Embedded - 206: Crushing Amounts of Snow

Episode Date: June 28, 2017

This week, we mix things up a bit. This joint show with the Don't Panic Geocast.  This episode explores what happens when electrical engineering meets geoscience in cold places. We’re joined by guest Dr. Sridhar Anandakrishnan of Penn State to talk about geopebbles, ice, climate, and more! Asimov Robot Series Anthropornis (giant penguins) Ice crystal structure Ice streams GeoPebble Propeller Programming (Book) Fun Paper Friday: The Boring Company

90% of all scientists that have ever been alive are alive today. That's a lot of information. But don't panic. It's not an exact science. Welcome to the Embedded Geocast, the podcast for people who love making science gadgets. I'm Alessia White, here with Christopher White and John Lehman. Wait, you're not Shannon. This is true. I'm also not Shannon. No. So what are we doing here? This is a little bit different form than normal for the Don't Panic Geocast or

Embedded.fm. That's true, but I believe we have a guest on who kind of crosses those lines, being an EE and then doing geoscience things. I feel like maybe we should all grill him. Yeah, it's great to have guests on and have them be interviewed by three people. It makes them comfortable. Absolutely. So in the spirit of true cross-disciplinary work, we're joined this week by Dr. Sridhar Nandakrishnan. Hi, John. How are you? Good. How are you doing, Sridhar? I'm doing well.

Hi, Sridhar. Welcome. Thank you. Could you tell us about yourself as though you were on some strange conference panel that involved geoscience and electrical engineering. I'd be delighted to. So I am a professor in the geosciences department. So that's the hat that I wear most of the time. But I started life as an electrical engineer. I got all my degrees in electrical engineering until I had the chance to go to Antarctica to deploy an instrument that I had built for some folks at the University of Wisconsin. And my life changed radically after that. I said, gosh, as an electrical engineer, they'll pay you to go off to wonderful places like Antarctica and Greenland.

And so I decided to go back to school and get a PhD in geosciences, and in particular in glaciology. And that has been my life since then, trying to understand the vast ice sheets of this continent using my electrical engineering background as best I can. Excellent. This is going to be fun. So on the Embedded podcast, we have this thing called Lightning Round, where we ask you pretty random questions just to get to know you a little bit. And we're hoping you're going to give us short answers. And if we are all behaving correctly, then we won't ask you for great in-depth answers. Sounds good.

Are you ready? I'm ready. Okay. Okay, so I'll start with one. What is your favorite mineral? Oh, that's an easy one. Ice. Ice is a mineral, and it is the best one on the planet. What is your preferred voltage? Huh. I would have to say my preferred voltage recently has been 128 volts. I'm working on an instrument that produces high voltage pings in the ice, and then we listen for those.

And boy, getting that 128 volts down to the bottom of the ice sheet is hard. I have many follow-up questions for that one. I guess I didn't do my job right. No, no, that's great. Favorite robot, real or fictional? It would have to be Isaac Asimov's Whole World, starting with iRobot and then following through all of the robot detective novels.

I devoured those as a young'un. Those were so good. So, Rockhammer, laptop, or laboratory? Definitely laptop, unfortunately. It's a most distressing thing to be in the middle of the Greenland ice sheet and watch the battery voltage on your laptop drop even as you're looking at it because it's cold. Well, yeah, because batteries don't like cold. They do not. I would imagine the keys don't like the cold either, right?

They're not bad. It's the blowing snow onto them that can be an issue. So you have to try and wrap them in your blanket or in your parka or something like that. Okay, next question. Favorite charismatic megafauna? Ah, well, penguins, of course. It's what gets me entree to many a dinner party, so I got to go with penguins. Were penguins ever a megafauna? I'm going with it. They were. Actually, there were giant penguins about a million or a million and a half years ago. They got up to about six feet high.

That doesn't seem so giant. Well, there's my nightmare for tonight. Well, I guess since we're on the topic, have you ever touched a penguin? One is not supposed to, and so I have not. It's not an answer. It's not an answer. So it is against the Antarctic Treaty and U.S. law to touch a penguin, period.

Your lawyer has advised you to answer that one. All right, so what about what is your favorite geophysical method? So we're going to get into that at some length down the road, but it would have to be seismic waves, measuring them, interpreting them and figuring out what's inside the earth from them. All right. How cold is too cold? There is a very specific answer to that. It's minus 35. I spent summer camping on the ice sheet, and minus 30 was okay. I'd crawl out of my sleeping bag okay.

Minus 40, I'd say, no, I think I want to stay in here for a little while longer until I was dragged out by my students. Is it Celsius? Yes. Well, once you get down that far, they kind of get pretty close to each other. At minus 40, they cross. Minus 40 is the magic one, yeah. Yeah. All right.

