This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. In this episode, we'll be talking about geodesy, and particularly about accurately determining positions on the Earth's surface and monitoring how these positions change over time. Over the past few decades, the technology involved has improved dramatically, and it has become much more affordable. How can we use this to address geological questions, such as how the Earth deforms during and after large earthquakes, or to see plate motions in high resolution? Tom Herring is a pioneer in high-precision geodetic analytical methods and applications for satellite-based navigation systems to study the Earth's surface. He is a professor in the Earth, Atmospheric and Planetary Sciences Department at the Massachusetts Institute of Technology. Tom Herring, welcome to Geology Bites.

Tom Herring: Thank you, Oliver, for inviting me to your show. It's a pleasure to be here.

Oliver Strimpel: Let's start with the technology of high-precision geodetic measurements, and then we'll talk about some geological applications. First of all, what kind of accuracy can we achieve with these systems now?

Tom Herring: Well, in the best cases, we can make position determinations at submillimeter accuracy horizontally and about 3 millimeters in height. The height is worse than the horizontal components just because of the geometry of satellite orbits being above us, and the atmospheric refraction effects. Getting centimeter accuracy with these systems is actually pretty easy these days. The accuracy does depend on the averaging time and the type of system that you use, and for geologic processes, we often use 24-hour average positions to reduce the noise. This is where the best accuracy is, but with systems such as GPS, we can also determine those positions every time a measurement is made, which could be at rates of up to 10 or 100 times per second. The accuracy of those measurements is generally around a few millimeters, if it's in a very good environment.

Oliver Strimpel: OK, what are the main types of geodetic measurement systems that can achieve such impressive accuracy?

Tom Herring: There are three main types. There are the satellite-based types, such as the GPS system. The Global Positioning System is one, and it uses radio waves. There are also laser systems that can send laser pulses up to satellites, and to the Moon with specially installed corner-cube reflectors. Those are very expensive systems. The GPS system is the US one, and these days we add to that the European Galileo system, which is of similar design, the Chinese BeiDou system, and the Russian GLONASS system, which is of the same era as the GPS system. Combined together, people refer to this as the Global Navigation Satellite System, or GNSS. The more satellites you have, the easier it is to make position determinations when you have obstructed views.

The other big radio technique that we use is called very long baseline interferometry, and it uses radio signals from extragalactic sources. Those need to be observed with very large telescopes because of the weak signals involved. And finally, we have synthetic aperture radar, which is a technique that can be used for imaging deformation on the surface, based on the returned signals from radars installed generally on satellites.

Oliver Strimpel: OK, can you give us a high-level explanation of how these technologies work? Let's start with satellite-based systems.

Tom Herring: Yes, these systems work by a technique called trilateration. It's very similar to triangulation, except that it uses distance measurements rather than angles, and that distance measurement is what allows us to achieve range measurements to satellites from the ground with millimeter-level accuracy. For GNSS, there are actually two types of measurements that are made, two types of signals that are interpreted. One of them is readily available in all positioning devices, such as cell phones, boats and car navigation systems, and there's a second version, which allows very high accuracy, and that used to cost typically about $20,000 per receiver.

The radio signals that are transmitted from the GNSS satellites have wavelengths of around 200 millimeters. They're encoded in a way that different satellites can be easily separated from each other, and your ground receiver correlates the arrival time of the codes from the satellites with a replica of that code and measures the time difference between them. That time difference is related to the range to the spacecraft, but also to the errors in the clocks in both the spacecraft and the ground receivers. The satellites can have very expensive clocks, but what we want to do is make the ground receivers have very inexpensive ones. And the great revolution that happened with GPS is that multiple satellites can be seen at the same time. So the position can be based on the measurements to three satellites, and then to estimate the clock, you need a fourth satellite. That allows you to have a very cheap clock in your receiver, which makes these receivers inexpensive. Typically, these days with GNSS, we can see 25 to 30 satellites at any one time, anywhere in the world. That level of correlation with the codes is called pseudorange positioning, and that's meter-level positioning.

