NGA seeks ideas for using data to predict illegal fishing

Illegal fishing harms the legal fishing industry and endangers the marine ecosystem. That is why the Coast Guard and other agencies spend so much time trying to detect and stop it. Now the National Geospatial Intelligence Agency is launching a prize challenge for data-based ways to get on top of illegal fishing. For details,  the Federal Drive with Tom Temin spoke with research and development scientist Mike Brady.

Interview Transcript: 

Tom Temin And let’s start with, well has the NGA ever had a role to date in the illegal fishing idea or is this something new for the agency?

Mike Brady Yeah. So this is a space where NGA is getting more into or maybe some of your listeners, if you’re not familiar. NGA is in the business of providing geospatial intelligence or GEOINT and for short. GEOINT tells us where something happens and when. NGA does have a role in, especially more recently in providing information on where fishing and illegal fishing is occurring around the world for GEOINT. And we support our combat commands so that they can keep track of what’s going on in their areas of responsibility. And this challenge is really about complementing the work that’s being done at NGA, focused on detecting fishing and illegal fishing. We’re looking to forecasting such activity, which goes hand in hand with detection.

        Join us May 2 at 2 p.m. ET for a discussion with agency and industry leaders on network modernization strategy, 5G, AI and plans for the future at the Naval Network Warfare Command and NASA, sponsored by T-Mobile. | Register now!

Tom Temin And just a detailed question, I think of NGA is mostly data based from satellite imagery. And is the resolution these days sufficient from satellites that you can see what’s going on at the level of a fishing boat?

Mike Brady Yeah, absolutely. So this is more on the detection side, but sure. Yeah, we detect boats.

Tom Temin And tell us about this prize challenge. You’ve got $1 million prize for a data based approaches using open source data to help forecast. What are you driving out here?

Mike Brady At NGA much of what we do is what we call anticipatory analysis. So we want to know what happens before it does. It’s one thing, it’s super important and necessary to be able to detect fishing activity to make a determination. Is it legitimate? Is it not legitimate. But it’s also extremely useful to anticipate where fishing activity is going to occur, so we can be more prepared to address those threats.

Tom Temin So the ID is the combatant command or the Coast Guard or somebody that you could alert could head off that fishing before it happens.

Mike Brady That’s correct.

Tom Temin All right. We need to preserve those herring stocks, that’s for sure. And so how does the contest work? What are you looking for from contestants or people entering this idea contest? What are you looking for?

Mike Brady Yeah. So this is a two phase challenge. We actually just started. We kicked off on Jan. 8. The first phase of the challenge is going to go through March 1. We’re asking participants to forecast what’s called fishing effort. We grid the world up by one square kilometer, and we count the time that fishing vessels spend in those square one kilometer cells. That’s called fishing effort. We’re looking at forecasting future fishing efforts so we can compare it against historical. That’s one way we’re evaluating contestants, but we’re also interested in a written summary of how they’re approaching the problem. And that’s something where we’re looking to learn about innovative data sets, new ideas. That’s one thing with this challenge. It’s a way that the NGA can complement its traditional acquisition process. Things we’re not thinking of. So great question. We’re looking to see what kind of data can be brought to bear on this challenge from participants.

        Read more: Big Data

Tom Temin And the contest notes that you need to use open source data, data available to anyone. You can’t use secret government data, I suppose, to try to find this out without giving away the contest. What are typical of the types of data sets that might be able to be enlisted in this process?

Mike Brady So the challenge, of course, we’re looking to what data sets are in the public domain. Also commercial data sets, obviously unclassified. We anticipate participants making use of what’s called automatic identification systems, data AIS. This is vessel broadcasts its location and inputs some characteristics into that data so we can know where the vessel is and where it’s going, where it’s been and a little bit about that vessel. So we anticipate that being part of this. One thing that comes up is when you’re looking to the future is where are the fish stock? Where are they going? That’s not typically you don’t care so much about that when you’re detecting,  you can just sort of see where the vessels are, where the activity is. But when you’re looking to where they’re going to go, maybe fish stock, locations, habitats may be more relevant in this case. But again, we’re really looking for I don’t want a hint off exactly what we’re looking for. We’re looking for the innovative solutions that come out of industry and academic partners.

