4 The maps
A noise map looks like a photograph of a city’s sound and is nothing of the kind. Before it is a planning document it is a computation, a model assembled from traffic counts and source data, propagated to receivers through equations, and governed throughout by rules about which sources to include, where to place the listener, and how quiet a sound may be before it stops counting. Tens of millions of Europeans carry a measurable cardiovascular risk, and a number on that scale cannot be acted on until it is located. The strategic noise map is the instrument that does the locating. What it renders visible, and what by its own construction it cannot show, is the matter of this chapter.
4.1 From statistic to street
The European Union’s answer to which millions, and where, is written into the Environmental Noise Directive. It requires each member state to draw strategic noise maps for its major roads, railways and airports and for every agglomeration above a hundred thousand inhabitants, to refresh them every five years, and to publish them and draw up action plans on their basis (European Parliament and Council of the European Union, 2002). What the Directive does not do is set a binding limit. It requires that noise be assessed and made public, not that it be kept below any European value. The map is, by design, an instrument of visibility rather than of control, built to show and classify rather than to prohibit. What happens after that visibility — in national transposition, in municipal ordinance, in the courts — is a separate question, and one the case below begins to answer.
The rules of inclusion are specific, and the specificity is the point. A road becomes a major road, and enters the mapping obligation, above three million vehicles a year; a railway above thirty thousand trains; an airport above fifty thousand movements; an urban area becomes a mappable agglomeration above a hundred thousand residents (European Environment Agency, 2025). Everything beneath those cut-offs — the secondary road, the smaller airport, the town of ninety thousand — is not mapped, and so, in the administrative record, scarcely registers as a source of exposure at all. The threshold does not merely set a workload. It draws the boundary of the visible.
Within that boundary the map does real work. It converts an abstract continental total into this junction, this façade, this bedroom, and hands a city a ranked list of where exposure concentrates and where action would remove the most of it. That is a genuine achievement, and nothing here is meant to diminish it. But the map earns its authority by looking like a measurement, a picture of how loud the city is, and it is not one.
The Directive imposes no limit, but a classification, once made, is not always so easy for the classifying authority to disown. In 2004 the European Court of Human Rights found that Spain had violated Article 8 of the Convention — the right to respect for private and family life and the home — over its failure to act against night-time noise suffered by a resident of Valencia, even though the city council had itself designated her neighbourhood an “acoustically saturated zone” years earlier. The Court held it “unduly formalistic” to require her to prove the noise levels inside her own flat, given that the municipal authority’s own classification, and its own officers’ repeated confirmation of the breach, already constituted official knowledge of the problem (European Court of Human Rights, 2004). The ruling predates the strategic noise map as a European instrument, and concerns a domestic zoning tool rather than an END-mandated map. But the principle it establishes travels: once an administration has made an exposure visible, on its own instrument and by its own hand, it becomes considerably harder to treat that visibility as though it committed the authority to nothing.
4.2 A map is a model, not a measurement
A strategic noise map is calculated. A city is not wired with microphones on every street. A modest number of monitoring points serve to calibrate and validate, but the smooth bands of colour that cover the whole territory are computed, source by source, from traffic and emission data pushed through propagation equations. Since 2019 those equations have been common across the Union. CNOSSOS-EU, the harmonised assessment method, was adopted so that a decibel mapped in one country means the same as a decibel mapped in another (Kephalopoulos et al., 2014). The map is a simulation disciplined by measurement, not a measurement itself.
A simulation runs on conventions, and the conventions are choices with consequences. The receiver is placed, by rule, four metres above the ground at the most exposed façade — a height that yields a higher figure than a listener would read at a ground-floor window over soft ground, and a defensible convention that is a convention nonetheless. Whether to add the few decibels a reflecting façade contributes, what reference year to model, how to treat the ground between source and listener: each is a decision taken before any colour reaches the map, and each moves the number. Change the settings in the calculator below and the mapped level moves while the sound itself stays fixed — the map answering not to the city but to its own assumptions.
