Methodology

Download the EU-27 AEI extract cited on this page: eea-aei-statistics.csv (CC BY 4.0). Also on Statistics.

This site currently publishes 28 country pages, 39 station pages, and 7 pollutant pages. Country PM2.5 uses the EEA Average Exposure Indicator (AEI). PM2.5 mortality uses the EEA Burden of Disease series. Other country pollutant means (PM10, NO2, O3, SO2) are still editorial composites and are labeled as such. Station pages use the classification-adjusted modeled rule in §5b.

1. Data sources

What is actually wired into the database today:

  • EEA Average Exposure Indicator (AEI) for PM2.5 — Table 3 of the Air Quality Status Report 2025, transcribed 2026-08-09 into countries.pm25_annual_avg for EU/EEA reporters. AEI is the three-year mean (2021-2023) of urban-background annual means. The United Kingdom is not in that table; its PM2.5 figure is an editorial composite, not EEA AEI.
  • EEA Burden of Disease (PM2.5 mortality, 2022) — applied by scripts/etl/10-correct-eea-figures.mjs. The UK is set to NULL because the EEA does not publish a UK row in that series.
  • Station identity rows — named EEA station codes and operators are carried as an editorial station register. The concentration values on station pages are not that station's own E1a time series; they are modeled from the country PM2.5 AEI (see §5b).
  • Not yet ingested — we do not currently download EEA Dataflow E1a hourly series, Copernicus CAMS grids, or live national-agency feeds into this database. Mentions of those agencies below are context for benchmarks and future ingest, not a claim that those feeds populate today's numbers.

2. Pollutants and benchmarks

PlainAirQuality covers the seven pollutants regulated under the EU Ambient Air Quality Directive 2008/50/EC and revised by the 2024 EU directive: PM2.5, PM10, NO2, ground-level O3, SO2, CO, and benzene (C6H6).

On every detail page we show two benchmarks side by side:

  • EU limit value — the regulatory limit currently in force under Directive 2008/50/EC. For PM2.5 the annual limit is 25 μg/m³ (falling to 10 μg/m³ by 2030 under the 2024 revision, Directive (EU) 2024/2881); for NO2 it is 40 μg/m³. Member states must take measures when limits are exceeded.
  • WHO 2021 guideline — the World Health Organization Global Air Quality Guidelines (2021 update), which describe concentrations below which no significant adverse health effect is expected. For PM2.5 the WHO guideline is 5 μg/m³ (one-fifth of the current EU limit); for NO2 it is 10 μg/m³ (one-quarter of the EU limit).

3. Measurement methods

All measurement methods cited on this site follow the European reference methods mandated by Annex VI of Directive 2008/50/EC:

  • PM2.5 / PM10: Gravimetric reference (EN 12341). Equivalent automated methods (TEOM-FDMS, beta-attenuation, optical particle counters) are permitted when calibrated against the gravimetric reference.
  • NO2: Chemiluminescence with photolytic converter (EN 14211).
  • O3: UV photometry (EN 14625).
  • SO2: UV fluorescence (EN 14212).
  • CO: Non-dispersive infrared spectroscopy / NDIR (EN 14626).
  • Benzene: Active or diffusive sampling on Tenax cartridges with thermal desorption / gas chromatography (EN 14662).

4. Aggregation

Hourly station observations are aggregated to:

  • Annual mean — arithmetic mean of all valid hourly values for the calendar year. Data capture ≥ 75% required per Directive 2008/50/EC.
  • Daily mean — for PM10 and SO2, where the EU 24-hour limit applies.
  • 8-hour maximum — for O3 (EU target value 120 μg/m³ over an 8-hour rolling mean, not to be exceeded more than 25 days per calendar year averaged over three years).
  • Exceedance day count — the number of calendar days on which the daily mean exceeds the WHO guideline (PM2.5: 15 μg/m³; PM10: 45 μg/m³).

5. Country-level statistics

Country PM2.5 on PlainAirQuality is the EEA-published Average Exposure Indicator (AEI) for 2023 from the Air Quality Status Report 2025 Table 3 — a three-year urban-background mean (2021-2023), not a simple arithmetic mean across every station in the country, and not a fresh pull of Dataflow E1a.

