It’s Always Cloudy in Novaya Zemlya

Jeffrey Lewis 0:01

Scott, work your magic. You are listening to the Arms Control Wonk Podcast, a podcast on arms control, disarmament, and nonproliferation. I'm Dr. Jeffrey Lewis, the founding publisher of the Arms Control Wonk blog and a professor at the Middlebury Institute of International Studies at Monterey.

Aaron Stein 0:23

And I am Aaron Stein, Director of Research at the Foreign Policy Research Institute.

Jeffrey 0:29

How you doing, Aaron?

Aaron 0:31

I'm okay. How are you?

Jeffrey 0:32

Yeah, not bad.

Aaron 0:33

Got to check out our new dorm that the Institute is building.

Jeffrey 0:37

Would you have liked to have had a dorm when you were a student here?

Aaron 0:41

No, actually, I wouldn't. My dorm days were over. At my university, we had to spend two years in the dorms, and that was enough for me.

Jeffrey 0:50

I think I would have said that. But, you know, housing prices here have gotten so crazy and it is so hard to find a short term rental that I think a dorm is going to be a really nice solution for students. Plus, it's next to the finest French bakery in the entire world.

Aaron 1:13

Is that the one up by Trader Joe's?

Jeffrey 1:15

Yeah, Parker Lusso, although their actual bakery bakery is located right next to the dorm. And there's a mountain Mike's in case you want some pizza.

Aaron 1:25

Oh, that sounds OK. All right. So, all right. I remember that the apartments I was looking were not very nice and very expensive. So maybe I'll recalibrate my thinking, taking myself back a decade.

Jeffrey 1:37

I mean, the other advantage of having the dorm is that we can recruit bright young things who have GIS skills and can play around with cool stuff like synthetic aperture radar.

Aaron 1:51

That's a great segue to what we want to talk about today, which is the case study of OSINT or open source intelligence and looking at the story of Russia's cruise missile and all the fun stuff that you guys are doing with synthetic aperture radar or SAR.

Jeffrey 2:06

Yeah, man, this was pretty wild.

Aaron 2:10

So I don't know the whole story. This is one of those pods where I'm kind of going to be surprised on air here as you walk us through this case study, because there were reports that they were going to test fire Skyfall again.

Jeffrey 2:22

Right. And we're the source of those reports. Although, you know, the Russians, you know, put down a lot of breadcrumbs themselves. So it isn't it isn't a pure act of open source sleuthery. But Russia has this nuclear powered cruise missile, which several people have dubbed a flying Chernobyl, which is a really crazy idea. I mean, it uses a standard rocket booster to accelerate the cruise missile up to speed.

But then in theory, you use the heat from the reactor in the same way that you would use the fuel to make combustion in order to, you know, suck in the air, combust it and then propel the cruise missile. So, you know, like if that works. That's a cruise missile with infinite range. The US looked at doing this in the 50s.

The US decided not to do this because at least our design, and we don't exactly know what their design is, but at least our design, you're blowing air through an open nuclear reactor, which seems undesirable from an environmental and human health perspective.

Aaron 3:28

That's right. But there's few humans up where the Russians are testing this.

Jeffrey 3:33

Right. So they tested up at Novaya Zemlya, which is this Arctic archipelago, which has a lot of interesting qualities. It's freaking cold. And as we'll see, it's cloudy.

Aaron 3:46

And it was the great love of our former Arms Control Wong co-host and Pellegrino.

Jeffrey 3:51

Yeah, that's right. So we really got excited about mapping Novaya Zemlya several years ago because it's also where Russia's nuclear test site is. And it was a really... Interesting place to look at because it's so far north and always cloudy. There isn't, or at least historically, had not been a lot of great satellite coverage of it. So we actually worked with Planet and got some of the first high-resolution color images of the site that anybody had seen in the open-source universe.

