Walruses are skittish. That’s why this scientist is using drones to count them.

Walruses gathered by the tens of thousands in September 2013 to rest on the shores of the Chukchi Sea near the coastal village of Point Lay, Alaska. (Ryan Kingsbery/USGS)

There are well over 150,000 walruses living in the U.S. waters of the Chukchi Sea.

But how do we actually know that?

It’s not an exact figure, and for the most part, researchers still have to count them, one by one. That means flying over, taking pictures and then spending months tallying up the massive pinnipeds.

New technology, though, has allowed walrus surveys to be conducted with drones, and hopes are that will provide more precise counts.

It’s something USGS research biologist and longtime walrus counter Tony Fischbach has been working on.

Fischbach says drones have proven to be safer, and they’ve helped with a long-running problem of understanding how many walruses just happen to be out of view underwater when researchers fly over.


[Sign up for Alaska Public Media’s daily newsletter to get our top stories delivered to your inbox.]

Tony Fischbach: In 2007, when the Chukchi Sea, the ice completely pulled back from the walrus foraging grounds offshore, they all came ashore in large numbers. And we realized, we can count animals on a shore. First we used people in aircraft. But we realized that it’s really hard to get survey aircraft near walrus in a safe way, because you have to fly so high or so far offshore that you can’t get good images of them. Or it’s just a very rare event when you have a beautiful bluebird day to count walruses.

Casey Grove: Remind us, what’s the risk there? I mean, if you disturb them, what could potentially happen?

TF: So walruses, they’re naturally a little bit skittish. They’re probably a little bit like a horse in terms of this mentality, in terms of responding to stimulus. So when walruses sense that there’s some some danger, they pack together as a herd, and they want to go someplace they know they’re safe, which is the water. So if they’re all resting on land, and there’s something that disturbs them — it could be a gull, it could be a bear, it could be dogs, it could be an aircraft — if they’re disturbed, they may get themselves moving. If they’re disturbed enough, they’ll move rapidly, and they may crush each other on the way to get to safety.

CG: So, enter the drone. How has that changed the work that you’re doing?

TF: Yeah, they’re very handy. They’re small, they’re portable. When the wind lies down and the rain stops, we get out of our tent, walk down the beach, and we fly a survey. The beauty of that is you can fly that survey even when there’s cloud cover that’s far too low for any occupied aircraft to survey. When we did this in 2018, 2019, we were able to complete 13 surveys in each year, without much of an effort. The data that comes back is basically a bunch of images that get stitched together in a geographic way. You can lay it on a version of the Earth, put a grid across that, subsample it and start counting. These haulouts are huge. They extend at times up to an entire mile of beach. At any given haulout that we surveyed, we found animals and numbers of animals in excess of 50,000. At times we found 58,000, almost 60,000, walruses together on the beach. It’s a phenomenal occurrence to have that many animals there together.

CG: It sounds like it’s probably less expensive and safer, probably, if you don’t have people flying.

TF: Safer for people, and much safer for walruses.

CG: Yeah, but what else are you able to learn from that, other than just counting? Like I imagine that for some studies you would want to land and maybe, you know, dart a walrus and check it out. But other than counting them with the drones, is there other information that you can grab that’s not as invasive as that?

TF: So one of the things that we realized we can do with drones is we can look at the walruses, and we were looking at them using a very structured flight, where we have very precise knowledge of the altitude of the drone. We’re using a survey grade drone and a good lens, we can actually measure the walrus. Very specifically, we can basically put a virtual tape measure on that walrus and know how wide it is at the hips, underneath the armpits, what its body length is. And by working first with captive walruses at zoos and aquariums throughout North America, and possibly even in Europe, we’ll build up enough of a sample size that we can get a calibration to take our measurements from photos taken by drone and understand what that means in terms of the walrus’s real measures and its real body condition.

CG: I think it was a year or two ago, I had interviewed a researcher who was developing artificial intelligence to be able to put pictures of sea ice, and they were looking for seals in this case, into this, you know, computer that would would count them and tell them, “This was a picture of a seal versus a rock,” or something like that. So when you’re doing walrus counts, is it possible to do that same sort of thing with walrus?

TF: When we did the counts off of the survey drone images we collected for walruses, we had to manually count them. And that was a very laborious process. Three and a half months, we sat in front of a screen and put dots over the walruses to get the counts down and make it accurate enough to be used for the population estimate. When we did that, we also created a training data set that can be used in developing automated mechanisms to do the counting through artificial intelligence. That has yet to be seen if that can really be done. It’s been really successful with animals like penguins or seals that have a little bit of space between each other. When walruses get together on the beach, they are in incredible densities. They like to be together. They like to be on top of each other. And it varies how they are. There’s density here, and there’s a little bit more or less density there, throughout the haulout. It may be a full mile of haulout. There’s a lot of variability in how dense they are. And not only variability in the density, but there’s also little ones mixed in with big ones. It’s not as simple as working with other organisms that are more regular in size and more regular in their personal space, because they’re just stacked in there. And that’s apparently how walruses like to be.

CG: I have just one last question for you, Tony. What do you think the walrus think about all this?

TF: You know, the happy thing about the drones is I don’t think they think about it, because I’m pretty darn certain they don’t notice it. We’re flying at an altitude where you can’t hear the drone and it’d be pretty hard to see the drone. So I’m really, really happy about that. I want the walruses to not to think about us. They have their own world to be thinking about.

CG: They have their own problems to deal with.

TF: They’ve got their own concerns.

Casey Grove is host of Alaska News Nightly, a general assignment reporter and an editor at Alaska Public Media. Reach him at cgrove@alaskapublic.org. Read more about Casey here

Previous articleDenali’s historic, lingering snow season is affecting the park’s wildlife
Next articleAlaska Supreme Court finds Republican gerrymander in Anchorage districts, orders new map