Tracking sperm whales with a towed acoustic vector sensor

• Published in: The Journal of the Acoustical Society of America 2010-11, Vol.128 (5), p.2681-2694
• Main Authors: Thode, Aaron, Skinner, Jeff, Scott, Pam, Roswell, Jeremy, Straley, Janice, Folkert, Kendall
• Format: Article
• Language: English
• Subjects: Acoustics; Acoustics - instrumentation; Animal Migration; Animals; Cetacea; Environmental Monitoring - methods; Exact sciences and technology; Fisheries; Fundamental areas of phenomenology (including applications); Marine; Models, Theoretical; Physics; Ships; Sperm Whale; Underwater sound; Vocalization, Animal
• ISSN: 0001-4966; 1520-8524
• DOI: 10.1121/1.3495945

Passive acoustic towed linear arrays are increasingly used to detect marine mammal sounds during mobile anthropogenic activities. However, these arrays cannot resolve between signals arriving from the port or starboard without vessel course changes or multiple cable deployments, and their performance is degraded by vessel self-noise and non-acoustic mechanical vibration. In principle acoustic vector sensors can resolve these directional ambiguities, as well as flag the presence of non-acoustic contamination, provided that the vibration-sensitive sensors can be successfully integrated into compact tow modules. Here a vector sensor module attached to the end of a 800 m towed array is used to detect and localize 1813 sperm whale "clicks" off the coast of Sitka, AK. Three methods were used to identify frequency regimes relatively free of non-acoustic noise contamination, and then the active intensity (propagating energy) of the signal was computed between 4-10 kHz along three orthogonal directions, providing unambiguous bearing estimates of two sperm whales over time. These bearing estimates are consistent with those obtained via conventional methods, but the standard deviations of the vector sensor bearing estimates are twice those of the conventionally-derived bearings. The resolved ambiguities of the bearings deduced from vessel course changes match the vector sensor predictions.