Detecting Missile Launches with Ionospheric Disturbances

Jeffrey Lewis 0:06

I'm Dr. Jeffrey Lewis, a professor at the Middlebury Institute of International Studies and the founding publisher of the Arms Control Walk blog. Aaron Stein isn't with me today. And in his place, I have with me Mike Nute and Tyler Nighswander, who were the two geniuses behind the thing we did the other day, which is creating an open source tool to detect at least some space and missile launches.

Mike Nute 0:33

Hey, guys. Hey. Hey.

Jeffrey 0:36

How are you? So for listeners, I'll just briefly describe the backstory to this idea. And then I'm going to ask Tyler and Mike how they went about doing this. Because for many years, I have been aware that GPS data records ionospheric disturbances, which means that when there are space launches and some missile launches, the ionospheric disturbance that is created by that event Will be recorded in GPS data if only we can go search it.

So knowing that in theory this is possible, but having like zero idea of how actually technically to do this, I whined and moaned on the Slack channel about how this surely must be possible. And you two took up the charge. So I guess I want to ask each of you in turn, maybe Tyler first and then Mike, what What did you think when you saw that this was possible and how did you then sort of start by going about doing this?

Tyler Nighswander 1:47

So, I mean, I think when I first saw it, I was like, I don't know. Like I was a little bit skeptical, but then you linked to several papers about people doing this for rocket launches. I was like, okay, like, I guess someone's done this before. So there's precedent and it's not like you made it up or something.

And I had a bit of experience working with some GPS stuff in the past. So like I was a bit familiar with how everything worked. So I, I kind of just started off reading, reading some of the papers, trying to figure out what they were doing. And then, you know, I, I'm a, I guess a programmer at heart, I suppose.

So just kind of jumping in and playing around, trying to, trying to see what. What sorts of things I could collect and kind of what it looked like, like what the raw signals from the GPS ground stations looked like. And if I could see anything for a few launches and just kind of starting from there.

Jeffrey 2:45

Okay. So for listeners, that's Tyler. Mike, what was your experience like?

Mike 2:50

So, so I had listened to that podcast where you mentioned it, I think a month after it had aired. And at the time I had just graduated, For my PhD in statistics and I hadn't yet landed on a new gig. So I had a little free time and I thought, well, surely I can at least learn something cool from doing this.

And when I jumped into the Slack channel, Tyler had already put together, I would say, like a pretty strong skeleton of what the eventual product became that I was able to kind of dig into and just start learning about the GPS data And the process of like trying to clean it and manage it and then how you sort of eventually get to the ionospheric content values out of that.

And that was just a process of it was like I'm sort of addicted to learning new things. And that was like one new thing after another. So that was just a like going through his code was a cool experience and. At least initially, my primary contribution was just documenting it by trying to figure out what each step did and what Tyler had done at that point.

Jeffrey 4:00

I do want to ask one of you to explain how this is working, because my sense is that there are actually a couple of different mechanisms that we could have picked. And so I don't know which of you wants to jump in, but how would you simply explain for listeners What's happening to the signal between the satellite and the ground station?

And how are we how are we using that to infer, you know, a disturbance that we can then correlate with a space launch or a missile launch?

Tyler 4:35

So I guess we can start with talking a little bit about how it's measured. So at the kind of simplest level, what happens is these GPS ground stations are a little bit fancier than the ones in like your phone or something like that. And instead of just measuring the time of flight from the GPS satellite to the ground station, they measure it on two different frequencies.

And this is super important because basically what happens is kind of something like a prism. If you send two different wavelengths of electromagnetic radiation through something and it has a charge, they'll split kind of, again, like a prism, which means they take different times to arrive from the satellite to the ground station. And so if you can measure the difference between how long it takes one signal, the signals on two different frequencies to hit the ground station, you get some notion of how much charge is in the ionosphere.

And so this is normally referred to as something like the total electron content. And so by figuring out the total electron content, you get some notion of just how to measure the ionosphere. So nothing about rockets or anything in general, but, or specifically, but just generally how, how the ionosphere is, I guess, instrumented. Yeah.

Mike 6:02

So like for perspective, I, I meant to look at some of these numbers for perspective. On like what, what we were working with, but like, you know, the distance from like a point on earth to a satellite is going to be in like the roughly millions of meters. And, you know, the, the two frequencies are, their wavelengths are something like 18 centimeters and like 25, maybe something like that, maybe 22.

