Background sound level selection in a noise assessment
The selected background level moves a BS 4142 result more than the plant data does. How to read the distribution, justify exclusions and document the choice.
Background sound level selection decides the outcome of a BS 4142 assessment more often than the plant data does. Pick 38 dB instead of 43 dB as the representative LA90 and a rating level of 48 dB moves from a +5 dB difference to +10 dB, which is the difference between a negotiation about an acoustic louvre and a refusal. The measurement is usually sound. The selection is where assessments fall apart under review.
BS 4142:2014+A1:2019 is unusually direct about this, and section 8 repays a second reading before you sign off a report.
What the background level is actually used for
The standard rates sound by comparing the rating level, which is the specific sound level with corrections for acoustic character, against the background sound level at the assessment location. The difference between the two is the indicator.
A difference of around +10 dB or more is an indication of a significant adverse impact, depending on context. Around +5 dB is an indication of an adverse impact. A rating level at or below the background is an indication of low impact. All three are qualified by context under section 11, which covers absolute levels, the character of the sound, the sensitivity of the receptor and the existing acoustic environment.
So the background is not a limit. It is the reference against which the arithmetic difference is calculated, and that difference then gets interpreted. Reviewers who treat the background as a target to be met have misread the standard.
Representative, and why that word replaced typical minimum
The 1997 version asked for the typical minimum background level, and practitioners interpreted that as the lowest measured LA90, or close to it. The 2014 version asks instead for a value representative of the time period of interest, and the change was deliberate.
Section 8.2 is explicit that simply taking the arithmetic average of the measured LA90 values is not appropriate, and neither is taking the lowest measured value. Both answers are available in the data and both are wrong. The lowest value may represent a single unusual quarter hour. The average is pulled upward by the noisier intervals, which is exactly what the assessment is trying to exclude.
What the standard wants is a judgement supported by the distribution of measured values, with the reasoning set out. Two acousticians looking at the same dataset can reasonably reach values a decibel or two apart. Five decibels apart means one of them has not done the analysis.
Reading the distribution instead of averaging it
The method that survives review is to plot the data and describe what it shows.
Produce a time history of LA90 by reference interval across the whole survey, then a histogram or cumulative distribution of those values for the period of interest separately for day and night. Then identify the modal value, or the value below which only a small proportion of intervals fall, and explain the choice.
| Approach | What it gives you | Why reviewers challenge it |
|---|---|---|
| Lowest measured LA90 | The single quietest interval | Not representative, often a measurement artefact |
| Arithmetic mean of LA90 values | A value inflated by busy intervals | Expressly discouraged by the standard |
| Modal value from the distribution | The level that occurs most often | Defensible, but needs the distribution shown |
| Lower percentile such as the tenth | A consistent and reproducible rule | Needs justification that the percentile suits the period |
The modal approach works well where the distribution has a clear peak, which is common in suburban locations dominated by distant road traffic. It works badly where the distribution is bimodal, for instance a location affected by an intermittent railway or by a nearby premises with fixed closing hours. In that case split the period and assess each separately rather than forcing one number onto two different acoustic environments.
A worked illustration, clearly illustrative. Night time LA90,15min values across five nights cluster between 34 and 37 dB with a mode at 35 dB, with a tail down to 30 dB on two intervals in the hour before dawn and a scattering above 40 dB when a delivery vehicle passed. Selecting 35 dB is defensible. Selecting 30 dB would be the old typical minimum reading of the standard, and selecting 38 dB would be the average dragged up by the deliveries.
Excluding atypical periods without excluding the truth
Exclusions are legitimate and must be documented, item by item, with the reason.
Rainfall and wind speeds above about 5 m/s at microphone height are the standard meteorological exclusions, and a logging weather station on site is the only way to evidence them. Beyond that, the defensible exclusions are events unrelated to the normal acoustic environment: road works, a one off local event, construction on an adjacent site, an alarm sounding.
What is not legitimate is excluding the character of the place. A location next to a pub garden is noisy at 22:00 on a Friday and quiet at 22:00 on a Tuesday, and both are the real acoustic environment. If the plant will run every night, the quieter nights are the relevant case and you say so in the selection reasoning rather than deleting the louder ones.
Survey timing carries the same discipline. Measuring an urban background during the school summer holidays, or during the week between Christmas and the new year, gives a lower result that will not hold. Note the dates and justify them.
