Background sound level: how LA90 is determined

The background sound level in a BS 4142 assessment is the LA90,T: the A-weighted sound pressure level exceeded for 90 per cent of a stated reference time interval, measured at the assessment location with the specific sound source absent. Get that single figure wrong and everything downstream moves with it, because BS 4142:2014+A1:2019 rates a proposal on the arithmetic difference between the rating level and the background sound level. A 3 dB error in the background is a 3 dB error in the result. And 3 dB is often the whole distance between a finding of low impact and an objection from the environmental health officer.

So the determination of a representative background sound level deserves more room in the report than most of us give it.

What the LA90 descriptor actually measures

LA90 is a statistical level. Over the measurement period the meter builds a distribution of short-term levels, and the LA90 is the level exceeded for 90 per cent of that period. In practice it strips out passing cars, dogs, doors and aircraft, and leaves the quieter underlying sound that a resident would notice a new plant item rising above.

That is the whole logic of the standard. A fan is judged against what the place sounds like when nothing much is happening, not against its noisiest minute.

Two consequences follow. First, LA90 is not the same as LAeq and the two can diverge by 15 dB or more on a road-influenced facade, so the numbers are never interchangeable. Second, because LA90 is a percentile, it is largely blind to brief events but very sensitive to steady sound: a distant motorway, a river weir, a neighbouring extract fan that runs all night will all lift it.

Choosing the reference time interval

BS 4142 works in reference time intervals, and the standard's defaults are one hour during the day (07:00 to 23:00) and 15 minutes at night (23:00 to 07:00). The reference time interval must match the period over which the specific sound occurs and the period the receptor is sensitive in.

Pick the interval before the survey rather than after it. A survey logged in 5 minute samples can be aggregated up to either, whereas a survey logged in 1 hour samples cannot be broken down to a 15 minute night-time interval, and that mistake costs a second visit.

Picking a representative value from logged data

Here the standard hands the work to professional judgement, and this is where reports diverge most. BS 4142 asks for a value representative of the background sound level during the periods of interest. It does not say take the lowest, the modal value, the tenth percentile of the LA90 set, or the average.

What defensible practice looks like:

1. Plot every LA90 sample as a time history and look at it before calculating anything.
2. Separate weekday from weekend, and school term from holiday, where the sound climate differs.
3. Identify the quiet plateau within each period rather than the single lowest sample, which may be one anomalous interval.
4. State the range across the survey alongside the value adopted, so the reader can see the spread you chose from.
5. Explain in one sentence why that value is representative of the periods when the specific sound will occur.

Adopting the arithmetic mean of all night-time LA90 values is the most common criticism raised on these reports, because the mean is pulled up by the busier hours either side of midnight and flatters the scheme. Adopting the single lowest 15 minute sample from a four-night survey is the opposite error and is usually indefensible too.

Illustrative example: a four-night survey at a rear facade logs LA90,15min values between 28 and 41 dB. The samples between 01:00 and 04:00 sit consistently at 28 to 30 dB, with the higher values confined to the hour after 23:00. If the proposed condenser runs continuously, the representative night background is around 29 dB, not the 34 dB mean of the whole night period. This example is illustrative only.

Night-time determination and where it goes wrong

Night surveys fail for practical reasons more often than analytical ones. A microphone at 1.2 m above a hard patio picks up reflections. An unattended logger records a sprinkler system, a neighbour's heat pump, or the survey vehicle idling. Insects on the windshield produce a level that looks like genuine background sound.

Attended spot measurements alongside the unattended logging solve most of this, because half an hour of listening at 02:00 tells you what the distribution is made of. Note the sources you can hear, in order of contribution, and keep that note in the appendix.

The other night-time trap is assuming the quietest period is the small hours. Near a bakery, a hospital, a 24 hour depot or a mainline railway it may not be, and the lowest background may fall in a short window that your reference time interval never samples.

Measurement positions and justifying them

Measure at the assessment location, which is the position where the effect on people is judged. For a residential receptor that normally means 1 to 1.5 m in front of the most affected window, or within the garden where amenity space is the concern. Where access to a private garden is refused, a proxy position is acceptable only with a stated correction and a stated reason.

Record for every position: height above local ground, distance from reflecting surfaces, a photograph looking towards the source, grid reference, and the meter serial number. Councils increasingly ask for that as a table. Where your position differs from the receptor because of access, say so in the same paragraph as the result rather than burying it in an appendix.

Data you should exclude, and why

Exclusions must be declared, not silent. The conventional grounds are wind speed above 5 m/s at the microphone, rainfall, and identifiable extraneous events such as construction next door or a fireworks display. Fit a windshield, log wind speed and rainfall locally rather than relying on an airport station 20 km away, and keep the raw data so that the exclusions can be audited.

A short table of excluded intervals with the reason against each is worth more to a reviewing officer than a paragraph of reassurance. Where more than about a fifth of the survey has gone, the honest answer is to repeat it.

Writing the judgement down

The assessment section is not the place to reveal the background figure for the first time. State it once, early, with the reference time interval, the period it applies to, the position it was measured at and the survey dates. Then carry it through the rating calculation without restating it differently.

The difference between rating level and background sound level is then interpreted in context: around +10 dB is an indication of a significant adverse impact, around +5 dB an indication of an adverse impact, and a difference at or below 0 dB an indication of a low impact, all subject to the context of the receptor and the character of the sound. Those are indications, not pass marks, and a Noise Impact Assessment that treats them as limits invites a challenge.

If you are assembling the wider evidence base around that acoustic work, the full set of 66 report types shows which documents usually travel together on a plant or change of use application, and the way national development management policies are drafted will shape how noise policy is cited in decisions over the next few years.

This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. A Noise Impact Assessment is a report councils expect a suitably qualified acoustician to author, and professional review is strongly advised.

Enter your postcode on PlanScope to see which reports your site is likely to need before you commission the survey.

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