Geostatistical Modeling of Environmental Sound Propagation

Abstract

The atmospheric influence on sound propagation along the ground surface is a critical issue for estimating the noise impact of industrial plants or road networks. Indeed, sound refraction in the surface layer has a dramatic impact on the geographical acoustic exposure. Many analytical and numerical models and studies based on the laws of physics are available in scientific papers whereas very few works in statistical analysis have been attempted. However several important practical issues need to be considered. Among these, time and space representativity of "in situ" measurements, sampling design, influence of meteorological and ground parameters on acoustic exposure show to be a few challenges. They need to be investigated with statistical tools taking into account space and time autocorrelation. A new protocol which includes ground impedance monitoring, spatial micro-meteorological and acoustical characterization has been applied to an experimental campaign from June to August 2005 in a case of sound propagation from a point sound source on a grassy flat ground. The first geostatistical study on such a multivariable experimental database is presented. It adresses both the issue of modelling space varying impedance properties of an homogeneous meadow and the issue of modelling the acoustic field itself. This latter includes an analysis of the spatial variogram of the acoustic field residual calculated from a basic physical model as an external drift.