I can't remember the, the second one, but so the, the lengths or the times of flight are really measured in, in wavelengths or cycles. And the different, like the scale, so, you know, figure it's probably in the four, the several million cycles order of magnitude. And in the ionosphere, like the difference that between what's measured on one frequency and another due to what's in the ionosphere is like on the order of like 10 cycles or a single digit number of cycles.

So you're talking about a signal that's, ... measured on a very very like it's it's... a small amount of a very very large number but if you control for all the other sources of noise in there you can... like isolate that you know to a Useful amount of precision.

Jeffrey 7:22

Okay. And so when a rocket launch occurs, then you're going to get an ionospheric disturbance. And so the total electron content is going to change. And then that's what we're really looking for, right?

Tyler 7:34

Yeah. So I think, you know, I'm not like a ionospheric scientist or anything, but my Understanding from reading these papers is that I think there's two main mechanisms. So I think the first one is, is what's usually referred to as ionospheric depletion. So if you think of something like, you know, to have those like ionospheric air filter type things where it's, You know, has like some weird thing in a box and, you know, you smoke into it or something.

And the smoke kind of gets stuck to these metal plates that are charged. So the ionosphere depletion is similar. So there are Charged particles in the ionosphere, hence the name. And when you launch a rocket, there's a bunch of exhaust from it, obviously. And that exhaust will kind of stick to these charged particles and deplete their charge or at least make them neutral.

And so that causes, you know, if you're looking at something like a graph of what the total electron content is, you'd see a big dip in it. And this is called an electron depletion event. And this, it's kind of, those are nice because they're kind of bigger signals, but we've had a lot of issues kind of detecting them just because it turns out You know, over the course of a day, the ionosphere values the total electron content goes up and down as things like solar storms come in, or as the UV radiation from the sun comes in or weather, like all these things affect it.

And so if you're looking at like a long timescale, it's pretty difficult to separate these These depletion events from just kind of normal activity, at least if you're doing it automatically. If you're looking at it by hand, it's a little bit easier. So the other effect, I guess I don't know exactly what the name of it is, but it's more like a shock wave.

So something like sometimes, I think the traveling ionospheric disturbances are sometimes used specifically to reference these things. But it looks a lot like if you drop a rock into a pond and you see the ripples moving out. So, you know, as a rocket, Hypersonic rocket, it's the ionosphere at a, at a high speed, you know, it's going to cause these shock waves as there's, you know, just like any acoustic wave of areas of high pressure and low pressure, and if there's high pressure is going to be more electrons there, low pressure less electrons there.

And so the nice thing is that you can apply some filters like a band pass filter that you might think of for audio kind of processing. And that will allow you to kind of select for just a few sets of these waves. And so right now, the main things that we detect are using these filters to try to look for these waves.

Jeffrey 10:35

I mean, I think one of the things that's really cool is if you look at the final graphic, and so I'm trying to paint this picture for listeners, but maybe we have posted on Twitter the actual graphic. The way that you're sampling the ionosphere is on the line of sight between the satellite and the ground station. And so the satellite's moving, which means that the line of sight is changing.

And so you have these Points, which represent a point somewhere in the ionosphere. But they sort of, you know, in our case, because you got the imagery or the data from Japan, it's like a map of Japan that's basically sort of moving consistently, or in one case, you know, several maps. So I wonder, you had a real issue with, at some point, I recall, I mean, the ionosphere goes from like, what, like 80, 90 kilometers up to like 600 kilometers.

And so there are some really difficult considerations, I guess, in sort of modeling where you're looking. Because you're not looking straight down. You don't actually know which portion of the ionosphere you're looking for the disturbance in, right?

Mike 11:49

So this is probably a good time to talk about the data that we're working with and some of the issues that we had to settle out. But one of them, so what you actually have is for every, These are called continuously observing reference stations that are these points on earth that you know, are used like their, their, their place in earth coordinates is, is known to like within, you know, fractions of a meter and like the drift per year is monitored.

And for each of those points, you have like a record of, of the, the signals coming from the satellite, I think like every 30 seconds generally or something like that. And that data, It has for both frequencies, it has a number of wavelengths. So you have to sort of, and so you also know at that time, at the time that the signal was sent, where the position, the satellite's position was in Earth coordinate space.

