How Underwater Drones Are Transforming Fisheries Data Collection in Remote Areas

How Underwater Drones Are Transforming Fisheries Data Collection in Remote Areas

0:00:02.1 John Sheehan: Underwater gliders, the data gathering autonomous vehicles that look a little like torpedoes, have been used increasingly across NOAA, probably most famously when they're launched into the center of hurricanes.

0:00:13.5 Speaker 2: New video tonight from inside that major hurricane, the National Oceanic and Atmospheric Administration, NOAA, sent an uncrewed vehicle into the storm. The vehicle collected data to help improve...

0:00:24.9 JS: But remote gliders present opportunities for all kinds of data collection and oceanographic research including for the surveys that are central to NOAA Fisheries operations.

0:00:35.7 Jennifer Walsh: We maintain a fleet of underwater gliders that we fly both in the California current and in Antarctica. They do allow us to collect a whole bunch of data, really important ecosystem data, in a much more cost effective way.

0:00:48.8 JS: This is Dive In with NOAA Fisheries. I'm John Sheehan, and today we're going to hear about one of the survey use cases, specifically, how gliders are used to survey Antarctic krill.

0:00:58.5 JW: Before we started using gliders for ecosystem studies, I would actually go out on our oceanographic surveys in Antarctica and collect zooplankton for biochemical analyses. But now that gliders can't collect physical samples, I'm mostly focused on the imagery that we collect with gliders.

0:01:15.2 JS: This is Jen Walsh, a research biologist for the Southwest Fisheries Science Center's Ecosystem Science division where she pilots underwater gliders.

0:01:23.2 JW: So over the last two and a half years and several glider deployments, we've collected over a million images of really small zooplankton and my job is to go through a subset of images and identify broad categories of animals. And these identifications will be used to train artificial intelligence models to rapidly identify animals in all of the images that we've collected.

0:01:45.4 JS: And why survey krill?

0:01:48.5 JW: Because krill are the base of the food web in Antarctica. Fish, sea birds, marine mammals, they all depend on krill for survival. And while krill are important for a lot of different animals, they're especially important for the penguins and the fur seals that we study. There is also a growing fishery for Antarctic krill, which are considered valuable for their high levels of Omega-3 fatty acids. So there's competition for krill between predators and the fishery. Uh, and then of course there's climate change which has affected the abundance and distribution of krill around the Antarctic peninsula. So it's really critical that we understand how the environment is changing and how these changes affect krill and the predators that depend on them.

0:02:29.2 JS: So what... What information are the gliders collecting specifically? Like what kind of data streams are you getting?

0:02:35.5 JW: Gliders can have a whole bunch of different sensors integrated into them depending on the questions that we're trying to answer. So for example, in Antarctica this year, we're gonna deploy two gliders and both of them will have sensors for measuring general water column properties like temperature, salinity, oxygen concentration, chlorophyll concentration. Both gliders will also have an echo sounder, which uses sound to estimate the density of Antarctic krill. So this is arguably the most important data we collect in Antarctica because this is how we calculate the biomass of Antarctic krill in important foraging areas for these krill dependent predators that we study. These fur seals and penguins. Unfortunately, gliders don't have the capability to collect physical samples yet, so we do miss out on those biochemical analyses and those food web structure studies. And we also don't get the ancillary data that we used to collect on our krill surveys like observations of at sea distributions of seabirds and marine mammals.

0:03:32.3 JS: And let's talk about what these gliders actually are. Can you describe what, what these things look like?

0:03:38.1 JW: Sure. So our gliders look like giant yellow torpedoes. They are tubes that are about 12 or so inches in diameter. They're approximately 8 feet long and they weigh around 200 pounds each. And they have a tail with a fin or a rudder on the back end. And they have two wings that provide lift so that it can dive and climb in the water column. And most of the sensors on the glider are integrated into it, but our echo sounders are visible sticking out of the bottom of the gliders. And umm, one of the camera systems that we use is actually external and mounted to the nose of the glider. So they might look a little bit different depending on what sensors we're using at that time.

0:04:21.4 JS: Okay. And how do they kind of move through the ocean? 'Cause glider sort of implies, you know, are they like coasting above or are they like going down into the water column?

