This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. The lower reaches of the Earth's mantle extend all the way down to the boundary with a metallic core, which lies about 2,900 kilometers below the surface. We knew almost nothing about this highly inaccessible region until good seismic measurements became available in the 1970s. That coincided with a rapid increase in computer power, which enabled seismologists to generate images, albeit at very low resolution, of the mantle. The images surprised us by revealing some dramatic features in the lower mantle.

Allen McNamara is a professor of geological sciences at Michigan State University. He uses computer-based fluid mechanical models to investigate the behavior of the mantle, working in tandem with the seismologists to understand the origin and dynamics of the features we see in the mantle. Allen McNamara, welcome to Geology Bites.

Allen McNamara: Thank you, Oliver. I'm happy to be here.

Oliver Strimpel: In my introduction, I mentioned some dramatic features that the seismologists have found in their images of the mantle. But before we talk about these features, how is it even possible to see into the depths of the mantle and create an image out of seismic waves?

Allen McNamara: Well, there are a couple different ways. We can image the inside of the Earth in kind of an analogous way as you would take a CAT scan. The physics is a little different, but the concept's similar. On the surface of the Earth, we have earthquakes, and we also have seismic stations located all around the globe. And whenever there's an earthquake, those seismic waves travel through the interior of the Earth to the seismometers all around the globe. And by looking at those travel times, how long it took to go from the earthquake to the seismometer, we can invert the problem and determine what the inside of the Earth looks like. Basically, we can find in which areas the seismic waves move faster than average, and in which areas the seismic waves move slower than average. And usually, if a seismic wave moves faster than average, that usually means the area is colder, and if a seismic wave moves slower than average, that usually means the area is hotter. But through that technique, we can put together a rough image of the Earth's interior.

Now, unlike a CAT scan, where the sensors and the detectors can be distributed equally around, in the Earth we're limited. Earthquakes occur at plate boundaries, and the big earthquakes, which are the most useful ones, mostly occur at subduction zones. And the Earth is only 30% covered by land, so that's where most of our seismometers are. So there are huge gaps, and the heterogeneity in receivers and earthquakes leads to a very blurry image of the Earth's interior. Now, there's another technique that we can use to image very small things at the bottom of the Earth's mantle. The sharpest contrast in material properties in the Earth is actually between the mantle and the core, and at that boundary, seismic waves reflect off very easily. Now, if you have other materials above that boundary, or even below that boundary, you'll often see little bounces that come back before or after that main core-mantle-boundary wave bounce, and we call those precursors or postcursors, and those can be used to detect the presence of very small objects along the core-mantle boundary.

Oliver Strimpel: What do we see in these seismic tomography maps?

Allen McNamara: In the regions where we've had subduction, where plates sink back into the mantle, we see that those seismic waves move faster than average, so that indicates that those areas are colder than average. And that's consistent with what we think is happening: at the surface, the plates are cold, they sink into the mantle, and they retain that coolness compared to the rest of the mantle for a very long time. So we see that in the seismic tomography. Between those areas, we have two large regions in the lower mantle that exhibit lower-than-average seismic wave speed, and we call those the large low-shear-velocity provinces, and those are located beneath Africa and the Pacific Ocean.

Oliver Strimpel: When researchers started working on understanding what these features mean and what's actually going on in the mantle, did they have any observational data other than the seismic data to go on?

Allen McNamara: This is a very interesting question. So before we had a good understanding of what the interior of the Earth looked like in a 3D sense, we had geochemical observations. And there are two main environments on the surface of the Earth where you get the formation of basalt. So if you melt the mantle, you form a rock called basalt, and the ocean floor is all just covered in basaltic rock. So one area where you get the creation of basalt is at mid-ocean ridges. This is where the plates are separating and new plates are being formed. And another area is in ocean islands called hotspots. Hawaii is the prime example of a hotspot. Iceland is also another example of a hotspot. And they also are made of basalt. And so those basalts, to first order, look the same, but if you do detailed geochemical analysis of these basalts, you'll find that they're quite different.

