Are We Measuring the Air We Actually Breathe?

Europe has a sophisticated air-quality monitoring system — fixed monitors, reporting networks and increasingly detailed models — that matters for law, health protection and public debate. The harder public-health question is how well ambient readings and modelled estimates stand in for the pollution people encounter where they live, walk, work and sleep. Those are related questions, not the same one. Confusing a station concentration with how far it can reasonably apply across space, or either of those with personal exposure, is how sophisticated numbers can still mislead policy and epidemiology.

What official networks actually measure

Official networks measure ambient outdoor concentrations at designated sites — not a personal dose, and not automatically the air on every street. The European Environment Agency (EEA) classifies monitoring stations by purpose. Background stations, in EEA’s wording, are intended to represent pollution levels characteristic of the average exposure of the general population. Traffic and industrial sites serve other purposes: capturing local source influence rather than pretending to be everyone’s day.

That design intent is useful and easy to over-read. A well-placed background station can be an excellent tool for tracking trends and checking compliance with limit values. It is not a miniature of every neighbourhood. When public conversation treats “the city’s reading” as if it were a personal meter on every resident, the network is asked to answer a question it was never built to settle alone.

None of this is an argument against fixed networks. Without comparable, quality-assured ambient series, Europe would lose the ability to show whether limit values are met, whether trends are real, and whether Member States are converging. The public-health mistake is narrower: treating a compliance instrument as if it already answered the exposure question in full. Those jobs can share infrastructure. They still need different sentences in reports and in headlines.

EEA’s Air quality status in Europe 2026 report underlines both progress and unfinished business. Air quality continues to improve, yet in up to 20% of monitoring stations across the reporting network pollution remains above current EU air-quality standards — especially for PM10, ground-level ozone (O3) and benzo(a)pyrene (BaP). That is a station-level finding about legal standards still in force, not a claim that monitors systematically underestimate what people breathe.

Up to 20 percent of monitoring stations in Europe still record air pollution above current EU air quality standards, especially for PM10, ozone and benzo(a)pyrene.

≤20%

of monitoring stations still above current EU standards (esp. PM10, O3, BaP)

Source: EEA, Air quality status in Europe 2026 (key messages).

Against the World Health Organization’s 2021 Global Air Quality Guidelines, the picture is stricter. EEA reports that for most pollutants concentrations remain well above WHO guideline levels, with more than nine out of ten Europeans exposed above those levels. The same report’s PM2.5 chapter for 2024 shows how yardsticks diverge at station level: 92.5% of PM2.5 reporting stations were above the WHO annual guideline (5 µg/m³), about 1% (23 stations) were above the current EU annual limit value (25 µg/m³), and 60.7% were already below the 2030 EU annual limit (10 µg/m³). Passing today’s EU limit is not the same as meeting a health-based guideline — and neither figure is a personal exposure diary.

WHO annual air quality guideline 5 micrograms per cubic metre; EU 2030 annual limit value 10; current EU annual limit value 25. Bars scaled relative to 25 as 100 percent.

5

WHO AQG

10

EU 2030 LV

25

Current EU LV

Source: WHO Global Air Quality Guidelines 2021; Directive (EU) 2024/2881 / EEA PM2.5 standards table. Values are legal or guideline thresholds, not observed station concentrations.

Spatial representativeness and hotspots

A monitoring station measures pollution at a specific location. That reading is essential — but it does not, by itself, describe every street nearby. Spatial representativeness asks a practical follow-up: over what area can that value reasonably be used? The question matters because pollution can change sharply over short distances, especially along traffic corridors and in narrow streets lined with tall buildings, where local peaks sit close to quieter pockets the nearest station may never “see.”

This is where modelling sits alongside fixed monitoring. Models help estimate what happens between monitoring points and across the city — filling the map that a sparse network cannot cover alone. FAIRMODE’s Working Group 4 guidance on microscale modelling emphasises high-resolution estimates at the height where people walk, and treats understanding local spatial variation — including how far a measured or modelled value can reasonably apply — as central to exposure-relevant assessment. Not every modelling approach captures those fine-scale urban patterns equally well; some smooth them away. Choosing tools suited to local gradients is therefore part of getting the spatial story right, not a detour into modelling theory for its own sake.

For public health, the pay-off is practical. Better spatial information can reveal hotspots and corridors where concentrations — and often the people living closest to traffic — diverge from city averages, and can help target interventions where they are most needed. It improves the step from an ambient reading toward a more credible picture of population exposure. It does not invent pollution, and it does not mean that official monitoring stations are “wrong” when a peak sits metres away from a site designed for a different purpose. Nor does a sharper map, by itself, clean the air: measurement and modelling refine what we know; emissions, traffic and heating choices still decide what people breathe.

From ambient air to the air people breathe

Personal exposure is shaped by place, time, indoor environments and mobility. Even a spatially perfect outdoor map would not equal what one person inhales across a day spent partly indoors, on transit, or near a busy road for only an hour. Ambient concentration at a monitor or model cell is the first step; spatial representativeness asks how far that value can reasonably apply; population and personal exposure ask who is where, when, and in what micro-environment.