0:04:32.5 JW: No, they go down into the water column. So they perform these, we call them sawtooth pattern dives, and they can dive up to a thousand meters. They're really, really cool because they are very, very energy efficient. They dive simply by changing their buoyancy. They have an oil bladder up in the front and they have a battery, we call it a pitch battery, kind of up in the front. And that battery can shift back and forth. So for a glider to dive, if you think about oil and vinegar salad dressing and you don't shake it for a while, the oil settles on top of the vinegar because it's less dense. So when we want a glider to dive, basically it retracts that oil into the body of the glider, which makes the nose heavier, so it points down. But the other thing that does is it confines the mass of the glider to a smaller volume, so it actually becomes more dense.

0:05:29.8 JW: The other thing it can do is shift that battery forward just by an inch. So that's making the nose heavy, its density is greater and it just dives, it just goes down. It has an altimeter in its nose that senses the bottom. So we have them turn around 30 meters above the bottom just so we don't risk smacking into the bottom. And then to climb, it just reverses that process. It shifts that battery back to make the nose lighter and then it pushes that oil out into the nose of the glider. So that also helps to make the nose lighter, but that also distributes the mass of the glider over a bigger volume. So the glider is technically less dense and then it just climbs through the water.

0:06:12.2 JS: That's super cool.

0:06:13.8 JW: Yeah.

0:06:14.8 JS: Yeah. It seems almost low tech, but that's actually very sophisticated.

0:06:16.4 JW: It is super sophisticated and our gliders do have a propeller or a, a thruster on the back. We rarely use the thruster. We will only turn on the thruster if the glider is in a sticky situation and we need to get it out of that situation because the propeller takes a lot of battery power. So we don't wanna burn through a whole bunch of battery using the thruster. But we have only needed it in maybe one or two cases. And otherwise, we just rely on this super duper energy efficient way for the glider to dive and climb.

0:06:47.2 JS: Okay. Then are you controlling where it goes or are you just sort of letting it out and go where it may?

0:06:54.1 JW: Oh, no. No, no. We control where it goes. We provide instructions to the gliders before they're deployed. We load up what we call a waypoint plan on them, so it has specific waypoints to go to. And we also load up instructions for how to dive and how to use the various sensors that are on it. So for example, do we wanna sample the water column on both dives and climbs or just on dives or just on climbs? They are autonomous vehicles in the sense that theoretically, if we never wanted to change those instructions over the course of a deployment, the glider could be considered entirely autonomous as long as nothing goes wrong. But in reality, that's, that's not how we do it. We actively pilot them throughout the entire deployment. We have a pilot on duty 24/7.

0:07:42.7 JW: We rotate through three or four pilots per deployment so everyone has a shift every three or four days. And that pilot is responsible for monitoring glider performance while the glider is deployed. So they'll look at things like battery life, speeds and angles of the dives and climbs to make sure those look pretty good. They'll look at some of the science data coming back from the glider, 'cause we do get very limited amounts of real-time data. In Antarctica, our biggest problem is sea ice. That's our, our biggest obstacle or hurdle that we don't wanna run the glider into sea ice or ships can also be a hazard, but we monitor ship traffic closely and we make sure we aren't flying the gliders in shipping lanes.

0:08:20.3 JS: And so these pilots, are you, are you sort of sitting like in front of a console with like all this data coming at you? Like are you, is it, is it like, uh, I'm imagining a flight simulator where you're kind of like plugged into what this thing is seeing and there's like numbers flashing at you. What... What's it like?

0:08:37.8 JW: Uh, I wish it were that cool. I feel like we should all be sitting in a big dark room with lots of screens that should look like a video game or something.

0:08:44.3 JS: It's not like that?

0:08:45.9 JW: No. But in reality, watching someone pilot a glider wouldn't make for great TV. We pilot glider through a web interface. So when a glider surfaces and it connects via iridium to satellites, we see a map of where the glider is and we see data related to the glider's flight performance just scroll down the computer screen. So, uh, what you said about numbers, yes, that's accurate. We're looking at numbers. Those can help us determine if there are any problems that need addressing.

0:09:15.2 JW: That can also help us determine if we need to give the glider any new instructions on where to go or how to dive or how to use its sensors to sample. And if the pilot needs to make any changes, the pilot prepares a file with those instructions that the glider grabs and it reads it in when it's at the surface. And then the glider will send us a little bit of flight data and a little bit of science data so that we can visualize plots of these various parameters just to make sure everything is looking good.

