Automated Detection and Localization of Fin and Blue Whale Calls Recorded with Distributed Acoustic Sensing on the Ocean Observatories Initiative Cables
Abstract
Distributed acoustic sensing (DAS) is a relatively new observational technique in academia that interrogates an optical fiber with repeated laser pulses and utilizes changes in the phase of backscattered light to measure the strain rate along the fiber. The method can work to distances of up to ~100 km and at this scale has a spatial resolution of tens of few meters. A DAS fiber optic cable thus, behaves similarly to a long line of closely spaced single-axis broadband seismometers.In November 2021, we participated in a 4-day DAS experiment on the twin cables of the Ocean Observatories Initiative (OOI) Regional Cabled Array (RCA) that extend offshore from Pacific City, Oregon. This experiment occurred during peak fin whale calling season and at least 20,000 20-Hz calls are visible throughout the experiment to ranges of tens of kilometers. A smaller number of 15-Hz A and B calls of the Northeast Pacific blue whale are also observed but only out to ranges of ~10 km. This proposal seeks support to understand the capabilities of submarine distributed acoustic sensing to record low frequency fin and blue whale calls and to develop automated tools to detect and localize baleen whale calls that can be applied to this and other DAS data sets. To accomplish this, we propose to recruit a postdoc with experience handling large data sets in acoustics or seismology to address the following four tasks:1.Develop approaches to automatically identify fin and blue whale calls in DAS data. We will investigate multichannel signal processing techniques such as f-k filtering to enhance the signal to noise ratio of marine mammal calls. We will apply established single channel techniques of spectrogram cross-correlation and matched filtering on both single channel and beamformed data. These will be compared to machine learning approaches using supervised and unsupervised algorithms based on both the distance-time images and single-channel time series.2.Develop techniques to locate fin and blue whales automatically and accurately with DAS data. We will apply cross-correlation techniques to either thetime series or spectrograms to estimate time difference of arrivals and localize the calls using either an iterative linearized inversion or grid search with travel times based on uniform velocity and more accurately the summation of multipaths obtained with the bathymetry and sound speed profiles for the region.3.Analyze the performance of DAS in recording whale calls at different conditions. We will utilize the large fin whale call data set, supplemented by the smaller data sets of blue whale calls and ship tracks, to understand the effect of gauge length, the directional sensitivity of DAS, source to receiver distance, call frequency, cable depth and source depth on the performance of DAS for marine mammal detection and localization.4.Estimate cable location using the whale calls. We will explore a simultaneousinversion for fin whale and DAS channel locations to accurately locatethe DAS channel on the seafloor, thus negating the need for an active source experiment to accurately locate the position of channels along the cable.Because the OOI DAS data set comprises 30,000-50,000 channels per fiber sampling at least 200 Hz, the data sets are very large (~ 2 GB / minute) and their analysis is challenging. We believe that character and scope of the work proposed above is best suited for a 2-year postdoc with prior experience working with large data sets either in acoustics or seismology. The postdoc#s research and career aspirations will be supported by an Individual Development Plan and by participation in UW#s eScience Institute and Photonic Sensing facility.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 15, 2023
- Source ID
- N000142312442
Entities
People
- William S.D. Wilcock
Organizations
- Office of Naval Research
- United States Navy
- University of Washington