Improving MAMBAT: Toward fully automated model-based tracking of multiple marine mammals

We developed MAMBAT (Multiple-Animal Model-Based Acoustic Tracking) to fully automate the process of model-based passive acoustic methods for tracking multiple marine mammals. We previously demonstrated MAMBAT on one- and four-animal datasets of clicking sperm whales recorded on bottom-mounted hydrophones at the Atlantic Undersea Test and Evaluation Center in the Bahamas. This presentation will review the motivation, theory, and application of MAMBAT and discuss the improvements that we made to MAMBAT to track different species in increasingly challenging scenarios (e.g., more animals and lower signal-to-noise situations). Tracking results using bottom-mounted hydrophones from the Pacific Missile Range Facility in Hawaii will be used to demonstrate the improvements. [Work supported by the ONR Marine Mammals and Biology program.].