0:09:41.0 JW: We don't get all the data back in real time because it would take a really long time to send that data back over satellites. And it's also generally good not to keep the glider at the surface for that long because that's where it's most vulnerable to being pushed around by currents or it could be hit by something. But once the glider sends us some data, then it just goes on its way. Um, and as for where we are when we're piloting, we can be anywhere with an internet connection and a laptop. So I'm usually at work or at home. But I have piloted from restaurants. Once I had to turn my phone into a hotspot and pilot from the car. I wasn't the driver, I was a passenger. But, um, we can pilot anywhere again with an internet connection and a laptop.

0:10:21.7 JS: And so after you bring the glider home, that's when you can sort of like access the trove of of data that's collected?

0:10:30.2 JW: Exactly. Yep. For the echo sounders, we get some very, very limited data back. But as I mentioned, that's one of the most important data streams that we get from gliders in Antarctica because that's how we estimate krill biomass. And if we don't get a glider back, then we lose the vast majority of that data. And if we do have a camera mounted to a glider, we don't get any imagery back in real time. So we would lose all images.

0:10:53.6 JS: And you mentioned that it's uncommon for problems not to arise and that two of them are sea ice and, and watching out for ships. Do things go wrong frequently that you've gotta kind of jump in and take the reins?

0:11:09.4 JW: You know, every deployment is different. We have had some gliders that haven't given us a single issue for a three month deployment, and then we have some gliders that checked out fine before we deployed them, and for whatever reason, they develop a problem when they're deployed that we then have to manage throughout the deployment. We have had to perform a few emergency recoveries for issues that we just couldn't overcome. We couldn't leave the glider deployed in that state.

0:11:38.7 JS: Let's talk about deployment for a second because I sort of also never thought about kind of the logistics involved in, in deploying something like this. You can't just pull up to the ocean next to the Southwest Fisheries Science center and plop it in the ocean and say like, we'll see you in a few months. It takes a bit more than that.

0:11:58.0 JW: It does. We like to have a certain depth of water that we deploy these in. So it's really helpful if we can be out in water that's a hundred meters or deeper. These are deep diving gliders and they don't perform well in shallow water. They're really susceptible to currents in shallow water. And because they don't fly terribly fast, they have a really hard time fighting current if they can't get the depth to make some horizontal progress. So when we deploy for test deployments, for example here in San Diego, we'll drive up a little north to a city called Oceanside because right out of their harbor, the water drops off pretty fast and we can get into deeper water pretty quickly because we're deploying from just a small inflatable boat. So it's not like we can go really, really far offshore. And then we have a glider deployment procedure that we follow so that we feel confident that the glider will perform as expected when the ship departs or when the inflatable leaves it out there.

0:12:54.0 JW: And then we'll put it in the water but we'll have it tethered to a buoy for its first couple dives just to make sure that if something goes wrong, we have the ability to pull it back up. And then we'll eventually start its mission. But we'll have it diving just shallow at first and surfacing frequently so that we can take a look at all the data, make sure everything is looking good, and then we'll take that buoy off and then we'll slowly increase the depth. And if we feel like the deployment is gonna go well, the glider is responding properly, then the inflatable comes home and the glider stays out for however long the deployment is supposed to be.

0:13:27.6 JS: How long is that generally?

0:13:30.3 JW: In Antarctica, it can be anywhere between 60 to a 100 days. And really our Antarctic deployments are dependent on when ships are available to deploy and recover. Uh, the deployments that we do in the California current, we generally aim to have them last about a month each.

0:13:46.8 JS: Wow. I mean that just, it, it does sound like really it can be pretty intense work. I mean, Antarctica seas are, are a little choppy.

0:13:56.7 JW: They are choppy. There are areas that have really, really strong currents. And as I mentioned, these gliders don't perform well when currents are strong. The gliders that we fly are pretty slow. They go about 20 centimeters a second or about eight inches a second. And so currents that move faster than that, the glider can't really fight very well. So we do our best to keep the gliders in deep water, 'cause again, in shallow water and especially in shallow water and strong currents, it's really, really hard to get control of the glider. And then you risk having it potentially wash up on a beach somewhere if you're really shallow. Antarctica's 9,000 miles away from our lab in La Jolla. So we ship our gliders by sea in a cargo container every year to Punta Arenas Chile, which means that we have to have gliders ready to go by early August every year.