So the first thing you notice is that at the mid-ocean ridges, where new plate is being created, those basalts tend to have very similar chemistry all around the world. The chemistry we're talking about is trace element chemistry, so just the elements that are a minor fraction of the composition of those basalts. So they seem to be relatively consistent around the world, and by understanding that chemistry, it seems that those basalts are coming from rock that has melted before. So it's not the first time that mantle melted to form those. Now, at the hotspots, the trace element geochemistry is quite different. First of all, between each hotspot, the chemistry is quite variable, unlike the mid-ocean ridges, where the chemistry is very similar from ridge to ridge. Another interesting thing is, if you look into the geochemistry, it indicates that those basalts were formed from mantle that has not melted before, or maybe melted less than the mid-ocean ridge basalts. So it seems like that material is coming from a different place.

An early idea was that the mantle was layered, with a lower mantle and an upper mantle, and the upper mantle caused all of the mid-ocean ridge basalts, whereas the lower mantle was isolated from the upper mantle, but there would be mantle plumes rising up from the lower mantle to the surface, bringing up that lower mantle material to the hotspots. Once the seismological observations started coming out, we noticed that there probably isn't a layered mantle. But it is intriguing to think that these large low-shear-velocity provinces beneath Africa and the Pacific could be the source of this different geochemistry that we see at the hotspots.

Oliver Strimpel: So actually, we have a kind of marriage between the geochemical and the seismic observations. What are the main scientific questions we're trying to answer about these large low-shear-velocity provinces?

Allen McNamara: I would say the big question is: what are they? That we don't know. So what we do know is they are seismic anomalies. Additional seismic studies that I haven't talked about here indicate that they probably have very sharp edges, meaning that likely they have a compositionally sharp edge to them. So we think they're a different composition than the rest of the mantle, but we don't know what they are, what they're made out of, how they got there, whether they are growing or whether they're shrinking. The main thing we don't understand is how they interact with the mantle to affect mantle convection. So one big question is, are they just passive features that are just kind of moving around in response to mantle convection, or are they active features that are participating in the mantle convection itself, and perhaps even guiding mantle convection? So those are the big questions that we have.

Oliver Strimpel: OK. So let's talk about the question of what these things are, particularly whether they're just temperature anomalies, or also temperature anomalies with different composition from the surrounding mantle. How do you go about shedding light on this question?

Allen McNamara: My research is in computational fluid dynamics, basically modeling the fluid dynamics of mantle convection. There are a couple things that we can help with in this process, and one is we can first test hypotheses we may come up with. For example, let's say there's a hypothesis that the large low-shear-velocity provinces are caused by denser-than-average compositional reservoirs in the mantle. We can put those ideas into a computer model. An observation will spark an idea, and oftentimes those ideas are incorrect. They're just not consistent with physics, and the modeling can help show that. The second thing we can do is, once we put those hypotheses into computer models, we can investigate what additional predictions those hypotheses would make. So again, using the example of hypothesizing that the large low-shear-velocity provinces are made of compositional reservoirs of denser-than-average material: if that's the case, does that work? And it does work in computer models. And then the question is, how can it affect the core-mantle boundary topography? How can it affect other things going on in the mantle, in ways that seismologists can later detect, to determine whether the hypothesis is a valid one or not?

Oliver Strimpel: Can you give an example of a specific prediction, say, of your models that could, at least in theory, distinguish low-shear-velocity provinces that contain denser material from those that do not?

Allen McNamara: Sure. I had a student several years ago named Teresa Lassak. Her main PhD research was to explore a question like this. She looked at two different conceptual models for what causes the large low-shear-velocity provinces. In one model, she modeled the hypothesis that these are areas of mantle plumes, small mantle plumes that are rising up beneath Africa and the Pacific, but when viewed through the blurry lens of tomography, they look like a large anomaly, as opposed to a lot of little plumes. She performed calculations on that hypothesis, and also on a hypothesis that the large low-shear-velocity provinces are denser-than-average compositional reservoirs. And for each of those models, she looked at the influence that those different ideas would have on the topography of the core-mantle boundary, and she found very stark differences between the two.