Source · modelled façade level using the geometric-divergence and atmospheric-absorption terms of ISO 9613-2:1996; 55/65/75 dB \(L_\text{den}\) bands and the 55 dB reporting floor per Directive 2002/49/EC (European Parliament and Council of the European Union, 2002); CNOSSOS-EU common method per Commission Directive (EU) 2015/996 (Kephalopoulos et al., 2014). Illustrative propagation, not a CNOSSOS-grade computation. Accessed 2026-07-13.
Beneath the conventions sits a harder rule. Strategic maps report exposure only down to fifty-five decibels \(L_\text{den}\), and fifty at night; below those values the map is not cautious but blank (European Environment Agency, 2025). A receiver modelled at fifty-three decibels does not appear as low risk — it does not appear. The same move appeared in the metric itself, a floor written into the instrument beneath which a real exposure becomes an administrative non-event, now inscribed a second time in the map the metric feeds (Chapter 2). The five-year cycle adds a temporal blankness to the spatial one. Every map is a portrait of a year already gone, often published later still, so the city being governed is never quite the city on the wall.
4.3 The citizen and the phone
If the official map is a model, the democratic question follows quickly: might people measure the city themselves? The phone in every pocket already holds a microphone, and the evidence on whether it can be trusted is more encouraging than expected. Testing a field of smartphone sound-measurement applications against a reference meter, Kardous and Shaw found a handful of iOS apps reading within two decibels of the reference on the built-in microphone, and within one decibel once an external calibrated microphone was fitted; the Android field, fragmented across hardware, fared worse (Kardous & Shaw, 2014, 2016). A phone is not a class-1 sound-level meter, but a well-chosen app is not a toy.
Nor is the result merely anecdotal. Evaluating participatory sensing directly against conventional technique, D’Hondt, Stevens and Jacobs concluded that, properly executed, citizen measurement can reach accuracy comparable to standard noise-mapping methods (D’Hondt et al., 2013). What such crowdsensing adds is exactly what the strategic map averages away: temporal density, the three-in-the-morning event a long-term mean dissolves; coverage of the places the official frame ignores, including the quiet the map never rates and the sub-fifty-five exposures it will not record; and a channel for the perceived, not only the modelled. Its weaknesses are the mirror image — uneven calibration, device-to-device variance, and the self-selection of who carries a phone where. Neither instrument is complete; each sees what the other cannot.
The frontier, then, is not measurement against model but their fusion, in permanent sensor networks that feed and correct the simulation in something nearer real time, and in observatories built to run both at once.
4.4 The same instrument, three cities
Nothing in the argument so far is peculiar to one place. It can be read in a mid-sized Andalusian city, in a national capital, and in a European metropolis that has pushed the mapping of sound further than most.
Granada’s strategic noise map, revised in 2016 by the city’s environmental office together with the University of Granada’s applied-physics department, does what the Directive asks. It models the exposed population band by band, at the four-metre façade, and identifies road traffic as the dominant source (Ayuntamiento de Granada & Universidad de Granada, Departamento de Física Aplicada, 2016). Read against the life of the city, the map’s emphasis is revealing. What residents complain of most is not the arterial traffic the map foregrounds but the noise of nightlife and neighbours — sources that sit awkwardly, or not at all, inside the Directive’s road–rail–air–industry frame (El Independiente de Granada, 2023). The map is not wrong; it is answering the question it was built to answer, which is not quite the question the city is asking.
