Field guide

How to record it

Noise, light and air, recorded so the readings hold up. You do not need expensive kit to start. You do need dates, positions and honesty about what your instrument can and cannot do.

The three grades

Every report on the map carries one of these. They are shown differently because they are not the same evidence, and pretending otherwise is how good campaigns lose arguments.

Noise

The official figure is usually A-weighted and hour-averaged — a metric built around how we hear speech. It under-weights the low-frequency hum that cooling fans, chillers and backup generators actually produce. A facility can read compliant on paper while nobody nearby can sleep, because low-frequency noise penetrates walls and makes rooms resonate. Worse indoors, worse at night — exactly where and when the daytime outdoor spot-check is not looking.

What to capture

• dB(A), dB(C) and unweighted at the same moment. The gap between C and A is the low-frequency flag. The map works that gap out for you.
• Octave or third-octave bands if your meter does them, with attention to the low bands.
• L_night (23:00–07:00), plus L_eq (average), L_max (peaks) and L₉₀ (the steady background the hum sits in).
• Tonality — is there an audible steady tone, and roughly what pitch.
• Indoors and outdoors. Bedroom readings matter most.

Kit

• Evidence grade: a calibrated Class 1 or Class 2 sound level meter with octave-band or FFT analysis, plus a field calibrator. Borrow or hire from a university acoustics department, or from a member of the Association of Acoustic Consultants of Ireland.
• A windshield for the microphone. Essential outdoors, not optional.
• Phone apps: dB(A) only, and even then indicative. Phone microphones roll off at low frequency, so they cannot capture the hum. Use them to decide where to bring a real meter. Never as your evidence.

The protocol

1. Fixed positions. Record GPS for each. Microphone 1.2–1.5 m high, at least 3.5 m from walls and reflective surfaces. Note distance and line of sight.
2. Time of day. Always include a night measurement, plus a daytime comparison. Log start and end times.
3. Duration. At least 15 minutes per position. Longer if the noise comes and goes.
4. Weather. Low wind only, under about 5 m/s, windshield on. Never in rain or high wind. Record temperature, wind and conditions.
5. Background comparison. Measure a spot away from the facility too, so you can show the difference it makes.
6. Calibrate before and after. If it has drifted, discard the session.
7. Log everything. Incomplete metadata makes data unusable.

Light

Data centres run security lighting all night, every night, often in places that had none. Light is easier to record than noise and much easier to show — and unlike noise, most planning permissions contain explicit lighting conditions you can hold up against what is actually happening.

What to capture

• Illuminance in lux at your position — how much light is landing where you stand, or on your window.
• Night sky brightness in mag/arcsec² if you can get a Sky Quality Meter. Astronomy clubs often have one and will usually lend it.
• Direction you were facing, and whether the light source is directly visible.
• Whether it enters a bedroom, and whether curtains stop it.

The photograph method

This is the one most people can do tonight, and it is stronger evidence than it sounds.

1. Pick a fixed spot and mark it so you can stand in exactly the same place again. Include a fixed object in the frame — a gate, a pole, a rooftop — for scale and comparison.
2. Lock your settings. Use manual or pro mode. Note the ISO, shutter and aperture, and use the same ones every time. Automatic exposure will silently brighten a dark scene and ruin the comparison.
3. Same time of night, every time.
4. Keep the originals with their timestamps intact. Do not crop, filter or send through messaging apps that strip the data — email them to yourself instead.

A run of identical photographs from the same spot over weeks or months is a record nobody can argue with. If you have a shot from before the build, that is gold.

Kit

• Evidence grade: a lux meter, or a Sky Quality Meter for sky brightness.
• Indicative: phone light-sensor apps. They read the sensor on the front of the phone, so they are rough and depend on how you hold it. Fine for a first look.
• Observation: the photographs, and a note of when the lights come on and go off.

Air

Be straight about this one: you cannot do reference-grade air quality monitoring as a citizen. The EPA's national network exists for that, and its stations are sited to assess population exposure, not to watch any particular facility. Anyone telling you a phone or a cheap sensor proves a data centre is polluting the air is selling you something that will fall apart under challenge.

What you can do is far more useful than people realise.

Log the generator tests

Backup generators are tested on a schedule. A test is an event, and events can be recorded precisely. A dated log of testing hours is a factual record that says nothing about anyone's health — and it is the thing most likely to be checkable against the site's own licence conditions.

• Date, start time, finish time. To the minute if you can.
• What you could hear — and take a noise reading at the same time if you have a meter. A generator test is when the site is loudest.
• What you could see — visible exhaust, and a timestamped photograph.
• What you could smell, and from how far away.
• Wind direction, because it explains why some days are worse than others.
• How long it lasted.

If you have a low-cost sensor

PM2.5 and PM10 readings from a consumer sensor are worth logging as indicative. They are good at showing change over time at one spot — a spike during a generator test, say — and poor at absolute accuracy. Record the make and model with the reading, and never present a consumer sensor number as a compliance figure.

Your own account

The measurements say what is happening. The accounts say what it is like. Both are needed, and the accounts are often what makes anyone read the measurements.

Write down

• Sleep — woken, unable to get to sleep, which room, what time.
• The hum — when you hear it, where in the house, whether it changes.
• Night lighting — where it falls, whether it comes through curtains.
• Traffic, dust, construction hours.
• Water pressure, and any restrictions in force locally.
• The view, and property.

Keep a dated diary rather than writing it all up from memory later. Dates and times are what turn an impression into a record. A line a night is plenty.

Safety, legality and privacy

• Measure only from public ground or your own property, or with explicit permission. Data centres are security-sensitive sites. No trespass, no confrontation.
• You need nobody's permission to measure on your own land or a public road.
• Do a daytime recon before any night session. Tell someone where you will be. Go in pairs at night.
• Keep it calm and lawful. The credibility of the data depends on how it was gathered.
• Your location is rounded before anything is published — to about 100 m, never a house number. You can leave your name off entirely, or give a first name and a townland.

What to do with it

• Send it in. It goes on the map once the source or the method checks out.
• Keep your raw data and log sheets. An unbroken record is what makes it usable later.
• For a planning objection or a formal complaint: lead with the night-time levels, the low-frequency bands and the C−A gap, with the protocol attached.
• Partner with an acoustician to interpret and validate anything intended for formal use. Citizen data opens the door; a professional walks through it.