Shall we declare lightning round done, or do you want to go once more, John? I think we can declare it done, yeah. All right, because I want to hear once more john i think we can declare it done yeah all right because i want to hear about this 128 volt pings could you explain that in more detail sure so uh this is a sonic logger uh what it's an instrument that i've been working on developing with a small company in Colorado called ESS. So, glaciologists go out there and drill holes in the ice sheet. So, they want to pull out the ice and look at the physical ice for evidence about former changes in temperature. But once they're done, they leave a hole in the ice that's about four inches, five inches wide, that goes all the way down three and a half kilometers through the

ice. And what I do is I lower an instrument down into it that has a piezoelectric transmitter, and then a string of five piezoelectric receivers. And the time it takes for the ping from the transmitter to get to the receivers tells you a lot about the structure of the ice. Ice is, as I said earlier, mineral, and it has a crystalline fabric to it that controls how strong the ice is. If the crystalline fabric is nicely aligned and all of the grains are all pointing the same way, ice will flow relatively easily. Whereas if all the grains are going every which way, ice is a little harder to shear. And measuring the time it takes between the transmitter and the receiver, you can turn that into a velocity within the ice, and you can turn that into a crystal orientation within the ice. And so then ice sheet modelers take that and they say, gosh, the ice in this part of the ice column will shear easily, and this part will shear less easily. And then they can use that to infer what the future of that ice sheet might be.

But the 128 volts comes in because the transmitter, it's quite an old technology. You simply hit it with a high voltage pulse, and then you let the piezoelectric crystal vibrate. And so then you have to get this high voltage down a hole, and that's quite hard because the cables are, we only have two cables to work with and all sorts of problems, and the ice sheet doesn't have a good ground, and blah, blah, blah. We've been fighting with it, but I think we're getting there. But, I mean, this is like three miles below.

Your transmitter is far below you, and then you're trying to push all of this voltage down a very, very long hole. Is that right? That's right. That's right. So we have a transmitter that's down at the bottom of the hole, and then we have a string of receivers that are also within the hole, but they're distributed up along the hole. And then we have a cable that then attaches to our winch and then to our electronic instrument up at the top.

And we can produce, we have to produce the high voltage up at the top and then send it down this long lossy cable, there's not a lot of room down in the hole to squeeze a lot of electronics in. So we've chosen to just produce the high voltage up at the top, and then the only thing down in the bottom is a switch that pulses the transmitter. Given that the cable is essentially a giant resistor, a very, very long and giant resistor, how much voltage does the piezo get? It's 109 volts. That's what we're aiming for down at the bottom.

It's still a lot. It's still a lot. Sorry, I'm just easily amused by such things. And this sounds a lot like, well, more like radar in World War II where they were looking at thermoclimbs in the ocean. Because it would bounce off the difference in temperature. And so you would be able to map areas areas but also you would not be able to like if there's a rock that gives you less information about the ice right or how do you deal with noise in this situation yeah so the there's a couple of sources of noise and one is um that the hole

itself isn't perfectly straight and perfectly vertical it kind of corkscrews down through the ice. And we don't know where our transmitter really is in space. We have to assume it's vertical. But in fact, it might be off five degrees to the right. It's not quite five degrees, about two or three degrees to left or right. But two or three degrees over three kilometers, it could be a long ways off. So that's a source of noise in the interpretation. You're assuming that your transmitter is at position X, whereas in fact, it's at position Y. So that introduces a problem. But other than that,

the ice is a wonderfully homogeneous material because it's formed by very, very similar processes year after year after year. It snows on top of the ice. It sits there. It snows again on top of the ice. It sits there. It snows again on top of the ice. It sits there. And that pile simply gets compressed down. And so there's very little rock or foreign material in the ice. There's gas bubbles, which change their size. There might be a little bit of dust or some ash or something like that. But for the most part, it's a very clean material. And that makes our life a little easier. So you're putting a lot of voltage down to this piezoelectric element that's generating a strain wave that propagates through the ice. So how big of a strain wave do you have to generate for it not to be attenuated out by the time it gets to your receivers? And what is that mean velocity for a column of solid ice?

So our transmitters and our receivers are separated by a fairly short distance, uh, about five meters. Uh, and so we average the velocity between the transmitter and the receiver, and then we lower that whole package down. So we're not transmitting from the bottom of the ice to the top of the ice or something like that. We're transmitting over a short section of the ice, and then we're moving that whole packet down in increments of about a meter,

and then measuring again, and then moving it down and measuring again. And the velocity, what's called the P-wave velocity or the compressional wave velocity, is just under 4 kilometers per second, 4,000 meters per second. But it can vary by as much as three or four percent. So anywhere from 3950 to 4100, depending on how the ice crystal is lined up. And we can measure that variation of velocity by looking at how long it takes for the ping to get from the transmitter to the receiver. The distance is fixed all the time, and as the time it takes changes, that means the velocity within the ice is changing. Okay, so that's the technology and how it works, but what does it tell you?