Country PM10, NO2, O3, and SO2 means on this site remain editorial composites. They are useful for relative orientation on the page but are not verified row-by-row against a named EEA release. Do not cite them as EEA figures. We will replace them when a verified ingest lands.

Mortality attribution — the "premature deaths attributable to PM2.5" figure on each country page — uses the EEA Burden of Disease series for 2022 (stage 10). We report the central estimate; the underlying uncertainty range is roughly ±30%.

5b. Station-level figures

The individual station pages on PlainAirQuality show a modeled estimate, not an independently re-derived aggregation of that station's own raw hourly readings. Each station's annual mean is calculated as its country's EEA AEI PM2.5 value (or, for the UK, the editorial country composite), adjusted by a multiplier reflecting the station's classification: traffic stations +15%, industrial +25%, background −15% (ozone uses a separate, smaller traffic adjustment). This means two stations of the same classification within the same country will show identical modeled values, even though their real-world readings would differ — our current dataset does not yet ingest each station's own independent time series from the EEA Dataflow E1a. Treat station pages as "what a typical station of this type looks like in this country," not a substitute for the station's actual reported reading (available directly from the EEA datahub linked on each station page). We are working toward ingesting genuine per-station historical series in a future data update.

6. Update frequency

The EEA publishes the Air Quality Status Report and AEI figures on an annual cycle. We refresh the transcribed AEI table when a new report lands and record the change on the data changelog. We do not currently auto-ingest Dataflow E1a or refresh within two weeks of every EEA release — that claim previously overstated the pipeline.

7. Limitations

Air quality at any street, building, or specific moment may differ substantially from the station-level annual means we publish. Personal exposure depends on local micro-meteorology, indoor sources, commuting patterns, and short-duration episodes (e.g. wildfire smoke, Saharan dust intrusions, agricultural ammonia events). We publish station-level annual means because they are the regulatory metric; we do not publish hourly real-time data, real-time AQI, or hyperlocal exposure forecasts. For real-time alerts consult the European Air Quality Index or your national agency.

7b. Data reuse, licensing & monetization posture

Readiness only, no paid products or affiliate units are live. This section inventories what is productized today so journalists, researchers, and future operators know the boundaries. Activating pricing, API keys, affiliate placements, or alternative ad networks requires explicit operator sign-off.

Downloadable & embeddable today

License & attribution

The underlying EEA, Copernicus CAMS, and national-agency figures are published under their respective open-data licences. PlainAirQuality's compiled comparison table, derived rankings, and WHO/EU-benchmark tiering are our own work — see Terms of Use for the personal, journalistic, and academic reuse terms already published there (attribution + link back; no bulk scraping or competing redistribution without written permission).

Affiliate & ad placement rules (if ever activated)

No affiliate units run today; display ads are limited to labeled Google AdSense slots when approved. If the operator later enables additional monetization: units must be clearly labeled; no units may run adjacent to the station-reading tables or health-benchmark copy in a way that could read as a personalized health or relocation recommendation — this is YMYL-adjacent health content, so that adjacency rule is strict; density must respect the fleet Module 36 ceilings shared with AdSense. We present official monitoring-station data, we do not sell air-purifier products or health advice.

Alternative network eligibility (paperwork readiness)

Mediavine Journey and similar networks typically require roughly 1,000 real sessions per 28 days, original reference content, and a clean policy stack (privacy, terms, editorial policy — all present here). PlainAirQuality carries an open country/* rendered-sibling-divergence class fenced through 2026-09-10 (STRUCTURAL, pending a dated pollutant-experiment read); alternative-network application is not paperwork-blocked by that — the readiness criteria above (session volume, content originality, policy stack) are the actual gate, and session volume is the one this site has not yet crossed.

8. Corrections

If you find an error on PlainAirQuality — an incorrect station identifier, out-of-date capital city, mis-attributed operator, or any data-source discrepancy — please email hello@plainairquality.com. We respond within two business days and publish a corrections log linked from the footer once a correction has been verified against the underlying source.