This is one real Thursday at Dublin Airport, what if they sent flights over your house instead?
As it is
Sent over here
Every overflight of the day, placed by the clock and drawn to its
peak level. The shaded ends of the strip are 23:00 to 07:00.
Hear a typical one
Hear the worst one
Spend an hour under them
Stop
Set your speakers to real loudness
Speakers only. Never headphones or earphones
This is a safety instruction, not a preference. Everything below
assumes a sound level meter standing in the room, and a meter cannot
be put inside an ear, so on headphones the calibration means nothing
and the level is unguarded. The setting you are about to make is
around 32 dB above normal listening. Sent to earphones, with a phone
at full volume, that arrives directly at your eardrum with no room and
no distance in the way. Headphones also lose most of what an aircraft
is, which is the part below 200 Hz, so they understate the very thing
this page exists to show. Use speakers you can stand back from, and be
careful even then.
Calibrating leaves your system about 32 dB louder than normal
listening, because this page keeps that much headroom in reserve for
the loud passes. Everything else on the device is then 32 dB louder
too. A notification, an advertisement, or whatever you open next
arrives in the nineties rather than at 70, with no warning and no
fade. That is loud enough to hurt you and loud enough to destroy a
tweeter.
So before you start, silence notifications and close anything else
that plays audio. When you have finished, turn the amplifier back down
before you use the system for anything else.
The page gets loud by design as well. The loudest departure in this
day reaches 90 dB(A) at St Margaret's, which is about 96 dB unweighted
in the room, and the hour is an hour of them. Treat it the way you
would treat standing next to the aircraft, because that is the claim
being tested.
A sound level meter, or a phone app, will tell you whether this page is
overstating anything. Play the reference below and turn the volume up or
down until the meter reads 70 dB(A) where you would be sitting. After
that every pass arrives at the level printed for it, and the loudest of
the day arrives about 32 dB above the reference you set.
Play the 70 dB(A) reference
Stop
Tick that box once the meter reads 70. Nothing you play is at true
level until you do, and everything will be about 14 dB too loud, because
ordinary playback is levelled for comfort and keeps no headroom in
reserve.
Set the meter to A-weighting, slow response, and Lmax or plain SPL. Do
not use its peak reading. Every figure on this page is LAmax, which is
an A-weighted average over a moment, and that is also what an airport
noise monitor reports. Peak is a different measurement of the same
sound: unweighted, and instantaneous rather than averaged. For this kind
of noise it runs about 20 dB higher, so a meter left on peak makes the
page look like it is understating by twenty decibels when it is not.
Expect a MAX hold to read about a decibel above the figure printed for
a pass. The level quoted is the mean, and MAX catches the loudest
moment of a signal that fluctuates, which for broadband noise is about
half a decibel up on its own. Calibrating the reference on MAX as well
cancels most of it.
The reference is the aircraft sound itself held steady, not a test tone,
and that is deliberate. No speaker and no room is flat, and an aircraft
here lives from below 40 Hz to a couple of kilohertz, so a volume set by a 1 kHz
tone could be ten decibels out. Calibrating on the same sound that will
be played makes the speakers and the room cancel out of the reading
instead of being measured into it.
The tones below are the page's own bands, each sent at the level that
band carries in an aircraft. The three lower ones leave within about a
decibel and a half of each other, so if the meter reads the 38 Hz one
far below the other two, that is your speakers and not the page. Most of
an aeroplane is rumble, and most small speakers cannot produce it at
all, which is worth knowing before you judge the hour by them.
Take the hour with you (MP3)
How this is worked out
The flights are the real ones: every departure recorded off runway 28R on
20 August 2026, with its aircraft type, its time off the runway and its
track. Turboprops and ten flights with no recorded type are left out, so
the count here is lower than the day's 372.
Where a place already lies under a departure route, no turn is applied and
the figures are simply that day. Turning the airport for someone who is
already under it would overstate what they live with, and they can check
the page against their own morning.
The turn is a rigid rotation of the routes about the departure end of the
runway. The runway does not move, because it could not be moved without a
planning permission; the routes move, because daa says they need none. Every
aircraft keeps the height it really held at that distance from the end of the
runway, so the turn cannot make one lower or louder than it was.
One simplification in the reader's favour, stated plainly: a real change of
route would turn the traffic over a few kilometres rather than pivot it at
the runway end. This page swings each track whole, to show where the traffic
would end up rather than to draw the departure procedure that would take it
there. Heights, speeds and power settings are the ones actually flown.
The noise is daa's own measurement. The level of each type at each distance
comes from a propagation law fitted to 364,831 noise-monitor events that daa
itself recorded and released under access-to-information request AIE 2611.
The figure is a single-event peak (LAmax) outdoors, which is what every
count on this page reports.