Yeah, so ice is a crystal. As I said before, it's my favorite mineral, and it forms this wonderful hexagonal crystal. And those hexagons, there's a billion of them within the ice sheet. And depending on how one is sitting next to its neighbor, it will either slide over its neighbor smoothly and easily. And the analogy everybody uses is a deck of cards. You can think of these ice crystals as the individual cards within a deck, and if they're lined up sitting one on top of the other, they'll slide past each other easily. But if they are catty-cornered each other, so if they're sitting with the edge of one card pressing against the middle of another one, so these two ice crystals are at an angle to each other,

then they won't slide past each other quite as easily. And so that is the hardness of the ice, whether it's going to deform easily or not in response to a given stress. And that has to go into the models to be able to accurately predict or project what the ice sheet might do as temperature changes and as conditions around the ice sheet change. Right, because as conditions around the ice sheet change, it's not even remotely a simple linear process, right? There's all kinds of feedbacks and dynamics that have to be taken into account. And this is an important piece of that puzzle in terms of the stability of the ice sheet as the whole, right? That's exactly right. The big picture is that the

ice sheet is one part of a cycle. Water evaporates from the ocean, it falls, there's snow on the ice sheet, and then that snow eventually turns into ice, which then flows back to the ocean, and the cycle is complete. Well, the speed at which that ice flows back to the ocean depends very much on its temperature. It depends on things like this fabric, this crystal orientation fabric that I was talking about, the interlocking of these ice crystals. It depends on the roughness of the bed that the ice is flowing over. Is it a slippery bed? Is it a rough bed? Is it a wet bed? Is it a frozen bed? All these things change how the ice would flow back to the ocean. Now you've got this wonderful cycle going, well kick the cycle,

change the temperature, change sea level, kick the cycle. Change the temperature. Change sea level. Change the amount of snowfall that you get. And to be able to project how the ice sheet and sea level respond to those changes, you need to be able to model the ice accurately. You need to know about the crystal orientation fabric. You need to know about what the bed is like, whether it's slippery or smooth, wet or frozen, things like that. So you have another tool that you have developed. I heard about

something called a geo pebble. What does that do? Is it a similar sort of measurement goal? It's not. Actually, that instrument, geoPebble, was developed by some colleagues of mine, Sven Belen, Julio Urbina, and myself here at Penn State. And I alluded to that just briefly a moment ago. The ice flows. It deforms. The ice itself deforms. And so you need to know the crystal orientation fabric to be able to model that deformation. But it also slides. It moves over its bed, and the bed itself is fixed in place. It's rock or sediments or something like that, and the ice is sliding over the bed. Well, the speed at which the ice slides over its bed depends very, very

much on what that bed is made of. Is it hard and bumpy? Is it soft and smooth and slippery? And as I said, is the ice frozen to the bed or is the ice thawed? All these things affect the speed at which the ice flows over its bed. Our knowledge about that interface, the bed that the ice sits on, is very, very limited. We can't see it. We can't just strip off the ice and peer at it. And so we have to infer the properties of that bed by geophysical methods. Geophysics is using things like sound waves or light waves or gravity waves, whatever we can to figure out what's in the places that we can't look at, inside the Earth, down within the bottom of the glacier in this case. And so what we do is, and some of your listeners will enjoy this,

the pyromaniacs among them, we drill a shallow hole in the top of the ice about 30 meters deep, and we put a one-pound charge of high explosive down into it, and then we detonate it. And the sound waves from that detonation travel down through the ice, bounce off of the rock, and then come back up to the surface where they're recorded by these geopebbles. Geopebbles have three-component seismometers in them.

Seismometers are devices that are sensitive to vibration of the Earth. And so when the geopebble vibrates because of this reflected wave, we measure that. And then we can infer something about the properties of the bed by looking at the quality of that reflection. Was it a strong reflection or a weak one? And other things about it. So just going back to ice formation for a minute, is there a difference? You mentioned that it's predominantly snowfall, which gets packed down over the centuries. Is there a difference between ice forming in that manner to, you know, what people traditionally think of as I put water in a cold place and a volume of water crystallizes. There's a huge difference, and I'm glad you asked me about that. So we call the ice that is in a glacier or on an ice sheet,

and an ice sheet, by the way, is simply a mass of ice that covers a large part of a place like Greenland or all of a continent like Antarctica, just to distinguish it from something like Blue Glacier or a glacier that you might have in the Alps. So an ice sheet is just a big chunk of ice, that's all. So the ice there is what we call meteoric ice. It falls from the sky as snow and then gets compressed and then flows. And that we distinguish from sea ice, which is ice that forms every winter on top of the ocean by the formation of freezing of seawater. And then that forms a layer on top of the ocean. So those are the two main types of ice in the natural environment,

meteoric ice and sea ice. And we have to distinguish between them. And so what I get out of my freezer is more akin to sea ice. Exactly. It has all the properties of sea ice. And if you put a little bit of salt in it, you have to get it colder to freeze it, which also happens if you put a little bit of salt into your fresh water