The big thing that came from GNSS innovation was that we can look at the phase of the radio signal that's being transmitted. And since we can continuously track the spacecraft, we can count the number of cycles of phase as it goes through. So what we end up getting is this range change measurement to the spacecraft, from when we first observe it and then as it moves through the sky. That phase change is accurate to a couple of millimeters, and it's from that that we can determine these millimeter-level positions down on the Earth's surface. We also need multiple frequencies because of the impacts of some of the atmospheric refraction effects, and that, again, is the way that we can eliminate some of the systematic errors of these types of measurements.

Oliver Strimpel: That's impressive, to get the cost down, and I can see that if you can have a low-cost clock, obviously that makes a big difference. But has the cost reduction also benefited from the same kind of thing as integrated circuits have, following Moore's law over the decades?

Tom Herring: Yes, they have, although not for precisely the same reason. The receivers for the current techniques that are used to do millimeter-level positioning typically used to cost around $20,000 each, and now they're less than about $700. That reduction happens because of the way the GPS signal is transmitted. In the main part of the GPS system, there are additional codes that are transmitted with the signals that are required for anti-spoofing protection for the US military. Those additional codes on the second frequency that's transmitted make it actually very difficult to track the signal, and hence the cost of the receivers that have the technology to be able to do that. But starting about a decade ago, the military realized the importance of civilian applications of high-precision positioning, and so they started to introduce a third frequency on the spacecraft, which allows civilians to use that frequency. They've also added some additional codes to the normal transmissions, which again make it easier for these receivers to receive these high-accuracy signals.

Now, GPS satellites get launched at about one to two per year, so over the last decade, we now have about 17 or 18 of the satellites that are transmitting these signals, readily available to civilians. These cheap receivers take advantage of that. They also take advantage of the fact that the Galileo system, which is now almost complete, transmits signals which are easily tracked, and the Chinese BeiDou system, which is now on its third generation, also transmits signals that are easy to track. And so the combination of changing the way the signals are encoded has allowed these very inexpensive receivers to drop the price down. And even today, with a bit of tweaking, the latest generations of cell phones are actually becoming capable of doing this millimeter-level positioning as well.

Oliver Strimpel: OK. Let's move on to the second of the three technologies you mentioned, very long baseline interferometry, VLBI. How do these systems work?

Tom Herring: Yeah. So VLBI is the sort of older, bigger, literally, sibling of GPS. In GPS, we know what the codes are that are being transmitted, and so your receiver can create a copy of that code to correlate with the incoming signals, to tell which satellite it's looking at and the range and phase to that satellite. In VLBI, we don't have that, but we have these extragalactic radio sources that can be seen from multiple telescopes at the same time. So you can take each pair of telescopes; that very weak extragalactic signal is recorded at the sites, and you can cross-correlate them to find the time difference between the arrival of one signal at one telescope and at the other. And again, that time difference can be translated into a range, and that allows you to do the relative positioning of these large radio telescopes.

Now, since these extragalactic radio sources transmit radio waves across a very wide bandwidth, these VLBI systems can make millimeter-accuracy differences in range measurements because of that really wide band that they have. So again, it allows submillimeter-level positioning of these systems. However, the radio telescopes have to be very large, they have to have very low-noise receivers, and because the clocks are completely separated and they can only see one thing at a time, they need to have very accurate clocks. So these systems typically cost $1 to $2 million per unit, and there are about 30 of them operating around the world at the moment.

Oliver Strimpel: OK. Let's talk about the third of the technologies you mentioned, the radar-based one, InSAR. In a previous episode of Geology Bites with Romain Jolivet about the Turkey–Syria earthquake of 2023, we discussed how InSAR was used to map the movements associated with that event. Can you remind us how this technology works?

Tom Herring: Yeah. So InSAR is a radar imaging technique, and it looks at signals which are reflected from the Earth and transmitted, normally, from satellites. But you can also put this in aircraft, and sometimes on balloons. The synthetic part of the name comes from the fact that the motion of the spacecraft is used to synthesize, effectively, a very large antenna, and so with that you get very high resolution on the ground. The pixel size in these images is typically of order one to two meters across. The interferometric part of the name comes from the fact that when you look at a pixel on the ground in one image, there are lots of little facets in there that reflect the signal back, and they sort of add together into the final amplitude and phase of the signal that's received back at the spacecraft. If you come back six days later, or 12 days later, and reimage that same pixel, and those little pieces have all stayed pretty much the same, then that phase is going to add to a very similar result. And if the pixel has changed position, you can actually see the change in position.