Position, height and the facade correction
The assessment location drives the value, and the standard expects measurement at the assessment location or at a location with a similar acoustic environment.
Free field measurement, with the microphone at 1.2 m to 1.5 m above ground and at least 3.5 m from any reflecting surface other than the ground, is the default. Where a facade measurement is unavoidable, the reflection raises the measured level by approximately 3 dB and the report has to state how that has been handled. Mixing a facade background with a free field specific sound level, or the reverse, is a frequent error and it always works in one direction.
Height matters more than people assume. In locations dominated by distant road traffic, upper floors often have a higher background than ground level, because the ground floor is screened by boundary walls and parked vehicles. Assessing a fourth floor bedroom against a ground level background understates the background and overstates the impact, which is a mistake that favours the objector rather than the applicant and still gets the report sent back.
Where the nearest sensitive receptor is a window on an upper storey, measure there or model the difference and justify it.
Night time and the quietest reference interval
The night period is 23:00 to 07:00 with a 15 minute reference time interval, against the 1 hour interval during the day.
Shorter intervals produce a wider distribution, so the spread of night values is always greater and the temptation to reach for the lowest is stronger. Resist it on the same grounds as during the day, and apply the same distribution analysis to the 15 minute data.
Two night time specifics. The dawn chorus from roughly March to July raises LA90 for an hour or more and should be identified rather than quietly left in. And the hour between about 02:00 and 04:00 is often the genuine minimum, so if the plant runs continuously that period is the relevant case and the selected value should reflect it rather than an all night mode.
Where plant only runs in the evening, the relevant background is the evening, and the night data is context rather than the reference.
Documenting the choice so a reviewer can audit it
The selection paragraph is the most read part of a noise impact assessment, and it should contain all of this in half a page.
- The survey dates, times and the reason for choosing them.
- Instrument type, class, serial numbers, and field calibration before and after with the drift recorded.
- Measurement position with a photograph, height, and free field or facade stated.
- Meteorological data logged on site, with the exclusion criteria applied.
- The distribution of LA90 values, shown as a plot, for each period of interest.
- The selected value for each period, with the reasoning in words.
- A sensitivity check showing the rating difference if a value two or three decibels either side had been chosen.
That last step disarms most objections. If the conclusion holds at 33 dB and at 37 dB, the argument about which is representative stops mattering. If it does not hold, you have found the real issue before the council's environmental health officer does.
Plant serving a kitchen extraction system raises the same selection question with an added character correction for tonality, and the detail of that sits in a ventilation and extraction statement. Where a site carries three or four separate plant items, the approach to cumulative effects applies to sound as it does elsewhere.
This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional 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.
Draft the structure of your assessment on PlanScope, then complete the survey analysis and sign it off yourself.
Frequently asked questions
Should I use the lowest measured LA90 as the background level?
No. BS 4142:2014+A1:2019 states that neither the arithmetic average of measured LA90 values nor the lowest measured value is appropriate. The standard asks for a value representative of the period of interest, supported by analysis of the distribution of measured values and explained in the report.
How do I identify a representative value from survey data?
Plot the LA90 values by reference interval as a time history, then produce a histogram or cumulative distribution for each period of interest. Identify the modal value or a justified lower percentile. Where the distribution is bimodal, split the period and assess each acoustic environment separately rather than forcing one figure.
Which measurement exclusions are acceptable?
Rainfall and wind speeds above roughly 5 m/s at microphone height, evidenced by a logging weather station, plus discrete events unrelated to the normal environment such as road works or an alarm. Excluding the ordinary character of a place, for example busier weekend evenings, is not legitimate and will be challenged.
Does it matter whether the background was measured at a facade?
Yes. Free field measurement at 1.2 to 1.5 m above ground and at least 3.5 m from reflecting surfaces is the default. A facade measurement is raised by roughly 3 dB by reflection, and the report must state how that has been handled. Mixing a facade background with a free field specific level biases the result.
What reference time intervals apply?
One hour during the day period of 07:00 to 23:00 and 15 minutes during the night period of 23:00 to 07:00. The shorter night interval widens the distribution, so the same distribution analysis must be applied rather than reaching for the lowest quarter hour of the survey.
How can I make the selection defensible?
Include a sensitivity check showing the rating difference if a value two or three decibels either side had been selected. Where the conclusion holds across that range, the argument about the exact figure stops mattering. Where it does not, you have identified the real issue before the environmental health officer does.
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.