Monitoring Missile Proliferation with Open Source Tools

Jeffrey Lewis 0:07

Hello, you are listening to the Arms Control Wonk podcast. I'm Jeffrey Lewis, the founding publisher of the Arms Control Wonk weblog and a scholar at the Monterey Institute. Aaron Stein's not with us this week. He's on vacation, sitting by the pool in Thailand. Instead, I've invited two of our research associates, Melissa Hanham and Catherine Dill, to come on the podcast and talk to us about their recent work.

How you doing?

Melissa Hanham 0:31

Hey, Jeffrey. Hi, Jeffrey. I'm doing well. Not in Thailand well, but well.

Jeffrey 0:37

Yeah, poor Melissa was complaining that she'd rather be at the pool in Thailand than sitting in the digital learning commons here at the Monterey Institute. So let's just jump into it. Normally we talk about my work, but often really my work is their work and it's the work we do together. So I was pretty excited about having Melissa and Catherine come talk about the research projects that they're doing and in some cases that we're doing together.

So I wanted to start with Melissa.

Melissa 1:09

Melissa, what have you been working on? I've been working on about 20 missile models for the Nuclear Threat Initiative, and we use students. I teach a class here that teaches students how to use open source tools. Including 3D modeling tools, which are free and available to the general public. And the very best students out of that course generally join us and start making missiles.

Jeffrey 1:36

Yeah, we should really thank the Nuclear Threat Initiative. They've been such a great supporter of the Monterey Institute over the years. So how many graduate students do we have making models at this point?

Melissa 1:44

I think we have about two or three right now because we're winding down our grant for the year, but we're about to publish 10 more missiles to the... Wait, 10 missiles? How many missiles have we modeled? We've done 10 earlier this year. I took a brief sojourn. Into motherhood. And now we're back and we're going to publish 10 more.

Jeffrey 2:03

Wait, wait. So you modeled 10 missiles. You made a baby. And then you modeled 10 more missiles. I made two babies. You made two babies. That's right. I'm so efficient. I made 20 missiles and two babies. What are the 20 missiles that we've done? How many countries does that represent?

Melissa 2:17

So we've tried to spread them around areas that have the most interest. So we have Russian missiles, we have Chinese missiles, we have North Korean missiles, and we have Iranian missiles. And I believe we still have the Iranian and Chinese to publish.

Jeffrey 2:31

Ah, so the East Asia Non-Proliferation Program will be proliferating models of missiles.

Melissa 2:36

Yep. Yep. It's a fun project. We use open source information. And I think you have spearheaded this project, too. I think you have a few images you shared with us of guys on their cell phones driving down a highway taking pictures of, oh, gee, what's this? Oh, are these the Chinese tells? We haven't modeled the tells yet, although I would love to if someone were to be interested in funding that.

But no, these are just the missiles now. And unfortunately, we don't model things that are just in canisters. Right.

Jeffrey 3:08

OK. Well, you know, the inspiration for me for this was I call this the Nick Caluso effect. Nick Caluzzo was this really great MIT intern I had one summer at the New America Foundation. And Nick wrote this beautiful paper about, Modeling the DF-31, taking ground truth images, and just picking apart the assumptions in the intelligence community's assessments.

Not saying they were wrong, but just basically being able to re-engineer them. Okay. And so at the end of it, he was able to say, well, I'm pretty sure that their assessment of the energetic, how energetic the fuel is, is X, right? And so Nick wrote this paper, I showed it to a bunch of friends, and then Nick totally disappeared.

I think he got sucked into the intelligence community apparatus.

Melissa 3:55

Well, it's interesting you said that because I never heard of Nick Caluzzo and I was trying to sound not dumb by racking my brain on who that was. And I feel better knowing that he's gone dark or whatever now.

Jeffrey 4:05

I have no idea what Nick is doing. Nick, if you're out there and listening to the podcast, drop me an email.

Melissa 4:09

Yeah, just put a big X on your mailbox or something and we'll find you.

Jeffrey 4:13

But it's great. It shows that we can get students jobs, right?