And so you actually do know to relative precision like what the distance should be. And so then you have a measure of the number of wavelengths along each frequency. But then there's a bunch of things that you have to iron out like the first one is, you have to for these wavelengths for whatever reason. Every time that you acquire the station acquires a lock on the satellite there's like an unknown integer number of wavelengths that it's ambiguous to so you have to sort that out and then.

Yes, so then you can measure the diffraction. But then the diffraction is a matter of how far through the ionosphere it's traveled. So if it's traveling, if it's looking at the satellite at a very oblique angle, it's got a much bigger diffraction just by virtue of the slant. That's not because the ion content along that path is any more or less.

So you have to sort that out. And so the way we go from sort of the slant, the oblique to... Vertical ion content measurement is just by, it's a simple model, it's called the thin shell model. So you basically assume that all of the ionosphere exists at a constant altitude above the center of the Earth. And that, you know, when, so you can just, you just do some trig and sort of back into what the vertical value would be at that particular point where the path pierces that shell.

Then there's another step in the process after that. So that thin shell model, it's not the most sophisticated, but for us it sort of gets the job done. So in that map, what you're seeing is the point where the path from the reference station to the satellite intersects the thin shell, which of course travels over time as the satellite orbits.

Jeffrey 14:39

Hence the beautiful sliding map effect.

Mike 14:44

Exactly.

Tyler 14:45

It's also interesting, you know, right now, I mean, I guess this is a forward looking statement. I think there's this kind of, you know, in some sense, we've reached a nice milestone where we can show these cool videos. But in another sense, there's still a lot more that can be done, I think, with this data and this concept.

And so one of the things is, so right now, as, as Mike was saying, we have this thin shell model where we assume that the ionosphere is basically fixed at something like 300 kilometers. But it turns out for some of the launches, we can actually see it, it looks like the data is maybe a bit more 3d than that.

So, so the effect that we'll see is that as the rocket or missile Gains altitude, it looks like it's the path that we see on the graph is moving something like down closer to Japan. And the reason is that it's actually moving higher and higher. And so when we try to do this calculation where we have this line looking out from Japan over to North Korea, The, the rocket is moving up.

So it's hitting kind of a higher thread, which is pulling it closer to Japan. And so if we were doing something a lot more complicated, you know, it should be possible to do something like say, okay, let me look at this from multiple angles or, or do something a little bit fancier. And you should be able to get a notion of the actual trajectory over time, which I would, I would be, I'd be very happy if we, if we did that.

Jeffrey 16:24

I think I would be super stoked if we could do that. I mean, this sounds, you know, maybe this is a funding pitch for people listening, right? I mean, we could do more. I wanted to ask a little bit, I mean, Tyler, you had some pretty funny stories about just our trying to get the data. Because right now, we're really focused on looking for launches from North Korea.

And we can use data from ground stations in Japan, But obviously, if we wanted to do this in other countries, we'd need data from other places. And I was struck at how many different skills this project took. I mean, there's the understanding the science of what's happening. There's coding in order to be able to process the data.

But just like getting the data and getting it whipped into shape, I think, and I think actually both of you have something to say about that process. Was really intensive in a way that I think people maybe don't appreciate.

Tyler 17:26

Yeah, I mean, I think a lot of people, at least who understand OSINT, would kind of respect some of this. It's a lot of trying to figure out Who runs these ground stations, right? There's ground stations, but if I have me as Tyler has 10 ground stations somewhere, I mean, how are you going to find out about it, right?

And so luckily, a lot of these are run by government organizations, usually for something like internal mapping of your country to figure out where stuff is or where tectonic activity is or things like that. But the first step is going through and trying to find this. So doing something like looking at papers that reference specific countries or trying to find specific government agencies that deal with mapping and then trying to look through their websites or find people and send them emails and be like, hey, you have GPS data that we can access for free.

For some people, this works great. It's super awesome that Japan makes this data available, and it's not so great that some other countries don't. I sent some emails to some folks from China, for example, And they kind of sheepishly responded like, hey, you know, I appreciate that you're trying to do ionospheric research. That's a really cool thing to do.

But unfortunately, our GPS data is classified as a state secret and we can't share it with you. And that's super bummer. There are similar things where there's an agency in Iran which has, I guess, for tectonic activity monitoring or something, and they maintain a network, and they never responded to any of my emails, which I guess I shouldn't be too surprised by.