and then put it in your freezer, it won't freeze until it gets a little bit colder than, than the temperature at which freshwater would freeze. And all of that is, has to be modeled by people who work on sea ice. So if you were to look at a piece of this sea ice or ice that you get out of your freezer and look at the orientation of the crystals that you were talking about, would you see pretty much a random orientation or is there some pattern that forms to even that just naturally frozen ice? So the naturally frozen ice tends to have a much more random orientation,

uh, because the, the, the crystals form in, in a relatively rapid manner, whereas the ice that falls on the ice sheet, it is formed under stress. And so as the crystals are being stressed,

they tend to rotate during that process of stressing. So you can think of this, again, going back to my deck of cards analogy, you throw a bunch of cards down on the ground and they might land every which way. up next to each other for some reason. If you step on it, well, that tent would collapse and they would tend to lie flat. I mean, that's not a perfect analogy, as a matter of fact. It's a terrible analogy, but for our purposes, it'll do. Because you're making meteoric ice under stress, these crystals tend to line up with each other,

whereas when you're just doing it without stress, then the crystals go every which way. That's funny. Mentally, I thought it would be the other way. The sea ice version, I would expect, because it's a smoother process, it doesn't involve having things step on you or large crushing amounts of snow on top. It seemed like it would be more crystalline. I guess it's always crystalline. Yeah, I guess the way to think about this for the meteoric ice is that crystals will tend to a low energy situation. And so when you've got two crystals that are at cross purposes to each other, if you will, and you put them under pressure, they'll slowly rotate until they get to the lowest energy configuration, which is where the two of them can slide past each other relatively easily.

So that just happens over time as the ice sheet gets thicker and thicker and thicker. Back to the GeoPebbles. You have a seismometer and a recording system. I would say probably some sort of flash device, like an SD card. And does it communicate automatically? Is it Wi-Fi? Does it store? Is it sat phone? Because, I mean, if you're on a Greenland ice sheet, it's not like you're going to be getting good cell service. Sure.

We actually get terrible cell service, except for Iridium phones, if you might have heard of those. Iridium phones are these satellite phones. And because they're in polar orbits, we actually are always in sight of one of those iridium satellites. And the same with GPS, by the way. GPS satellites converge at the poles, and so we always have very good sight of, and there's no trees on the Antarctic or buildings, and so we always see good signal.

Back to the geopebbles. So to go back just a step, the reason we developed these instruments in the first place was the Antarctic can have some dangers, or I should say obstructions to travel uh for example crevasses which are the most uh most difficult thing to work around and so there's places that we can't go by just driving over the surface and the instruments that we'd been using for a number of years all had cables connecting the seismometer to the recording device. Long cables, sometimes as much as five or six kilometers long. Because you could keep all of the electronics in one place and then just have the sensor out there, or hundreds of sensors in this case, out along this long cable that stretched out over the ice. So that was how work is done

out in Oklahoma or Texas or in other places like that. But where we worked, we would have interruptions on the surface where we couldn't cross over these margins where there were big crevasses. And so we needed some way to put these seismometers out without having these long five kilometer long cables. So that was the impetus for developing these geopebbles. The name is supposed to refer to the fact that these things are small and can be deployed out of an airplane or out of a helicopter or something like that. And we've done that.

So we had to put not just the sensor, the seismometer, but also the data logger and the digitizer and the data logger all into one box. Once you've done that, obviously you need power. So we had to put the battery in there as well. And then we said, gosh, how do we get the data off of these? One way to get the data off is to walk up to it, pick it up, take it home, pull the memory device out, and then download the data. But the difficulty there is that these glaciers can be a little bit unforgiving. It can be hard to go back to the same place again to pick it up because a crevasse might have formed nearby, and so you couldn't land an airplane there anymore.

So then we made the decision to put a Wi-Fi on there that we could interrogate the geopoble from up to a kilometer or two kilometers away. And we did that in Switzerland quite successfully, actually. We put a bunch of these geopebbles up on a glacier, and the mountaineer who put them out, he was actually slung from underneath the helicopter. So he was on this long line underneath the helicopter, and he had five of these geopebbles on his belt. And they would hover over a spot, and he would dump the thing down. And then they'd move, and they'd hover to a different spot, and they'd dump it down.

And he said, I'll put them out as many as you want, but I ain't going back to get them out. So he didn't want to actually—so he just threw them down onto the surface from a little ways off. And then we had, on the other side of the valley, we had a Wi-Fi antenna that was pointed at these. And then we could interrogate them and get the data off. And then we just did this until the glacier swallowed up the devices and they disappeared forever. So having put out a more normal traditional seismic instrument like a broadband seismometer, that's a much different process. You're digging this hole that's a couple feet deep and you're tamping sand and you're getting the instrument precisely level. And then you align it with north very carefully so you know the direction of all three components of the seismometer.

