The European Noise Directive requires "noise action plans" to be formulated, relying largely on a software prediction approach. But a new technology could help. Matthew Cand investigates.
Environmental noise assessments are routinely undertaken to study the impact of various developments; for example, a new residential development or the proposed modification of an existing road.
The impact of existing noise (mainly from transportation sources) on sensitive receptors must also be considered.
To evaluate the baseline or future noise environment for the area of interest, both measurements and predictions are used, often in combination, and either approach has different strengths.
But measurements studies tend to be limited in their extent and duration mainly given the associated time and costs constraints.
National noise mapping and action plans
The UK's current noise management policy has, in effect, been defined by the EU Directive on Environmental Noise. It requires that "noise maps" are produced for the main sources of environmental noise (ie major roads, railways, and airports) and agglomerations of a certain size using noise modelling software.
The first round of implementing the Directive saw noise maps published in 2007.
These maps are used as input to a second phase in the strategic planning and management of environmental noise, where competent authorities draw up Noise Action Plans (NAPs), with a view to "preventing and reducing environmental noise where necessary" and "preserving environmental noise quality where it is good".
The main benefit of using these predicted noise maps is for objectively targeting particularly affected areas, as is required by the legislation, over a wide area (the whole country or an agglomeration). This is done by identifying population centres exposed to the highest levels of noise.
Figure 1: Example of a predictive noise map covering a large area
For example, in England, the recently published NAPs have designated "Important Areas" as well as "First Priority Locations" which relevant authorities will focus on.
Similarly, NAPs established in Scotland or Northern Ireland for road and rail sources have identified Candidate Noise Management Areas (CNMA), where people are most likely to be annoyed by noise.
A CNMA may be promoted to a Noise Management Area (NMA), and for which mitigation measures are studied, following a validation process which includes an audit of the input data to the noise mapping. In Wales, the equivalent is a (Candidate) Noise Action Planning Priority Area.
Due to the complexity of noise prediction over large areas, it is necessary to make some simplified assumptions in the models.
Whilst these may then be improved or refined when considering a particular area of interest, this is limited by practical or technological considerations: the necessary input information may not be readily available, or require significant resources to capture in the required level of detail; the calculation or processing time could become prohibitive.
There are also inherent limitations in the scope of the noise predictions, which use simplified models and only predict the averaged metrics used in the strategic assessment, and neglect key aspects of the noise such as temporal variation or certain spatial features.
The first generation of NAPs have generally not prescribed a role for noise measurements to evaluate ‘important areas'. This is despite the considerable value that carefully designed measurement surveys can have in refining and supplementing noise predictions.
Value of array noise measurements
The first value of an array noise measurement is to complement and validate the strategic noise metrics given by the predictions, such as the average day-time noise level (LAeq,16hr) which can be compared with those derived from long-term measurements.
Close to the source, a discrepancy may indicate that the input noise emission data is incorrect: for example, the amount of road traffic has been underestimated.
Further away, information on the rate of decay of the noise level, and potential reflection or screening effects, can be obtained.
The evidence may also point to additional sources and features not included in the models. By increasing the number of measurement points, and therefore the spatial resolution, local features of the noise environment can be resolved in more detail; for example the effective screening effect of a noise barrier or a building might be of interest.
Environmental noise levels continually vary on a moment by moment basis, hourly, daily or seasonally and thus there is usually a time dimension that needs to be considered.
For example, two sites with equal average predicted noise levels could ultimately experience markedly different variations in levels, with associated differences in subjective response. Only measurement data can show these temporal patterns.
For example, statistics on the distribution of the noisiest events and quietest periods can be obtained, and the typical variability of noise levels examined in detail.
Long-term measurement data can provide this additional information, which is not available from the predictions alone. It provides a useful context which may be factored into strategic decision-making.
For example, the English NAP for major roads recognises the deficiency of the first round of noise mapping in modelling the real impact of night-time road traffic noise, along with the increased emphasis placed on this subject by the World Health Organization (WHO), and therefore aims to improve the assessment in this regard.