So if the large low-shear-velocity provinces are caused by thermal anomalies, you would expect to see kind of a spidery ridge pattern of core-mantle boundary topography, where beneath each plume you would have the core-mantle boundary uplifted a little bit, and it would lead to a spidery pattern, because if you have a lot of plumes upwelling, a plume has roots, kind of like the roots of a tree, and so that spider-web pattern. On the other hand, if the shear velocity provinces were just piles of compositional reservoirs of more dense material, the core-mantle boundary topography beneath them was a very flat plateau, so a very different pattern of topography altogether. And along with that plateau, you had a little ridge of elevated topography along the margins of that plateau. So if seismology can look at the core-mantle boundary topography at that level of detail, to detect the difference between those shapes, that would provide a lot of insight into which of those hypotheses are more credible. Unfortunately, the seismic results aren't there yet.

Oliver Strimpel: These low-shear-velocity provinces, in fact, they're called large low-shear-velocity provinces because they're thousands of kilometers across, with one of them extending under the whole of southern and eastern Africa, and the other under about a 3,000-kilometer-wide region of the Pacific. Can we actually see any structures smaller than that in the lower mantle?

Allen McNamara: There are other structures called ultra-low-velocity zones, and these, compared to the large low-shear-velocity provinces, are very teeny. So they may be around 100 kilometers wide, anywhere from 20 to 40, maybe up to 60 kilometers thick. So really teeny structures compared to the size of the mantle, that are on the core-mantle boundary. Now, they're probably not very important in terms of affecting mantle convection in any way, but what they can be very useful for is to act as little tracers, to help us see what's happening on the larger scale. So we detect these ultra-low-velocity zones by these precursors and postcursors, the reflections that I mentioned before. They've been known for quite some time, since probably the mid-'90s. One idea is that they're just regions of the lower mantle that are undergoing partial melting, and another idea is that they're ultra-dense materials. And I think that's probably my favorite hypothesis. And as for where they are: just like if you had a swimming pool with water flowing around in it, and you have a little sand or dirt in the swimming pool, you'll find that if there's currents in the water, that sand will form little patterns. And that's what dense stuff does in a fluid dynamical environment: dense material will collect in the convergent areas of flow. And one thing that we hope for is that we can use ultra-low-velocity zones to tell us what the mantle flow is doing in the lower mantle, just like sand in the swimming pool.

Oliver Strimpel: So if ultra-low-velocity zones are made of a different, denser material, where would that have come from? And is there anything in your models to support or refute these ideas?

Allen McNamara: So just like the surface of the Earth is very heterogeneous in composition (we have low-density sedimentary rocks at the surface, we have granitic crust, we have basaltic ocean floor, and it seems like the closer you look, the more heterogeneity you see), we expect the same thing to happen in the lower mantle as well. So the lighter things float to the top of the mantle, and the denser things will sink to the bottom. So one thing we don't know is, are all ultra-low-velocity zones the same thing? Just like the surface of the Earth is complicated, it's not unreasonable to think that the lowermost mantle of the Earth is complicated in terms of its chemical heterogeneities. One idea is that these are reaction products from interaction of mantle rock with the iron core. Another idea is that these are very dense rocks that formed at the surface of the Earth billions of years ago. In fact, two and a half billion years ago, there was the formation of a lot of iron-rich rocks called banded iron formations, which are very dense rocks and contain a lot of iron. Some people have proposed that the ultra-low-velocity zones that we see at the bottom of the mantle are actually those rocks that formed at the surface, that subducted into the mantle with sinking plates, and are now in the lower mantle.

Oliver Strimpel: Do you have a personal favorite?

Allen McNamara: My personal favorite is that they're banded iron formations that formed on the surface of the Earth two and a half billion years ago. And it's my favorite because that's the most tangible. I've seen those kinds of rocks. The idea's intriguing. You know, all of the different other hypotheses regarding chemical reactions and stuff, that's a little bit less tangible.

Oliver Strimpel: How well can you model the mantle dynamics? When we're trying to model the core, we can't even come within orders of magnitude of the correct viscosity without overtaxing the available computing resources.

