the interim air quality maps for 2024 for the area of the EEA member and cooperating countries( 1 ) (and the three microstates of Andorra, Monaco and San Marino). These maps are based on the non-validated up-to-date (UTD) measurement data (as available in the E2a data set of the AQ e-reporting database) and the CAMS Ensemble Forecast modelling results, together with other supplementary data. The reason for the production of these interim maps is their earlier availability. The interim mapping approach was previously developed and evaluated, and consequently the interim maps of PM10, PM2.5, NO2 and ozone (O3) were recommended for regular production, see Horálek et al. (2021a, 2021b). To overcome data gaps in the E2a data in some areas, so-called pseudo stations data are used in areas with a lack of E2a stations, based on the regression relation between the E2a data from a year Y and the validated E1a data from a year Y-1, together with the ratio of the modelling results from years Y and Y-1. The use of the pseudo station data in the interim mapping is applied for PM10, PM2.5 and NO2. For O3, the data coverage of the E2a data is larger and the interim O3 maps might be constructed without the use of the pseudo stations. The interim maps are not produced for the area of Türkiye, due to the lack of E2a monitoring data from Turkish stations. In this report, interim 2024 maps for the PM10 annual average, the PM2.5 annual average, the NO2 annual average and the O3 indicator peak season average of maximum daily 8-hour means are presented. Also, the difference between the five-year mean 2019-2023 (where available) and 2024 and the inter-annual difference between 2023 and 2024 are discussed. In addition, population exposure estimates based on the concentration maps are briefly shown. Based on the analysis presented in Horálek et al. (2023a), in this report only basic exposure estimates are provided, without detailed information for individual countries. The exposure estimates are presented for five large European regions (northern Europe, western Europe, central Europe, southern Europe and southeastern Europe) ( 2 ), for the EU-27 and for the whole mapping area. Apart from this, the evolution of the overall population-weighted concentration in the 20-year period 2005-2024 is also shown, where available. ( 1 ) The EEA member countries are the 27 Member States of the European Union (EU-27), Iceland, Lichtenstein, Norway, Switzerland, and Türkiye. The EEA cooperating countries are Albania, Bosnia and Herzegovina, Montenegro, North Macedonia, Serbia, and Kosovo under the UN Security Council Resolution 1244/99. In this report, Kosovo is considered individually, without prejudice on its status. ( 2 ) Northern Europe: Denmark, Estonia, Finland, Iceland, Latvia, Lithuania, Norway, Sweden; Western Europe: Belgium, France (metropolitan) north of 45°, Ireland, Luxembourg, Netherlands; Central Europe: Austria, Czechia, Germany, Hungary, Liechtenstein, Poland, Slovakia, Slovenia, Switzerland; Southern Europe: Andorra, Cyprus, France (metropolitan) south of 45°, Greece, Italy, Malta, Monaco, Portugal (excl. Azores, Madeira), San Marino, Spain (without Canarias); South-eastern Europe: Albania, Bosnia and Herzegovina, Bulgaria, Croatia, Kosovo, Montenegro, North Macedonia, Romania, Serbia, Türkiye. ETC HE Report 2025/5 7 Apart from the 2024 interim maps, this report also presents the validation of the interim maps for 2023 as presented in Horálek et al. (2024b), based on the validated E1a data for 2023. It also presents the comparison of the exposure tables based on the interim maps for 2023 (Horálek et al., 2024b) against the exposure tables based on the regular maps (Horálek et al., 2025). Where applicable, the maps and population exposure estimates are presented with respect to the current EU LVs (EC, 2004; EC, 2008), the new EU LVs to be attained by 1 January 2030 (LV2030; EU, 2024) and the current WHO AQG levels (WHO, 2021). Table 1.1 presents a comparison of these different standards. Table 1.1 Overview of the current EU limit values (LV; EC, 2008) and new EU limit values to be attained by 1 January 2030 (LV2030; EU, 2024) and WHO Air Quality Guideline (AQG) levels (WHO, 2021) for the pollutants assessed in this report Pollutant Averaging period LV [µg/m3] LV2030 [µg/m3] WHO AQG [µg/m3] PM10 calendar year 40 20 15 PM2.5 calendar year 25 10 5 NO2 calendar year 40 20 10 O3 peak season (average of maximum daily 8-hour means) - - 60 Chapters 2, 3, and 4 present concentration maps and basic estimates of population exposure for particulate matter, O3 and NO2, respectively. Chapter 5 brings the conclusions. Annex 1 describes briefly the methodological aspects (including the geographical distribution of the considered area into five large regions) and Annex 2 presents the input data applied. Annex 3 provides the technical details of the maps and their uncertainty estimates. Annex 4 provides the validation of the interim maps and exposure estimates for 2023. ETC HE Report 2025/5 8 2 Particulate matter 2.1 PM10 annual average Map 2.1 presents the interim map for the PM10 annual average in 2024, as the result of interpolation and merging of the separate map layers as described in Annex 1 Section A1.1 (for technical details of this map, see Annex 3, Section A3.1). Red and dark red areas indicate concentrations above the EU annual LV of 40 µg/m3 (EC, 2008). Dark green and light green areas show concentrations below the revised EU annual LV of 20 µg/m3 (the new EU LV for the PM₁₀ annual average to be attained by 1 January 2030 (LV2030); EU, 2024). Dark green indicates the areas where the PM10 annual average concentration is below the WHO AQG level of 15 µg/m3 (WHO, 2021). Map 2.1 Interim concentration map of PM10 annual average, 2024 The map shows concentrations above the annual LV only in urban areas around some Balkan cities (in Bosnia and Herzegovina and Serbia). In addition to these countries, there are areas in the Po Valley, in Italy, and smaller disconnected areas in Bosnia and Herzegovina, Serbia, Albania, North Macedonia, Greece and Cyprus where PM10 concentrations of 30-40 µg/m3 have been estimated. Most of Europe shows concentrations below 20 µg/m3 (LV2030) with levels below 15 µg/m3 estimated for most of western and northern Europe. The relative mean uncertainty (i.e. the relative root mean square error, RRMSE) of this map is 19% for rural and 17% for urban background areas (Annex 3, Table A3.2). However, these uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations (i.e. not for regions covered by the pseudo stations only). The complete validation of the interim PM10 map can only be done when the validated E1a data for 2024 are available. For validation of the interim PM10 map for 2023 as presented in Horálek et al. (2024b), refer to Annex 4, Section A4.1. Map 2.2 shows the difference