Conceptual three-step pathway: ambient concentration at a monitor or model cell; spatial representativeness of that value; population or personal exposure shaped by place, time, indoors and mobility.

1. Ambient concentration at a monitor or model cell
2. Spatial representativeness of that value
3. Population / personal exposure shaped by place, time, indoors and mobility

Conceptual framing for this article (informed by FAIRMODE WG4 microscale guidance). Not a statistical result.

Recent studies illustrate how large that last step can be — without licensing a blanket claim that fixed monitors always underestimate exposure. A 2025 Journal of Exposure Science & Environmental Epidemiology study of older adults in London (Wood and colleagues) found that personalised, time-activity-adjusted exposure estimates were substantially lower than residence-based ambient estimates. That is a directional finding in that cohort and setting: time spent indoors and away from the outdoor residence field can pull personal estimates down relative to a home outdoor assignment. It is not a universal rule for every city, pollutant or age group.

A 2026 npj Clean Air student panel in Guangzhou (n = 37 across three seasons) adds a pollutant-specific caution: the association between fixed-station and personal measurements was stronger for PM2.5 than for ozone. Small panels outside Europe cannot settle EU-wide policy, but they remind us that “how well does the network track personal air?” is not one answer for every pollutant.

AAQD 2024/2881 and Implementing Decision 2026/1208

Directive (EU) 2024/2881 — the revised Ambient Air Quality Directive — entered into force on 10 December 2024. It tightens the path toward 2030 standards, strengthens expectations around modelling and indicative measurements, and requires spatial representativeness information when modelling is used, with reassessment on a multi-year cycle. Article 10 also requires urban monitoring supersites at a density of at least one per 10 million inhabitants. These are institutional investments in how air is known and reported.

Commission Implementing Decision (EU) 2026/1208 of 9 June 2026 sets out detailed rules on modelling applications and spatial representativeness assessment. It applies from 30 June 2028. On spatial representativeness, the Decision applies a tolerance of 15% of the central value or a minimum tolerance in Annex I — whichever is greater — above and below that central value to define a concentration interval of “similar” locations. That is a methodological rule for how far a modelled or measured value may be treated as representative, not a promise of emission cuts.

Read carefully, this regulatory turn is an admission that ambient numbers need an explicit spatial story attached to them. That is good measurement governance. It should not be mistaken for the separate work of reducing emissions, redesigning traffic, cleaner fuels for heating, or protecting the people who live in the noisiest chemical micro-environments. Better maps can make unfair air visible; they do not, by themselves, make it cleaner.

Why exposure misclassification matters for epidemiology and equity

Epidemiology that assigns the same outdoor concentration to everyone in a large area will misclassify people whose true exposure differs — upward or downward — from that assignment. Classical measurement-error thinking predicts attenuation of associations toward the null when error is largely random; when error is structured by neighbourhood, housing quality, commuting routes or occupational micro-environments, bias can also distort who appears most affected. Either way, the policy risk is real: under-powered or mis-located estimates of harm can weaken the case for targeted protection.

Equity enters because spatial and personal gaps are rarely randomly distributed. People living beside heavy traffic, working outdoors, or spending more time in poorly ventilated housing can sit systematically away from “average” background sites even when city-wide indicators look acceptable. Improving the fidelity of exposure assessment is therefore partly an equity tool: it reduces the chance that the most exposed become statistically invisible inside a smooth map. It is not, however, a substitute for the interventions those maps should then justify.

Better measurement is not the same as cleaner air

The central test this article applies is simple to state and easy to blur in practice: do we know ambient concentrations, or do we know what people breathe? Europe’s answer is increasingly sophisticated on the first half — stations, models, supersites, spatial representativeness rules under the revised Directive and Decision 2026/1208. That sophistication deserves credit. It also deserves boundaries. Monitoring accuracy and modelling fidelity are means. Population-health gains still depend on whether emissions fall, whether hotspots are treated as policy objects, and whether exposure assessment is honest about mobility, indoors and inequality.

I keep returning to the three-step chain: ambient value, spatial representativeness, lived exposure. When those steps stay distinct, official numbers remain powerful without pretending to be something they are not. When they collapse into one slogan about “the air we breathe,” public health loses precision exactly where precision would protect people. Better measurement is necessary. Cleaner air is a different achievement — and it is still the one that matters most.

Sources & further reading

1. EEA — Air quality status in Europe 2026
2. EEA — Particulate matter (PM2.5), Air quality status in Europe 2026
3. EEA — Monitoring station classifications and criteria
4. Directive (EU) 2024/2881 on ambient air quality and cleaner air for Europe
5. Commission Implementing Decision (EU) 2026/1208 on modelling applications and spatial representativeness
6. FAIRMODE WG4 — Best practice guidance for microscale air-quality modelling
7. WHO — WHO global air quality guidelines (2021)
8. Wood et al. — Personalised time-activity adjusted exposure estimates (JESEE, 2025)
9. npj Clean Air (2026) — Fixed-station versus personal exposure associations in a student panel