Ocean ambient noise variation and non-linear signal analysis

Abstract

Snapping shrimp sounds are a significant component of the ambient noise background in subtropical regions such as Hawaii. Quantification of this noise source is important to a variety of underwater acoustic applications, especially passive acoustic monitoring of harbors and ports. Seasonal variations of snapping shrimp spectra have been reported. Snapping shrimp noise production has been shown to increase at night and decrease during the day. However, the spectral characteristics of sound during these periods are not well understood. For passive acoustic monitoring, changes in the background sound levels over hours to minutes are of significant interest. This study examines changes in the ambient noise spectra over the diurnal period of data taken at two locations on Oahu (Kilo Nalu Ocean Observatory and Coconut Island). The correlation between light levels and noise intensity is also investigated. Non-linear methods such as recurrence plots and detrended fluctuation analysis are applied to the ambient noise data. Detrended fluctuation analysis provides a measure of self-similarity. Recurrence plots are applicable to short, nonlinear and nonstationary data, and allow for visualization of higher order phase space trajectories. These methods may provide further insight into the nonlinear characteristics of ambient noise dominated by snapping shrimp. Recurrence plots are used to investigate the acoustic characteristics of nearshore boating activity. Using a root mean square (RMS) signal, boating activity is analyzed from a long term data collection, and cross referenced with optical video records. Recurrence quantification analysis provides a means of discerning repeated trajectories of a single boat from the RMS of an acoustic time series record. Recurrence plots are applied to short acoustic time records of different boats, and their differences are explored in terms of dynamical invariants such as entropy, laminarity, and determinism.