AI tracks whale calls in Mariana Trench for research

In 2014, researchers recorded a mysterious sound in the Mariana Trench, initially thought to be from a ship or military source. This sound, later identified as a call from Bryde’s whales, prompted further investigation into the elusive species. Ann Allen from NOAA recognized the potential to track these whales by analyzing over 180,000 hours of underwater recordings. The challenge of manually sifting through this data led to a collaboration with Google researchers, who developed an AI model capable of identifying whale vocalizations. The AI was trained to recognize patterns in sound spectrograms, allowing it to identify calls from eight different whale species, including Bryde’s whales. This technology functions similarly to music identification apps, enabling researchers to detect the presence of these whales without direct observation. The findings revealed that biotwang calls were most prevalent among a specific population of Bryde’s whales in the western Pacific, suggesting a link to their migratory patterns. The study also indicated that these whales might be following a transition zone chlorophyll front, an area rich in prey due to varying water temperatures. As climate change leads to more extreme weather events, the distribution of this feeding zone may shift, forcing whales to travel further for food. This could have significant implications for the health and survival of Bryde’s whale populations in the future. Overall, the integration of AI in marine research not only enhances our understanding of whale behavior but also highlights the challenges posed by environmental changes, emphasizing the need for ongoing monitoring and conservation efforts.