How aircraft noise is calculated: ECAC Doc 29 and the FAA's AEDT

Every official noise map you see around an airport, from Heathrow's contours to an FAA DNL contour, is the output of a computer model. Nobody stood in each street for a year with a sound meter. Knowing how the model works tells you what the map can and cannot say about a particular house.

Noise maps are calculated, not measured

To measure a year's average noise directly, you would need microphones working without a break for twelve months, in acceptable weather, at thousands of points. ECAC's guidance says plainly that this is not normally possible. So airports and governments calculate instead.

In Europe the method is set out in ECAC Doc 29, the European Civil Aviation Conference's report on computing noise contours around civil airports. It was first published in 1986, and its 5th edition was endorsed in May 2026, bringing it closer to ICAO's equivalent guidance, Doc 9911. In the US, the FAA requires its own software, the Aviation Environmental Design Tool (AEDT). The EU's strategic noise maps use a common method, CNOSSOS-EU, that covers road, rail, industrial and aircraft noise and was written into the Environmental Noise Directive in 2015 and updated in 2020.

The details differ, but the calculation follows the same broad steps.

Step 1: where each plane flew

The model needs each flight path in three dimensions: the track over the ground, and the height, speed and engine power along it.

Ground tracks come from radar records where they exist, or from the published departure and arrival routes. Real planes spread either side of a route, so a model uses a central track plus several side tracks. The Civil Aviation Authority's model for Heathrow, Gatwick and Stansted, called ANCON, uses up to 12 side tracks for each departure route, drawn from large samples of radar data.

Height, speed and power are either calculated from the aircraft's performance data, its weight and the procedure it flies, or taken from radar. Radar shows position and speed but not engine power, which has to be estimated. Doc 29 warns that using default profiles that do not match what pilots actually fly is a major reason why different models give different answers.

Landings and take-offs need very different profiles. Landings descend steadily on a shallow slope, so they stay low for many kilometres. Take-offs climb much more steeply, at high power.

Step 2: how loud the aircraft is

For each aircraft type, the manufacturer supplies noise-power-distance (NPD) tables. They give the noise directly below the aircraft, in steady flight, for different engine power settings, at ten standard distances from 200 ft to 25,000 ft. They come from the extensive noise testing each type goes through for certification.

In Europe these tables sit in the international Aircraft Noise and Performance (ANP) database. EASA has managed it under an EU regulation of 2014; before that the data were collected by EUROCONTROL and the FAA. For aircraft covered by both, the data match those of the FAA's older model, INM, which AEDT replaced. Aircraft not in the database are modelled with a similar type and a correction.

Step 3: how the sound travels to the ground

The model then works out how much of that sound reaches each point on the ground:

• Distance. Sound spreads out as it travels, losing 6 dB each time the distance doubles.
• Air. The atmosphere absorbs sound, more at high frequencies, and more or less depending on temperature and humidity.
• The ground. Sound travelling at a shallow angle to the ground, off to the side of a flight path, is weakened by interference with its own reflection. Directly beneath the aircraft, this extra loss does not apply.
• The aircraft's shape. Engines mounted under the wings radiate differently to the side than engines on the rear fuselage, and more noise usually goes backwards than forwards.
• Local features. Hills can matter a lot at airports in mountains. The 2026 edition of Doc 29 adds an adjustment for places where the line of sight to the aircraft is blocked.

What models cannot capture well is the weather on a given day. Doc 29 notes that in practice only average surface temperature and humidity can be taken into account, while wind and temperature changes with height bend sound in ways that vary from day to day.

Step 4: adding up the flights

Each flight is split into short straight segments. The model calculates the sound energy each segment delivers to a point on the ground, adds up the segments into that flight's total, then adds up every flight in the period. Repeated over a grid of points (the UK uses spacings of 50 to 500 m), this gives a surface of noise levels, and the contour lines are drawn through it.

The result is an average: LAeq over a summer day in the UK, Lden in the EU, DNL in the US. Averages add energy, not events. The FAA gives an example: one flight a day at a sound exposure level of 114.4 dB, ten at 104.4 dB or a hundred at 94.4 dB all give exactly 65 dB DNL. Our guide to Lden, DNL and Lnight explains the day, evening and night weightings.

Averages also blend the days together. At Kew Green, 10 km from Heathrow, our sample report counts 92 loud landings a day when the wind is from the west and 4 loud flights a day when it is from the east. The average day has 66.

The same model can report other figures, such as the number of flights above a given level. Doc 29 lists these as useful extra information, and accepts that average contours are often criticised for hiding how many flights there are and when.

ECAC Doc 29 and the FAA's AEDT compared

AEDT also calculates fuel burn and emissions, and it is the main tool for the worldwide analyses behind ICAO's environmental standards. It replaced INM for FAA noise work on 29 May 2015.

Models vs noise monitors

Many airports run permanent noise monitors, and their readings will not match the map exactly. That is expected. A monitor measures real flights at one spot. It picks up other sounds, can confuse two planes passing at once, and misses the quieter flights that fall below its trigger level, which pushes its average up. Doc 29 says 50 or more measurements of each aircraft type on each route may be needed for a reliable average.

The two work best together. Each year the Civil Aviation Authority analyses hundreds of thousands of monitor readings and radar tracks around Heathrow, Gatwick and Stansted and adjusts its model's aircraft data to match. Monitors keep the model honest; the model covers every street, not just the few with a microphone.

How our reports use the same method

Official maps show an average over a whole period, drawn in broad bands and often a few years old, as explained in our guide to airport noise maps. A Flight Noise report takes the method airports use for official maps, ECAC Doc 29, and applies it to every recorded landing and take-off of the last twelve months, at the one address you choose. The flight paths come from ADS-B, the position signals aircraft broadcast, collected by ADSB.lol; our guide to ADS-B explains how. Instead of one average, you get how many planes were louder than 65 dB outdoors, by day and at night, and real days replayed plane by plane. It is an independent estimate, not an official noise map.