between the five-year mean 2019-2023 and 2024 and the inter-annual difference between 2023 and 2024 (using the regular maps for 2019-2023 and the 2024 interim map) ETC HE Report 2025/5 9 for PM10 annual average. Orange to red areas show an increase of PM10 concentration in 2024, while blue areas show a decrease. Compared to the five-year mean 2019-2023, the highest increases in annual mean PM10 concentrations (> 5 µg/m3) are observed mainly in parts of Greece and Italy. Increases bigger than 2 µg/m3 are observed in different parts of southern, south-eastern Europe and central Europe. The highest decreases (> 5 µg/m3) were observed in parts of some Balkan countries. Moderate decreases (> 2 µg/m3) were observed across a large area of France and in several scattered regions throughout Europe. No change or slight increases/decreases of about 2 µg/m3 are observed in the rest of the mapped area. Based on the map of the inter-annual difference between 2023 and 2024, increases bigger than 2 µg/m3 are observed in parts of southern, south-eastern Europe (parts of Portugal, Spain, parts of different Balkan countries and Cyprus), central Europe (much of Hungary, parts of Slovenia, Austria, Slovakia, Czechia, Poland, Germany) and northern Europe (the Baltic states). Decreases were observed in parts of France, Italy, Switzerland and Serbia. No change or slight increases/decreases of about 2 µg/m3 are observed in the rest of the mapped area. Map 2.2 Difference in concentrations between five-year mean 2019-2023 (left) or 2023 (right) and 2024 (based on the interim map) for PM10 annual average Based on the mapping results and the population density data, the population exposure estimates have been calculated. Table 2.1 gives the population frequency distribution for a limited number of exposure classes and the population-weighted concentration for five large European regions, for EU27 and for the total presented area. The exposure estimate for individual countries is not presented, due to their high uncertainty. As presented in Horálek et al. (2023a), the exposure estimates based on ETC HE Report 2025/5 16 3 Ozone 3.1 O3 – peak season average of maximum daily 8-hour means In September 2021, the WHO introduced a new AQG for O3. The new long-term AQG level for O₃ is set at 60 µg/m³, expressed as the average of maximum daily 8-hour mean O₃ concentration of the socalled peak season. The peak season is defined as the six consecutive months of the year with the highest six-month running-average O3 concentration (WHO, 2021). Map 3.1 presents the interim 2024 map for the O3 indicator peak season average of maximum daily 8hour means. The map is a result of merging separate rural and urban interpolated map layers as described in Annex 1 Section A1.1 (for technical details of this map, see Annex 3, Section A3.3). Red and purple areas show values above 100 µg/m3, while the dark green areas show where the values of the peak season indicator are below the WHO AQG level of 60 µg/m3 (WHO, 2021). Generally, southern Europe shows higher concentrations of the O3 peak season indicator than northern Europe. Map 3.1 Interim concentration map of O3 indicator peak season average of maximum daily 8hour means, 2024 The map shows that in 2024 areas with a peak season average of maximum daily 8-hour means exceeding 60 µg/m³ cover the entire considered region. Lowest values (< 80 µg/m³) are observed in northern Europe (parts of Sweden, Finland, Estonia and Latvia), as well as in Ireland, and in a large area of France. Higher values (> 100 µg/m³) are recorded in northern Italy, central Spain, and various parts of the Balkan countries and Cyprus. Values of 80–100 µg/m³ are observed in the remaining mapped area of Europe. The relative mean uncertainty (RRMSE) of this map is 7% for rural and 8% for urban background areas (Annex 3, Table A3.5). However, these uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim O3 map can only be done when the validated E1a data for 2024 are available. ETC HE Report 2025/5 17 Map 3.2 shows the inter-annual difference between 2024 and 2023 for the O3 indicator peak season average of maximum daily 8-hour means. Orange to red areas show an increase in O3 concentration in 2024, while blue areas show a decrease. Compared to 2023, most regions in Scandinavia, Ireland, central Europe, Spain and Italy experienced no significant changes or only slight variations (±4 µg/m³) in the O3 peak season average of maximum daily 8-hour means. A noticeable increase in the O3 peak season average is observed in the Balkan countries, particularly in Greece, Bulgaria and Romania, Cyprus, southern Poland and parts of Hungary. Conversely, a decrease is evident in large areas of France, parts of Germany, Italy, Switzerland, Portugal, Spain, the Baltic states and parts of Scandinavia. A decrease greater than 10 µg/m³ is observed in large areas of Croatia and Bosnia and Herzegovina. Map 3.2 Difference in concentrations between 2024 (based on the interim map) and 2023 for O3 indicator peak season average of maximum daily 8-hour means Based on the mapping results and the population density data, the population exposure estimate has been calculated. Table 3.1 gives the population frequency distribution for a limited number of exposure classes and the population-weighted concentration for large European regions, for EU-27 and for the total mapping area. The exposure estimate for individual countries is not presented, due to their high uncertainty. As presented in Horálek et al. (2023a), the exposure estimates based on interim maps give good results for the total area and the EU-27, but somewhat poorer results for individual countries. Based on the interim map for 2024, it has been estimated that 100% of both the considered European population and EU-27 population lived in areas where the O3 concentration was above the peak season average of maximum daily 8-hour means of 60 µg/m3. The population-weighted concentration of the O3 peak season average of maximum daily 8-hour means for 2024 for both the considered European and the EU-27 population is estimated to be around 89 µg/m3. ETC HE Report 2025/5 18 Table 3.1 Population exposure and population-weighted concentration, O3 indicator peak season average of maximum daily 8-hour means, 2024, based on an interim map Area Population [inhbs·1000] O3 – peak season indicator, exposed population, 2024 [%] O3 – peak s. indic. < 60 µg/m3 60 -80 µg/m3 80 -90 µg/m3 90 -100 µg/m3 100 -120 µg/m3 > 120 µg/m3·d Pop. weighted [µg/m3inhbs] Northern Europe 34 237 70.45 29.50 0.05 78.3 Western Europe 86 767 45.90 53.40 0.69 0.01 81.1 Central Europe 166 272 0.24 53.97 44.54 1.25 90.2 Southern Europe 143 256 5.38 24.25 45.09 25.27 0.00 94.3 South-eastern Europe 46 096 19.13 38.57 38.59 3.71 87.7 Total 476 628 16.6 41.7 33.3 8.4 0.0 88.7 EU-27 444 765 16.6 41.7 32.8 8.9 0.0 88.8 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. 