Artificial intelligence and high-tech hydrophones: tracking whales in Gitga’at territory
From the depths of Squally Channel in traditional Gitga’at territory west of Kitimat, B.C., a fin whale launches into a song too low for human ears to detect.
Almost immediately, a laptop at SWAG’s nearby Fin Island research station comes to life. “We listen to the whales and the computer spits out the locations,” explains Ben Hendricks, the project’s acoustic scientist.
That data is helping us understand how whales are using this habitat — information that can bolster conservation efforts and help manage the impacts of increased ship traffic through the area.
It all starts with four underwater microphones that monitor 200 square kilometres of ocean. Spaced about 10 kilometres apart in a box pattern, these hydrophones capture the sounds of the sea around the clock. They transmit the acoustic data to battery-and-solar-powered relay stations on land, which radio the information to Fin Island.
Operating this sophisticated infrastructure in such a remote area is impressive in its own right. The hydrophones and relay stations have to perform 24/7 no matter how choppy the water or wild the weather. But collecting the audio is only step one.
The system then intelligently picks out whale noises from the mishmash of other ocean sounds, whether it’s a curious seal, a passing ship or a bit of seaweed brushing against the equipment. Meanwhile, vocalizations differ from one whale species to the next and from one purpose to the next. Orcas use echolocation clicks for hunting, for example. Humpbacks produce trumpet-like feeding calls. Fin whales emit low-frequency moans to attract mates.
Leveraging the power of AI
To tease all that information apart, SWAG has brought together experts in acoustics, marine biology, computer programming, engineering and more. Together, we’ve created a detection program powered by artificial intelligence. By feeding it data from a huge acoustic library, we’ve taught the system to identify which noises belong to whales, as well as to distinguish between different species and types of vocalizations.
Meanwhile, each hydrophone comes equipped with a GPS clock, which logs exactly when it registers each acoustic signal. Based on the whale’s position, that signal will arrive at each hydrophone at slightly different times, allowing us to triangulate the animal’s whereabouts.
The result is a powerful monitoring tool that alerts users when it detects probable whale vocalizations. It specifies where those sounds are coming from and how precisely they can be pinpointed — within 20 metres, under the right conditions. It also describes how likely it is that a whale created the noise — and if so, what species of whale and what activity that noise represents. And it does it all in near-real time.
“A big chunk of the work goes into making sure that what comes out of the program is something we can rely on and trust,” Ben says.
Creating data-driven solutions
This system opens the door to many exciting possibilities. For starters, it is already producing valuable insights about local whale populations and how and when they use their habitat. This gives Gitag’at First Nation, NGOs and governments hard data to inform marine management and conservation best practices.
Now we’re discussing how to make the data most useful to managers and mariners. SWAG Project co-lead Chris Picard from the Gita’at Oceans and Lands Department explains that the data are being integrated into Gitga’at Enhanced Maritime Situational Awareness Pilot project with Transport Canada. This project provides Gitga’at Managers and Guardians with real-time information on the location of vessels — and now whales from the SWAG Project.
We’re also exploring collaborations with the Canadian Coast Guard and the WhaleReport Alert System to find ways to deliver real-time information on whales directly to ships transiting the area, so that they can slow down for whales or avoid them.
This will become particularly valuable when the new natural gas plant in nearby Kitimat comes online, increasing the number of large commercial vessels sharing these waters with whales.
But Ben is excited the project is already producing usable data that will help ensure at-risk whales in the area continue their recovery. “We really feel it’s accelerating,” he says. “And it’s accelerating in the right direction.”
We’re interested in collaborating with mariners who operate in this area. To find out more, please contact the SWAG Project.