And that's much different than tossing the thing down from hanging underneath a helicopter so how do you handle all of the the uncertainty and the orientation of the instrument the fact that it could be even changing orientation throughout time as ice moves to get back to the normal either you know northwest up or radial transverse up components of a seismometer? Sure. That's an excellent question. So a seismic wave originates someplace and then it travels through the, you know, we set off this explosive and it travels through the ice and then it comes up. And so we need to know where our sensor is.

That's the very first thing we need to know. Where did you put this thing down? And so we use GPS. So we immediately knew we needed GPS to locate our instrument. And we needed to be able to locate the instrument to within sort of a few centimeters or or maybe tens of centimeters at worst because these instruments are only five or ten meters apart and so knowing its position to within three or four meters wasn't good enough we needed to know it much more accurately than And so we use GPS unlike what you might have in your handheld, within your car, within your iPhone. Most of those will give you your position to within about two or three or four meters.

And we had to use a special chip that gives out the raw data from the GPS chip. So we don't actually get a bad position from that. We get the raw data and then we process the data afterwards to get the accurate position that's down to about 10 or 15 centimeters. So that's the first thing that we had to do that was a consequence of moving away from a cabled system. On a cable, each cable is precisely, let's say, five meters away. Each geophone is precisely five meters away from the next geophone and seismometer. And so its positioning is very easy. In our case, because we've cut the cable, now we need to know what our position is. So that was the first problem we had to overcome.

The second one was that what you talked about is what the data from the magnetometer and the accelerometer. And that way we know relative to the box which way is up and which way is north. three geophones are or we have three geophones in there and they're oriented orthogonal to each other we can take the data and rotate them to north and up in after the fact it once we've taken the data from the magnetometer and accelerometer to tell us which way it is in fact up does that make sense yeah absolutely it sounds like there's quite a bit of post-processing that has to go on to the data when it comes back from these, though. And that's a lot of data, because you're probably recording, especially for active surveys, relatively fast, right? That's right. So we have three seismometers, three geophones.

I keep using the word interchangeably, sensors that record ground motion. So we have three geophones, one pointed nominally vertically, one pointed nominally east-west, and one pointed nominally north-south. So we've got these three geophones, and we record them at about 4,000 samples per second, anywhere from 2,000 to 10,000, but generally we run at 4,000 samples per second, anywhere from 2,000 to 10,000. But generally, we run at 4,000 samples per second. And we run continuously because if we know that we're setting off an explosive at this time, then we could indeed put a timer on the G-Pebble and say,

okay, go to sleep until noon to noon tomorrow. And then and then we're going to be setting off an explosive and then wake up at that point and record it for 10 seconds and then shut down again. But we're actually setting off explosives all through the day at different locations. And so we just leave the geopad on and recording all the time. And that has the advantage that it will also record natural seismicity, just small events, creaks and groans, if you will, from within the ice sheet, small earthquakes, small crevasses opening, anything, any natural noise that is generated by the ice sheet, we record that as well. So that's just an extra bonus. But it does end up producing many tens of megabytes of data and eventually gigabytes

of data over the course of the season, which all has to be post-processed. We're fortunate in that we have a small Linux computer effectively running within the GeoPebble. So we try to do as much of the processing within the GeoPebble as we can. So we combine channels and try and process some of the magnetic and acceleration data within the GeoPebble, but that has a trade-off. We've got a battery in there that we'd like to have last for as long as possible. So it's kind of a difficult problem to solve. How much work do you do within the device, and how much do you do outside? And at the end of the day, we've sort of plunked down on the side of preserving

battery life. And so all we do is record the data and do the minimal amount of processing and just store everything onto SD card and then pull the data off once we're back in camp and then do the heavy crunching on our laptops. Explosive events tend to be very fast. What sampling rate are you using on your accelerometers? Yeah, so the explosive events have a frequency content generally between 100 and 500 hertz. There's obviously energy above that, but the energy above 500 hertz gets attenuated within the ice after it's traveled three kilometers down to the bottom and three kilometers back up. And so all the energy above five, six, 700 hertz is gone by the time it's gone that long six, seven kilometer path. And so we record at 2000 hertz, or 4000 hertz, 4000 samples per second, to be sure that we're getting good fidelity in in the signal, because we don't really have to worry about that ultra high frequency energy. And so 4000 samples per second over 24 hours,

you really do have to do some local processing or you're going to blow out any storage media pretty quickly. Yeah, that's right. So we have a 16-gigabyte card, or 16, and I think in some of the ones that we've just put out, we put a 32 gigabyte card in but that really will only last you 10 or 12 days so we we can't we we do have to solve this problem the the most the simplest thing to do is to have some kind of event detection where you would watch the data stream

and then you would only record data when there is a signal on there, either an explosion or a natural event, something like that, something that rises above the noise. I kind of don't want to do that because there's this absolutely wonderful development in the earth sciences in seismology called noise correlation or sort of noise analysis. These folks have taken just the background noise from seismometers, and they take one in Northern California and one in Southern California, and you just look at the noise. And it turns out that there is coherent signal in there if you sit and listen to the noise long enough.