So in InSAR, you take little groups of pixels, maybe 9 or 16 of them, and you look at what the phase difference is between two images. And if the whole group of little pixels, which again is quite small in size, only 7 or 8 meters across, has moved, you can tell that it's moved. That can typically be done with accuracies, again, between a millimeter and a centimeter or so; the noise is about the millimeter. What you worry about with InSAR is that you're looking through the Earth's atmosphere when these signals come down, and the refractive index of the atmosphere changes the range as weather conditions change. That makes it difficult to know exactly whether you're looking at the Earth having moved, or whether there was an atmospheric effect. And there are many techniques now of using multiple images that try to reduce that atmospheric effect.

Oliver Strimpel: So I suppose one difference is that the radar technique actually produces a spatial map, whereas the other two technologies are really more point-to-point measurements, where you have your sensors.

Tom Herring: Yes, these two are very complementary in a sense, because with the GNSS-type systems, we can get relatively high density, stations spaced by 5 to 10 kilometers in areas of tectonic activity, and we can make measurements literally continuously. Most of the systems that run make measurements at least once every 30 seconds, so you get very high time resolution. But as you say, the spatial resolution is quite coarse. You can combine that with these InSAR measurements, which have very high spatial resolution, but we can only acquire an image of a region typically once every 12 days or six days. So you have high spatial resolution and low time resolution, or high time resolution and low spatial resolution. And as we try to combine them, one of the areas of research actually is how to generate a product which is essentially high resolution in both time and space from this combination of InSAR and GNSS measurements.

Oliver Strimpel: OK, let's move on to the application of these technologies to geological questions. How exactly does it help us understand what happens before, during and after a large earthquake?

Tom Herring: Yes. So what we were able to see very early on is plate motion. And as we monitor something, we can see the large coseismic offset that happens at the time of an earthquake, either through InSAR or the GNSS measurements. And then we look very carefully at the accumulated strain that's building up until the time of the earthquake, and the strain release that happens during the earthquake. One thing which is very clear for many earthquakes now is that there is also a continued strain release after the earthquake, which is called the postseismic deformation phase. That phase is very well measured with these modern techniques, and it can typically be of order 10 to 15% of the size of the coseismic offsets. One thing we have noticed is that this postseismic part of the earthquake seems to be a much deeper process, and so it spreads over a larger distance. We think it also potentially transfers stress from the region where the earthquake happened to areas where stress is still building but no earthquake has happened yet.

The other thing we can do with GNSS is see what happens during the earthquake itself, when we have these high-rate measurements. So we can see the arrival of seismic waves. One of the first interesting cases of this was the Denali earthquake up in Alaska, which happened a couple of decades ago. The direction of that earthquake rupture essentially sent seismic waves right across the whole of the US, and all of the GNSS stations in the continental US region that were tracking data at 1-hertz rates could see these 1- to 2-centimeter waves propagating across the US after that earthquake. Now, that type of process is quite commonly observed in large earthquakes.

The other phenomenon, which happens before some of these earthquakes, depending on where you are, is that we have strain accumulation before the earthquake. But particularly in subduction zones, we often see this process called slow slip, where there seems to be a relaxing of that strain accumulation. Geodetically, it looks like you've had an earthquake in the region, but instead of lasting only a few seconds, this earthquake takes maybe 10 to 20 days to happen. That is something, again, very well observed in large subduction zone areas. What's not clear at the moment is the implication of these slow slip events for the future occurrence of earthquakes. One might imagine that they release stress, so they're like little earthquakes that stop the big earthquake from happening. But what's more likely is that they're relieving stress in one region but transferring it to an area where that earthquake is more likely to happen. We just don't have enough examples of large earthquakes in fully instrumented subduction zones at the moment to be able to address the question of what the impact of these types of events is before the earthquakes.

Oliver Strimpel: Does this mapping of strain work as well in highly mountainous regions, like, for example, the Himalaya in the 2015 Gorkha earthquake in Nepal?

