Choosing a representative background on an industrial estate

On an industrial estate the representative background sound level decides the BS 4142 outcome more often than the plant specification does. Pick 38 dB and a rating level of 48 dB sits at plus 10. Pick 43 dB from the same dataset and the same plant sits at plus 5. Nothing about the development changed. The difference is a judgement made during data analysis, and it is the part of the assessment a reviewer should interrogate first.

BS 4142:2014+A1:2019 asks for a background sound level that is typical of the period being assessed. It does not ask for the lowest value measured, the arithmetic mean, or whatever the logger happened to record on the quietest night.

Why the choice decides the outcome

The standard compares a rating level against the background, and the initial indication of impact is read from the difference. A difference of around plus 10 dB or more indicates a significant adverse impact, around plus 5 dB an adverse impact, and a rating level at or below background a low impact, all subject to context.

Those bands are narrow. A five decibel shift in the background moves an assessment from one band to the next, and on an industrial estate the spread between the lowest and the modal night-time LA90 is routinely more than five decibels.

So the acoustician is making the determinative judgement before any mitigation is considered. That is why the standard requires the choice to be explained, and why a report that states a single figure with no supporting distribution should be sent back.

Presenting the full distribution of measured values

A defensible selection starts with the data presented in full. For each assessment period, a reviewer needs to see every measured LA90 interval, not a summary.

What makes a submission reviewable:

• A time history chart of LA90 and LAeq across the whole survey, with weather and the operating status of nearby sources annotated.
• A table of LA90 values by interval for each night, or a cumulative distribution for the night-time period.
• A histogram or ranked plot showing where values cluster.
• The modal value, the lower quartile and the minimum, stated side by side.
• A statement of which value was adopted and why.

Where the data appears only as a single number in a results table, the reviewer cannot test the judgement, and the honest response is to ask for the raw intervals rather than to argue about the conclusion.

Identifying recurring low values

The useful question is not what the quietest interval was but what level the location reliably falls to. A background that recurs on most nights of the survey is typical. One that appeared once, at 03:40 on a Sunday during a plant shutdown, is not.

A workable approach: look at the lowest level that is reached on the majority of survey nights, within the hours the proposed plant will operate. If five of six nights reach 40 dB for a sustained period and one reaches 35 dB, the typical figure is close to 40 dB, and the 35 dB value belongs in the report as an observation with an explanation rather than as the adopted background.

Where the use is intermittent, the relevant period narrows. Plant that runs only during a 06:00 to 08:00 delivery window is assessed against the background in that window, not against the quietest hour of the night. Assessors who adopt a whole-night minimum for daytime-only plant are being conservative in a way that costs their client money for no acoustic benefit.

Periods to exclude and why

Exclusions have to be stated and justified. The ones that are normally accepted:

1. Intervals affected by rainfall, which raises measured levels.
2. Intervals where wind speed at the microphone exceeded the threshold in the survey methodology, typically around 5 m/s.
3. Intervals contaminated by an identifiable one-off event, such as an alarm, a nearby works, or roadworks.
4. Periods when the existing site plant being replaced was operating, where the assessment concerns a replacement rather than an addition.

What is not acceptable is excluding intervals because they were inconvenient. Removing every low value as an outlier until the remaining distribution supports the desired conclusion is visible in the data, and a reviewer comparing the excluded intervals against the weather record will find it.

Keep the weather record. On-site measurement of wind speed, direction and rainfall through the survey period is what makes an exclusion defensible after the event.

Separating neighbouring plant from the true background

This is the hardest judgement on an industrial estate, and it is where reasonable acousticians disagree.

The background is what remains in the absence of the specific sound under assessment. Sound from other operators on the estate is part of the residual environment and legitimately contributes to it. But if the adjacent unit runs a chiller that may be decommissioned next year, basing an assessment on a background propped up by that chiller builds in a risk the receptor bears.

Three practical responses:

• Identify and describe the dominant contributors to the measured background by listening and by frequency content, rather than reporting a number alone.
• Where a single neighbouring source dominates, report the background both with and without it, and discuss the consequence for the assessment.
• Where the estate's operating pattern is known to be changing, say so, and consider assessing against the quieter future condition.

Receptor location matters as much as source identification. A measurement taken at the estate boundary is not the background at a dwelling 80 metres beyond it, and a measurement at the dwelling may be dominated by a road the estate does not control.

Sensitivity testing the selection

The strongest reports do not defend one number. They show what happens across a plausible range.

A short sensitivity table, giving the initial indication of impact at the modal background, the lower quartile and the minimum, tells a reviewer and an environmental health officer exactly how firmly the conclusion holds. Where the conclusion is the same across all three, the argument is effectively over. Where it flips between bands, the report should say so plainly and the mitigation should be designed for the worse case.

That honesty tends to shorten the determination. An officer who can see the uncertainty quantified is far more likely to accept a conclusion than one who suspects a convenient selection and has no means to test it.

Documenting the judgement for a reviewer

A reviewer of a Noise Impact Assessment is checking a chain: survey design, data quality, background selection, specific sound prediction, character corrections, and context. The background selection is the link most often asserted rather than argued.

What belongs in the report at that point is a short passage of reasoning naming the adopted value, the statistic it corresponds to, the nights on which it recurred, the exclusions applied with their justification, and the sensitivity result. Three paragraphs is usually enough, and it removes the single most common ground of objection from a competing acoustician.

Context then does the rest of the work. The standard is explicit that the initial indication of impact is modified by context, including the absolute level, the character of the area and the design of the development. A rating level at plus 5 dB in an area already dominated by industrial sound is a different proposition from the same figure in a quiet residential setting, and the report should address that rather than stopping at the arithmetic.

Where the proposal also raises emissions or extract questions, the same discipline applies to the Air Quality Assessment and the Ventilation and Extraction Statement, because officers read the three together. For survey design itself, the related note on unattended noise monitoring duration, position and validity covers what has to be right before any of this analysis is possible.

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