So there might be a storm off the coast of, I don't know, Washington State, and it will generate what appears to be incoherent noise, which travels down through Northern California into Southern California. And if you compare the signals from Northern California and Southern California, boy, lo and behold, there's actually a signal in there, even though it all just looks like noise. But you need to gather together vast amounts of noise to do that. So as the old saying goes, one person's noise is another person's signal. And so I kind of want to just record everything and store it if I can. That means we've got to go back and download the data more often.

It's a tradeoff. Everything's a tradeoff. Yes. Do you do any onboard compression? So what we do is we store the data in HDF5. I don't know if you know what that is, but it's a data storage format that's quite popular within the meteorological community because they have to deal with large data volumes and so it has a built in compression algorithm so you get about a three to one compression out of that and that helps a lot it takes energy you gotta spend yeah you gotta spend you gotta spend juice to do that compression and so we have turned off that compression algorithm for now just because again we want to run these for a little bit longer at the expense get them up to the uh to the longest time that they will run that both keeps

the battery alive and doesn't fill the hard disk and And so the trade-off up till now has been to keep the compression off and just let the battery do it because that still has not filled up the disk. I'm surprised. I would think that keeping the compression on means you write less to the disk, which is somewhat expensive, but more importantly, it would mean less data over the air, which is probably one of your big power sources. Yeah, that's right. Power sinks. Power sinks. The data over the air is the biggest power sink, and so we minimize that.

We store all the data in all of its glory to the sd card within the device itself and we only go over the air when when we know that we have good power when we have the ability to um when we have uh fully charged batteries when the sun is shining and the solar panels are charging everything, or we've physically picked up the device in those cases that we're doing that, physically picked up the device and brought it back. In the case of this experiment that we did in Switzerland, where we put the instrument out on the glacier and we were transmitting the data back, we kind of knew that the SD card had no purpose because we weren't going to be able

to recover the device anyway. And so we had to transmit everything over the air. And we probably should have gone with compression in that case. But being conservative engineers, my engineers were a little bit worried about how the, let's say you drop some packets. If you're sending raw data, then you would simply drop the packets and lose the information during that time. If you're doing compression, you might end up losing a lot more because the compression algorithms tend to need long runs of data to be able to do the decompression properly. So in the end, we again decided to be somewhat conservative.

As we do more tests, we might end up going with doing compression in the transmission over the air. Because I think you're absolutely right. We lose most of our power in the Wi-Fi. Well, so I think everybody listening to both shows knows that dealing with equipment in the field is always difficult. And dealing with prototype equipment and developing equipment is always difficult and dealing with prototype equipment and developing equipment is always difficult, especially the scientific instruments where you have very tight noise constraints and timing constraints and all of this going on. So my question was, what's it like operating

development scientific instrumentation in Antarctica? So I've been doing this for, gosh, 30 years now now so that was the first time i went to antarctica was uh in 1985 uh and i have learned only one thing in over in all that time which is it's the connector stupid everything that's ever broken has always been the connector uh and whether it's the connector that goes to your battery, whether it's the connector that goes to your sensor, whether it's the connector within your device, 99% of the failures are that.

So I've tried to, over time, make connectors that are robust and, more importantly, as few connectors as possible. And that's one of the advantages of these GeoPebbles, is that they have no connectors. Everything is inside. The sensor's inside, the battery's inside, the digitizer's inside, the Wi-Fi, the GPS, etc. Everything's inside this one relatively featureless box and and then the the data come out over Wi-Fi and and the charging has to happen through a connector on the side so we do have a connector on there

in an early prototype I actually urged some of the engineers I was working with to try and fit a wireless charger into these, you know, like you have on your toothbrush, where you just stick it into a socket and it charges the battery through the wall. They told me to go away and do something geophysical while they did engineering. So one day, one day, I would like to have absolutely no connectors and have this perfectly featureless snowball. I'll paint it black so that it doesn't disappear on the surface of the ice sheet. And just we'll be able to toss it out there and it'll charge through the walls. This is my dream.

That's fair. Inductive charging is awesome. But it does require space. It does. So why? I mean, okay. is awesome but it does require space it does so why i mean okay so ice goes ice slips ice slides deck of cards i got all that and i understand why it's important how fast it goes along a stream bed so how how it interfaces with the bottom of the ice and into the earth. But what's the overall reason for this? Why are you studying this?

I mean, it's ice. Who cares? Yeah. So that's an excellent question. that I like to work with my local folks here because, you know, they pay taxes, and I use some of their tax money to take me off to Antarctica. And although it is a wonderful vacation for me, it has a serious purpose. And here's that purpose. I think I alluded earlier to this cycle.