5-year mean, i.e. 5-year mean 2019-2023. Diff., i.e. difference concentrations between 2024 and 5-year mean 2019-2023. Figure 3.1 shows, for the whole considered area, the frequency distribution of the O3 peak season average of maximum daily 8-hour means for population exposure classes of 1 µg/m3. The highest population frequency is found for classes between ca 75 and 95 µg/m3. For classes above 95 µg/m3, a sharp decline of the population frequency can be seen. Figure 3.1 Population frequency distribution, O3 indicator peak season average of maximum daily 8-hour means, 2024, based on the interim map. The WHO AQG level (60 µg/m3) is marked by the green line The map of the O3 peak season indicator has been prepared for the third year only, so no evolution in a longer period can be shown. It is estimated that the mean population-weighted concentration of the peak season indicator shows a slightly lower value for 2024 (89 µg/m3), compared to both 2023 (90 µg/m3) and 2022 (92 µg/m3). ETC HE Report 2025/5 19 4 Nitrogen dioxide 4.1 NO2 annual average Map 4.1 presents the interim map for the NO2 annual average in 2024, as the result of interpolation and merging of the separate map layers as described in Annex 1 Section A1.1 (for technical details of this map, see Annex 3, Section A3.4). Red and purple areas indicate concentrations above the annual LV of 40 µg/m3 (EC, 2008). Green areas show concentrations below the revised EU annual LV of 20 µg/m3 (LV2030, to be attained by 2030). Dark green areas indicate concentrations below 10 µg/m3, being the WHO AQG level (WHO, 2021). According to Map 4.1, no areas where NO2 concentrations exceeded the annual LV of 40 µg/m³ were observed. Some cities, particularly capitals in southern and south-eastern Europe, show NO2 levels above 20 µg/m³ (new LV2030). The urbanized areas of several major cities fall within the range of annual average NO2 concentrations between 10 and 20 µg/m³. Most of Europe shows NO2 levels below the AQG level of 10 µg/m³, while a larger region with concentrations between 10 and 20 µg/m³ is found in the Po Valley. It should be noted that the interpolated map is created at 1 km resolution only. Although the urban traffic map layer is used in the map creation, the traffic locations are smoothed in the final map at 1 km resolution. Thus, the map as such refers to the rural and urban background situations, while the values above the NO2 annual LV occur mostly at local hotspots such as dense traffic locations. Such concentrations (although they occurred at some urban traffic locations in 2024) are not visible in the 1 km resolution Map 4.1. Map 4.1 Interim concentration map of NO2 annual average, 2024 The relative mean uncertainty (RRMSE) of this map is 30% for rural areas and 24% for urban background areas (Annex 3, Table A3.7). However, these uncertainty estimates are based on the nonvalidated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim NO2 map can only be done when the validated E1a data for 2024 are available. For such ETC HE Report 2025/5 20 validation of the interim NO2 map for 2023 as presented in Horálek et al. (2024b), see Annex 4, Section A4.1. Map 4.2 shows the difference between five-year mean 2019-2023 and 2024 and the inter-annual difference between 2024 and 2023 (using the regular maps for 2019-2023 and the 2024 interim map) for the NO2 annual average. Orange to red areas show an increase of NO2 concentration in 2024, while blue areas show a decrease. Compared to the five-year mean 2019-2023, no change or slight increases/decreases of about 2 µg/m3 are observed in most of the European mapped area. On the other hand, relatively continuous areas in northern Italy, northern Germany and the Benelux countries show a decrease in annual average NO2 bigger than 2 µg/m3. Based on the map of the inter-annual difference between 2024 and 2023, there is no change or a slight increase/decrease (±2 µg/m³) in annual average NO2 concentrations in almost the entire considered (i.e. presented) European area. Nevertheless, increases greater than 2 µg/m³ are evident in parts of Greece, Bulgaria, North Macedonia, and Albania, while decreases of more than 2 µg/m³ were observed mainly in northern Italy. Map 4.2 Difference in concentrations between five-year mean 2019-2023 (left) or 2023 (right) and 2024 (based on the interim map) for NO2 annual average Based on the mapping results and the population density data, the population exposure estimate has been calculated. Table 4.1 gives the population frequency distribution for a limited number of exposure classes and the population-weighted concentrations for large European regions, for EU-27 and for the total mapping area. The exposure estimates for individual countries are not presented, due to their high uncertainty. As presented in Horálek et al. (2023a), the exposure estimates based on ETC HE Report 2025/5 21 interim maps give good results for the total area and the EU-27, but somewhat poorer results for individual countries. Based on the interim map, it is estimated that less than 0.1% of both the considered European and the EU-27 population has been exposed to concentrations above the EU annual LV of 40 µg/m3. In comparison, 10% of both the considered European and the EU-27 population were exposed to annual average concentrations above the revised LV2030 of 20 µg/m3. Most of them lived in southern and south-eastern Europe. Around 63% of both the considered European and the EU-27 population has been exposed to concentrations exceeding 10 µg/m3 (being the WHO AQG level). The populationweighted concentration of the NO2 annual average for 2024 for both the considered European and the EU-27 population is estimated to be 12.4 µg/m3. Table 4.1 Population exposure and population-weighted concentration, NO2 annual average, 2024, based on interim map Area Population [inhbs·1000] NO2 – annual average, exposed population, 2024 [%] Population-weighted concentration < 10 µg/m3 10-20 µg/m3 20-30 µg/m3 30-40 µg/m3 40-45 µg/m3 > 45 µg/m3 2024 5-year mean Diff. Northern Europe 34 237 84.1 15.5 0.5 6.8 7.8 -1.0 Western Europe 86 767 44.7 49.3 4.9 1.0 11.1 13.8 -2.7 Central Europe 166 272 34.6 61.0 4.3 0.1 12.0 14.2 -2.2 Southern Europe 143 256 29.3 51.2 18.1 1.2 0.1 0.1 14.2 16.3 -2.0 South-eastern Europe 46 096 22.5 62.1 14.6 0.8 14.4 16.0 -1.5 Total 476 628 37.0 52.9 9.3 0.7 0.0 0.0 12.4 14.3 -1.9 EU-27 444 765 37.4 52.4 9.3 0.7 0.0 0.0 12.4 14.5 -2.1 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 2019-2023. Diff., i.e. difference concentrations between 2024 and 5-year mean 2019-2023. Figure 4.1 shows, for the whole considered area, the population frequency distribution for exposure classes with a width of 1 µg/m3. One can see the highest population frequency for classes between 6 and 17 µg/m3, continuous decline of