Behind the video below the surface
Capturing underwater footage from a tag placed on the back of a whale is just the beginning.
Ask Dave Cade about his research, and he will be quick to pull up a video. “Watch this,” he says. Blue-green water fills the screen for a few seconds, then the gray head a minke whale dips into view, then another. The camera seems to speed up, and the two whales open wide — their feeding pouches inflating like massive hot air balloons.
Captured by a specialized camera tag placed on the back of one of the whales, the footage is remarkable — but it’s only the beginning of Cade’s research.
As a PhD candidate in the Goldbogen lab at Hopkins Marine Station, Cade studies the biomechanics of baleen whales. How fast and efficiently do they move when feeding? How fast can they accelerate? How many giant gulps of food can they consume in one dive? It’s a matter of asking difficult physics questions, made more complicated by the fact they occur in the underwater world.
The video tags, manufactured by CATS (Customized Animal Tracking Solutions), help Cade find answers to some of these questions.
In addition to video footage, each tag includes four basic instruments — a pressure sensor records depth, an accelerometer tracks both the animal’s acceleration and the animal’s orientation based on gravity, a three-axis compass measures its heading, and a gyroscope generates values that correspond to the animal’s pitch or roll.
For each deployment, Cade spends a few hours converting the raw data (volts) into different physical units to generate useful metrics.
“I think a lot of people underappreciate the work that goes into translating raw data into useful information,” says Cade’s PhD advisor and director of the lab, Jeremy Goldbogen. “Some people assume we slap the tag on, recover it, grab the data, and we’re good to go. But there is a lot of work that goes on behind the scenes.”
This work includes syncing all the underwater footage using custom scripts Cade wrote using Matlab. As he scrolls through the video of the minke whales feeding, a graph underneath it illustrates the corresponding data. The green line tracks speed, showing each feeding event, while the blue line shows depth and the pink line captures body position and how much the animal is moving.
But to have the video and movement metrics line up in this way takes many, many hours of processing enormous amounts of data. The accelerometer, for example, can log data 400 times a second.
“You can calculate the speed of the whale based on the jiggle of the tag,” Cade says. “The faster the animal moves, the more the tag vibrates.”
One of the first steps for translating the data involves correcting for the orientation of the tag. “The tag gets slapped on in an unknown orientation,” Cade says. “We don’t care which way the tag is facing. We care which way the animal is facing.”
When Cade finishes converting and correcting the raw data, he starts syncing the videos with the line graphs, which typically take ten times as long to process as the length of the video. A tag that recorded eight hours of footage, for example, takes roughly 80 hours to sync up with Cade’s custom scripts. “That’s why I have three computers running at all times,” he says with a laugh.
“One reason Dave has been really successful with developing these tag tools is he has a background in math, and he’s applied that mathematical thinking to oceanography,” Goldbogen says. “He wrote all the programming from scratch to optimize these tags.”
This kind of skill set is absolutely critical to the future of marine biology, and grasping a better understanding of our oceans, according to Goldbogen.
“When I meet high school students who say they want to be marine biologists, I tell them they should continue to pursue marine biology,” Goldbogen says. “But I also tell them to major in math, statistics, physics, or engineering. Also, learn a programming language and learn it well.”
New technologies, like these CATS tags, can collect an incredible amount of data. “But our pace of discovery has stalled,” Goldbogen argues. “Because we don’t have the analytical tools to extract meaningful information from the data.”
Unless you have the analytical mind — plus the patience and persistence — of a researcher like Dave Cade.
On this trip to Antarctica, Cade has the huge task of processing 18 different tag deployments, from 10 humpback whales and eight minkes. In addition to constantly converting data, Cade also works tirelessly to revise his custom programming.
“We keep getting new versions of tags so we have to update our programming,” he says. “While it’s exciting to get new instruments, it’s not possible to test them before they’re used in the real environment.”
In this way, Cade and Goldbogen are tasked with being both scientists and test drivers. Every minute of data gained from a tag is the result of an immense amount of technical, physical and analytical effort — and a bit of luck.
“You’re putting a $7,000 instrument on a wild animal in the Antarctic — every time we get a tag back and plug it in and start uploading data, it’s a huge relief.”
A relief, and a step forward for the marine science community at large. The Goldbogen lab has collaborated with a dozen different groups from around the world, from Holland to South Africa.
“The goal is to make a stable platform for other scientists doing this kind of research,” Cade says. “But everything is still in development.”
While the technology of the tags and their corresponding programming tools are still evolving, and though the process can feel tedious at times, Cade feels grateful to be on the forefront of his field.
“For hundreds of years, people could only study whales from the surface — where they spend just a fraction of their time,” he says. “Every time we get a tag back, we learn something new.”