Objective classification of North Atlantic right whale (Eubalaena glacialis) vocalizations to improve passive acoustic detection
Author and article information
Abstract
Passive acoustic monitoring is playing an increasing role in the detection of endangered North Atlantic right whales (NARW). Previous acoustic monitoring has relied on a single stereotyped vocalization, the upcall. Here the entire repertoire produced by NARW during the winter and early spring in Cape Cod Bay, Massachusetts is described. An objective sound classification scheme and automatic classification algorithm were developed. Nine days of acoustic recordings were used for the data analysis and a total of 9,611 right whale sounds were identified. The objective classification scheme of right whale sounds allowed for rapid identification of a diversity of right whale sounds. These sounds were assigned to 6 classes of narrowband upcalls, downsweep, complex and high frequency calls, wideband gunshot sounds and complex sounds. Results indicate that the prevalence of upcalls varied from 28% of detected calls in January to 80% in April. Other classes of signals were also well represented in the repertoire including the narrowband complex(10-36%) and high frequency calls (1-26%), wideband gunshot sounds (4-25%) and wideband complex sounds (0 – 25%). The prevalence of non-upcall signals suggests that including more signals classes may improve rates of detection for right whales in the Cape Cod Bay habitat.
Cite this as
2014. Objective classification of North Atlantic right whale (Eubalaena glacialis) vocalizations to improve passive acoustic detection. PeerJ PrePrints 2:e322v1 https://doi.org/10.7287/peerj.preprints.322v1Sections
Supplemental Information
Figure 1: Map of the sensor array geometry. MARU locations areshown as black, numbered circles.
Figure 2: Spectrogramsof narrowband upsweep FM signals (NU).
The
spectrograms were obtained using 1024 point FFT with Hann window, 75% overlap, and
sampling rate of 5 kHz.
Figure 3: Spectrogramsof narrowband downsweep FM signals (ND).
Figure 4: Spectrogramsof narrowband complex FM signals (NC).
Figure 5: Spectrograms of narrowband high calls (NH).
Figure 6: Spectrogramsof wideband gunshot sounds (WG).
Figure 7: Spectrogramsof complex wideband signals (WC).
Figure 8: Block-diagram of the automatic classifier.
Figure 9: Daily distribution of detected NARW calls and their classes.
Figure 10: Empiricaldistribution of classes of all detected NARW vocalizations.
Additional Information
Competing Interests
There are no competing interests.
Author Contributions
Ildar R Urazghildiiev conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, wrote the paper, prepared figures and/or tables, reviewed drafts of the paper.
Susan E. Parks contributed reagents/materials/analysis tools, wrote the paper, reviewed drafts of the paper.
Funding
The funding source of this work is The Lab of Ornithology, Cornell University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.