population frequency for classes between 18 and 25 µg/m3 and continuous mild decline of population frequency for classes between 25 and 40 µg/m3. Figure 4.1 Population frequency distribution, NO2 annual average 2024, based on the interim map. The WHO AQG level (10 µg/m3) is marked by the green line, the revised EU annual LV2030 (20 µg/m3) is marked by the yellow line, and the EU annual LV (40 µg/m3) is marked by the red line ETC HE Report 2025/5 22 For changes in the population-weighted concentration of the NO2 annual average in the period 20052024, see Figure 4.2. For the previous years, mapping results as presented in Horálek et al. (2025 and references therein) have been used. As the regular maps are not available for all years, in addition an alternative mapping results prepared based on a subset of stations for the purpose of trend analysis 2005-2019 (Horálek et al., 2022) are also presented. Again, the population-weighted concentration for the whole area including the United Kingdom is presented for the whole period including 2024, for consistency reasons. The NO2 concentration (in terms of annual average) shows a decrease of about 0.7 µg/m3 per year. One can see that the interim results for 2024 show the lowest estimates in the presented period. Figure 4.2 Population-weighted concentration of NO2 annual average in 2005-2024, based on both the regular (red) and the trend analysis (blue) mapping results (where available), and with interim results for the year 2024 ETC HE Report 2025/5 23 5 Conclusions The report presents the interim 2024 maps for PM10 annual average, PM2.5 annual average, NO2 annual average and the O3 indicator peak season average of maximum daily 8-hour means. The maps have been produced based on the non-validated E2a (UTD) data of the AQ e-reporting database, the CAMS Ensemble Forecast modelling data and other supplementary data. Together with the concentration maps, the difference maps between five-year mean 2019-2023 and 2024 and between the years 2023 and 2024 are presented (using the 2019-2023 regular and the 2024 interim maps), as well as basic exposure estimates based on the interim maps. Regarding PM10 annual average, the map indicates that concentrations above the annual LV occur mainly in urban areas of Balkan cities. Additionally, elevated PM10 levels (30-40 µg/m3) are estimated in parts of the Po Valley in Italy and other scattered areas in south-eastern Europe, while most of western and northern Europe shows concentrations below 20 µg/m3 (the revised LV2030). PM2.5 levels above the annual LV occur in scattered urban areas of Bosnia and Herzegovina. Exceedances of the ILV are observed in the Po Valley (Italy), the Krakow–Katowice–Ostrava industrial region, and parts of the Balkan countries. In central and south-eastern Europe, concentrations range mainly between 5-15 µg/m³, while in southern Europe they are mostly 5-10 µg/m³, with some areas below the WHO AQG level (5 µg/m³). Central and western Europe show predominantly values below 10 µg/m³ (the revised LV2030), and northern Europe mostly below 5 µg/m³. In the case of O3, the map shows that in 2024 areas with a peak season average of maximum daily 8hour means exceeding 60 µg/m³ (with the peak season defined as the six consecutive months of the year with the highest six-month running-average O3 concentration) cover the entire considered region. Lowest values (< 80 µg/m³) are observed in parts of northern Europe as well as in Ireland, and in a large area of France. Higher values (> 100 µg/m³) are recorded in northern Italy, central Spain, and various parts of the Balkan countries and Cyprus. Values of 80–100 µg/m³ are observed in the remaining mapped area of Europe. In the case of NO2, no areas exceeded the annual LV of 40 µg/m³ for NO2, based on the 1 km resolution final map (although concentrations above the annual LV occurred at some urban traffic locations, which are smoothed in this final map resolution). Some cities, especially capitals in southern and southeastern Europe, show NO2 levels above 20 µg/m³ (the revised LV2030). The urbanized areas of several major cities fall within the range of annual average NO2 concentrations between 10 and 20 µg/m³. Most of Europe shows NO2 levels below the AQG level of 10 µg/m³, while a larger region with concentrations between 10 and 20 µg/m³ is found in the Po Valley. Uncertainty estimates based on the cross-validation of the E2a data have been performed for all interim maps, showing quite satisfactory results in general. However, these uncertainty estimates are based on the non-validated E2a data and are valid for areas covered by the E2a measurements only. The complete validation of the interim maps should carried out when the validated E1a data for 2023 become me available. In the report, population exposure for only large European regions, EU-27 and the total considered area has been presented. The more detailed exposure estimates for particular European countries will be presented in 2026, in the ETC HE regular mapping report on the 2024 air quality maps created based on the validated data E1a. ETC HE Report 2025/5 24 List of abbreviations Abbreviation Name Reference AQ Air Quality AQG Air Quality Guideline CLC CORINE Land Cover https://land.copernicus.eu /pan-european/corineland-cover CORINE Co-ORdinated INformation on the Environment https://land.copernicus.eu /pan-european/corineland-cover CTM Chemical Transport model 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 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 ILV Indicative Limit Value JRC Joint Research Centre https://ec.europa.eu/info/ departments/jointresearch-centre_en LV Limit Value http://eurlex.europa.eu/LexUriServ/L exUriServ.do?uri=OJ:L:200 8:152:0001:0044:EN:PDF LV2030 revised limit values to be attained by 1 January 2030 https://eurlex.europa.eu/legalcontent/EN/TXT/PDF/?uri= OJ:L_202402881 NILU Climate and environmental research institute https://www.nilu.no/ NO2 Nitrogen dioxide O3 Ozone ORNL Oak Ridge National Laboratory https://www.ornl.gov/ PM10 Particulate Matter with a diameter of 10 micrometres or less PM2.5 Particulate Matter with a diameter of 2.5 micrometres or less R2 Coefficient of determination RIMM Regression – Interpolation – Merging Mapping RMSE Root Mean Square Error RRMSE Relative Root Mean Square Error UTC Coordinated Universal Time WHO World Health Organization https://www.who.int/ ETC HE Report 2025/5 25 References CAMS, 2025, 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 22 September 2025. 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. 