Caller ID for Risso's and Pacific White-sided dolphins | JoVE Visualize

Caller ID for Risso's and Pacific White-sided dolphins

Mahdi H Al-Badrawi^1,2, Yue Liang³, Kerri D Seger⁴

¹Center for Acoustics Research and Education, University of New Hampshire, Durham, NH, 03824, USA. Mahdi.AlBadrawi@unh.edu.

Scientific Reports|March 17, 2022

View abstract on PubMed

Summary

New variational mode decomposition (VMD) methods improve tracking of marine species like Risso's and Pacific white-sided dolphins. This acoustic analysis enhances conservation efforts for expanding populations in the Arctic Ocean.

Area of Science:

Marine biology
Bioacoustics
Conservation science

Background:

Temperate marine species, including Risso's (Gg) and Pacific white-sided (Lo) dolphins, are expanding their ranges northward into the Arctic Ocean.
Accurate monitoring is essential for conservation, but these species produce similar pulsed sounds, complicating identification by humans and algorithms.
Automated acoustic detection and classification are needed to efficiently monitor these expanding populations.

Purpose of the Study:

To develop a novel visualization tool using variational mode decomposition (VMD) and a statistics-based classifier.
To differentiate between the similar pulsed vocalizations of dolphin species with expanding ranges.
To improve the efficiency and accuracy of acoustic monitoring for marine mammal conservation.

Main Methods:

Applied variational mode decomposition (VMD) for feature extraction from pulsed dolphin signals.
Utilized a statistics-based classification algorithm to analyze VMD-processed audio data.
Tested the method on low signal-to-noise ratio (SNR) audio files, without concurrent visual data.

Main Results:

The proposed VMD-based method achieved 81% accuracy in differentiating similar pulsed dolphin signals.
Demonstrated robustness to background noise and effectiveness in low SNR conditions.
Provided a visualization tool that aids human analysts in signal interpretation.

Conclusions:

Variational mode decomposition (VMD) is a promising technique for analyzing and classifying complex bioacoustic signals.
This automated approach can significantly enhance the monitoring of marine species with overlapping acoustic features.
The method supports conservation efforts by improving our understanding of ecosystem dynamics in a changing climate.