Spectral probability density as a tool for marine ambient noise analysis

Abstract

The empirical probability density of the power spectral density has been successfully applied as tool to assess signal variability and sensor system performance in the seismic literature. This paper presents the application of this analysis method to underwater ambient noise measurements, and demonstrates its utility in assessing the field performance of passive acoustic monitoring systems and the statistical distribution of noise levels across the frequency spectrum. Using example datasets from an autonomous passive acoustic recorder in the Moray Firth, Scotland, UK, and a cabled subsea observatory in the Strait of Georgia, British Columbia, we show how this method can reveal data limitations such as persistent tonal components and insufficient dynamic range, and phenomena such as bimodality and outliers which may be undetected by standard analysis techniques. We then combine this approach with conventional percentiles and spectral averages, illustrating how the underlying noise level distributions influence these metrics, and propose this technique as a standard, integrative presentation of ambient noise spectra.