Tom Temin We’re speaking with Mike Brady. He’s a research and development scientist at the National Geospatial Intelligence Agency. So at the basic level, if the world is divided into kilometer squares and a vessel moves through three of them diagonally, you probably know what direction it’s going in. So what you want to do is build on that kind of basic idea and to really get predictive about where is it going to turn right or where is it going to turn left, that type of thing.

Mike Brady That’s right. Yeah. So I mentioned the AIS data. You can have a ship that identifies as a fishing vessel, but that doesn’t mean it’s fishing when it’s traveling diagonally through three cells. So it’s really to be able to get at that, we need to assess the behavior. So what is the vessel doing? Is it meandering? Is it sort of having the behavioral characteristics that tells us, oh it’s probably fishing. And then you start counting up the time it spends in those square kilometer cells.

Tom Temin And I guess there’s probably no limit to what you could feed into algorithms to try to determine these things. For example, what if that boat was also docked alongside some type of military vessel or other unknown vessel? They were both in the same dock, you know, at the same time. And one goes one way, one goes to the other, the other is headed toward a war zone and maybe has munitions aboard. But it could be the fishing boat that has the munitions or this kind of thing. I’m just making this up. But there’s a lot of parameters, actually, that could conceivably come into understanding and tension.

Mike Brady Yeah. No. Terrific point. So in addition to forecasting capabilities, we’re interested in putting these forecasts into relevant context systematically so we can serve it up as not just, here’s data about where fishing is likely to occur, but what are some useful context information that can help us interpret what that means. And one way we kind of bake this into the way we’re evaluating the challenge is we want to hear about ideas to summarize information like, where is the activity occurring within relevant fisheries management zones, such as exclusive economic zones, regional fisheries management organization zones. That’s information that helps analysts say, ok, this information is relevant for these particular regulations or laws, and it helps sift through the data and information to determine is this legitimate or illegitimate activity.

Tom Temin This sounds like there could be some real wild card types of people entering. I mean, lots of people know marine activity, and they know about the automatic identification system. Ships sometimes trade transponders to fool people tracking this kind of thing. They all know this goes on, but I would think this is where some wild card thinking could really come to bear.

        Want to stay up to date with the latest federal news and information from all your devices? Download the revamped Federal News Network app

Mike Brady Absolutely. Yeah. So I mean, I’m really interested to see who comes to participate in this challenge. We have a first engagement with potential participants. We’re doing an Ask Me Anything session. It’s going to be my first sense of who’s going to participate, or maybe maybe thinking about participating. As far as eligibility, the challenge is open to US owned companies. It’s open to US citizen students, and it’s open to other US based industry, academic and other organizations. The one key thing is all participants need to have an active registration of sam.gov. But really, yeah, I’m interested. I’d hope we bring new faces. One thing we get out of this challenge is expanding partnerships. We hope that this connects us with individuals that can do things for NGA that aren’t in our existing orbit through the traditional acquisition process.

Tom Temin And this first round will result in a certain number of finalists that will go on to develop their ideas further.

Mike Brady Yeah, absolutely. So the first phase, which as I mentioned, ends March 1. We’ll have an inter-agency panel of expert judge the performance on this forecasts, and they’ll select up to ten finalists who will be awarded each $25,000 a piece. And they’ll be invited to participate in phase two, which will start, as soon as phase one ends on March 1. And then the participants will go through another round of fishing forecasts. They’ll, refine their written concept of operations where they’re explaining the way they’ve approached the problem. Phase two will culminate in a pitch event to the interagency panel. We will then use that information to select first, second and third place. First place gets 500,000, second place 200,000 and third place $50,000.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.