But yeah, I mean, it's nice because at least you don't need to have... The GPS receivers right where the launch is. You know, luckily we don't have to message Kim Il Sung University or something and ask if they have any receivers we can have access to. It's pretty cool that you can take receivers from Japan and overlook something pretty far away.

And so, you know, there's still some hope here. There are some Other stations and things like Mongolia or India or places where we can try to look over China or similarly some other Middle East countries. But it's interesting just kind of the philosophies that different people take and different countries take and it's Yeah, just as a first step, it's not so fun.

Although I guess on the flip side, it is nice because if there's other people who want to contribute to this project, going out and trying to find these data sources, so things like GPS data in RINEX format, hit me up if you find anything from anywhere very interesting.

Mike 20:25

The GPS satellites in particular are very, very high orbits like their period, the period of their orbit is like 12 hours as opposed to like low earth orbit, which is like 90 minutes, but because of that, like any given reference station. You know the path can intersect the ionosphere like way to the east or west or north or south of its actual location, so I think that's.

I guess it's Japan or is it South Korea reference data that we're using to see over North Korea? It's Japan.

Tyler 20:51

So it's the Geospatial Information Authority of Japan. They operate GeoNet, so we're using that. So again, thank you, GeoNet and Geospatial Information Authority of Japan.

Jeffrey 21:04

And I should say, we debuted this because the tool saw very clearly the North Korean Hwasong-17 launch, which was awesome. But I think we didn't debut when we were still sort of trying to figure out to make sure it worked was we also saw the South Korean space launch about, about what was that? Maybe 18 months ago?

Tyler 21:27

Something like that. Yeah. I mean, it's, there's quite a few things that we've detected. I mean, the downside is at least so far, you know, it's usually space launch or ICBM class launches. So I don't know if this is something that with more processing, maybe looking beyond these kind of acoustic wave type things and looking more at these ionospheric depletions that we talked about before.

Maybe it's possible that we'll be able to find smaller things like IRBMs or something like that. But it's not like it's sniffing out anything in particular from North Korea. Anything in the region that's spouting exhaust or going very quickly into the ionosphere should be, at least in principle, something that we detect.

Jeffrey 22:16

So I'm wondering for each of you, was there a point in this process where you were like, ah, crap, this isn't going to work? Or was there... Was there a point in this process where like, oh, wait, actually this might work? Because it's really been gestating for a long freaking time. I mean, I was so happy that it worked and we were able to announce it.

But boy, this has been, you two have been at this a long time.

Mike 22:45

I mean, for me, you know, with the early part of what Tyler put together, there was enough signal there that it was, Probably going to work that I never really thought, Oh, this isn't going to work. But there were a lot of points where I thought like the to-do list and the things that we needed to like whip into shape and like the, the errors in the signal that we needed to iron out or like the, the data sources that we needed to obtain, like those things, I could project the number of like person hours that would be required for those.

And it was just clear that like, I wasn't going to, I wasn't going to have enough to push over the goal line. So at one point I wrote up like a big set of documentation in the hopes that it would like recruit more people to the project. I don't, I don't know if that was actually ever successful, but that was always to me, like the big stumbling block was just whether there would be enough time to get it all Get it all done.

There were points also where we were looking at certain rocket launches that we couldn't pick up, and that was frustrating. I felt like that was a bummer. I wasn't sure why.

Tyler 24:00

I think it's similar. I mean, I think, you know, when I started out, I was hoping this was going to be kind of like a weekend project. So I kind of pulled it up one weekend, you know, pulled down some GPS data and was like, all right, I'll look for a little disturbances in the signals when I see some rocket launches.

And, you know, that didn't, didn't exactly work. Like I was maybe hallucinating some things to try to convince myself that it was going to work, but You know, as Mike's kind of alluded to, there's a lot of stuff that you need to do from the time when you get the raw GPS signal until you can kind of distill it down to just this ionospheric measurement.

And, you know, that took a long time. And there were... I think now, in retrospect, we understand a bit better why some of the other launches that we've looked at weren't working. So when we started out, we were looking mostly at data from the US because it's very readily available. But part of the main issue is that a lot of the launches in the US, they normally happen from a coastal area right over the water.