Melissa 4:16

Yes. We just never hear from them ever again. I'm also excellent at giving background checks these days.

Jeffrey 4:21

So what did you learn when you modeled the missiles? Anything interesting you found?

Melissa 4:25

Well, so when you're doing these 3D models, you're working from often 2D images and you spend a lot of time staring at them.

Jeffrey 4:34

So this would be like a picture somebody takes of a missile on a highway.

Melissa 4:38

Right. So I'm Joe Blow Public. I don't have access to top secret images anywhere. So everything comes to me from social media. Little birdies sometimes forward me things that they've found in social media. We go on Google Earth. We spend a lot of time there. We have an excellent relationship with Google. And I would like to thank Google Earth Outreach for sharing with us many images that they update the Google Earth with.

And from those images, we can gather a lot of information. And you've been doing this yourself for a long time.

Jeffrey 5:14

Yeah, but I'm old and lazy now. Now we have the young people who are actually willing to sit down and make the little missile models. Are you using SketchUp?

Melissa 5:23

Right. So we use a software called SketchUp, which is owned by Trimble. And I think it's designed primarily for architectural firms, but it offers a lot of tools, even in the free version, that people can use for our purposes. So you can take a 2D image and do essentially a photo match technique where you're setting horizon lines.

And so even if you have no idea how big or small the object is that you're making, if you model it and you know the size of one object in the image, something standardized, then you can scale it up and figure out how large the rest of the object is. So if you remember when we did the North Korean...

Tell case where the truck arrived from China and we were trying to discover whether this was modified by the North Koreans or ready to go straight from China. I do. It was a subject of an earlier podcast. Right. So I won't go into all the details. Excellent podcast. But because we knew the specifications of the truck, the company who had marketed them broadly just published them.

We were able to... It was like the tires, right? Right. Tires, everything. They published all the specifications of the truck. Then we knew how big the missile was. Great.

Jeffrey 6:39

So which North Korean missiles have you done?

Melissa 6:42

I think we've already put those up. So we have the Taepodongs. We have a Nodong. We have a Musudan. And...

Jeffrey 6:53

Did you do a KNO2? Not yet. No. That's a future podcast. I've been really excited lately about geolocating the extended range KNO2 launches. But that's a whole other conversation.

Melissa 7:05

Geolocating is how I got into this field, though. So when Google Earth was still really, really new, I think back in like 2004, maybe, I You know, they had launched, they were tested, I think a Moose, no, not a Moose Adon. They had tested a Taepodong-1.

Jeffrey 7:27

The North Koreans had fired it.

Melissa 7:28

North Koreans had tested a Taepodong-1. And I, you know, curious, wanted to find out where they had launched it from.

Jeffrey 7:35

Wait, you weren't working for CNS at this point?

Melissa 7:38

No, I was working for International Crisis. I was a summer intern at International Crisis Group. So you were working for Gareth Evans?

Jeffrey 7:45

I was, actually, yes. Gareth and I had a fantastic sushi breakfast in Tokyo.

Melissa 7:52

Oh, I had a fantastic Peking duck dinner with him several times. Yeah. No, it's good to eat with Gareth.

Jeffrey 8:00

So you were at the crisis group and you, I wanted to find it.

Melissa 8:04

And so I went online and you know, through various banter online and various blogs we had a guess of where it was launched from, went there, started looking all around Musa Don, found the launch pad and found the missile on the launch pad because lucky us, it just snapped an image right then. And I was sold on Google earth forever.

Yeah. And now you're indoctrinating our students. Yes.

Jeffrey 8:29

So what's the what's the big takeaway? I mean, as much as you and I might like modeling the missiles, you know, maybe maybe it's a good idea to be explicit for listeners about the kinds of things that you can learn when you take a two dimensional picture and then you extract a three dimensional model out of it.

Melissa 8:49

Well, like I said, the most important thing is that you are looking at detail that is beyond what you would look at for a 2D image. And it causes you to ask more and more questions that inform you about how this whole process of building a missile works. And then from that, you can go down various interesting avenues, including, you know, what components make this missile?

Where could they have come from? What other missiles look like this?