Water evaporates from the ocean. It falls as snow on top of the ice sheet and then eventually turns into ice and then it flows as a solid material back to the ocean where it melts. that cycle is perfectly in balance year in, year out, then what happens is you evaporate the same amount of ice or same amount of water from the oceans every year, and you dump that water on the ice sheets, and the ice sheet returns exactly that same amount of water back to the ocean every year. We call that ice sheet imbalance. And a consequence of being imbalanced is the ice sheet doesn't grow, the ice sheet doesn't shrink. It's the same size year in, year out, because there's as much water or snow falling on it as there is snow leaving it every year. Well, the same thing is true of the ocean. If the ice sheet is in balance, then the ocean is in balance.

And so if the ice sheet is in balance, then the ice sheet doesn't grow or shrink, but the ocean doesn't rise or fall either. It stays at the same level. If the ice sheet's not in balance, if the ice sheet is shrinking, in other words, if the ice sheet is dumping more ice into the ocean every year than it's receiving, that excess is in the oceans, and then the oceans rise. And this isn't just oceans rising around Antarctica or around Greenland, because the oceans of the world are all connected, this means that the oceans rise around New Orleans and Florida

and the Chesapeake Bay and Bangladesh and the islands of the Pacific. Every place around the world will see sea level rising if the ice sheets shrink. And the reasons ice sheets shrink is very simple. They get warm, and they lose more mass into the ocean. They lose more ice into the ocean. As we all know, ice doesn't like warm temperatures. And this is, I mean, this is something we all should be thinking about and understanding. So thank you for that.

I want to switch topics to something a little cheerier, I hope. When John was in grad school, because he was one of your grad students, if I understand correctly. He was, an excellent grad student. And he decided, you know what? I want to do a podcast about geoscience. What was your response to that? Were you like, yes! Or were you just like, okay, that means he's never finishing school.

No, John is one of the most focused people I know. If he decided he was going to do the podcast and continue with his studies and build instrumentation on the side and get married and have a family, I knew that he would do it all and do it all excellently. So I was absolutely delighted when I heard that he was doing this podcast

because it is really crucial that the general public listen to and understand that science matters, that science is fun, that science is something that anybody can be involved in and working with and become a part of. If they're young people who want to choose a career, well, this is one way of doing it. And the kinds of podcasts that he does kind of, I think, show that playful, fun, and engaging side of geosciences in particular. But there are obviously podcasts for all types of science. And so I was absolutely delighted. And I knew that there was no worry about John not finishing, that even though he would indeed be a rock star of the podcasting world, it still would not pull him away from his true vocation.

Sorry. Rock star. Sorry. Sorry. Sorry. So, Sridhar, you also wrote a book as well that involved more of the embedded side of your work, right? That's right. So we have inside these geopebbles, we've got a board that was developed by one of my colleagues here that has a small microprocessor on it called a propeller. It's an extraordinary little device that is well used and known and loved in the hobby community, but I think is also starting to be used in commercial projects out there. And it has

eight parallel microprocessors and running in parallel, and they call them COGs. And then there's a central supervisor that kind of partitions the jobs within the COGs. And it's a really wonderful little device. And so that's the sort of embedded part of the device. That's what speaks directly to the various chips, the digitizer and the magnetometer and so on. And it then hands data up to the Linux machine, which stores the data onto SD card

after doing some processing and so on. And the propeller can be programmed in a high-level language called SPIN. And there's a lot of resources for programming the propeller in be programmed in a high level language called spin. And there's a lot of resources for programming the propeller and spin. But to get the most performance out of it, you have to program it in assembler. And there is a propeller assembler language. And there aren't as many resources for that. So in the process of learning, I had an engineer who worked with me, Peter Burkett, who did a lot of the programming for it, but I knew that he wouldn't be around forever.

And so that we didn't lose his knowledge, I had him teach me what he could about programming the propeller in the process of doing that, in kind of learning this from scratch with his tutelage, I kept writing things down. And over time, it got to the point where I was like, gosh, this is how a beginner would get into propeller assembler programming. And maybe there's somebody else out there who would like to know about this. And so I took all my notes and I turned them into a book. And you can purchase that book on leanpub.com, which is a self-publishing website. So if you go on there and look for propeller programming, you can search for it and this will pop up. And it's a donation or purchase type of thing. And you download a PDF and you can look at it on your computer. So I encourage you to go to lean pub.com and search for propeller programming. And there will be links in our show notes,

of course, on both sets of show notes. I imagine. Yes. And if you're a fan of trains, you will not be disappointed by the figures and analogies in this book. It's a really fantastic book that you wrote, Tredar. Thank you, John. Appreciate that.