The character of the noise, or so-called "soundscape" of an area, represents another key parameter. For example, low frequency noise in urban environments can be more intrusive and annoying than is otherwise suggested by the A-weighted noise level alone.
Future strategic noise action plans may wish to consider how low frequency can be reduced, for example by the restriction of HGV movements in sensitive areas, or by the restriction of emissions from public transportation sources. But this is unlikely to be reflected in predicted noise level maps which focus on A-weighted noise level only, and thus measurements could prove useful in these instances to quantify and demonstrate the benefits of such strategies.
Data credibility can also be an issue. Relying on theoretical prediction models, regardless of the quality of the input data, can sometimes lead to mistrust in the affected population, because of the lack of confidence in the output data reflecting reality.
The use of extensive measurement results can help to reassure stakeholders of the robustness and realistic nature of the analysis. The effectiveness of mitigation measures could also be demonstrated by showing tangible improvements in the measured noise levels.
This would be consistent with the legislative requirements to monitor the progress of the action plans and provide periodic updates.
For noise measurement campaigns, practical issues such as security, accessibility, data transmission and power requirements should be considered.
They also require careful interpretation. But one of the main factors limiting the use of measurements in environmental impact studies has been the associated costs. Long-term, distributed measurement systems employing large numbers of sensors (10 to 100) have been restricted to specialised applications having large financial resources available to them.
For small to medium scale environmental noise impact studies, conventional measurement approaches tend to rely on short-term/temporary measurements at a limited number of locations.
The DREAMSys project
A collaborative research project was therefore proposed to develop a low-cost measurement system, with appropriate level of performance for noise mapping applications, to make wide-scale distributed measurement readily accessible. This would promote the use of noise measurement ‘arrays', consisting in a large system of measurement points deployed around the area of interest at more or less regular intervals, mainly to validate and complement noise prediction maps.
Figure 2: DREAMSys prototype units with different mountings
The project, involving the National Physical Laboratory, Castle Group, QinetiQ and Hoare Lea Acoustics, started in October 2007. The measurement system developed is known as DREAMSys (Distributed Remote Environmental Array Monitoring System).
It is based on the MEMS (Micro-Electro-Mechanical Systems) technology, originating in the micro-electronics industry, and used in Apple's iPhone and Sony's Playstation. The resulting microphones are low-cost with a very small form factor.
Following validation of its performance in the laboratory, the effectiveness of the system was tested by carrying out noise measurement trials at a range of outdoor sites.
In particular, access to a disused brownfield land in the Docklands area of East London was secured.
The area is very close to the direct take-off path of London City Airport, as well as the Docklands Light Railway, a large road and some light industry; this provided a rich acoustical environment for investigation. Forty units were deployed between September 2009 and June 2010.
Based on the extensive set of measured data, a refined noise prediction model was then produced for this area (see Figure 3 below).
It captures the spatial distribution of environmental noise during the daytime for the area in more detail than publicly-available strategic maps for the area.
For example, a light-industry source was found to dominate the noise environment over a limited area to the east of the site.
Only limited information was available on the noise sources at night, particularly as the airport doesn't operate during this period.
A prediction model was nevertheless developed by matching the measurements closer to the road and rail sources to the south of the site (see Figure 4).
It was observed that the decay of noise further away from these sources is lower than expected. This "noise floor" effect is formed by the contribution of multiple distance sources, which is typical of dense urban locations but difficult to represent in a computer model.
A wealth of additional information on the noise environment was obtained and is currently being analysed.
Figure 3: DREAMSys measurements and refined prediction model (daytime)
Figure 4: Measured average long-term night-time noise level (Ln) and corresponding predicted noise levels
In this way using measurements to supplement noise predictions can provide added value to environmental noise assessments, in particular in the context of noise actions plans which focus on key affected areas.
The development of a low-cost array measurement system makes the more widespread use of such assessments feasible in practice. This was illustrated by an example of a deployment of the system which uses MEMS microphones.
Acknowledgments
The authors acknowledge the financial support provided for this work by the UK Department of Business, Innovation and Skills, the National Measurement Office and the Technology Strategy Board, as well as the help of the London Development Agency.