Jeffrey 9:18

So you see some hints of shared design heritage, something you're desperately trying to avoid saying.

Melissa 9:23

Well, what I'm hoping going forward is if, you know, Josh Pollack, who has done a lot of work on this, has laid the groundwork. And I would love to start taking out these North Korean missiles and comparing them, the Nodong, and comparing them to the Shahabs.

Jeffrey 9:40

So these are the Iranians? So we've done the Iranian missiles and we're about to release the North Korean missiles?

Melissa 9:46

We've done the North Korean missiles. We're about to release the Iranian missiles. Oh, okay.

Jeffrey 9:48

I've got it backwards. And then the hope is maybe in the new year, if we can find a little funding, then we can do a closer look at the individual missiles and see if there's some shared design heritage.

Melissa 9:58

And even more, I would like to go beyond... The missiles themselves and look at the gantry towers that are in North Korea and in Iran. Yeah, they're big, right? Really, really big. Way bigger than any quote-unquote space launch vehicles we've seen so far.

Jeffrey 10:15

Yeah, I've been noticing that the North Koreans are releasing a lot of statements about the importance of space to the development of their economy. So I get the feeling we're going to see another space launch. And maybe it'll be... What would that be? Would that be a Taepodong-3?

Melissa 10:31

Well, I mean, that's what we would name it. I don't know what they're going to name it. But the Kim Jong-un special, I don't know. The Kim Jong-un-ah. There you go. Yeah, they have a suggestion box. I'll pop that one in. But, you know, the people have been calling David on X, I guess. And we've seen some sort of...

Small, actually handcrafted models floating around in North Korea. People have snapped a few pictures of those.

Jeffrey 11:01

Oh, right. It's at a museum, right? And it's a really, really big unha suggesting that that's what we're going to get. Yeah, if you go and look at the... Satellite images of Sohei. I mean, it's not just the new Gantry Tower, which is huge, but there's now that big, fancy building that looks like something out of a Walt Disney 50s space adventure.

Melissa 11:25

Are we allowed to talk about this now?

Jeffrey 11:26

Yeah, I think 38 North has published it, haven't they?

Melissa 11:30

Yes. But maybe not the definitive. You've been off-raising the little guys. Proliferating.

Jeffrey 11:36

Yeah, I am proliferating. Sorry, Catherine, you were going to say something.

Catherine Dill 11:41

Oh, I was going to make a joke that a former colleague of ours made about it, that when you had twins, you said, oh, you know, she had to... She had to... Merv her twins.

Melissa 11:54

Oh, thank you.

Jeffrey 11:56

I'm going to use that. Okay. So, yeah, so I don't know. So we're looking at the Gantry Tower. You think maybe we might be headed for a Marshal Edger North Korean missile in the near future?

Melissa 12:11

Well, I mean, I think once you get that big, it's a space launch vehicle. It is indeed a space launch vehicle.

Jeffrey 12:17

Oh, you're right. I called it a missile. Yes. I mean, I blame the North Koreans. Anytime they do anything, I default to missile. But you're right. At that size, it should be a space launch vehicle. Right.

Melissa 12:27

But no, it's still interesting nonetheless. Maybe the North Koreans will put a person in space.

Jeffrey 12:33

That's got to be like the worst job in the world, right? North Korean astronaut.

Melissa 12:36

Oh, my God. Do you even get food? Well, and who knows? Maybe NASA is going to have to hitch a ride on North Korean Express.

Jeffrey 12:45

Oh, yeah. Well, that's a whole other a whole other set of conversations. So that's really cool. So just to close out that conversation about the missile models, we talked a little bit about what's next, but I don't know that I asked you explicitly. So how would the models change in the next year if we can do that work?

Melissa 13:06

Well, we're going to be adding. So we've been working with a company called Sketchfab that's very generously given us pro accounts and they host the 3D models so that a user, if you go to the NTI webpage right now and go to North Korea's missile page, you can open up. It's nti.org. It's nti.org.

Jeffrey 13:24

And this is the Nuclear Threat Initiative, one of our longest and best supporters. We love them.