And we were using a combination of their three-meter doves and then their higher-resolution skysats. And as we started mapping out the nuclear test site, lo and behold, Vladimir Putin announces that Russia is testing this nuclear-powered cruise missile and – It's actually a slightly painful story. We went and figured out we geolocated the test that he showed on video to a particular site in the Novaya Zemlya and then didn't publish it.

And then, you know, some Russian on live journal beat us to it, which was.

Aaron 5:09

Yeah, the blue containers of death.

Jeffrey 5:11

Yeah, that's right.

Aaron 5:12

Not to be confused with the bouncy, what are we calling those things?

Jeffrey 5:15

The bouncy castles of death.

Aaron 5:17

The bouncy castles of death. That's in China. Not to be confused with windmills and the blue containers of death. And I believe, correct me if I'm wrong because I've been about a week, in the CNN story that you guys helped put together, you have some pretty juicy imagery.

Jeffrey 5:33

Yeah, we were the origin of that story. We actually once we got the images, we worked with Zach Cohen at CNN, who, you know, we like work with all the time because he's awesome. But he got I mean, it was very carefully worded his sourcing, but he got a source briefed on the matter who confirmed that the U.S. was monitoring preparations for a test.

Aaron 5:57

Yeah. So, I mean, there's a kind of like whole story.

Jeffrey 6:00

So we, because we had always looked at this site, we actually, we followed the site really closely. And then Russia had this Period where they basically disassembled the test site. And that, by the way, is the same period in which they tried to recover one of the airframes off the coast of Severodvinsk. And there was an explosion and it killed five people from Russia's nuclear complex.

And then we watched them last fall, like put the site back together. So they had tested it a bunch. They had taken the site apart. Clearly something had gone wrong. They were going back to the drawing board. They had this terrible accident. And then in the fall, we saw them put things back together. So this summer, we were like ready, right?

So we're like monitoring this site all summer. And we noticed that ships start going up there again. And we can see in optical images from satellites, from planet satellites, we can see that the support areas are filling up, right, with shipping containers and things. And so, like, people are now on site and, you know, they're bringing them food and equipment.

And we can see that they're kind of gearing up for a test sometime this year. And then that cloudy thing comes in. Yeah. In August, how many days a month do you think Novaya Zemlya historically has of cloud cover?

Aaron 7:25

I have no idea, but I just remember, I mean, I think we talked about mapping the via assembly on the podcast. There really isn't any days without cloud cover.

Jeffrey 7:34

Yeah, the historical average is 25 days a month.

Aaron 7:39

Where it's covered in clouds.

Jeffrey 7:40

That's right. And if you don't get a picture on one of those days, so we got one picture on, I think, June 15th that was not cloud covered. And then every single high-resolution image we got afterwards was cloud-covered. I mean, it was some crazy number. Planet were total champs about it. Like, we just kept at it and kept at it and kept at it.

But it's just freaking cloudy there, and we just could not get a picture. Because we wanted to see if they were going to, like, put the launch site back together and do anything. So... Then things like heated up and there were a bunch of little things we saw. But the big one was Russia issued notices to mariners and to airpersons saying that there were going to be dangerous activities between August 15th.

And I think it was originally August 20th and then extended to like the 25th. And so like then we were like, OK, they're going to do this. Right. Like that. Like we got we got to be ready. But it's like cloudy, cloudy, cloudy. And so I reached out to Capella Space, which takes radar pictures and, you know, radar can just see through clouds.

And what'd you see? Yeah. Boy, were we surprised. We knew we would be able to see through clouds, and so we thought we would get something. But we saw three things. First, we saw a bunch of things around the launch pad, which I'm guessing are shipping containers and vehicles. The resolution wasn't all that great. Actually, the resolution was fantastic, but it's just hard to make out a car from a radar return.

There's an object sitting on the launch pad, which could have been a launcher. But then what was really cool was they used to have this environmental shelter on rails that would like sit over the launcher and then they would retract it when they would do the launch.

Aaron 9:48

Yeah.