At the scale of the capital the computation is vast. Madrid’s map estimates levels at tens of millions of points across the city from its receiver model, and its most recent revision reports the large majority of residents — on the order of ninety-eight per cent by day and ninety per cent at night — below the regulatory limits, with the residual concentrated along the great traffic arteries where action is then aimed (Ayuntamiento de Madrid, 2022). Yet the map states the boundary of its own knowledge in its title. It maps the noise of traffic. A source the Directive does not list is a source the map does not hold, however much it may trouble a particular street — the completeness of the picture is bounded, in advance, by the closed list of what counts. The map is also, in aggregate, a record: Madrid’s first-round strategic map already charted the evening level across the Centro district in 2006, the hours when the quarter’s social life is loudest, and reading that early evening surface against today’s figures turns a snapshot into a two-decade time series of the same streets (Ayuntamiento de Madrid, 2006).
The Paris region shows where the instrument is heading. Bruitparif, the regional sound observatory, fuses CNOSSOS-style modelling with a dense network of permanent sensors, including a device that resolves the direction a sound comes from. On that combined basis it puts the health toll of transport noise in the dense zone at roughly a hundred and eight thousand healthy life-years lost each year — about ten and a half months per resident over a lifetime, and far more for the most exposed (Bruitparif, 2019). This is the model with measurement added back — more honest about what a static five-year snapshot cannot hold, and a glimpse of the map the coming decade may make ordinary.
Scale is not the only way these cities differ, and it may not be the most consequential. They differ in form — and the strategic map, computed on a uniform method across a whole territory, meets that form unevenly. Madrid and Paris are, for the most part, flat and broad-streeted; their arterial noise spreads and decays much as a propagation model expects. Granada is not. Like Las Palmas and a hundred hill-towns, it folds a narrow flat strip of wide avenues against slopes onto which much of its densest housing climbs, and there the acoustics change character. A narrow street flanked by tall façades is not an open field. It becomes an acoustic canyon, channelling sound between hard surfaces so that reflections pile up, the usual fall-off with distance flattens, and levels along the walls can run several decibels above what an open-street model would predict. Building and street geometry alone have been shown to swing pedestrian exposure by up to around seven decibels (Echevarria Sanchez et al., 2016). Add a gradient, and the sources themselves grow louder: an engine labouring uphill, a bus dropping a gear, the extra tyre and brake noise of the descent.
These effects compound precisely where a city is at its most picturesque and least tractable. In Granada’s Albaicín — a historic quarter of steep, narrow, cobbled lanes reachable, for a vehicle, only by a handful of climbing routes — the small municipal buses that serve it share a single hard-surfaced channel with cars, delivery vans and clusters of pedestrians, each reflecting and reinforcing the others between close stone walls. Cobbles raise rolling noise above smooth asphalt; the incline raises engine noise; the canyon returns both to the façades several times over; and because the route is one of only a few, the traffic that elsewhere would disperse across a grid is concentrated onto it, hour after hour. A wide three- or four-lane avenue in the flat lower town, carrying far more vehicles, can present a quieter façade than a one-lane street on the hill above it.
None of this is invisible to the method in principle. CNOSSOS-EU carries correction terms for road gradient, for road-surface type and for the reflections a built-up street returns (Kephalopoulos et al., 2014); the physics is, in that sense, accounted for. But a strategic map is computed for a whole agglomeration from generalised inputs: modelled traffic on categorised streets, a bounded number of reflections, terrain smoothed to what the data support. The fine-grained reality of one labouring bus in one cobbled canyon at one hour is exactly the resolution such a map cannot hold. The correction enters the average; it is not resolved in the particular. So the hillside quarter tends to read, on the strategic surface, as calmer than it sounds to the person climbing it — one more place where the map’s necessary abstraction and the city’s lived acoustics quietly part company.
What the three share is the instrument’s central bargain: to make a harm visible and governable, it must first decide what to admit, and it admits the loud, the listed and the long-term average. The strategic map is, above all, a map of traffic — the old giant (Chapter 5). By its source list and its fifty-five-decibel floor it is structurally blind to the low-frequency, tonal and intermittent sources that fill the rest of Part II: the heat pump on the party wall (Chapter 6), the data-centre hum (Chapter 7), the machinery that sits below the floor the map will not cross. To see those, look precisely where the map does not. That is where the argument turns next.