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, 1-44 (http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:EN:PDF) accessed 26 May 2021. ECMWF, 2025, CAMS Regional: European air quality analysis and forecast data documentation (https://confluence.ecmwf.int/display/CKB/CAMS+Regional%3A+European+air+quality+analysis+and +forecast+data+documentation) accessed 22 September 2025. EEA, 2024, Air Quality e-Reporting. Air quality database (https://www.eea.europa.eu/data-andmaps/data/aqereporting-8). Data extracted in March 2024. EEA, 2025, Air Quality e-Reporting. Air quality database (https://www.eea.europa.eu/data-andmaps/data/aqereporting-8). Data extracted in February 2025. 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 10 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. Gilbert, R. O., 1987, Statistical Methods for Environmental Pollution Monitoring, Van Nostrand Reinhold, New York. Horálek, J., et al., 2021a, Potential use of CAMS modelling results in air quality mapping under ETC/ATNI, Eionet Report ETC/ATNI 2019/17 (https://doi.org/10.5281/zenodo.4627762) accessed 15 June 2021. ETC HE Report 2025/5 32 Table A2.1 Number of stations used in interim mapping 2024 per station type, for PM10 (left) and PM2.5 (right) Station type PM10 PM2.5 E1a 2023 E2a 2024 E1a 2023 E2a 2024 Total For regression For pseudo 2024 Mapping 2024 Total For regression For pseudo 2024 Mapping 2024 Rural background 410 304 106 319 285 200 85 220 Urban/suburb. backgr. 1512 1178 334 1232 1032 760 272 817 Urban/suburb. traffic 803 663 140 696 495 399 96 432 For the PM2.5 mapping, in addition to the PM2.5 stations, 95 rural background, 70 urban/suburban background and 271 urban/suburban traffic PM10 E2a 2024 stations (at locations without PM2.5 measurement for 2024) have been also used for the purpose of calculating the pseudo PM2.5 station data (see Eq. A1.5). Table A2.2 shows the number of stations used in the interim mapping of O3 and NO2. In the O3 interim mapping, only E2a 2024 stations are used. No pseudo stations for O3 are used due to quite complete spatial coverage of the E2a O3 data. Table A2.2 Number of stations used in interim mapping 2024 per station type, for O3 and NO2 Station type O3 NO2 E2a 2024 E1a 2023 E2a 2024 Mapping 2024 Total For regression For pseudo 2024 Mapping 2024 Rural background 498 460 398 62 407 Urban/suburb. backgr. 1113 1376 1231 145 1277 Urban/suburb. traffic – 1231 881 350 913 A2.2 Chemical transport modelling (CTM) data The CAMS Ensemble Forecast data as provided by the Copernicus Atmosphere Monitoring Service (CAMS) at a regional scale over Europe have been used. The European regional production consists of an ensemble of eleven air quality models run operationally. All models use the same CAMS-REG anthropogenic emissions and current meteorology from the operational ECMWF IFS forecast. The models provide (along with other products) a 96-hour forecast made available at 08:00 UTC the day of the forecast. The forecast data product is available on an hourly time resolution and at a spatial resolution of 0.1° x 0.1°, which corresponds roughly to 5-10 km (W–E) x 10 km (S–N). Each model forecast is combined into an ensemble forecast by taking the median of all used models. For further details see ECMWF (2025). In this report, the CAMS Ensemble Forecast data (for the lead hour 0-23) for 2023 and 2024 have been used (CAMS, 2025). All the models used in the ensemble were run using the CAMS-REG-v5.1 REF2 v2.0.1 emissions (corresponding to year 2018) for most of the year 2023; in November 2023 emissions were updated to CAMS-REG-v6.1 corresponding to year 2022 and in November/December 2024 to CAMS_REG.v7 corresponding to year 2023 (ECMWF, 2025). For more information on emissions, see Kuenen et al., 2024. All modelling data have been aggregated into the annual statistics and converted ETC HE Report 2025/5 33 into the reference EEA 1 km (for PM and NO2) and 10 km (for O3) grids. The pollutants and parameters used are the same as those used for the monitoring data, as described in Section A2.1. A2.3 Satellite data Data from the TROPOspheric Monitoring Instrument (TROPOMI) onboard of the Sentinel-5 Precursor satellite (Veefkind et al., 2012) were used. Their spatial resolution is approximately 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 nitrogen dioxide (NO2), i.e. a vertically integrated value over the entire troposphere. All overpasses for a specific day were then mosaicked and gridded into the reference EEA 1 km grid in the ETRS89 / ETRS-LAEA (EPSG 3035) projection. The daily gridded files have been subsequently averaged to an annual mean. The annual mean has been aggregated from cloud-free high-quality (qa_value > 0.75) daily data only. The parameter used is NO2 – annual average tropospheric vertical column density (VCD) [number of NO2 molecules per cm2 of earth surface], years 2023 and 2024. A2.4 Other supplementary data Meteorological data The meteorological data used are the ECWMF data extracted from the CDS (Climate Data Store, https://cds.climate.copernicus.eu/cdsapp#!/home). Specifically, the hourly data of the reanalysed data set ERA5-Land in 0.1°x0.1° resolution have been used. In the coastal areas (where the data from ERA5-Land are not available), the same parameters from the reanalysed data set ERA5 in 0.25°x0.25° resolution have been applied. The hourly data have been derived into the parameters needed, aggregated into the annual statistics and converted into the reference EEA 1 km (for PM and NO2) and 10 km (for O3) grids. For details, see Horálek et al. (2025). Meteorological parameters used are wind speed (annual mean for 2024, in m.s-1), surface net solar radiation (annual mean of daily sum for 2024, in MWs.m-2) and relative humidity (annual mean for 2024, in percentage). Altitude The altitude data field (in m) of Global Multi-resolution Terrain Elevation Data 2010 (GMTED2010) has been used, with an original grid resolution of 15 arcseconds coming from U.S. Geological Survey Earth Resources Observation and Science, see Danielson and Gesch (2011). The data were converted into the EEA reference grids in 1 km and 10 km resolutions. Next to this, another aggregation based on the 1 km grid cells has been executed, i.e. the average of the circle with a radius of 5 km, calculated as a floating average for all 1 km grid cells. Land cover CORINE Land Cover (CLC) 2018 – grid 100 m, Version 2020_20 (EU, 2020) is used. The 44 CLC classes have been re-grouped into the 8 more general classes. In this paper, five of these general classes are used, namely high density residential areas (HDR), low density residential areas (LDR), agricultural areas (AGR), natural areas (NAT), and traffic areas (TRAF). For details, see Horálek et al. (2024a). Two aggregations are used, i.e., into 1 km grid and into the circle with radius of 5 km. The aggregated grid value represents for each general class the total area of this class as percentage of the total area of the 1 km x 1 km square or the circle with radius of 5 km. Population density and Road data Population density (in inhabitants/km2) is based on JRC-Geostat 2018 grid dataset, Eurostat (2020). For regions not included in the JRC-Geostat 2018 dataset, GHS population grid for 2020 (JRC, 2023) scaled to the reference year 2018 is used. For details, see Horálek et al. (2024a). GRIP vector road type data is used (Meijer et al., 2018). Based on these data (i.e., buffers around the roads), traffic map layers (Section 2.1) are merged into the final maps (Horálek et al., 2024a). ETC HE Report 2025/5 34 Annex 3 Technical details and uncertainties of interim maps This Annex 3 presents different technical details on the interim maps presented in this report. Sections A3.1, A3.2, A3.3 and A3.4 give technical details and uncertainty estimates of the 2024 interim maps for PM10, PM2.5, O3 and NO2, respectively. A3.1 Particulate matter PM10 This section presents the technical details and uncertainty estimates of the PM10 2024 annual average interim map as presented in Map 2.1. Like in Horálek et al. (2021b), first, the pseudo stations data have been estimated. The estimates have been calculated based on the E1a measurement data for 2023, the CAMS Ensemble Forecast modelling data for 2023 and 2024, and the regression relation with the E2a measurement data for 2024. Table A3.1 presents the regression coefficients determined for pseudo stations data estimation, based on the 1482 rural and urban/suburban background and 663 urban/suburban traffic stations that have both E1a 2023 and E2a 2024 measurements available (see Sections A1.2 and A2.1). Next to this, it presents the statistics showing the tentative quality of the estimate. Table A3.1: Parameters and statistics of the linear regression model for the generation of pseudo PM10 data in rural and urban background and urban traffic areas, for PM10 annual average 2024 c (constant) 0.7 0.5 a1 (PM10 annual mean 2022, E1a data) 0.578 0.734 a2 (PM10 annual mean 2022 * CAMS ratio 2023/2022) 0.394 0.242 Adjusted R20.88 0.87 Standard Error [µg/m3]2.0 2.2 Linear regression model (LRM, Eq. 2.4) PM10 – Annual average Rural and urban background areas Urban traffic areas Based on the E2a data and pseudo data, CAMS Ensemble Forecast modelling data and other supplementary data as used in the regular mapping, the interim PM10 annual average map for 2024 has been created (see Map 2.1). Table A3.2 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 A3.2 shows that the uncertainty of the interim map of PM10 annual average expressed by RMSE is 2.5 µg/m3 for the rural areas, 3.1 µg/m3 for the urban background areas and 3.5 µg/m3 for the urban traffic areas. The relative mean uncertainty (Relative RMSE) of this map is 19.2% for rural, 17.4% for urban background and 17.9% for urban traffic areas. However, these uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim PM10 map can only be done when the validated E1a data for 2024 are available. ETC HE Report 2025/5 35 Table A3.2 Parameters and statistics of the linear regression model and ordinary kriging in rural, urban background and urban traffic areas for the interim map of PM10 annual average 2024 Rural areas Urban b. areas Urban tr. areas c (constant) 2.34 1.01 1.78 a1 (log. CAMS-ENS FC model) 0.776 0.71 0.511 a2 (altitude GMTED) -0.00016 a3 (relative humidity) -0.043 a4 (wind speed) -0.019 -0.057 a5 (land cover NAT1) -0.0009 Adjusted R20.68 0.41 0.40 Standard Error [µg/m3]0.22 0.25 0.25 nugget 0.010 0.014 0.021 sill 0.043 0.041 0.023 range [km] 100 160 320 RMSE [µg/m3]2.5 3.1 3.5 Relative RMSE [%] 19.2 17.4 17.9 Bias (MPE) [µg/m3]0.2 0.1 -0.1 R2 of cross.-val. regr. equation 0.73 0.70 0.70 Slope of cross-val. regr. equation 0.81 0.76 0.69 Intercept of cross-val. regr. equation 2.6 4.4 5.9 Linear regresion model (LRM, Eq. 2.1) Ordinary kriging (OK) of LRM residuals LRM + OK of its residuals PM10 Annual average A3.2 Particulate matter PM2.5 This section presents the technical details and uncertainty estimates of the PM2.5 2024 annual average interim map as presented in Map 2.3. Like in Horálek et al. (2023b), the pseudo stations data of two types have been estimated at first, i.e. based on the PM2.5 E1a measurement data for 2023, the CAMS Ensemble Forecast modelling data for 2023 and 2024, and the regression relation with the PM2.5 E2a measurement data for 2024 (see Eq. A1.4) and based on the PM10 E2a measurement data for 2024, different supplementary data, and the regression relation with the PM2.5 E2a measurement data for 2024 (see Eq. A1.5). Table A3.3 presents the regression coefficients determined for these pseudo stations data estimations. Table A3.3 Parameters and statistics of the linear regression model for the generation of pseudo PM2.5 data in rural and urban background and urban traffic areas for PM2.5 annual average 2024, using PM2.5 E1a data for 2023 (top) and PM10 E2a data for 2024 (bottom) c (constant) 0.6 0.5 a1 (PM2.5 annual mean 2023, E1a measurement data) 0.487 0.514 a2 (PM2.5 annual mean 2023 * CAMS ratio 2024/2023) 0.462 0.457 Adjusted R20.89 0.90 Standard Error [µg/m3]1.3 1.2 c (constant) 20.8 37.8 b (PM10 annual mean 2024, E2a measurement data) 0.616 0.470 a1 (surface solar radiation 2024) -0.0030 -0.0039 a2 (latitude) -0.237 -0.493 a3 (longitude) 0.096 0.116 Adjusted R20.84 0.75 Standard Error [µg.m-3]1.5 1.9 Linear regression model (LRM, Eq. 2.4) PM2.5 - Annual average Rural and urban background areas Urban traffic areas Linear regresion model (LRM, Eq. 2.5) ETC HE Report 2025/5 36 The estimates based on the PM2.5 E1a data for 2023 have been calculated using 960 rural and urban/suburban background and 399 urban/suburban traffic stations that have both E1a 2023 and E2a 2024 data available, while the estimates based on the PM10 E2a data for 2024 using 964 rural and urban/suburban background and 394 urban/suburban traffic stations that have both PM10 and PM2.5 E2a 2024 data available. Similarly as in Horálek et al. (2023b), the estimates based on the PM2.5 data for 2023 show stronger correlation with the PM2.5 data for 2024, compared to the estimates based on the PM10 data for 2024. Leading from this, the pseudo data estimates based on the PM10 data for 2024 have been applied only in places with no pseudo data estimates based on the PM2.5 data for 2023. For the number of PM2.5 data and pseudo PM2.5 data of both types applied in the interim map creation, see Section 3.1. Based on the E2a data and pseudo data, CAMS Ensemble Forecast modelling data and other supplementary data as used in the regular mapping, the interim PM2.5 annual average map for 2024 has been created. 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 the linear regression model and ordinary kriging in rural, urban background and urban traffic areas for the interim map of PM2.5 annual average 2024 Rural areas Urban b. areas Urban tr. areas c (constant) 0.76 0.68 0.81 a1 (log. CAMS-ENS-FC model) 0.728 0.72 0.673 a2 (altitude GMTED) -0.00022 a3 (wind speed) -0.055 a4 (land cover NAT1) -0.0011 Adjusted R20.60 0.50 0.60 Standard Error [µg/m3]0.27 0.26 0.23 nugget 0.030 0.015 0.013 sill 0.075 0.044 0.055 range [km] 1000 90 1000 RMSE [µg/m3]1.5 2.1 2.1 Relative RMSE [%] 20.5 20.1 20.4 Bias (MPE) [µg/m3]0.2 0.0 0.0 R2 of cross.