Jeffrey 9:48

And when they packed everything up, they took down the shelter. And then when they put everything back together in the fall, they still didn't build the shelter again. It was like no shelter. And this summer, when we saw all the stuff flooding back in, there was still no shelter. So we were like, well, okay, they're probably not going to do a test without putting up the shelter, because you've got to protect the missile and the people working on it.

And we could actually see the skeleton... Of the shelter in the radar image. It was really freaking cool. I have this friend, Todd Goodman, who's an artist.

Aaron 10:23

I think he's made an appearance on the podcast before, at least in third-person retelling.

Jeffrey 10:28

I know. Todd's great. And Todd does these really classic Southern California surfer scenes. Yeah. So, like, lots of pictures of Santa Monica and the pier and Venice Beach. And he does this whole series, which, you know, I tease him about being sort of cheesy. It's superheroes surfing.

Aaron 10:49

Yeah.

Jeffrey 10:50

And he's like, no, man, because if you've ever been on a board, that's how you feel. You feel like a superhero. Like, that is how I feel when I'm looking at radar images. Like, I'm staring through clouds. I'm staring through thin roofs. I don't know if Todd Goodman's going to paint me at my computer. On a surfboard.

With a Batman mask. There you go.

Aaron 11:12

Actually, we should commission that. We'll auction it off to our top Patreon supporters. Here's a picture of Jeffrey in the Batman mask surfing, looking at SAR images.

Jeffrey 11:25

Yeah. And I have to say, you know, interpreting SAR images is really pretty tough because it's a radar return. And so it isn't really a, I mean, it's a picture, but there's a lot of noise. And so you really understand what you're looking at in a radar image. You have to do a lot of processing. And so we use Envy, which is this program made by this company.

L3Harris has a has a plugin called SARScape. And so we've really been working to use SARScape to enhance our ability to analyze SAR images, which, by the way, has been a huge effort. Like we've had so many research assistants involved. Work on this. So we, we started with this really talented woman named Hannah Rigdon who helped us set up our you know, ability to analyze radar data.

And then she went off this summer and did something else. And so we've hired Steven Della Fuente and this other guy Ben Mueller. And hopefully we'll bring Hannah back in the fall if she wants. And like, this has been like a real effort to learn how to process these images and, and extract the data from them.

Aaron 12:38

And so do we know if they actually tested it or not? I'll admit I've been on vacation for the past three days. So I haven't been intently following the news. But I don't think that we have any indications that it's actually been flown in the time of – because there was a NOTAM, if I'm remembering that correctly.

Yeah, they were a pair.

Jeffrey 12:56

They were both mariners and airpersons.

Aaron 12:59

Right. So, yeah. So – In the time period where those were both issued, I don't – I haven't seen any stories that they've actually flown.

Jeffrey 13:10

Yeah. I mean the thing is that the US government traditionally hasn't been very leaky about this issue and the Russians themselves haven't announced the tests.

Aaron 13:22

I would say because the last time this ended in an explosion, I think five people died.

Jeffrey 13:27

Yeah. Well, and, and from the, when Putin announced it, there was then a series of leaks and it indicated at the time that all of the tests had failed. And we saw a lot of open source evidence that corroborated that, like them taking down the test site, which suggested they were taking a break to go rework it.

But then also the recovery efforts to get the cruise missile airframes and reactor units up off the seafloor, which, you know, one of which led to that, that you know, terrible tragedy. And so we don't really know. I mean, what I would say is that we looked for collateral. So, you know, we try to watch U.S. aircraft and see if they go up.

But the problem is you just – they do it a lot. And it's not clear to me they would go up. They would probably go up after a test, but it's not clear to me how long they would go up after a test because, you know, you want any – You know, you want stuff to drift in the right space, and we just don't fully understand that.

But there is an aircraft that Rosatom uses, and we don't quite know how that aircraft relates to these tests, but we have seen it move and participate in the tests. And that thing disappeared from its airport outside of Moscow for a couple of days during the NOTAM and now is back again. The problem is all the airfields it would have gone to are cloudy.

