Design methodology

The initial objective of the Life Harmonica project was to develop a simple index for presenting environmental noise levels, that is:

  • easy for the public to understand, because it is based on a scale of 0 to 10, rather than decibels ;
  • easy to calculate using the type of measurement data usually collected by noise measurement devices, namely the LAeq,1s levels (A-weighted, equivalent continuous sound level for 1s);
  • possible to calculate for one-hour time slots;
  • relevant in terms of acoustic physics, taking into account the two major components that affect the noise environment: background noise and noise events that exceed this background noise (noise peaks);
  • more representative of people's perceptions of their noise environment than the environmental noise indicators currently used in French and European regulations.

For this, a three-stage methodology was designed for the creation of the index.

First step: selection of the parameters for the index

The choice of the parameters involved in the construction of the index was based on a preliminary statistical analysis carried out on a database of elementary acoustic measurements (LAeq,1s) from 24 sites that are representative of the eight main types of noise exposure (land transport noise, air traffic noise, and quiet areas) in a variety of environments (urban, suburban, and rural). The aim was to choose acoustic parameters that would account for a significant proportion of the variability of many of the acoustic indicators.

Around 60 different energy- and event-based descriptors used in the available literature or proposed by the teams were calculated on an hourly basis using this elementary data from all 24 sites. The correlation between the 60 descriptors was studied in order to evaluate the descriptors that were interrelated (redundance), thereby eliminating those that measure the same thing and defining families of descriptors that are uncorrelated and, therefore, which provide complementary information. This method of statistical analysis is called Principal Component Analysis (PCA).

Using this PCA, it was possible to define two main families of interdependent and complementary physical descriptors, which provided most of the information:

  • The background noise
  • The noise peaks that emerge from the background noise

Then the teams proposed different suggested indices based on various parameters wich describe these two noise dimensions (background noise and noise peaks).

Second step: Consideration of the public's opinion and perceptions

The suggested indices were compared with the public's impressions using face to face interviews with 246 residents from 8 areas chosen for the diversity of their exposure to noise, and tests in laboratory conditions (individual interviews following binaural playback) on three groups of respondents: general public, associations, and local authorities (130 people in total).

The aim of this research phase was to evaluate the comprehensibility, the acceptability, and the relevance of the suggested indices and to take into account people responses in the final construction of the index.

Third step: Development of the formula of the index

The formula for the index was then set up by testing the different selected descriptors in an operational way, on all the operational measurement data stored by the noise observatory Bruitparif, in order to take into account the diversity of environmental noise situations. These tests led to eliminate some descriptors that have proved to be insufficiently robust and to adjust the coefficients of the index formula, in order to present the variations in hourly noise levels on a scale of 0 to 10. This last stage allowed us to propose the following formula :

Hourly Harmonica Index = background noise sub-index (BGN) + peak noise sub-index (EVT)

The Harmonica index, therefore, is divided into two parts:

  • A component related to the background noise: BGN sub-index
BGN = 0.2 x (LA95eq - 30)
LA95eq is the equivalent background noise level during the hour period, the background noise being evaluated every second by the noise level exceeded 95 % of the time during the 10 minutes period before
  • An event-related component : EVT sub-index that represents the acoustical energy provided by noise peaks that emerge above the background noise
EVT = 0.25 x (LAeq - LA95eq)
LAeq is the equivalent noise level during the hour period

Nota bene:
Until the end of the Harmonica project (31/12/2014), the calculation of the index is in an experimental phase on Bruitparif's and Acoucité's measurement sites, which are in the Ile-de-France region and Lyons metropolitan area respectively, the demonstration areas for the Life environment programme's Harmonica project. The formula for calculating the index may be amended by the end of the project.

To go further

How to characterize environmental noise closer to people's expectations Acoucité, Bruitparif, Internoise, September 2013