-val. regr. equation 0.76 0.72 0.71 Slope of cross-val. regr. equation 0.79 0.74 0.71 Intercept of cross-val. regr. equation 1.7 2.7 3.0 Linear regresion model (LRM, Eq. 2.1) Ordinary kriging (OK) of LRM LRM + OK of its residuals PM2.5 – Annual average Table A3.4 shows that the uncertainty of the interim map of PM2.5 annual average expressed by RMSE is 1.5 µg/m3 for the rural areas and 2.1 µg/m3 for both the urban background and the urban traffic areas. The relative mean uncertainty (Relative RMSE) of this map is 20.5% for rural areas, 20.1% for the urban background areas and 20.4% for the urban traffic areas. However, these uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim PM2.5 map can only be done when the validated E1a data for 2024 are available. A3.3 Ozone Similarly as in Horálek et al. (2024b), no pseudo stations for O3 have been used, due to a quite complete spatial coverage of the E2a data. Based on the E2a data, CAMS Ensemble Forecast modelling data and ETC HE Report 2025/5 37 other supplementary data as used in the regular mapping, the interim map of the O3 indicator peak season average of maximum daily 8-hour means for 2024 has been created (see Map 3.1). Table A3.5 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 the regression and the kriging of its residuals. Table A3.5 Parameters and statistics of the linear regression model and ordinary kriging in rural and urban background areas for the interim map of O3 indicator peak season average of maximum daily 8-hour means for 2024 c (constant) -9.1 18.3 a1 (CAMS-ENS-FC model) 1.10 0.80 a2 (altitude GMTED) 0.01 a3 (wind speed) -1.4 a4 (s. solar radiation) n. sign. 0.0009 Adjusted R20.57 0.44 Standard Error [µg/m3·d]6.5 8.1 nugget 24 20 sill 41 41 range [km] 790 80 RMSE [[µg/m3·d] 6.3 6.8 Relative RMSE [%] 6.9 7.6 Bias (MPE) [µg/m3·d]0.0 0.1 R2 of cross.-val. regr. equation 0.60 0.60 Slope of cross-val. regr. equation 0.62 0.62 Intercept of cross-val. regr. equation 34.3 33.4 Linear regresion model (LRM, Eq. 2.1) Ord. krig. (OK) of LRM residuals LRM + OK of its residuals O3 – Peak season average of maximum daily 8-hour means Rural background areas Urban background areas Table A3.5 shows that the uncertainty of the interim map of O3 indicator peak season average of maximum daily 8-hour means expressed by RMSE is 6.3 µg/m3 for the rural areas and 6.8 µg/m3 for the urban background areas. These uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim O3 map can only be done when the validated E1a data for 2024 are available. A3.4 Nitrogen dioxide As a first step for the interim NO2 annual average 2024 map creation, the pseudo stations data have been estimated, based on the E1a measurement data for 2023, the Sentinel-5P satellite data for 2023 and 2024, and the regression relation with the E2a measurement 2024 data. Table A3.6 presents the regression coefficients determined for pseudo stations data estimation. Table A3.6 Parameters and statistics of the linear regression model for generation of pseudo NO2 data in rural and urban background and urban traffic areas, for NO2 annual average 2024 c (constant) 0.4 0.4 a1 (NO2 annual mean 2023, E1a data) 0.691 0.932 a2 (NO2 annual mean 2023 * Sentinel-5P ratio 2024/2023) 0.249 n.sign. Adjusted R20.93 0.92 Standard Error [µg/m3]1.5 2.2 Linear regression model (LRM, Eq. 2.4) NO2 – Annual average Rural and urban background areas Urban traffic areas ETC HE Report 2025/5 38 The pseudo stations data estimation is based on the 1629 rural and urban/suburban background and 881 urban/suburban traffic stations that have both E1a 2023 and E2a 2024 measurements available (see Sections A1.2 and A2.1). Apart from this, it gives the statistics showing the tentative quality of the estimate. Based on the E2a data and pseudo data, CAMS Ensemble Forecast modelling data, Sentinel-5P satellite data and other supplementary data as used in the regular mapping, the interim NO2 annual average map for 2024 has been created (see Map 4.1). Table A3.7 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 A3.7 Parameters and statistics of the linear regression model and ordinary kriging in rural, urban background and urban traffic areas for the interim map of NO2 annual average 2024 Rural areas Urb. b. areas Urb. tr. areas c (constant) 4.4 11.5 17.25 a1 (CAMS-ENS-FC model) 0.329 0.176 n.sign. a6 (satellite Sentinel-5P) 1.16 1.554 2.041 a2 (altitude) -0.0054 a3 (altitude_5km_radius) 0.0050 a4 (wind speed) -0.72 -1.635 -1.923 a7 (population*1000) 0.00082 0.00016 a8 (NAT_1km) -0.0401 a9 (AGR_1km) -0.0214 a10 (TRAF_1km) 0.0557 a11 (LDR_5km_radius) n.sign. n.sign. 0.0991 a12 (HDR_5km_radius) n.sign. 0.1920 a13 (NAT_5km_radius) -0.0271 Adjusted R20.70 0.50 0.39 Standard Error [µg/m3]2.0 3.8 6.1 nugget 3 8 14 sill 4 11 28 range [km] 990 90 9 RMSE [µg/m3]1.7 3.2 5.4 Relative RMSE [%] 30.4 24.0 24.4 Bias (MPE) [µg/m3]0.0 0.0 -0.2 R2 of cross.-val. regr. equation 0.76 0.63 0.52 Slope of cross-val. regr. equation 0.74 0.63 0.53 Intercept of cross-val. regr. equation 1.4 4.9 10.3 Ordinary kriging (OK) of LRM residuals LRM + OK of its residuals Annual average NO2 Linear regresion model (LRM, Eq. 2.1) Table A3.7 shows that the uncertainty of the interim map of NO2 annual average expressed by RMSE is 1.7 µg/m3 for the rural areas, 3.2µg/m3 for the urban background areas, and 5.4 µg/m3 for the urban traffic areas, respectively. The relative mean uncertainty (Relative RMSE) of this map is 30.4% for rural areas, 24% for urban background areas and 24.4% for urban traffic areas. However, like for other pollutants, these uncertainty estimates are based on the non-validated E2a data and are valid only for areas covered by the E2a stations. The complete validation of the interim NO2 map can only be done when the validated E1a data for 2024 are available. ETC HE Report 2025/5 39 Annex 4 Validation of 2023 interim maps and exposure estimates This Annex 4 presents the validation of the 2023 interim maps produced using the up-to-date E2a data (EEA, 2024) as presented in Horálek et al. (2024b), against the validated E1a data (EEA, 2025). Next to this, it presents the exposure tables calculated using the interim 2023 maps and validates them against the exposure estimates calculated using the regular 2023 maps as presented in Horálek et al. (2025). A4.1 Concentration maps This section evaluates the concentration interim maps against the E1a data, using cross-validation. PM10 Table A4.1 presents the evaluation of the interim PM10 annual average 2023 map, against the E1a station data for 2023. Additionally, it also presents the cross-validation evaluation of the regular PM10 annual