Trains is a good analogy for propeller because there's all these things that go on in software. They would be in separate threads. That's a really good analogy. I look forward to checking it out. I'd be happy to get some feedback from you or anybody out there who does download it or download the teaser text, and I can make any changes or improvements as needed. Well, before things run too long, I think we should probably transition on over to everybody's favorite part of at least the Don't Panic geocast and possibly the first time on Embedded.fm, Fun Paper Friday. Whoa! usually choose a paper that's slightly humorous or has an odd twist to it, and then they read it and they talk about it and they encourage people to read academic papers. And we did suggest a paper, but given the embedded folks,

I know that they won't go read it. Not an academic, serious, I mean, some of, but we submitted a video, a one minute long video. So, yeah. John, what did you think of The Boring Company? So, this was the first time that I'd heard about this, but this is basically the concept of underground transportation of cars on hyperspeed sleds. Yes, yes, indeed. Almost as good as giant shrews excavating tunnels. Yeah.

Last time I was mentioned on their podcast, they noted that I submitted that paper. All right. Anyway, so Elon Musk is building tunnels under cities. Do you build tunnels? Or do you just, I mean... Do you just find them? If you take something away, you're not building something. Well, subtractive manufacturing, right?

That's... All right. Yeah. So he's making these tunnels to transport cars at very high rates of speed. And this concept video is very polished, but it's kind of humorous to watch. It's hilarious. Where are these cars going from and to? Well, it's in a city and they're showing a busy city street and then you pull over to the side and you drop down below the busy city street and go on to a sled that is going in the same direction.

So it's sort of like the Underground Expressway. I'm pretty sure the actual inspiration for this is the 405 freeway in Los Angeles, since Elon Musk has to drive that all the time, apparently. So I think he wants to go around where all the freeways are in Los Angeles. I think Los Angeles is actually the target. And they have toll roads that make it go faster, skip over some of the traffic. Yeah, you've driven down there. I feel like the concept should then be called, you know, like 405 freeway not found. My question for the geoscientists here is, is this doable?

I mean, this just looks... You already dug a tunnel. I mean, there certainly are the big TBMs, the tunnel boring machines. And I think the challenge here is scale. Though anytime Elon Musk has challenged scale, he's come out pretty good. But in the video, you know, it zooms out and you see many, many, many of these tunnels crossing. And so I think the scale of digging all of those tunnels and also the scale of not interfering with existing infrastructure and making all of these

elevators to take the cars up and down and with the tunnels themselves would be quite a challenge and there's obviously all the safety risks that go on with earthquakes with flooding and everything else as well yeah i've done a lot of work of work in Norway because they have glaciers there. And so I've done work on the various parts of the Norwegian ice caps. Boy, they love their tunnels there. And it is no surprise that their national symbol is the troll because they are down there

digging like little elves all the time. It's pretty extraordinary. You go to Norway, and scale is not an issue. There are hundreds and hundreds, literally hundreds of miles of tunnel in quite remote parts of the place. So tunnel digging, I think, has become easy, and they can do that just because they need to connect from this valley to that valley, and the easiest way to get there is to go through the mountain rather than over it.

Well, I would maybe rather go over 405 than under it, given that that is definitely earthquake country. Yeah, it's true. It's like when we were in San Francisco. Every time you get on the BART, you kind of hold your breath for a little bit. That's right. You are underwater, under the sea,

under the rock and earthquakes. It's just very exciting. It's been through a big earthquake. It has. It did just fine as well. So yes, thank you for the opportunity to have Fun Paper Friday or Fun Video Friday as it turned out. Yes. No, I think this was a really great thing. I hadn't heard about it until you mentioned it

and sent this video over. But it's an interesting concept. And I think that there's a lot of geo challenges that would be posed by it. But also, as you're digging these tunnels, you're going to be collecting some data as you go with these TBMs. So it would be interesting to see what this dense network of data looks like on the subsurface. Cool. Yeah. Absolutely. So, Sridhar, are there any thoughts you'd like to leave us with?

Yeah. So go out there and next time you're at the beach, spare a thought for the poor glaciers in the far north and the far south that are suffering from our use of the atmosphere. And so try and limit your carbon footprint if you at all can. Thank you. Our guest has been Dr. Sridhar Anandakrishnan, professor of geoscience at Penn State and author of Propeller Programming. Thank you so much for being with us, Sridhar. Oh, it's been my pleasure.

And thank you for doing these podcasts. Thank you also to Christopher for producing and co-hosting. Thank you to Digilent for their 15% off coupon, Embedded FM, good only for the month of June. And of course, thank you for listening. Also, some of you requested that we have the cat on occasionally for just a meow now and then. I think the rooster was a good improvement.

Oh, absolutely. And for the listeners of the Geocast, if you would like any, we've got show stickers. If you'd like show stickers or have some feedback on the episode, we would love to hear from you. You can get a hold of us, show at don'tpanicgeocast.com. We're on Twitter at don'tpanicgeo. And we have a great community of programmers, scientists, and just generally interesting folks, including Sridhar, on the Software Underground Slack community,

which you can find at swung.rocks. And remember, until next week, don't panic. It's not an exact science. Any opinions, findings, conclusions, or recommendations expressed are solely ours and do not necessarily reflect the views of our employers or funding agencies.