Spatial variation in sound frequency components across an urban area derived from mobile surveys

\ua9 2019 The Author(s). Continuous exposure to noise can lead to premature hearing loss, reduced cognitive performance, insomnia, stress, hypertension, cardiovascular diseases and stroke. Road noise affects the health of >125 million people in the European Union and Member States are required to map major noise hotspots. These strategic noise maps are usually derived from traffic counts and propagation models because large- scale measurement of the acoustic environment using conventional methods is infeasible. In this study, the authors surveyed the entire city of Southampton, UK using a mobile survey technique, capturing spatial variations in street- level sound characteristics across multiple frequencies from all sound sources. Over 52,000 calibrated and georeferenced sound clips covering 11 Hz to 22.7 kHz are analysed here to investigate variations in sound frequency composition across urban space and then applied to two issues: the definition of naturalness in the acoustic environment; and perceptions of social inequity in sound exposure. Clusters of acoustic characteristics were identified and mapped using spectral clustering and principal components analysis based on octave bands, ecoacoustic indices and dBA. We found independent patterns in low, mid and high frequencies, and the ecoacoustic indices that related to land use. Ecoacoustic indices partially mapped onto greenspace, identifying naturalness, but not uniquely, probably because urban anthropogenic sounds occur at higher frequencies than in the natural areas where such indices were developed. There was some evidence of inequity in sound exposure according to social deprivation and ethnicity, and results differed according to frequency bands. The consequences of these findings and the benefits of city-wide sound surveys for urban planning are discussed