Page 40 - Acoustics Today Summer 2011

Fig. 2. An ambiguity surface representing agreement between measured and modeled arrival patterns for hypothesized source locations on a vertical range/depth slice along bearing 200o from the Bahamas receiver. The marked peak indicates the best estimate of source location. With permission of Canadian Acoustics.
far enough to use acoustic propagation models and multipath arrival information (echoes) for more accurate localization estimates, but these usually require use of an array of under- water sensors. However, Christopher Tiemann at University of Texas at Austin has demonstrated that a full three-dimen- sional estimate (range, depth, and bearing from a sensor) of a whale’s location can be made using acoustic data from just a single sensor.
In two separate experiments in Alaska and the Bahamas, the echolocation clicks of endangered sperm whales were recorded on mid-water and bottom-mounted hydrophones. A single broadband click is often heard multiple times at a receiv- er due to different acoustic paths that echo off the sea surface or sea floor, attenuating with each bounce. It is these echoes that enable single-sensor localization, but the late arriving echoes can be very difficult to detect. Spectrograms are com- puted from the full measured time series. Integrating spectral power in the expected click frequency band over small time windows of acoustic data creates a time series where peaks mark the arrival of broadband clicks. When excerpts from this series are time-aligned with the first arrival of consecutive click events, persistent peaks from echoes can be recognized; even extremely faint ones from paths reflecting several times through the ocean waveguide (Fig. 1). The spacing of these arrivals serves as a unique fingerprint of whale location which can be compared to predicted arrival patterns from modeled sources at all ranges, depths, and bearings. A score represent- ing their agreement can be viewed as an ambiguity surface (Fig. 2) where the location of the modeled source with the best score becomes the estimate of whale location. Repeating this process every time a whale clicks allows tracking of its motion over time. (Fig. 3)
Applying diffraction theory to measure acoustically the in-situ orientation of marine animals
A significant goal in the use of “active” acoustics for observing marine animals is to infer their size and type from the measure- ment of the reflected acoustic “pings.” Using the relative amount of reflected energy has yielded important information; however there are substantial complications because the reflected energy is not only a function of size, but also orientation and composition. In order to explore the use of methods that could improve on the more traditional nar- row band echo sounders, Jules S. Jaffe, Paul L. D. Roberts, and their group at the Marine Physical Laboratory, Scripps Institution of Oceanography, have been measuring the angular dependent wide-band backscatter from live fish in a laboratory environment. Using various techniques to process these data have yielded interesting insights into the utility of this additional information while also highlighting the significant challenges that occur for using advanced methods. This
work explores the implementation of an algorithm to compute animal orientation from wide-band backscatter data (Jaffe and Roberts, 2011) collected in the lab from live fish. Orientation estimation is an important parameter in interpreting echo- sounder surveys as the amount of reflected energy can vary by several orders of magnitude as a function of orientation and hence, yield incorrect estimates of animal size.
36 Acoustics Today, July 2011
Fig. 3. Range, depth, and bearing estimates relative to a single receiver for a clicking sperm whale. Estimates were made using acoustic data from just the one sensor. With permission of Canadian Acoustics.