Validating noise maps


Noise maps, calculated using standardized source and propagation models are today commonly used to guide environmental noise policy. This approach has its shortcomings for example:

  • Noise maps are as accurate as the source information that is entered in the model. Often traffic information for lower category streets is poor or lacking completely.
  • The majority of currently available methods for calculated noise maps do not accurately model urban canyon to canyon propagation and long range propagation over urban settlements. This results in an underestimate of noise levels in shielded areas. New approaches such as the model developed in the Life+ project QSIDE aim at improving prediction of the quiet side level.
  • Some important sources of environmental noise disturbance such as noise from people leaving pubs are hardly predictable.

Hence there is an interest in urban noise monitoring. Several larger cities have deployed such networks, some examples: Madrid (30 fixed stations), Brussels (17 fixed stations), Lille (90 fixed stations), … The classical strategy to deploy these networks consists in choosing very reliable – and thus expensive – sensors. New, alternative approaches use consumer electronics to construct the sensor nodes. IDEA follows this new line of thinking.


IDEA’s innovation and strengths

IDEA promotes the “just-accurate-enough” strategy in selecting sensor nodes. Given the application, accuracy of single nodes can be reduced: since most urban environmental sound contains neither very high nor very low in frequency, flat spectral response is less critical towards the extremes of the spectrum; noise levels are also sufficiently high to relax requirements on noise floor; finally occasional failure is acceptable as long as reestablishing operation involves low cost. The “just-accurate-enough” strategy for sensor nodes requires stronger back office quality checking. IDEA intelligence was developed for this purpose. This implies higher demands for communication and computation, both however becoming increasingly cheap over the years.


Detailed description of the demonstrator

Any of the noise sensor nodes deployed in IDEA can be part of a validation of the noise maps produced for compliance with the European Environmental Noise directive (END). The demonstrator checked this however in particular in the city of Gent, Belgium. During the demonstrator, it became clear that many of the sensor nodes outperformed expectations. The earliest nodes installed have been operational since the end 2010. At particular places the long term observations were compared with high-end equipment. Differences in noise levels between equipment are below 2dB(A) in actual field deployment. It should be noted that differences between high end equipment are slightly below 1dB(A) in the same conditions. Quality of the observations is assessed online and faulty or unexpected measurements are removed. The resulting overall Lday, Levening, and Lnight seem to correspond reasonably with the END maps along larger streets. However for smaller streets and shielded areas the noise maps largely underestimate noise levels. In view of the added benefit for the population of having access to quiet side of the dwelling or an (urban) quiet area, noise levels at these shielded locations are nevertheless important. Therefore the QSIDE model has been developed. This model was also validated against IDEA measurements. Noise maps are source related thus it is important to eliminate measurements of sound from other sources. The IDEA technology allows identifying the dominant source and thus eliminating evidently unwanted measurements. However, care is needed in this process not to eliminate quiet periods that are more easily dominated by other sound.





Prof. Dr. Ir. Dick Botteldooren
+32 (0) 9 264 99 68

Technologiepark-Zwijnaarde 126, 9052 Gent



©IDEA-Project 2009