Right? It's pretty far north. It's really far north. And so we think they probably tried just because they issued the NOTAM, so why wouldn't they – We could see all the stuff out on the pad getting ready. So why wouldn't they? And then the airplane disappeared. So I, I think they probably tried, but whether it worked or not, I don't know.

And you know, that's just, it's just the nature of the open source beast sometime, which is we can do really, really cool things. But you know, like, Unless we have a telemetry aircraft to be able to, you know, track it. It's still going to be a bit of a mystery.

Aaron 15:45

To track that airplane, are you using like open source transponder tracking or just seeing it on the pad?

Jeffrey 15:50

Yeah, we just look at the pad because they can turn the transponder off or, yeah.

Aaron 15:56

I was heading off a potential caveat that somebody would point out on Twitter, but yeah.

Jeffrey 16:01

Oh, yeah. That's right. Yeah. You mean an Afghanistan expert taking a break to become an open source expert?

Aaron 16:08

Yeah. One of those. Yeah. Windmill expert. No, no.

Jeffrey 16:11

We did that totally with optical imagery. We talked about maybe trying to get some radar imagery to see if there were more planes at certain airfields. But the issue is... It's a standard make Russian aircraft. And so it's really hard to tell other than it has this like really distinctive livery that you can see even in planets three meter data.

Got it. So we really wanted a color optical image. And, you know, like we say, it left and it came back. And so I know it was gone for a couple of days. I mean, maybe they were just, you know, maybe they were just going for some Thai takeout.

Aaron 16:47

Up there, Thai? Yeah.

Jeffrey 16:50

Well, from Moscow. All we know is it left Moscow. We don't know what direction it went in.

Aaron 16:53

It could have gone anywhere. Yeah, yeah, that's true. I mean, just listening to all this, SAR is obviously a valuable tool. And now that I think more and more people have access to it, or at least that you have access to it, I think it bears a fuller conversation about what this will mean to the field.

Jeffrey 17:13

Yeah, I mean, just as a general observation... We are really trying to shift the focus of our research from just relying on high resolution images that we look at with our eyes to emphasizing non-optical imagery and computer processing of that imagery. So, you know, that's everything from multispectral imagery like Landsat. I mean, people hate Landsat because the resolution is so low, but...

It's multispectral and that gives you a different kind of resolution all the way to SAR. Right. And so radar images are, are kind of a perfect example of a capability that's very, very useful, but it doesn't lend itself to self-interpretation, you know? So like one cool thing we just did, which we'll probably be releasing a little bit later is We used Sentinel 5-meter radar imagery of one of the Chinese missile fields, the one Matt Korda found near Hami.

And we just did, I think, an every 12-day time series. And we did coherence change detection on each pair of changes in order to watch the entire field, like all the roads and construction, which pops out really clearly. Yeah. And so you get a beautiful time series showing this silo feel taking shape over the course of a few months.

And so it's really good for stuff like that because if you looked at those images optically, it is so hard to make out the roads. You know, and I know a company like Planet is spending a lot of resources and making a lot of – I think they're making a lot of money. I have no idea. But they've made a really cool building and road – Detection data set.

They can tell you every new building that's been built in an area over a particular period of time. And they've been able to do that with optical images, which I think is nearly miraculous. I think it is easier to do with radar. It's certainly easier for us to do with radar than it is for us to do with optical imagery.

So it's just entering this era where we're Processing these images to extract the data and much more than we are, you know, just like it's, it's not the old days of like, oh wow, there's a, there's a satellite image attached to this email, I'm gonna open it up and stare at it, right? It's, it's like a more computationally intensive endeavor. I mean, and that

Aaron 20:06

That gives rise to the work that you guys are doing, in particular, training people how to do this. I think that there's something perhaps kind of intuitive about staring at an image, even if it does take processing and skills to become quite good at it. This is an entirely different ballgame.

Jeffrey 20:24

Yeah, this is really the balance that I'm trying to strike in my open source class, which is how much of it do I want to be actual skills building, right? Like, I will teach you how to process an image, and how much of it do I want to be the kind of technical education I got in my PhD program, which I really liked, was I don't expect you to be able to process an image, but I do expect you to be able to understand how this information is collected, how this information is processed, and how it is interpreted.

