Passive acoustic localization and tracking provide valuable information for characterizing the behavior of marine mammals. When using time-difference-of-arrival (TDOA) localization on a wide-baseline array, certain factors such as stereotyped vocalizations from multiple callers and sparse array configuration, can make the call association problem challenging. A "track-before-detect" localization and tracking framework is presented in which this challenge is addressed by taking a hard-association approach and performing the tracking in the TDOA domain. The framework is demonstrated on low-frequency stereotyped calls produced by Rice's whales (Balaenoptera ricei) recorded along the shelf break offshore of northwest Florida. Using the framework, a total of 398 whale tracks were extracted between May and August 2021. A median inter-call interval of 5.8 min was estimated from the calls identified in each set of TDOA tracks. Analysis of 1731 h from 368 two-dimensional tracks showed that whales primarily swam along, rather than across, the continental shelf break, with a mean horizontal swim speed of 0.9 m/s (standard deviation ± 0.6 m/s). By providing quantitative estimates of Rice's whale behavioral parameters, these results illustrate the utility of the developed passive acoustic localization and tracking framework.
Tenorio-Hallé et al. (Thu,) studied this question.