And so we have this kind of issue where you'll get some GPS receivers that are kind of well in the mainland and they have to look way out over the ocean. And so, you know, kind of just the way the trajectories are matching up, it doesn't quite work out as well as having, you know, a nice island nation, a little offset looking over.

But it was pretty frustrating for a while because we kept looking and like, I don't see... Anything for this launch? And should I keep looking at this data? Should I just move on to another launch and try to understand what's going on? So I think the first one that we were able to detect was some SpaceX launch out of Vandenberg.

And I just remember the first time we saw it and we got a graphic, kind of like the one that Jeffrey posted on Twitter, but a little bit more A little bit crappier. And I just remember looking at that and being so thrilled that there was some, there was a clear signal where, where the the launch intersected the ionosphere and it was something like a polar launch or something.

So it was a little bit closer to the mainland and yeah, that was really nice.

Jeffrey 26:22

Yeah. It's funny that you say that the grappy looks crappy. The graphic looks crappy because I recall early on in the process, You sort of noting that the graphic looked a bit like a potato and I liked it. Like I, I liked it. I, I thought it was, you know, very evocative. You have some implicit graphic design skills.

Tyler 26:44

Yeah, I mean, I think there's something to be said for, you know, we're presenting some relatively unprocessed data. I mean, there's, you know, you can pull this down and you can more or less replicate it yourself. Although you need the Japanese data, you've got to fill out some forms to get access to that. But it's, you know, we're not...

We're not someone trying to sell anything and be like, oh, we've produced the biggest thing ever. It's just a tool. We're happy. We'd love it if people looked at it. We're not trying to be kind of honest about what it is.

Jeffrey 27:18

Yeah, I think I want to talk a little bit about what we should do next. I mean, I think this is our chance to try to encourage people to volunteer because this is a lot to ask of the two of you for a weekend project that has now stretched into what, like two and a half years?

Tyler 27:35

Yeah, I think it's been off and on for about two and a half years, yeah.

Jeffrey 27:39

Yeah. But I'm curious about what both of you think is... Feasible for us to do in the future, what changes we'd like to make, what kind of help and support we might need. Because I think I've seen people say how amazing and how cool this is. And so it strikes me that it's probably not something that we should just let drop.

But, like, we need some help.

Tyler 28:12

Yeah, I mean, I think the... Probably one of the biggest things that would be like kind of a next step is, so right now the data that we're using from Japan, they post it every hour. And there are actually other signals that are posted basically real time, which is used by, Things like agricultural equipment that need like real time corrections when they're moving in a field.

But anyway, you can get access to that. And instead of having something like an hour delay, you'd have, you know, a second delay or less. And it would be really cool, I think, Probably all of us have had the same dream in the back of our minds, you know, some sort of like Twitter or I guess Mastodon, whatever bot that sits in the background.

And every time there's a rocket launch or an ICBM launch or anything, it's like, oh, launch detected. And then it posts this cool little video.

Jeffrey 29:04

Oh, you're not going to do it on Gab?

Tyler 29:06

Oh, probably not. But I think that's a place where we could get to. I think the data is sufficiently advanced that you could see that. I think... A lot of the core of what we have is kind of there. And I think everyone thinks that would be fun. You know, right now, a lot of, I mean, basically by the nature of how we get the data, we're always going to be a little bit behind something like Yonhap, but it'd be really cool to beat them to detecting a launch.

I think that would be a really cool milestone.

Jeffrey 29:45

You know, I think we could beat them. And Mike, I'll let you jump in in a second, but I will just observe that there have been times where the South Korean government is displeased that a launch or an event has happened and they have not wanted to say anything about it. And the famous case I think about is in 2019 when the North Koreans resumed short-range ballistic missile testing and the South Korean government denied that any test had taken place and we ended up with a satellite photograph of the exhaust Plume.

So I actually think that in a situation where governments may not necessarily want to own up to an event happening, we could beat Yonhap. Mike?

Mike 30:31

What I was going to say is that there's a lot of applications of this outside of the OSINT world that I find make me curious. I mean, NOAA just put out an SBIR call for applications related to Predicting space weather. And I mean, what you have here really is like a technique for using public data to get like a very high resolution ionosphere map, high resolution in both time and space.

And so there's a lot of things that that could be used for that are interesting, but they're probably outside of the scope of this podcast.