Because, you know, someday you might be on the receiving end of a briefing. In which someone is showing you a particular piece of technical information. And even if you can't do that analysis yourself, you ought to be able to read it and understand it. And I mean, we weren't doing SAR back then, but it was like there was an expectation that we would be savvy consumers of technical information.

And so I, you know, I try to split the difference in the class. You know, I try to show people enough of the techniques, which, by the way, like there are so many like I'm not good at them all. You know, like like we just learned how to use Starscape properly. And it's something like we're all doing together.

But I try to split the difference between teaching, like giving people enough that they're like. Interested and able to then kind of go to that next level of like making this something they want to do. And then this other world of like, well, maybe you don't have any interest in doing this yourself, but like someday somebody is going to show you a radar picture.

And, you know, you probably need to know why it looks like the buildings are laying down.

Aaron 22:01

Yeah. I was going to make a joke about a windmill, but I decided not to do it. But I guess I still do.

Jeffrey 22:07

Oh, God. You know what I what I'm what I'm waiting for is for the Chinese to actually add some windmills to the silo field just so we never hear the end of it.

Aaron 22:16

Well, you can turn the windmill with the plume from your ICBM.

Jeffrey 22:21

I wanted to make a little gif of a missile coming up out of a silo and like a little tiny pinwheel just spinning like crazy. And then Chinese state media going, see, it's a wind farm.

Aaron 22:35

It's a wind farm. That's right. Well, if you want to take this class, you can go to M-double-I-S-dot-E-D-U. No, I'm just kidding. But it's true because – It's changed so much even in 10 years, but that was the way it was presented to us. It was, you would be savvy consumers, not that you would necessarily have to do it.

And I think that that perhaps is something that it's even worth reiterating, is that even just having a basic idea so that you can interpret what you're looking at is a valuable skill, even if you can't do every aspect of the analysis that goes into that final image.

Jeffrey 23:08

Yeah, because there's a lot of bad... Photo interpretation out there.

Aaron 23:13

Yeah, no, it's bad OSINT, I think is the hashtag.

Jeffrey 23:16

It's a thing. And so, you know, even if you never plan to be one of those people just being able to look at something and go, you know, I kind of think this is ****. Like, that's a good spot to be in. Because, hey, look, if there's nothing else about OSINT, the one thing that you can say about it is it is...

Methodologically open, you know, there's no, I think Russia did a test because I say so. And there are classified things I can't tell you. It's, here's the picture, here's the radar returns, I, you know, does that prove that there's going to be a test? No, but you know, using one's faculties and all the available information, that's sure what it looks like. And so you know, like, that's a good thing, I think.

Aaron 24:07

And you know where they do incredible sort of peer review and checking of all of this OSINT work? I do, Aaron. It is in our wonderful Slack channel. And you know how you get access to that Slack channel?

Jeffrey 24:20

You pay between $3 and $5 a month, depending on your educational status. And do you know how you get there?

Aaron 24:28

You go to your bookmark that you've saved so you don't have to remember the address. Patreon.com slash ACWpodcast. For those of you who don't have it bookmarked and for Jeffrey who can't remember, Patreon.com slash ACWpodcast. $3 for a student. $5 if you're not. You can be part of this crew. You can get yourself featured in the Washington Post or The Economist because that's how cool it is.

Jeffrey 24:56

Yeah, I just really quickly, I dared the Slack channel to find a North Korean facility that we had been unable to find for years. And Decker, of all people, one of our members, found it in about 30 minutes. So I think we're going to have to give him a discount.

Aaron 25:14

Give him a discount. Well, if we give him a discount, other people are going to want discounts. I'm just kidding. But no, it's just a fun place to hang out. And so check it out. With that, Jeffrey, as always, it's a pleasure to talk with you. That was great to talk to you, Aaron.