Choosing the background sound level for a noise assessment
The background LA90 decides whether a plant noise scheme passes. Here is how to choose a representative value, where to measure it, what to exclude and how to present it so an environmental health officer can agree with it.
The background sound level in a BS 4142 assessment is the LA90 at the assessment position, and choosing it is the single decision that most often decides whether a plant noise scheme passes or fails. Pick a representative value and the rating level lands where the acoustician expects. Pick the lowest figure recorded in a five-day survey and you have set yourself an unnecessarily hard target. Pick a value measured in the wrong place, or during an unrepresentative period, and the environmental health officer will ask for the survey again.
What LA90 measures and why it is used
LA90 is the A-weighted sound pressure level exceeded for 90 per cent of the measurement period. Because it discards the loudest tenth of the time, it strips out the car door, the passing lorry and the dog, and describes the underlying level that persists when nothing in particular is happening.
That is the point. BS 4142 compares the rating level of a specific industrial or commercial source against the level that would exist without it. The residual, meaning everything except the source, is what the neighbour is already accustomed to. An assessment that used LAeq for the background would be comparing a source against itself plus the traffic peaks, and the comparison would lose meaning.
Two derived numbers matter. The excess of the rating level over the background is the headline: around 5 dB indicates an adverse impact, around 10 dB an impact of greater concern, and a difference around zero or below indicates a low impact depending on context. The standard is explicit that context modifies all of this, which is why section 11 of BS 4142 exists and why a bare arithmetic difference is not an assessment.
Choosing representative measurement positions
The assessment position is at the noise-sensitive receptor, not at the site boundary, unless the boundary happens to be where people are exposed. For a residential facade, measure free-field where you can, at 3.5 m or more from any reflecting surface other than the ground, or measure at the facade and apply the correction the standard sets out. State which you did. Mixing the two between the source measurement and the background measurement is a common error.
Height matters where the receptor is a flat. A ground-floor garden and a third-floor bedroom window can have background levels several decibels apart, because road traffic screening falls away with height while the source may be roof-mounted and closer. Where the scheme is a rooftop plant installation above flats, measure at the worst affected window, not in the car park.
If the source is already operating, the background survey has to happen with it switched off, or at a comparable location where it is not audible. Getting an operator to shut down for a night is a negotiation, but a background level contaminated by the source under assessment is worthless.
Survey duration and the periods to cover
Cover every period the plant will operate. If the kitchen extract runs to midnight and the chillers run continuously, the survey needs daytime, evening and night data, with the reference time intervals the standard uses: one hour during the day and 15 minutes at night.
Length is a judgement. A single attended evening on a quiet suburban street may be defensible for a small installation. A mixed-use city site with variable weekday and weekend patterns needs unattended logging over several days with attended periods inside it, so that the logged data can be interpreted against what the acoustician actually heard. The attended observations are what let you explain the dataset later.
Weekday and weekend behave differently. So do term time and school holidays, and the guidance on avoiding holiday periods for parking and traffic surveys applies for the same reason here: the street is not itself.
Excluding atypical events from the dataset
Not every measurement belongs in the analysis. An aircraft overflight, a passing emergency vehicle, a neighbour's party, roadworks two streets away and heavy rain on the windshield all produce data that does not describe the ordinary condition.
Exclusion has to be justified and logged. Keep a written record of the time, the nature of the event and the reason for the exclusion, and keep the raw data so the reviewing officer can see what was taken out. An assessment that simply presents a cleaned dataset without the audit trail invites the suspicion that inconvenient periods were removed.
The counter-error is over-cleaning. Traffic on the road outside is part of the acoustic environment, not an atypical event. Removing it because it is loud produces a low background and a harsh assessment that the operator then has to design to.
Selecting the representative value rather than the lowest
The standard asks for a representative background sound level for each reference time interval, and that is a matter of judgement rather than an arithmetic rule. The usual approaches are to look at the distribution of measured LA90 values for each period and take a typical or modal value, sometimes the lower end of the typical range, and to explain the reasoning.
Taking the single lowest measured LA90 over a week is conservative to the point of distortion. It describes a condition occurring for a few minutes on one night, and designing plant attenuation to that figure can add cost with no benefit to anyone. Taking the arithmetic mean is the opposite error, because it is pulled upward by noisier periods.
Whatever you choose, show the distribution. A simple table of the measured LA90 values per period, or a time history plot, allows an officer to agree or disagree with a specific number rather than reject the whole report. If the scheme is finely balanced, say so and propose a condition setting a limit at the receptor rather than arguing over a decibel.
Weather, calibration and the survey record
The mechanics are unglamorous and they are the first thing a reviewer checks. Use a Class 1 instrument. Calibrate before and after, record both, and note the drift. Abandon or exclude data recorded in rain, and record wind speed, because measurements in wind above about 5 m/s are unreliable and generally excluded.
Record the make, model and serial numbers of meter, calibrator and windshield, the laboratory calibration dates, the microphone height, the coordinates of each position and a photograph looking in each direction. That record is what turns a set of numbers into evidence.
Common criticisms from environmental health officers
The recurring ones are short: a survey too brief to show the range, the wrong position, an unstated free-field or facade convention, no attended observations, exclusions with no log, a background contaminated by the source, and no consideration of acoustic character or context at all.
A Noise Impact Assessment that deals with each of those in a paragraph tends to get through consultation without a holding objection. Where the outcome will be secured by condition, the wording used for noise insulation and glazing specifications shows how much precision officers expect, and the relationship between planning control and statutory nuisance is set out in the piece on what a condition cannot fix.
This is general guidance, not legal advice; check the position with your local planning authority or a qualified acoustician before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. Enter your postcode on PlanScope to see which reports a noise-sensitive site is likely to need.
Frequently asked questions
What does LA90 mean?
It is the A-weighted sound pressure level exceeded for 90 per cent of the measurement period. Discarding the loudest tenth of the time removes passing vehicles, doors and dogs, leaving the underlying level that persists when nothing particular is happening. BS 4142 uses it as the background against which the rating level of a specific source is compared.
Should I use the lowest measured background?
No. The standard asks for a representative value for each reference time interval, not the minimum recorded. The lowest LA90 over a week may describe a few minutes on one night, and designing attenuation to it adds cost without benefit. Present the distribution of measured values and explain which figure was chosen and why.
How long should a background survey run?
Long enough to cover every period the plant will operate, including night if it runs at night. A small installation on a quiet street may need one attended evening. A mixed-use urban site usually needs several days of unattended logging with attended periods inside it, so the logged data can be interpreted against what was actually heard.
Where should the microphone go?
At the noise-sensitive receptor rather than the site boundary, unless people are exposed at the boundary. Measure free-field at least 3.5 metres from reflecting surfaces, or at the facade with the correction the standard sets out, and say which convention you used. For rooftop plant above flats, measure at the worst affected window.
Can measurements be excluded from the dataset?
Yes, where an event is clearly atypical, such as an aircraft overflight, roadworks or heavy rain. Each exclusion needs a logged time, description and reason, and the raw data should remain available. Ordinary road traffic is part of the acoustic environment and removing it produces an artificially low background.
Sources and further reading
This article is general guidance about the planning system in England and, where stated, the rest of the UK. It is not legal or professional advice. Check your local planning authority’s validation requirements and take professional advice for your specific site.