Spatial variation of exponentially modified Gaussian parameters in an urban setting

Abstract

The distribution of urban acoustic noise levels varies spatially because of differences in traffic volume, distance to other sound sources (e.g., construction sites), urban canyon geometry, and several other factors. We want to connect these environmental variables to the parameters of a parametric probability density function of the acoustic levels. Previously, the exponentially modified Gaussian (EMG) distribution matched experimental observation more closely than other distributions for urban sound levels. Using the framework of common engineering methods (e.g., ISO 9613-2), the received pressure level is the source power minus several attenuation factors. The exponential distribution could be due to the attenuation from geometrical spreading because it is a good approximation for a source uniformly distributed on a line segment or within a circle. The Gaussian distribution could be due to multiple factors, including weak turbulent scattering, reflections and diffractions, multiple sources, and source power variability. We measured the sound levels at 37 locations in the North End of Boston (USA), plotted the spatial variation in the EMG parameters, and compared these maps to the environmental variables. The traffic volume, proximity to stationary sources, and scene geometry all seemed to impact the EMG parameters with the traffic volume being the most important.