Tom Herring: In high mountainous areas, we have a couple of issues. Logistically, if you're trying to install GNSS, it's tricky. That has been done, again with 50- to 100-kilometer spacings across many of these types of areas. With InSAR in high mountains, you have sort of two problems. One is that the high mountains can actually shadow the radar returns, because, again, the radars tend to look at the surface not directly down but from an angle, and that angle is somewhere between 45 and 65 degrees off nadir. So you can get shadowing. The other problem you have in the mountains is that you can get a lot of snow. When snow falls on the ground, the reflection that comes back is a completely different type of reflection from what it was before, and so there is what we call no coherence in the image. Nearby pixels now just have random phase variations, and so you lose the information on the deformation itself. But in general, InSAR has been very successful in mountainous regions.

Oliver Strimpel: Can one turn that problem on its head and use the variation in snow depth to measure how ice and glaciers are actually thinning out?

Tom Herring: Yes, that's another very big application of the InSAR technique, and of GNSS as well. The polar regions are really very well observed, because all of the satellites that make these InSAR measurements are in high-inclination orbits. So the paths of the spacecraft, the ground tracks, cross very close together up in the polar regions, and that gives you much more opportunity to look at the deformation that's happening in those regions. Also, some satellite systems, the European Sentinel systems, for example, have pairs of satellites. You have one satellite making a measurement, and then very shortly after, another satellite can measure the same region. And since glaciers can move very fast, up to kilometers per year and several tens of meters per day, you can actually see the deformation between those fairly rapid acquisitions.

This has been one of the great applications of InSAR: looking at the velocities of ice flows and how they change. Throughout the year, we've seen tidal effects on glaciers which actually move out to the sea. The ocean tide sort of accelerates and decelerates the motion of the glaciers, and that has lots of properties that give us insights into the mechanics of the ice. And that, in turn, allows us to have a much better sense of how things like water temperature and air temperature changes change the nature of the ice dynamically: how fast it can move, what the limits are on how fast a glacier can change its motion, and the amount it moves based on things like temperature changes.

Oliver Strimpel: In the previous Geology Bites episode with Chuck DeMets, we talked about high temporal and spatial resolution measurements of plate motions, and particularly the northward movement of the Indian plate and the effect of the plate's collision with Eurasia. Can we use the high spatial resolution of geodesy to get a better handle on plate motions?

Tom Herring: Oh, definitely. We can measure plate motions with submillimeter-per-year accuracy these days, but it is over the interval of time over which we have collected data. Take places like India, for example: the India–Eurasia collision is happening at around 50 millimeters per year. One of the interesting results that came out of early GNSS deployments in the late 1990s and early 2000s was that the motion of India into the Himalayan mountains, which is reflected in the building of that large mountain range, is only half accommodated at the Himalayan front. The Himalayan mountains themselves are actually being pushed back towards Eurasia at a rate of about 20 millimeters per year, and that 20 millimeters per year is accommodated in deformation that's happening on the other side of the Tibetan Plateau, in that large-scale pushing of the whole Tibetan Plateau by the Indian plate. That was not fully appreciated until we had GNSS measurements that gave us fine spatial-scale variations in velocity across these regions, so that we could interpret the mechanics of how this happens.

Oliver Strimpel: Wow, that's incredible. So really only about half the northward movement of India is actually accommodated by the building of the mountains, and the other half is actually transmitted to the north.

Tom Herring: Yes, and you sort of see that in the topography. I suppose that's one of the things we've learned from these GNSS measurements: wherever you have topography, you typically have deformation still ongoing. So if you look at North America, for example, once you get to Colorado and start moving across towards the Pacific, all of that material out there is actively deforming right now. And it's the same in the Tibetan Plateau, et cetera. Wherever you have topography, we tend to see deformation happening within those regions, again at rates of a few millimeters per year, up to a centimeter per year, depending on where we are.

Oliver Strimpel: You've also used high-precision geodesy to measure changes in the rotation of the Earth. Can you tell us how you did that and what you found out?

Tom Herring: Yes, the Earth's rotation is actually pretty wobbly, as I like to say. There are essentially three aspects to it. One is that the rotation axis of the Earth is actually not fixed to the solid body of the Earth. It moves around, and that movement is about 10 meters in a year. It's a circular sort of motion, because of some annual effects in the seasonal movement of fluids on the planet. It also has a drift to it, of about 10 meters per century.