0:14:45.3 JW: 'Cause that's when we'll load them in the container. And it takes about two months on various cargo ships all down the west coasts of North America and South America all the way to Punta Arenas. And then there's the travel. Once the gliders have been in a shipping container for two months, we don't know what state they're in when we get them out of the shipping container. So we go down and we run through a whole bunch of checks to make sure they survive the journey okay. We put them outside and make sure that they can talk to satellites. Um, so the, the travel logistics is something that's even separate from the glider logistics, but really, uh, interwoven into all of that.

0:15:25.1 JS: Are they... Are they sturdy or if they hit something under the water like sea ice, will they get damaged?

0:15:31.9 JW: Not generally. Um, the fact that gliders move really slow is actually really good for us because they don't like catastrophically crash into things. They just kind of bump into things. So we have flown a glider by accident under an iceberg, and we were really lucky that the glider, you know, it attempted to surface. It couldn't 'cause it ran into the bottom of the iceberg. And what it did was it just kind of bumped its way out from under it and eventually surfaced. So we were able to see that in the depth data when we got the glider back, but it wasn't damaged and it was actually some pretty cool data under the iceberg. So while we don't purposely try to fly gliders under icebergs, as long as it's not a huge iceberg, it doesn't cause a ton of damage to the glider to do that.

0:16:20.6 JS: And once you retrieve the gliders and get the data back, what do you do with it? How does it play into management decisions?

0:16:27.7 JW: Yeah. So we are still working on all of our various procedures and pipelines for analyzing these data. When we were going out on ships, we would get maybe a gigabyte of data per survey. With these gliders, we can get over a terabyte of data per deployment, which is a huge amount of data for us to work with. So we've started leveraging cloud-based storage and computing to manage large data sets and also working on various code to automate processing as much as possible. Again, the main data that we collect in Antarctica are those Antarctic krill biomass estimates, and those, uh, eventually are used by the commission for the conservation of Antarctic marine living resources, which is an international commission with 27 members.

0:17:17.2 JW: And the commission uses all the best available science from national Antarctic programs all over the continent to determine how to use the living marine resources in Antarctica. So this is, includes establishing regional catch limits for the krill fishery, um, that ensure that the krill predators that we study in particular, penguins and fur seals have the krill that they need and also that the fishery can catch what it needs. Um, so that's how we contribute to krill fishery management.

0:17:46.1 JS: Are you finding the sensors or the cameras that you can apply to these gliders? Are they getting more sophisticated over time?

0:17:54.8 JW: Yes and yes. So our program has been involved with helping to develop the echo sounders that we use and those cameras that we mount on the front of the gliders to collect images of the shadows of animals. Um, so we really have been at the forefront of helping to develop these more sophisticated sensors, and particularly for collecting imagery using gliders. We have found that more and more companies are modifying existing sensors so that they're easily integrated into gliders.

0:18:21.2 JW: And I really think that in the future, gliders will be able to collect a whole suite of information. For example, one of the things that I'm really interested in is phytoplankton species composition. So I can start to look at how krill are using their environment and what they're feeding on just based on glider data. I anticipate that using these gliders for ecosystem studies, it's only gonna expand as the technology advances.

0:18:46.0 JS: Yeah. It strikes me that these, the gliders seem so useful for especially that baseline data that you mentioned, and you also mentioned how they're kind of more cost efficient than sending teams out on these big ships.

0:18:58.1 JW: Exactly. Um, gliders, you know, they're, they're a big initial investment, but when you consider that chartering a vessel can be tens of thousands of dollars per day, and then you're looking at a 30 day, 60 day, 90 day survey, and then on top of that, you have to pay the scientists to be on the vessel, gliders are incredibly cost effective and they can remain deployed longer than a ship can be out. They can go places where a ship can't go. So the benefits of gliders to kind of expand our temporal and spatial coverage, uh, of the ocean is really, really great.

0:19:34.6 JS: And you can pilot them from restaurants.

0:19:36.8 JW: That's right. You can pilot them from your couch in your pajamas, which I have done.

0:19:42.6 JS: Jen Walsh, thanks so much.

0:19:44.2 JW: Uh, thanks for having me. This has been great. A lot of fun.

0:19:46.3 JS: Jen Walsh is a research biologist for the Southwest Fisheries Science Center's Ecosystem Science Division. You can read her blog about her adventures in piloting underwater gliders or you can also see pictures and read more about that story of getting trapped under an iceberg. It's at our website, fisheries.noaa.gov. I'm John Sheehan and this has been Dive In with NOAA Fisheries.