Technology on Tyuleny
Getting ready
The film crew, myself included, scrambled to get ready that morning. We were pushing through the chaos of the end of the trip- exhaustion, scattered gear, and the slight hangover that comes from time spent on a ship with old (and new) friends.
Dr. Vladimir Burkanov, our unexpected guest on the journey, was already drifting into the fog in a zodiac. It was laden with 42-gallon barrels filled of water that would supply the research station on Tyuleny Island.
We struggled to catch up with him, but a few minutes later, I was on another zodiac speeding off into the mist towards the island. All around me the heads of dozens, if not hundreds, of northern fur seals snuck out of the water surface. They silently stared at us, curious at the sight of the vessel passing by, before dipping back into the depths.The smell of nature
Renan Ozturk
Soon after the island came into view. The Tyuleny Island Research Station sat behind the slope of a narrow, boulder filled beach. At first glance I remember thinking that the boulders were moving. And sure enough, they were, but they weren't boulders- almost every corner of open space was covered with sprawling masses of the northern fur seals and Stellar sea lions.
It wasn't long after that I noticed the smell- not an unfamiliar one to me, but… one that seemed stronger than I had experienced before. Much stronger.
It was the smell of not tens, or hundreds, or even thousands of northern fur seals and Stellar sea lions, but tens of thousands. 50.000 of them, to be precise.
If, at this point in the story, you are wondering how and why so many large animals can make a living on an island that doesn't even reach 650 meters at its longest point- you are asking the right question.
The changes in the population of northern fur seals and Stellar sea lions here, along with their behavior and mating habits, are the exact reason Vladimir and his team spend months at the Tyuleny Island research station.
No kitchen, no heating - pure research
Chris Burkard
As I walked into the Tyuleny Island Research station, a building without heating or a working kitchen, I was surprised to see Vladimir's students and colleagues inside one of the rooms with computer screens dotted with aerial imagery and complex computing software.
The team inside the room was a small one: besides Vladimir and his colleague and the field station leader Ivan Usatov, efforts on Tyuleny are built upon the work of Anya Kirillova, a researcher from Nizhniy Novgorod; Dasha Gerasimova, a veterinary student from Irkutsk, Egor Vasyukov, a student from Kirov, and Sasha Igitov, volunteer from Kirov.
The Challenge: Studying 50.000 marine mammals
Taylor Rees
Studying such a large population is a daunting task. Simply navigating through the cacophony of sounds and smells while weaving through the moving maze of seal and sea lion bodies was a challenge for us as one-day visitors to the island. As an ocean scientist myself, I couldn't imagine the amount of labor that would be needed to get regular counts of the population here and how it changes on a week-to-week and day-to-day basis. On top of that, the team on the island is a small team with limited resources, supported only by occasional supply runs, making it all seem like an impossible task. Each survey would take days of intense work, and in order to get the best quality of data, this process would need to happen ad infinitum during each field season.
But Vladimir and his team have found a way to survey the populations anywhere from four to six times a day. How?
The technology behind the research
Chris Burkard
It all started when Vladimir's colleague Ivan taught himself to use U-Net: a type of convolutional neural network, originally made for medical purposes, which is designed to work with limited numbers of images as a training set.
Using U-net alongside existing drone technology, Vladimir and his team can capture aerial imagery suited to specific research questions. In some cases, for example, the team wants counts of Stellar sea lions and northern fur seals by age/sex (pups, juveniles, mature adults). Data related to other behaviors is also captured, like how many males and females have territory, tracking specific individuals with brands or injuries, or estimating body size.
Taylor Rees
Each of these surveys requires a huge amount of data, and getting the surveys right takes practice. Test flight paths have to be developed in order to find the best altitude, speed, time of day, and image overlap that maximizes the image quality of the drones. Anywhere from one thousand to three thousand images are collected per survey - again, with 4-6 surveys a day. Once the images are all normalized to the same scale and stitched together -often called an orthophoto plan, in technical terms- an application in the statistical programming language R created by Ivan Usatov automatically processes the images and collects the relevant information.
With that, a survey that might take days of labor can now be processed from start to finish in just six hours after the images are collected, all with an error rate in the range of 4-8%.
So what are the implications of this new integration of drone technology into Vladimir's work?
While we chose to integrate technology can be complicated, it became clear that the use of drones and modern computing techniques on Tyuleny has an outsized impact in their capacity to understand marine mammals in the region. With a bit of luck and a lot of effort, the strategic use of technology by the team here may one day help conserve the natural beauty in this tiny corner of the planet.