Melissa 13:29

Yes. And so if you open up their page on North Korean missiles, you'll be directed to a series of 3D models and you can actually take your mouse and use your cursor and twirl them around and play with them. But Sketchfab, this company that hosts the missiles, is going to let us annotate. They've just announced a few weeks ago that we're now able to annotate.

So we're going to start adding annotations to these missiles, which will make them more interesting to the general public. So you could write like cable duct and turbo pump. Came directly from Tehran. Whatever you want. Yeah. And we're also going to be doing some cutouts so that people can look inside the missiles. And to the extent we know what's inside them, we'll start modeling that too.

Catherine 14:15

I've actually been working on one of those cutouts for a tutorial also for the Nuclear Threat Initiative on chemical weapons. But we modeled a binary chemical warhead. It goes on like a...

Melissa 14:30

That's not export controlled.

Catherine 14:34

This one went on an Honest John rocket, a U.S. rocket back in the 80s. But we annotated it so you could see the different things that comprise the inside of that binary warhead.

Jeffrey 14:52

Well, Catherine, I wanted to move over to you and start talking about our little China open source project.

Catherine 14:59

We have lots of little China open source projects. We do, we do.

Jeffrey 15:02

It's hard for listeners. We have to pick among the various ones that we've started working on in the past month or so. But I wanted to ask you about the Korla Missile Test Complex. Sure. And see if you could explain a little bit what we've been up to. And I'll probably find it irresistible to jump in and interrupt you a lot.

Catherine 15:29

The genesis of this was actually with you. Back in the summer, China conducted a missile defense test. Some people wondered if it was an anti-satellite test or if it was truly a missile defense test and also wondered where that test occurred.

Jeffrey 15:47

Yeah, because you can't look at those things. For listeners, Catherine won't go anywhere near the WikiLeaks stuff. But in one of the WikiLeaks cables, there's a reference to the Corla Missile Test Complex, which is not a missile test complex I had ever heard of before. Sorry, go ahead.

Catherine 16:03

Right, so this test happens, and Jeffrey says, I wonder if that was out of Corla. I wonder if I can find Corla. And he found a facility that... Could possibly be the Korlov Missile Test Facility.

Jeffrey 16:16

Yeah, it had been floating around on the internet. These guys had potentially ID'd it, but it was just a guess.

Catherine 16:22

Right.

Melissa 16:22

So, is that Wikimapia had it?

Jeffrey 16:27

No, no, no, no. It was, there's this Chinese language blogger who goes by, I guess, KKTT or KKT, who has a, yeah, yeah, I can never remember. This is the problem when you're anonymous, I can never remember the handles, right? Like if his, if his name was just, you know, Tim or Doug or, you know, Xiaoming, like I'm fine.

I could deal with it. So he, he had this site but there was some debate about whether that was really the site. And then we got this email from Ian Johnston. I think we can mention Ian, right? Sure. We're very grateful to Ian. Ian finds this alumni video. Do you guys... Melissa, do you know this?

Melissa 17:12

No, I haven't heard this. She was on maternity leave. She was on maternity leave.

Jeffrey 17:15

Oh, Catherine, tell her about the alumni videos. The Chinese military, they make alumni videos.

Catherine 17:19

It was an alumni video of a unit in the general armaments department. And they were reminiscing about their time at this one facility. And the facility looks very... Familiar to us. And from the images in it, we're able to geolocate it to this one suspected missile test facility in Shkorla.

Jeffrey 17:42

Yeah, these crazy guys, they have... These really sappy alumni videos with music, you know, like really sad Chinese songs.

Melissa 17:52

Like KTV sing-alongs. Yeah, exactly.

Jeffrey 17:54

Yeah, listeners should know both Catherine and Melissa speak much better Chinese than I do, which is why we hire them, so that I don't look like such a fool all the time. Yeah, so they have these really sappy alumni videos, and they post selfies. But they always post the selfies of themselves at the base, And they show them going in and out of the front gate of the base.

And so the... The images are good enough not only to identify the main base in Korla, but also the secondary facility constructed in 2010 for the ASAT slash missile defense testing. Because they're like selfies of them in front of the buildings next to the freshly planted trees and everything.

