Choosing a representative background sound level period

Background sound level measurement decides more planning noise cases than the source measurement does. The comparison in BS 4142 is between the rating level and the background sound level, expressed as LA90 over the reference time interval, so a background chosen 3 dB too high turns an adverse impact into a low one. Environmental health officers know this, which is why the survey design is scrutinised harder than the arithmetic that follows.

The standard asks for a background that is representative of the period during which the source operates. Representative is doing a lot of work in that sentence, and the report has to explain how the figure was arrived at.

Why the background drives everything

LA90 is the level exceeded for 90 per cent of the measurement interval, so it describes the quieter underlying condition rather than the average. It rises with distant road traffic, falls in the small hours, and moves with season, weather and day of the week.

Because the assessment is an arithmetic difference, every decision about the background moves the conclusion by the same amount. Choose a period with a passing lorry route and the excess disappears. Choose the quietest 15 minutes of a still Sunday night and the same plant fails by 8 dB.

A Noise Impact Assessment that states a background figure without the dataset behind it cannot be reviewed, and a reviewer who cannot check it will assume the worst.

Choosing dates, days and duration

Match the survey to the source. If the plant runs 24 hours, survey across day and night. If a delivery yard operates from 06:00 on weekdays only, the relevant background is early weekday mornings, and a Sunday afternoon survey tells you nothing.

Avoid the periods that are not typical. Bank holidays, school holidays, local events, road closures and diversions, and harvest or other seasonal activity in rural areas all distort the picture. Note the school term dates for the area, because a background measured in half term on a residential road is measurably lower than in term time.

Season matters more than people expect. Leaf cover changes screening, heating and ventilation plant in nearby buildings runs in winter and not in summer, and rural backgrounds shift with farming activity.

On duration, a single hour is almost never enough. Several days of continuous unattended monitoring, covering the relevant days of the week, gives you a distribution rather than a point, and it is the distribution that makes the chosen figure defensible.

Attended and unattended monitoring

Each answers a question the other cannot.

Unattended monitoring gives you the statistical spread, the night-time data and a long enough record to show which values are typical. It cannot tell you what was making the sound.

Attended monitoring gives you the source identification, the subjective characterisation and the sanity check. An hour on site at the receptor tells you that the LA90 is being set by a distant motorway, or by a stream, or by an extract fan on the building next door that nobody mentioned.

Do both. Use attended periods at the start and end of the unattended survey, and at least one attended period covering the time of day that matters most for the assessment.

Record the equipment properly: a Class 1 sound level meter, laboratory calibration within date, field calibration before and after each deployment with the drift stated, and a windshield fitted at all times. Microphone position, height and distance from reflecting surfaces should be given, along with whether the measurement is free field or facade and what correction was applied.

Excluding atypical events

Something will always happen during a survey. A dog barks for an hour, a neighbour uses a hedge trimmer, an ambulance passes.

Handle it by exclusion with a record, rather than by quietly deleting. Log the period, the reason and the effect, and present the dataset with and without the exclusions where the difference matters. Audio recordings triggered by threshold, or continuous audio where site conditions allow, make this straightforward.

Two disciplines keep exclusions credible. Exclude the event rather than the whole night. And apply the same criteria across the dataset, so the loud events during quiet periods are not removed while the quiet events during loud periods are kept.

State what proportion of the data was excluded. A survey that discards a third of its measurements is telling the reviewer something about the location rather than about the events.

Night periods and the lowest level argument

This is where most disagreements happen. The reference interval at night is 15 minutes, so the dataset contains many more values, and the lowest of them will be considerably below the typical value.

BS 4142 asks for a background representative of the period, and does not ask for the lowest measured value. Environmental health officers often press for something close to the lowest, on the basis that residents are affected on the quietest nights.

The defensible approach is to show the distribution and reason from it. Present a histogram or a cumulative plot of the night-time LA90 values, identify the modal value and a low percentile, and explain which you have adopted and why. Where the difference between the modal value and a lower percentile changes the conclusion, run the assessment against both and say so.

Do not present a single number with no derivation. That is the pattern that produces a request for further information.

Weather and the data you must discard

BS 4142 measurements should not be affected by adverse weather. In practice that means discarding periods with rainfall, and periods with wind speeds high enough to generate microphone noise or to change propagation from the source.

Record weather throughout. An on-site weather station is the better option, and where a nearby official station is used, say which one and how far away it is. Wind direction relative to the source and receptor matters as much as speed, since downwind conditions raise measured levels and upwind conditions lower them.

Give the discard criteria explicitly, apply them mechanically, and show the discarded periods on the time history so a reviewer can see the gaps and their causes.

Construction noise sits under a different standard, BS 5228, with its own baseline conventions, so keep that assessment separate and cross-refer it from the Construction Management Plan rather than mixing the datasets.

Presenting the data

Include the time history for the whole survey showing LAeq, LA90 and LAmax, with weather overlaid and exclusions marked. Tabulate the LA90 values by reference interval for each assessment period. Add the distribution plot that supports the adopted figure. Provide photographs of the monitoring position with the source direction indicated, a location plan, and the calibration certificates.

That package is perhaps six pages of appendix, and it removes the most common reason an assessment is sent back.

Where a scheme turns on a marginal background figure, an independent check before submission costs far less than a refusal. PlanScope's professional review route is one way to arrange that, and you can browse the report types to see what else the site is likely to need.

This is general guidance, not legal advice. Noise surveys and BS 4142 assessments should be carried out by a suitably qualified acoustician, and for specialist technical reports qualified professional input may be required.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.