Cornell University
Library
Cornell UniversityLibrary

eCommons

Help
Log In(current)
  1. Home
  2. Cornell University Graduate School
  3. Cornell Theses and Dissertations
  4. Passive Acoustic Monitoring of Whales and Ocean Ambient Sound to Inform Management

Passive Acoustic Monitoring of Whales and Ocean Ambient Sound to Inform Management

File(s)
Parry_cornellgrad_0058F_14746.pdf (25.51 MB)
Permanent Link(s)
http://doi.org/10.7298/agpj-br21
https://hdl.handle.net/1813/117197
Collections
Cornell Theses and Dissertations
Author
Parry, Dawn
Abstract

Passive Acoustic Monitoring (PAM) provides reliable, long-term data to support and plan effective management of marine ecosystems and their living resources. Baseline data and continuous soundscape monitoring are crucial to identifying long-term trends and sudden environmental shifts. Federal agencies in the United States and other countries consider the impacts of human-generated sounds on marine ecosystems when granting permits for human activities in the ocean as they may affect whale communication, behavior, and long-term fitness. This work examines whale presence and sources of anthropogenic noise in the Gulf of Mexico (GoMex) and Bermuda before proposing a method to accelerate acoustic analysis, facilitating faster reporting for management. In the GoMex, we used hurricane periods as baseline soundscapes with minimal human activity and found that vessel noise and seismic exploration for oil and gas increased sound levels by over 16x. In Bermuda, we replicated sound analysis studies from 1966 and 2013 that found low-frequency (< 1000 Hz) ambient sound levels have increased 5 – 20 dB re. 1 μPa since the 1960s, during which time vessel traffic in the Atlantic has increased 2.5x. Additionally, we provide daily occurrence data for fin, sei, blue, sperm, and beaked whale in Bermuda and confirm established humpback migration patterns, valuable information to the government of Bermuda. Analyzing these long-term acoustic data is time-consuming. Though automated whale detection using machine learning can increase the pace of acoustic analysis, it requires large amounts of training data, ideally from the target soundscape. We propose a method that creates synthetic training data by inserting a known call onto background noise from a new environment using latent diffusion. It eliminates the need to search for whale calls in the target soundscape and improves model performance by incorporating the noise profile of the new site into the detector's training data. This method accelerates acoustic analysis, facilitating faster reporting to inform management. Results from this work can be used to inform permitting of anthropogenic activities in the U.S. and monitoring of a new MPA comprising 20% of Bermuda’s waters.

Description
183 pages
Date Issued
2024-12
Keywords
bioacoustics
•
cetacean
•
management
•
ocean
•
sound
Committee Chair
Klinck, Holger
Committee Member
Knuth, Barbara
Morreale, Stephen
Degree Discipline
Natural Resources
Degree Name
Ph. D., Natural Resources
Degree Level
Doctor of Philosophy
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International
Rights URI
https://creativecommons.org/licenses/by-nc-nd/4.0/
Type
dissertation or thesis
Link(s) to Catalog Record
https://newcatalog.library.cornell.edu/catalog/16921986

Site Statistics | Help

About eCommons | Policies | Terms of use | Contact Us

copyright © 2002-2026 Cornell University Library | Privacy | Web Accessibility Assistance