Automated model-based localization of marine mammals near California

Abstract

In a previous work, we developed an algorithm for acoustically tracking singing humpback whales near Hawaii. Pair-wise time-differences in arrival of whale calls as measured by a phase-only correlation process are compared to time-lags predicted by an acoustic propagation model. Differences between measured and modeled time-lags defined an ambiguity surface that identifies the most probable whale location in a horizontal plane around an array. In this work, we describe the application of this technique to a very different environmental scenario involving blue whales off the coast of California. The whale calls are much lower in frequency and the receivers are ocean bottom seismometers. Again the algorithm performs extremely well, providing the capability for real-time, automated monitoring and alert.