The spin rate of the Earth, how fast we spin, also changes dramatically, at least in a geodetic sense, throughout the year, and for periods of less than about a year, that is driven largely by changes in wind speeds in the atmosphere. That's extremely well measured and documented now, based essentially on our weather forecast models, which allow us to calculate the angular momentum of the atmosphere to extremely high precision. And in standard mechanics, the angular momentum of the whole system has to be conserved. So if the atmosphere's angular momentum increases, the solid Earth's angular momentum must decrease, and it decreases by changing the rotation rate.

We can also see that there's a long-term trend in our rotation rate as well: we're slowing down in our rotation. This has been happening ever since the Earth formed, essentially, and it is due to tidal forces, mostly from ocean tides coming in and out. We can also measure that indirectly through the laser measurements that have been made to the Moon's retroreflectors, mostly put there by the Apollo missions and by the Russian missions that put rovers on the Moon back in the 1970s. That's about 38 millimeters per year of motion, and that's been well measured, and it's consistent with the slowing of the rotation rate.

The other big influence on the rotation rate is the fluid core of the Earth. We actually need motion in the fluid core to generate the Earth's magnetic field, which is quite critical to protecting us from solar wind particles coming from the Sun. If that motion of the fluid core changes, then that also changes the rotation rate. And then finally, the Earth's rotation axis moves in inertial space, mainly due to the torque applied by the Sun and the Moon on the Earth's equatorial bulge, and the presence of the fluid core in there has a dramatic effect on that. So one of the very early very long baseline interferometry measurements was to determine the flattening of the fluid core boundary, based on how the Earth was rotating. That showed that the fluid core was not in hydrostatic equilibrium. It was off by about 500 meters in the difference between the polar and equatorial radius from what would be predicted. And that was a measurement that came directly out of Earth rotation variations.

Oliver Strimpel: So what are the time scales of these rotation period changes? You said the atmospheric one is about annual. What about the ones that we can trace back to the movements of the core, or the Moon, or other celestial impacts?

Tom Herring: Yeah. The atmosphere has a variety of frequencies in it, and we do see a strong annual signal in there. We also see periods down to just 10 days; there are natural resonance modes in the atmosphere that show up. On the longest time scales, longer than a year, those variations in rotation rate are up to about three milliseconds per day change in the length of the day. Now, 3 milliseconds doesn't sound like very much, but at times we are consistently 3 milliseconds slower, so every day it accumulates. After about 300-odd days, if you're running 3 milliseconds per day, you're up to almost a second of change in time. And we do try to account for that in the way that we relate our civilian timekeeping to atomic time.

The other time scales that are there are over decades, and there, again, we see multi-millisecond changes in the length of day. But if we were to try to explain that with the atmosphere, we just don't have enough angular momentum in the atmosphere for those large variations. So the place where we believe all of that is coming from is the fluid core of the Earth. Over the last 20 years or so, we've seen the fluid core losing angular momentum and the Earth gaining it, and the Earth spinning up over those periods. And then, as I said, over really long time periods, we have this tidal dissipation, due to the energy loss that happens as water flows in and out during the tidal sequence of the Earth.

Oliver Strimpel: I think you implied from what you said earlier that you've been able to measure the change in the geographic pole, namely that the axis of rotation of the Earth wanders with respect to the crust. We're used to magnetic poles wandering quite a bit. How much does the geographic pole move?

Tom Herring: Yeah, the geographic pole typically moves around 10 meters over a year, and over a century, it will also secularly drift by about 10 meters. That drift over that period of time is due to the Earth's response to the removal of ice at the end of the last Ice Age, which ended about 10,000 years ago. When you put all that ice on the Earth's crust, it pushes it down, and the material in the mantle as well. And when you take the ice away, there's an elastic response that happens instantly, but then there's a viscoelastic response, when material slowly comes back in. That process is still happening around Hudson Bay. For example, all of the land around Hudson Bay is moving upwards at about 15 millimeters per year at the moment due to that rebound. And that change in the height of the surface and in the material in the Earth's mantle is changing the moments of inertia of the Earth. The Earth's rotation axis essentially wants to rotate centered on the maximum moment of inertia of the Earth, and that internal change is moving the maximum moment of inertia. That's what we're able to see in these measurements.