Melissa 18:40

Fantastic. Someone could have forwarded that to me. I would have put down the babies for that.

Jeffrey 18:47

So, I don't know. Catherine, what did you take away from the analysis? Because I guess we ended up getting a grant from, was it from the GOI Foundation? They're not called that anymore, right? I have to get this right. Digital Globe Foundation.

Catherine 19:04

Digital Globe Foundation, they gave us a very generous imagery grant. So what happened is that we looked at this facility and there was no imagery in Google Earth of the facility on the day of the suspected test over the But we have press accounts of other suspected tests over the past few years. And one of them was in January of 2013.

And as it turns out, there is an image in... Well, it's not in Google Earth, but we went through all of the Digital Globe catalogs and found that there is a satellite image on... That day of an area with that facility.

Jeffrey 19:41

So it's a day of a previous test.

Catherine 19:43

There's a nice clean picture. We can look at the facility and see what kind of preparations have gone into the test just to also make sure that this is in fact the facility that is involved in this testing. So the Digital Globe Foundation gave us a very generous imagery grant and we were able to look at this image and there's Lots of movement and activity at the facility?

Jeffrey 20:08

Vehicles, instrumentation, and so the place... You can't actually see the... Unfortunately, you can't see the transporter erector launcher on the pad, but you can see... And we should specify that this facility is where the hit-to-kill vehicle was launched from.

Catherine 20:23

The test missile was launched from... Do you think it's Jiuchuan or Taiyang? It was Jiuchuan. It was Jiuchuan. And we actually had... David Wright do an intercept analysis for us to make sure that the hit-to-kill vehicle could actually intercept the missile launched from...

Jeffrey 20:45

Right. So we've got the launch site of the target missile. We've got the launch site of the missile defense, which is Korla. And so David... Demonstrated that, yes, given the reported type of target vehicle and given what we know, just do a little calculus. Yeah, well, literally it's not a little calculus, it was calculus. It was and and it matches, right? And and the nice thing is, so we've got the picture on one of the days of an asserted test, and it all the instrumentation is out, so you can tell that the base, because it's not the main base, right? The the soldiers live in town in Corolla, and then they like deploy to this place out in the desert.

But it's all instrumented. There are vehicles. You can tell there's a ton of activity. So clearly they were doing something that day. It's totally frustrating, though, right? Because of the radar.

Catherine 21:37

Right.

Jeffrey 21:37

That was a bummer, right?

Catherine 21:38

So the picture... We thought some of the instrumentation was... Oh, no.

Jeffrey 21:44

I mean, there's a phased array radar. There's a phased array radar in Coraline.

Catherine 21:48

Sean O'Connor had identified it a few years ago.

Jeffrey 21:51

It turns out it swivels.

Catherine 21:53

Which we noticed in the historical imagery in Google Earth.

Jeffrey 21:56

And we wanted to see if it was swiveled to watch the test on the day. The satellite photocopies off right there.

Catherine 22:02

Yeah, it's about two meters under.

Melissa 22:04

So it's not a resolution problem.

Jeffrey 22:05

It's just not there. It does not exist. They just didn't take the picture. Wow.

Catherine 22:08

Well, get them back out there.

Melissa 22:11

I know. Digital globe, go back in time. I forgot all about that. I blocked it out of my memory.

Jeffrey 22:17

You blocked that terrible, painful experience out. But I don't know, Catherine, what you thought about it, but I really thought it was cool that we were able to identify a location of a missile facility that the Chinese don't acknowledge existing.

Catherine 22:31

Yeah.

Jeffrey 22:32

Except for they acknowledge that a unit exists and they post selfies of themselves.

Catherine 22:36

I think what was also nice is we were able to match it with a specific unit under the General Armaments Department, which is involved in some of the more experimental and research and... Development work for space and missile programs.

Melissa 22:54

So do you think this is destabilizing? Like, do you think that the fact that regular public can discover this information is going to make China nervous? Or do you think this is actually going to make things a little more transparent, a little safer?

