Localization and tracking of fish sounds with a 4-element underwater passive acoustic array

Abstract

Localization and tracking are parameter estimation problems. As such, estimates are evaluated by their statistical distribution, in particular, their bias (equal to the mean of estimated positions minus the true position) and variance (variability in the measured positions about their mean). A variety of methods are used to localize sounds recorded by acoustic arrays. Here, we quantitatively compare the performance of four algorithms: (1) plane-wave-front frequency-domain beamforming; (2) curved-wave-front frequency domain beamforming; (3) time-difference-of-arrivals (TDOAs) using waveform cross-correlation; and (4) TDOAs using spectrogram cross-correlation, for fish sounds generated with a controlled underwater source at various GPS-located positions around a 4-element array. The array consisted of a SoundTrap ST4300 (Ocean Instruments, Auckland, NZ) four-channel acoustic recorder, equipped with four HTI-96-MIN hydrophones (High Tech, Inc., Long Beach, MS). The hydrophones were arranged in a tetrahedral-shaped configuration with a 20-m inter-element spacing. The objective is to optimize the localization and tracking of individual soniferous fish to better understand their small-scale spawning movements and reproductive behavior. Research supported by the Office of Naval Research, the Scripps graduate department, and a Natural Sciences and Engineering Research Council of Canada (NSERC) Postgraduate Scholarship-Doctoral (PGS D-3).