Photo credit: www.sciencedaily.com
Innovative Monitoring Method for North Atlantic Right Whales
Research from Cornell University has introduced a novel approach for estimating the population numbers of North Atlantic right whales through the integration of underwater microphones and machine learning (ML). This methodology promises to enhance the safety and cost-efficiency of monitoring efforts for this species, which is critically endangered.
Published in Endangered Species Research, the study illustrates the effectiveness of combining audio recordings with machine learning and established aerial survey techniques to monitor right whale populations in Cape Cod Bay, an essential feeding area where these whales congregate every spring.
Current methods for tracking these endangered whales often involve expensive and perilous aerial surveys or sound recordings that provide insights only into their presence or absence.
Lead author Marissa Garcia, affiliated with the Cornell Lab of Ornithology’s K. Lisa Yang Center for Conservation Bioacoustics, stated, “While using sound to monitor whale populations is not a new concept, our research distinguishes itself by enabling the estimation of the number of whales in a specific area, rather than merely confirming their presence.”
The research team deployed a series of marine autonomous recording units (MARUs) throughout Cape Cod Bay to capture sounds made by right whales.
After setting up the MARUs, the researchers trained and validated a deep-learning model capable of detecting right whale vocalizations with an accuracy rate of 86%.
Garcia noted, “By analyzing their unique upcall sounds, we can continuously monitor their presence, regardless of time or weather conditions. This persistent observation is something traditional aerial surveys cannot achieve, as they are limited to daylight hours and favorable weather.”
While there remain some uncertainties regarding the counts, the research team is enthusiastic about the potential of acoustic monitoring to provide valuable estimates that can inform conservation and management strategies.
The capacity to broaden monitoring to larger ocean regions will enable scientists to gain a better understanding of the right whale population across their entire habitat. Garcia emphasized that, although these whales are primarily viewed as a conservation challenge in New England, they inhabit coastlines along the entire East Coast.
“Employing passive acoustics and deep-learning techniques allows for an increased monitoring range, thereby facilitating the tracking of this critically endangered species,” Garcia added.
This research coincides with a pivotal moment for North Atlantic right whales, whose population has dwindled to fewer than 370 individuals, primarily due to threats such as ship collisions, entanglement in fishing gear, and shifting ocean conditions impacting their food availability.
Source
www.sciencedaily.com