This paper describes a method for real-time, autonomous, joint detection-classification of humpback whale vocalizations. The approach adapts the spectrogram correlation method used by Mellinger and Clark [J. Acoust. Soc. Am. 107, 3518-3529 (2000)] for bowhead whale endnote detection to the humpback whale problem. The objective is the implementation of a system to determine the presence or absence of humpback whales with passive acoustic methods and to perform this classification with low false alarm rate in real time. Multiple correlation kernels are used due to the diversity of humpback song. The approach also takes advantage of the fact that humpbacks tend to vocalize repeatedly for extended periods of time, and identification is declared only when multiple song units are detected within a fixed time interval. Humpback whale vocalizations from Alaska, Hawaii, and Stellwagen Bank were used to train the algorithm. It was then tested on independent data obtained off Kaena Point, Hawaii in February and March of 2009. Results show that the algorithm successfully classified humpback whales autonomously in real time, with a measured probability of correct classification in excess of 74% and a measured probability of false alarm below 1%.

No abstract

A new discriminant based on the scintillation of normal mode amplitudes is introduced for the problem of passive surface/submerged source classification in a shallow water waveguide. The scintillation of modal energies is often used to characterize and understand acoustic wave propagation in a randomly fluctuating ocean waveguide [D. B. Creamer, J. Acoust. Soc. Am. 99, 2825 (1996)]. This paper proposes a variant of the traditional modal scintillation index to treat the discrimination problem in a typical littoral oceanic waveguide. The approach is based on a modal decomposition of fluctuations in the received pressure field associated with the temporal modulation of the depth of an acoustic source about its mean value. Source depth fluctuations are the result of a platform’s response to surface or internal wave motion. The rms mode excitations due to source depth modulation are shown to exhibit a depth dependent signature that may be exploited to statistically separate surface and submerged source classes. In this work, the modal scintillation index (SI) is defined as the variance in the estimated magnitude of the modal excitation normalized by its expected value over some observation interval. The statistic is self-normalizing, so knowledge of source level and source range is not required to separate the two source classes. Estimation of the modal excitation statistics requires only knowledge of the water depth and the sound speed profile at the array. Classification performance predictions in terms of receiver operating characteristic (ROC) curves will be presented based on KRAKEN Monte Carlo simulations under conditions of known and unknown source depth and range in spatially white Gaussian noise. An ad hoc decision criterion, which compares the minimum scintillation index across all modes to a threshold, was used to illustrate the phenomenology. Vertical line array and horizontal line array endfire geometries were considered. The modal scintillation approach may provide a robust alternative to matched field processing for the problem of binary source depth classification.

No abstract

A solution to the problem of acoustic source depth discrimination in a downward refracting, shallow-water ocean waveguide is presented for the case of a horizontal line array at endfire. The approach exploits the phenomenon of mode trapping, wherein a shallow acoustic source cannot excite the lowest order waveguide modes due to its evanescent amplitude dependence near the surface. The important implication of this "trapping" behavior is that, given sufficient spatial aperture, it provides a mechanism for differentiating a shallow acoustic noise source from one at depth. The method does not require array cant, or physical vertical aperture of any kind, but instead relies only on the sensitivity of a line array at endfire to differences in horizontal wave number to resolve low and high order mode subspace excitations. The only inputs to the algorithm are an approximate sound speed profile, water depth, and bottom type. The theoretical basis for the test statistic is first reviewed, followed by discussion of key requirements, and illustration of the concept using results from a RAM PE simulation for a downward refracting environment. Finally, the algorithm is experimentally demonstrated using data from a bottom-mounted HLA deployed in the moderately cluttered continental shelf environment of the Florida Straits.