Oliver Strimpel: The dramatic lowering of cost that you talked about for the satellite-based systems means we can use them for a much larger range of applications. Can you give us some examples?

Tom Herring: Yes, we have lots of different applications. One thing we've been doing is putting GPS receivers on top of very tall buildings to see how they deform throughout the year. We've seen thermal responses of the buildings. We've seen buildings moving, shaking during earthquakes, with motions of up to 20 centimeters. It turns out, for the buildings we've looked at, that all of these large motions are fully within the specifications of the safety standards for the building, so it's not an immediate danger to any of these buildings. When we build large dams, there are also two processes that can be studied. One is that as the dam fills, the land around the dam actually gets pushed down by the weight of the water. That's very well observed. We can also put GPS receivers on the faces of dams to make sure that the dam is not being pushed out excessively by the water as it fills. There have been dams in Southern California which had this GPS monitoring put on them just to verify that the dam face is remaining stable and not suffering any damage from earthquakes.

There are also some rather unanticipated things that we've used these systems for. GPS antennas need to be able to see the whole sky above them, but they can also see reflections from below the antenna. That's what we call multipath, and for positioning the antenna, it's considered noise, and we'd like to eliminate it. But people have taken advantage of that reflected signal to tell them something about what is actually causing the reflection at the ground surface. People have used it to do soil moisture measurements, because typically as soils get moist, they reflect signals better, so the reflected signal becomes stronger. We can detect if there's snow in the region: as the snow builds up, there's a reflection off the snow, and you can see the snow height. If vegetation is growing, if you put this in your field, you'll be able to see it growing.

One thing we worry about and have to take into account with all of these geodetic measurements is that we have a refractive medium in the Earth's atmosphere, and we also have a layer of electrons high above the atmosphere called the ionosphere. We have to monitor those to do accurate positioning, and so we can use these measurements to help with weather forecasting. For the ionosphere, we can monitor the strength of magnetic storms, and we've even been able to monitor the impacts on the atmosphere of tsunami waves coming from large earthquakes, et cetera.

The areas that have been very viable commercially are things like precision farming, where large combine harvesters, for example, can be controlled by GPS. More importantly, as the field is harvested, the harvester knows precisely where it is, and it can use the multispectral scanner technologies that we often put in spacecraft to get a measure of the yield coming from that particular part of the field. The next year, when that field is used, the system can determine where it needs to put more fertilizer to maximize the yield. We've also seen lots of drone displays recently. There was a massive one of 1,500 drones that set a Guinness World Record for Christmas 2024. These drones, which map out wonderful patterns in the sky with lights, are all controlled by GPS receivers; the drones have to be able to position themselves while they're flying to within a few centimeters of their designated position to make the image look correct. So in general, I think we're seeing this ability of personal positioning to get us to within a few centimeters while we're walking around. In the future, when you go someplace, it's almost certain that there will be no debate as to which door of the building you have to enter through. It's not just knowing the building, but actually knowing which door to go to. I think there's a lot of potential growth for those types of things.

Oliver Strimpel: What are you working on at the moment?

Tom Herring: We're very interested in these slow slip events that are occurring in large subduction zones, getting better temporal and spatial correlation from them, and extracting as much noise out of there as we can, to get a better sense of their relationship to earthquakes. Fluids moving around on the planet, primarily water on the surface, are something which affects our position determinations, because they can load the crust and expand soils. So we want to get a better sense of how we can exploit these measurements to understand what subsurface water is doing. As a rule of thumb, a one-millibar change in atmospheric pressure actually moves the Earth's surface by about half a millimeter, and pressure changes of 10 millibars are very common. So we see this in our stations: as weather systems move across the stations, we can see those types of motions.

Oliver Strimpel: Tom Herring, thank you very much.

Tom Herring: And thank you, this was fun.

Oliver Strimpel: To see pictures and illustrations that support this podcast, go to geologybites.com, where you'll also find a subject matter index of all the episodes. There you can also give me feedback, which I welcome, as well as sign up to get my emails about new episodes.