Jeffrey 31:06

I don't know. Maybe we need three or four volunteers who decide that they want to make themselves billionaires by creating a space weather company. And then they can just, you know, they can have the billion dollars. I just want to be invited on the yacht and then, you know, maybe get notifications of missile launches every now and again.

How far down the value chain do you think we're going to be able to go with this data? I mean, you said, and it's definitely been our experience, that space launches we definitely see, big ICBM launches we definitely see, but we have not had great luck with like intermediate range ballistic missiles. And So I guess one question is how much of that do you think is a hard limit of what we're going to be able to see just given the way in which the rocket interacts with the ionosphere and how much of it is sensitivity and coding and processing data?

Tyler 32:03

So, and this is all kind of speculation, but I think, you know, something like a, let's say SRBM, which I guess I'm not the best missile expert, but you know, those are, those probably aren't really going to go more than 60 kilometers in altitude.

Jeffrey 32:19

So certainly not the way the North Koreans fire them.

Tyler 32:21

Yeah. So, you know, something like that, I think it's pretty unlikely that we're going to see. But some of the IRBMs certainly reach the point where they should be interacting with the ionosphere. And, you know, I think there's kind of a number of things which we don't know, you know, For example, right now we're using data from Japan and there is of course, some kind of loss in, I guess, fidelity when you're looking that far away, you know, the, the data can kind of bend around small disturbances or something like that.

And so you can imagine if we're using data from South Korea, which we used to have access to, but somehow they, they closed down the website or something. But You know, data from a closer source, even from different angles can help a lot in kind of reducing some of these, these sources of noise or these sources of uncertainty.

So I think something like a, like an IRBM, I really think we should see them. And, you know, maybe there's some additional processing software stuff that we need to handle, but, you know, it, something like that, I think seems doable. Short, short range. I'm, I'm skeptical.

Mike 33:35

Yeah, I felt when I last spent time on the project that there was still some small sources of noise that we could iron out that maybe would make the signal a little bit more precise and maybe that would reveal a little bit more or maybe it would give you some way to measure the scope or the scale of the disturbance that would be useful in measuring the power of it or something to that effect.

But that's a good question. Certain questions. The next question, I think, is certain things that we thought we would be able to detect that we weren't. Can we iron out the signal neatly enough to suddenly be able to see those? But if not, there's plenty of reasons why that just might not be feasible that we're not aware of.

We're still amateurs, really, dealing with this data. We're a lot further along than we were when we started, but we've taken a lot of lumps in learning.

Jeffrey 34:35

So now comes the traditional time at the end of any podcast, often which gets cut, which is where I ask you if there's anything I didn't ask, if there's anything that you felt compelled to say that I should have mentioned.

Mike 34:50

Yeah, not much. This was a lot of fun to work on for anyone who is thinking they might want to get involved and does this sort of thing for a living.

Tyler 35:01

So, you know... Go ahead. Oh, I was going to say... Yeah, I... It's also worth mentioning that there are people, papers have been published about this sort of thing. There are experts in this field who have done this. We're not the first in the sense of other people have detected Rocket launches from the ionosphere, although usually they're doing something like looking at data that's a couple years old for some other reason, and maybe they happened upon it and they're like, oh, that's weird.

So, you know, it's not like we're some crazy geniuses who invented this technique. I think that's an important caveat. But the flip side of that is if there are any of those people listening and they want to help out, we'd love some of your expert advice.

Jeffrey 35:46

So the last thing I'll say is there's a place for this, which is, I mean, this all started on the Arms Control Wonk Slack channel, which makes me just endlessly happy that there is a community of people who are talking about this stuff. And so I will, you know, I don't know. I feel like, I don't know, it's weird for me to, it's weird for me to like pitch the Slack channel when I have the, The two the two stars, like you, weirdly you paid me for this privilege, yeah. You know, for a long time I was a postdoc and I was still paying the grad student rate, so, so, so. For listeners who, who don't know you can join the Arms Control Wonk Slack channel.

For $5 a month. It helps support the podcast, but also gives you access to this incredible community of people like Tyler and Mike. And it's only $3 if you're a grad student. And, you know, Mike, I think you've earned a discount. Anyway, guys, thanks for talking to me. And I'm looking forward to keeping on with this project.

Mike 37:01

Yeah. It's been a pleasure. Thanks a lot.