Detection of long term trends in underwater noise levels

Abstract

The risk of adverse impact to marine ecosystems from underwater noise pollution is increasingly recognised by scientists, policymakers, and wider society. Deep water measurements from the Northeast Pacific indicate that ocean noise has increased substantially over recent decades. Policymakers are now considering establishing noise monitoring programs to determine noise levels and trends in their waters. However, the ability of noise monitoring to detect statistically significant trends is a function of the temporal extent, variance, and autocorrelation of the time series. This has implications for the feasibility of evaluating quantitative policy targets within prescribed time frames and hence should inform the formulation of such targets. The present work demonstrates that methods developed in other environmental science disciplines (e.g., atmospheric temperature measurement) to design long-term monitoring networks and assess their statistical power can be applied to noise monitoring programmes. Example datasets are used to show the application of these methods both to assess the significance of long-term trends in ambient noise, and the required monitoring period to detect a given magnitude of trend (e.g., 3 dB per decade). The implications for the design of noise monitoring networks and target setting for policy purposes are then discussed.