Exploring spatial relationships among soundscape variables in urban areas: A spatial statistical modelling approach

Abstract

Noise maps based on sound pressure levels are limited in accurately representing how people perceive the sound environment. As a complementary approach to noise maps, soundscape maps can be useful tools for urban planning and design because they provide more information than conventional noise maps to reflect perceived acoustic environments. This study provides an overview of soundscape maps and explores the influence of spatial contexts on soundscapes in urban spaces. Physical, acoustic, and perceptual data were collected on sound environments in various urban areas in Seoul to create soundscape maps. Sound source types (traffic noise, human sounds, water sounds, and birdsongs), psychoacoustic parameters (loudness and sharpness), and perceived soundscape quality maps were generated and it was found that soundscape characteristics were varied in accordance with the primary functions of a space. Based on the collected soundscape data, global and local spatial regression analyses were conducted to examine the spatial autocorrelation on prediction on soundscape quality, and spatial dependency of soundscape quality was found in both models. In particular, a local spatial regression model based on geographically non-stationary relationships in variables was found to be more effective in understanding soundscape quality in multi-functional urban spaces. These findings could provide useful knowledge for soundscape planning strategies.