average 2023 map (Horálek et al., 2025) for the same subsets of the E1a station data, for comparable reasons. Table A4.1 Validation of interim (left) and regular (right) map of PM10 annual average 2023 showing RMSE, RRMSE, bias, R2 and linear regression from validation scatter plots in rural background (top), urban background (middle) and urban traffic areas (bottom), against two validation sets of stations. Units: µg/m3 except for RRMSE and R2 RMSE RRMSE Bias R2Regr. eq. RMSE RRMSE Bias R2Regr. eq. E1a stations with E2a data 2.3 18.2% 0.3 0.742 y = 0.831x + 2.4 2.3 18.5% 0.2 0.739 y = 0.831x + 2.4 E1a stations with no E2a data 3.1 22.1% 0.1 0.649 y = 0.726x + 3.9 3.0 21.8% 0.0 0.664 y = 0.765x + 3.2 E1a stations with E2a data 3.2 18.3% 0.1 0.696 y = 0.754x + 4.3 3.3 18.9% 0.3 0.684 y = 0.789x + 3.9 E1a stations with no E2a data 4.0 20.3% -0.5 0.554 y = 0.551x + 8.4 2.9 14.8% -0.2 0.530 y = 0.582x + 8.2 E1a stations with E2a data 3.3 17.6% -0.2 0.691 y = 0.681x + 5.8 3.3 17.6% 0.1 0.693 y = 0.711x + 5.5 E1a stations with no E2a data 4.0 19.2% -0.6 0.539 y = 0.613x + 7.4 4.0 19.6% -0.1 0.526 y = 0.652x + 7.1 PM10 – Annual Average Urban traffic Interim map Area Regular map Validation set Rural Urban background One can see that, in general, the uncertainty of the interim map is the same or only slightly worse compared to the uncertainty of the regular map. Additionally, the validation of the E2a data and the pseudo station data used in the interim PM10 mapping has been performed. Table A4.2 shows the validation of the E2a and the pseudo data against the E1a station data in the locations of these stations. Table A4.2: Validation of E2a and pseudo station data showing RMSE, RRMSE, bias, R2 and linear regression from validation scatter plots for rural background (top), urban/suburban background (middle) and urban/suburban traffic stations (bottom), PM10 annual average 2023. Validation by E1a station data. Units: µg/m3 except for RRMSE and R2 Station type Evaluated set Validation set N RMSE RRMSE Bias R2Regr. eq. E2a stations E1a stations with E2a data 286 0.7 5.9% 0.1 0.974 y = 1.015x - 0.1 Pseudo stations E1a stations located at pseudo stations 95 1.5 11.1% 0.2 0.876 y = 1.037x - 0.3 E2a stations E1a stations with E2a data 1152 1.1 6.1% 0.0 0.964 y = 0.982x + 0.3 Pseudo stations E1a stations located at pseudo stations 270 2.1 10.4% -0.5 0.897 y = 0.834x + 2.8 E2a stations E1a stations with E2a data 649 1.1 5.7% -0.1 0.968 y = 0.955x + 0.7 Pseudo stations E1a stations located at pseudo stations 81 2.1 10.2% -0.6 0.858 y = 0.917x + 1.1 PM10 – Annual Average Rural background Urban/suburban background Urban/suburban traffic ETC HE Report 2025/5 40 In general, the results show similar or better agreement of the pseudo data with the E1a data, compared to the validation of the pseudo stations presented in Horálek et al. (2021b), which recommended the use of the pseudo stations. Map A4.1 shows the difference between the interim and the regular maps of the PM10 annual average 2023, for rural and urban background map layers. One can see the greatest differences in Balkan and Cyprus, i.e. in the areas with the lack of the E1a stations. Map A4.1 Difference between interim and regular map for PM10 annual average 2023 in rural (left) and urban background (right) areas. Urban map layer is applicable in urban areas only PM2.5 Table A4.3 shows the evaluation of the interim PM2.5 annual average 2023 map, against the E1a station data for 2023. Additionally, it also presents the cross-validation evaluation of the regular PM2.5 annual average 2023 map (Horálek et al., 2025) for the same subsets of the E1a station data, for comparable reasons. ETC HE Report 2025/5 41 Table A4.3 Validation of interim (left) and regular (right) map of PM2.5 annual average 2023 showing RMSE, RRMSE, bias, R2 and linear regression from validation scatter plots in rural background (top), urban background (middle) and urban traffic areas (bottom), against two validation sets of stations. Units: µg/m3 except for RRMSE and R2 RMSE RRMSE Bias R2Regr. eq. RMSE RRMSE Bias R2Regr. eq. E1a stations with E2a data 1.4 18.5% 0.2 0.814 y = 0.723x + 2.3 1.4 18.3% 0.2 0.808 y = 0.775x + 1.9 E1a stations with no E2a data 2.9 31.7% -0.6 0.615 y = 0.591x + 3.2 2.6 28.3% -0.5 0.691 y = 0.666x + 2.6 E1a stations with E2a data 2.8 26.9% 0.2 0.704 y = 0.781x + 2.4 2.1 20.9% 0.1 0.694 y = 0.778x + 2.4 E1a stations with no E2a data 2.7 22.8% 0.1 0.645 y = 0.759x + 3.0 2.5 21.0% 0.2 0.698 y = 0.801x + 2.6 E1a stations with E2a data 2.0 20.4% -0.1 0.774 y = 0.718x + 2.7 2.0 20.5% 0.0 0.754 y = 0.706x + 2.9 E1a stations with no E2a data 2.3 20.7% -0.3 0.671 y = 0.769x + 2.4 2.3 20.3% -0.2 0.682 y = 0.788x + 2.2 PM2.5 – Annual Average Urban traffic Interim map Regular map Validation set Rural Urban background One can see that the uncertainty of the interim map is at the similar or only slightly worse level as the uncertainty of the regular map. Additionally, the validation of the E2a data and the pseudo station data used in the interim PM2.5 mapping has been performed. Table A4.4 shows the validation of the E2a and the pseudo data, against the E1a station data in the locations of these stations. Table A4.4 Validation of E2a and pseudo station data showing RMSE, RRMSE, bias, R2 and linear regression from validation scatter plots for rural background (top), urban/suburban background (middle) and urban/suburban traffic stations (bottom), PM2.5 annual average 2023. Validation by E1a station data. Units: µg/m3 except for RRMSE and R2 Station type Evaluated set Validation set N RMSE RRMSE Bias R2Regr. eq. E2a stations E1a stations with E2a data 185 0.6 7.9% 0.0 0.963 y = 0.955x + 0.3 Pseudo stations E1a stations located at pseudo stations 89 1.4 14.7% -0.1 0.915 y = 0.888x + 1.0 E2a stations E1a stations with E2a data 759 0.8 8.0% 0.0 0.955 y = 0.981x + 0.2 Pseudo stations E1a stations located at pseudo stations 232 1.6 12.9% -0.2 0.879 y = 0.907x + 1.0 E2a stations E1a stations with E2a data 396 0.5 5.2% -0.1 0.984 y = 0.961x + 0.3 Pseudo stations E1a stations located at pseudo stations 78 1.4 11.7% -0.2 0.902 y = 1.001x - 0.2 PM2.5 – Annual Average Rural background Urban/suburban background Urban/suburban traffic In general, the results show similar or better agreement of the pseudo data with the E1a data, compared to the validation of the pseudo stations presented in Horálek et al. (2023b), which recommended the use of the pseudo stations. Map A4.2 shows the difference between the interim and the regular maps of the PM2.5 annual average 2023, for rural and urban background map layers. One can see the greatest differences in Balkan and Cyprus, i.e. in the areas with the lack of the E1a stations. European Topic Centre on Human Health and the Environment https://www.eionet.europa.eu/etcs/etc-he 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.