You are here

ECE PhD Defense: "Temporal-spectral characterization and classification of marine mammal vocalizations and diesel-electric ships radiated sound over continental shelf scale regions with coherent hydrophone array measurements," Wei Huang

10
Aug

318 Stearns

August 10, 2017 10:00 am
August 10, 2017 10:00 am

Abstract:

The passive ocean acoustic waveguide remote sensing (POAWRS) technology is capable of monitoring a large variety of underwater sound sources over instantaneous wide areas spanning continental-shelf scale regions. POAWRS uses a large-aperture densely-sampled coherent hydrophone array to significantly enhance the signal-to-noise ratio via beamforming, enabling detection of sound sources roughly two-orders of magnitude more distant in range than that possible with a single hydrophone. The sound sources detected by POAWRS include ocean biology, geophysical processes, and man-made activities. POAWRS provides detection, bearing-time estimation, localization, and classification of underwater sound sources.

The volume of underwater sounds detected by POAWRS is immense, typically exceeding a million unique signal detections per day, in the 10–4000 Hz frequency range, making it a tremendously challenging task to distinguish and categorize the various sound sources present in a given region. Here we develop various approaches for characterizing and clustering the signal detections for various subsets of data acquired using the POAWRS technology. The approaches include pitch tracking of the dominant signal detections, time-frequency feature extraction, clustering and categorization methods. These approaches are essential for automatic processing and enhancing the efficiency and accuracy of POAWRS data analysis. The results of the signal detection, clustering and classification analysis are required for further POAWRS processing, including localization and tracking of a large number of oceanic sound sources.

Here the POAWRS detection, localization and clustering approaches are applied to analyze and elucidate the vocalization behavior of humpback, sperm and fin whales in the New England continental shelf and slope, including the Gulf of Maine from data acquired using coherent hydrophone arrays. The POAWRS technology can also be applied for monitoring ocean vehicles. Here the approach is calibrated by application to known diesel-electric ships present in the Gulf of Maine and in the Norwegian Sea from their underwater sounds received using a coherent hydrophone array.

  • Professor Purnima Ratilal-Makris (Advisor)
  • Professor Stratis Ioannidis
  • Professor Douglas Cato