Both indoor settings are daa's own assumptions, and the pair is the point.
Bickerdike Allen Partners, in the aircraft noise modelling methodology for
Dublin Airport dated 31 March 2020, derive two night-time thresholds from
one World Health Organization indoor guideline of 30 dB. They get 45 dB
outside using "the estimated difference between indoor and outdoor levels
of 15 dB for 'tilted or half open' windows given in the WHO 2018
document", and 55 dB outside "if windows are closed (assuming 25 dB
difference between indoor and outdoor noise levels)". The report then
adopts 55 dB as "the onset of sleep disturbance".
So the sourced figure produced the stricter threshold and was not
adopted, while the unsourced one, introduced by the word "assuming",
produced a threshold ten decibels more permissive and was. The
open-window assumption survives only in the N60 count, which the same
report calls supplementary and for which it says there is "no
conventionally accepted method of assessing impacts". The same consultant
assumes a window open when counting how often you are woken, and a window
shut when deciding whether being woken counts.
Two independent figures, for context. The World Health Organization's own
Night Noise Guidelines for Europe use a year-average difference of 21 dB,
chosen because "even in well-insulated houses windows may be open a large
part of the year", which is the averaging step the consultant skipped.
Locher and colleagues measured 10.0 dB for a genuinely open window across
115 measurements in 102 Swiss residences, and the Ballyboughal noise
survey used that figure for this area in 2023.
The window and the room are two different things, and only the second is
ours. Locher measured an open window as frequency independent, a hole
being a hole, and a closed one as not: the reduction is larger between 400
and 4,000 Hz. So the closed setting splits daa's total across the bands on
a mass law of 4 dB per octave, indicative rather than measured, and not
carried below 150 Hz because a mass law does not hold through the
resonance of a glazed unit.
Both indoor settings then pass through a room. A furnished room absorbs
high frequencies far more than low and lifts the bottom end near its
walls: domestic absorption runs about 0.12 at 125 Hz against 0.45 at
4 kHz, which is nearly 6 dB of extra loss at the top. It is why an
aircraft indoors sounds duller than the same aircraft outdoors even with
the window wide open, which has nothing to do with the window.
That shaping changes no number on this page. Each setting is re-solved
after it so the overall reduction still lands exactly on daa's figure, 15
dB and 25 dB, and a meter reads what it read before. The same total is
simply distributed differently across the spectrum. Adding attenuation
nobody measured would be inventing evidence; moving it about within a
sourced total is not, and the tests hold both figures to a tenth of a
decibel so it cannot drift into the first thing.
Treat the closed figure with suspicion and check it against your own
room. It assumes a sealed room with ordinary vertical windows. A
rooflight faces the aircraft directly and is usually the thinnest glazing
in the house; a wall vent is a hole in the facade and the housing under
these routes has one in every room; letter boxes, chimneys and single
glazing all let more through. On daa's assumption a 61 dB(A) departure
arrives indoors at 36 dB(A), close to the noise floor of a quiet room. If
that does not match what you hear, it is the assumption that should be
doubted rather than your hearing, and the assumption is daa's.
Choosing where you are listening changes only what you hear. The counts,
the median and the loudest are outdoor levels throughout, so they still
match the study the figures come from.
The tilted-window figure is one daa's own noise consultant works from.
Bickerdike Allen Partners, in the aircraft noise modelling methodology for
Dublin Airport dated 31 March 2020, took the World Health Organization's
45 dB LAmax indoor guidance for sleep and wrote that "accounting for
sleeping with a bedroom window slightly open (and a reduction from outside
to inside of 15 dB), this translates to an outside sound pressure level of
60 dB LAmax", and that "N60 contours are therefore used in this
assessment". The N60 count exists because of that assumption. Setting the
window to tilted is a way of hearing what it assumes about your night.
Ordinary playback is levelled for comfort rather than for measurement. It
runs through a compressor, so a distant aircraft stays audible without a
close one clipping, and the browser's compressor is not specified closely
enough to give the same result in two browsers. Setting the level switches
to a path with no compressor in it, corrects the output so that what a
meter reads tracks what the page prints to within a tenth of a decibel,
and fixes the scale once at a 95 dB(A) ceiling for every place on the map.
The scale deliberately does not follow the place you pick: a quiet street
has to read quiet on a meter nobody has touched since it was set.
Two simplifications worth knowing. Ground height is taken as a constant, so
places on higher ground are treated as slightly further from the aircraft
than they are. And positions arrive about every sixteen seconds, so the
closest approach is interpolated between them.
Nothing you click is sent anywhere. The place you pick is held in the part
of the web address after the # symbol, which browsers never transmit to the
server, so the link is shareable and the server still never learns it.
North Runway Technical Group. Data and method:
dublin-north-runway.com .