Jeffrey 23:08

You know, so this is like the number one reaction we always get when we do these presentations. It's, you know, the foreign governments are kind of like, you know, you can see people having heart attacks. It was great when Tamara Patton gave her presentation showing the models of the Pakistani production reactors. Yeah. Well, I can't say who, but a very senior at the time elected official in the United States Looked at it and said, wow, well, it's really wonderful, but how do you tell the Pakistanis?

You know, my response to that seriously, though, is that This is a little aspect of a huge change that is almost certainly threatening to governments, but that that change is so broad and so unavoidable that in a, in a sense we don't matter. So, you know, it's one thing for us to be able to identify the location of the site at Corla But that's not what really freaks the Chinese out about social media.

What freaks them out about social media is Ai Weiwei organizing several thousands, several tens of thousands of people to get worked up about an earthquake. And so there are all of these technological changes happening that probably are pretty disruptive and probably are pretty threatening. And this is just like one tiny instance. So my guess is that... The bad news is, yeah, it's really hard to keep secrets if you're a government.

But the good news from our perspective is they're probably going to be so freaked out about the other 50 things they can't keep secret that us burning one of their little missile bases out in the desert probably doesn't even rate. I don't know.

Melissa 24:55

What do you think? I guess I would say it's probably a I believe in transparency. I'm a little bit... This is because Melissa is Canadian.

Catherine 25:06

It's true. Yes. Starry-eyed Canadian. Perhaps it is a bit destabilizing, but perhaps that's also not a terribly bad thing in some sense, because it does lead to increased transparency.

Jeffrey 25:18

It does seem to me that if we're concerned about disarmament, arms control, and nonproliferation, you know, one of the big questions we always have is verification. And, you know, there's this huge discussion about societal verification spurred in part by, God, we're going to do it again, right? Corey Hinderstein, we had on the previous podcast from the Nuclear Threat Initiative.

She's been a huge advocate of thinking... Trying to think systematically about how we do verification in this new kind of age. And yeah, I think the conclusion that it's pretty easy to reach is it's just going to be harder to lie about stuff and it's going to be harder to hide stuff. And so, you know, I think when this started, I don't know if any of you had interactions where people were just convinced that no one would put sensitive national security information online.

You know, there was just this total level of disbelief, right? Like, well, you know, No soldier would post the location of his unit to his Facebook account. And it's like, yeah, actually, lots of them do. Just to impress his girlfriend. Right. Gosh, you know, I don't. Have I spared you guys of going through the... The contact? Yeah, the Russian Facebook.

Catherine 26:38

No, I was not spared.

Melissa 26:39

Oh, yeah, yeah.

Jeffrey 26:40

I'm sorry about that.

Melissa 26:41

Yeah, yeah. I heard much moaning in the hallway outside my office.

Jeffrey 26:45

Yeah. Young Russian men who are conscripted into the military, in addition to posting lots of sensitive national security information, also post some really embarrassing pictures. Yeah. We have a clean rating on iTunes, so unfortunately I can't even begin to explain the things. Well, I will say, I asked Jerry Davidoff, who's one of our research associates who's now left us, who speaks Russian, and did he grow up Did he grow up abroad or here?

Catherine 27:15

He was born in Russia, but he moved here as a small child.

Jeffrey 27:19

Okay, well, so I asked him what the deal with Russian men taking pictures of each other on the toilet was, and he just kind of shook his head and walked away. But you get a lot of those on V Contact. Lots of pictures of Russian men going to the bathroom. I don't know why.

Melissa 27:33

On behalf of women everywhere, I would just like to say that this does not impress us.

Jeffrey 27:39

Well, this has been a really fantastic conversation. I'm very grateful to both of you for hiking down to the Digital Learning Commons here at the Monterey Institute and letting me talk over you and interrupt your stories, which is generally be as annoying. You're signing our paychecks. Yeah, right. Be as annoying on the podcast as I am to work for.

But no, seriously, I really appreciate you coming down.

Melissa 28:05

Thanks for having us. Our pleasure.

Jeffrey 28:07

Great. Well, and thanks for listening. Have a good day.