When an unattended noise survey is enough

Unattended logging over a week is the right answer for most residential schemes near a road, and the wrong answer for a plant noise assessment against a tonal source. Unattended noise survey duration is the question reviewers get asked, and the useful reframing is that duration and attendance are separate decisions. A seven day logged data set with no attended period can be unusable. A two hour attended survey with no logging can be equally unusable. Most competent surveys are both.

What unattended logging captures and misses

A logged data set gives you time history. Statistical indices at fixed intervals across days and nights, weekdays and weekends, which is the only way to identify the representative background sound level for the period of interest and to show how stable it is.

What it cannot do is tell you what made the sound. A logger records 47 decibels at 2am on Tuesday. Whether that was road traffic, a refrigeration unit two properties away, a railway, or a bird roosting near the microphone is not recoverable from the data, and the temptation to attribute it retrospectively is where assessments go wrong.

It also cannot identify acoustic features. Tonality, impulsivity and intermittency are characteristics of the specific sound that determine the rating level correction under BS 4142, and while a narrow band analysis of logged audio can support a tonal assessment, the subjective observations that back it up come from someone standing there.

When attended periods are necessary

Five situations where attended measurement is not optional.

Where the assessment applies a correction for an acoustic feature. Tonal and impulsive characteristics need to be observed and described, and an inspector or environmental health officer will ask who heard them.

Where the specific sound has to be separated from other sources. A plant item running against a busy road background cannot be measured directly, and the approach taken, whether measurement at a closer position with propagation calculation, or measurement during a quiet period, or derivation from sound power data, has to be described by the person who did it.

Where the background is being determined at a position with intermittent activity nearby. A logger next to a car park records the car park.

Where the source is not operating continuously and its cycle has to be captured. Refrigeration, defrost cycles, extract fans on time clocks.

And on any site where the microphone position needs to be justified. Photographs taken during an attended visit showing the microphone, the source and the nearest receptor answer more reviewer questions than a paragraph of description.

Survey duration and representative days

There is no single correct duration, and the honest reviewer position is that the duration has to be justified against the source and the receptor.

Representativeness matters more than length. A seven day survey during a school holiday, or in the week a neighbouring factory was shut, is seven days of the wrong data. Note the period in the report and say whether anything atypical was happening, including road works, seasonal closures and unusual weather.

Where the conclusion sits close to a threshold, extend the survey rather than arguing about it later. The cost of three extra days of logging is trivial against a further information request that delays determination by a month.

Weather criteria and data exclusion

Weather is the commonest reason logged data has to be screened, and the commonest omission from a report.

Record wind speed and direction, rainfall and temperature for the survey period, from an on site station or from the nearest official station with the distance stated. Exclude periods of rainfall and periods when wind speed exceeded the applicable criterion, conventionally around five metres per second, and exclude data affected by wind induced microphone noise.

Then show what was excluded. A table of excluded periods with the reason, and a statement of how much of the data set remains, lets a reviewer judge whether the survey still supports the conclusion. Reports that present a clean seven day data set with no mention of weather invite the assumption that nothing was screened, and a reviewer who checks the local weather record for that week will find the rain.

Downwind and upwind propagation matters for the specific sound too. Measuring a plant item with the wind blowing from receptor to source understates the level, and the wind direction during the measurement period belongs in the report.

Calibration and instrumentation records

This section takes ten lines to write and settles a surprising number of arguments.

• Instrument make, model and serial number, with its class
• Date of the last laboratory calibration, traceable, and the certificate available on request
• Field calibration before and after each measurement period, with the drift recorded
• Acoustic calibrator details and its own calibration date
• Windshield fitted, and microphone height and distance from any reflecting surface
• Measurement descriptors and time intervals logged, stated explicitly

Drift beyond the acceptable tolerance between the start and end checks invalidates the data in between, so record both figures rather than asserting that calibration was satisfactory. And state the descriptors. A report quoting background levels without saying whether they are LA90 values, and over what interval, cannot be checked.

Reporting both data sets together

The structure that reads well puts the two together rather than in separate sections.

Present the logged time history as a chart with the excluded periods marked, then the attended observations against the same timeline, then the derived values used in the assessment with the reasoning for each. A reviewer should be able to trace every number in the assessment back to a measurement or a stated calculation.

Two things to include that are often left out. The reason each measurement position was chosen, named against the receptor it stands for. And the professional judgement applied in selecting the representative background level, because that selection is a judgement rather than an arithmetic output, and a report that presents it as a calculation is concealing the step a reviewer most wants to see. Where a barrier is proposed on the strength of the survey, the limits of what one achieves are set out in our note on what an acoustic barrier actually achieves.

Common reasons a survey is rejected

From the reviewer's side, the same faults recur.

1. No attended period, with acoustic feature corrections applied anyway.
2. Daytime data only, with a night time assessment conclusion.
3. No weather record and no evidence of data screening.
4. Background determined at a position that is not representative of the nearest receptor facade.
5. Wrong descriptor, or the interval not stated.
6. Calibration drift not recorded, or laboratory calibration out of date.
7. The specific sound not separately characterised, so the rating level is derived from a total that includes the background.
8. The survey period unrepresentative, with no acknowledgement of it.

Most of these are recoverable by returning to site, which costs a fee and a fortnight. The one that is not recoverable is a survey carried out in a season or a period that cannot be repeated before the determination date. That is worth thinking about at the point the survey is commissioned rather than when it is reviewed. Servicing and delivery noise has its own survey timing problem, covered in our note on night deliveries and servicing noise.

Work of this kind belongs in a Noise Impact Assessment prepared or reviewed by a suitably qualified acoustician. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. For specialist technical reports, qualified professional input may be required. This is general guidance and not legal advice, so check the position with your local planning authority or a qualified professional before relying on it.