Allen McNamara: You know, that used to be a big problem in the mantle geodynamics community. For example, when I was a graduate student, our models were very simple. We had to model fluids that were many orders of magnitude more viscous than what we think the Earth's mantle is, just because we didn't have the compute power. But in the last 20 years, computers have increased their speed quite a bit, and with the advent of parallel computing, we can now do what we consider to be realistic problems of Earth's mantle convection. People that model the Earth's outer core are still many, many, many years away from being able to do that. So our problem right now is not the compute power. Our main problem right now is the unknowns.

So there's a lot about the mantle that we don't know, a lot about its material properties. The biggest thing that we don't know is the viscosity. Many different studies have been done that try to tackle what the average viscosity is in the lower mantle, and so we have some rough numbers. I would say our error bars are several orders of magnitude, and that's pretty big. And probably the biggest issue is we don't understand the grain size of the mantle, particularly the lower mantle, and any formulation of viscosity depends on the grain size. So we don't know what the average grain size is, and more importantly, we don't know how heterogeneous that grain size is throughout the lower mantle. So this really limits the types of problems that we can study. And so what I like to do, instead of trying to simulate the Earth's mantle, is to just tackle scientific questions, and with a particular scientific question, try to design the simplest experiment I can that can give us understanding of the process. Normally, the question is related to some dynamical process that's occurring. So by making the simplest experiment possible to understand that process, we can take that information and then add it to all of the other information that we know, to put together our story of how the mantle works.

Oliver Strimpel: Can we actually do any experiments in the lab that maybe recreate the pressures and temperatures of the mantle and help nail down one of these things, such as the viscosity or the grain size?

Allen McNamara: Yes, we do. So there's a branch of geophysics called mineral physics. Now, there are a couple issues with that. All of the experiments have to be done at many orders of magnitude higher strain rates than the Earth is convecting at. To do the experiment at Earth deformation rates would just, you know, take way too long, take lifetimes to complete. So we're already a little bit out of the range of reality when we do those experiments, but still, that's the best we can do. And I think the information we get from that is very enlightening. The biggest issue still is we don't understand the grain size. So even if mineral physicists finally came up with the best formula for the viscosity of mantle materials in the Earth as a function of temperature, pressure and grain size and all of that, we're still limited by the fact that we don't know what the grain size is. Well, now, in the upper mantle we have some samples. There are some places where there have been volcanic explosions that rip up part of the top of the upper mantle and bring it to the surface, and so we can see those. I have one of those rocks on my desk at work, and their grain size is around a millimeter or so. But in the lower mantle, we have no idea what the grain size is at all.

Oliver Strimpel: When you say we can't reproduce the strain rates, roughly how fast are these motions? How fast is the strain rate going on in the mantle?

Allen McNamara: So in the Earth's upper mantle, at least, the estimates for strain rate are around 10 to the minus 15 per second. Most people don't deal with strain rates, so that's not a very tangible thing to put in your mind. But keep in mind that the plates are moving at a rate of a few centimeters per year. So for example, North America and Europe are separating at about the same rate as your fingernail is growing, and the minerals are being deformed over millions of years. So the strain can be quite high. But to do a mineral physics experiment where you're straining things at the rate that your fingernail's growing, it can take a long time to come up with enough strain that you would need to put together a formula for the viscosity.

Oliver Strimpel: A lot longer than the duration of a typical PhD, I'm sure.

Allen McNamara: Yeah, absolutely.

Oliver Strimpel: You mentioned that the plate motions tells us something about the mantle motions, of the order of centimeters per year. Does the fact that we really actually know the motion of the plate pretty well, from geodesy and so on, does that provide a constraint, or like a top boundary condition, on your mantle models?

Allen McNamara: That's helpful. That's probably one of the most powerful constraints we have to understand convection in the Earth's mantle. It actually provides a number for us that, especially now with GPS, we know with high reliability. So this number of centimeters per year, it's the only place that we have in the Earth's mantle that provides any kind of number that will tell us how vigorously the mantle is convecting. Now, we don't know how that extends downward. For example, we have reasons to think that the lower mantle has a higher viscosity than the upper mantle, so maybe the lower mantle's flowing at much lower than centimeters per year. Just the fact that plates are moving: that was the main motivation to suspect we have mantle convection in the first place.

Oliver Strimpel: OK. I want to ask you about something you mentioned earlier, which is that one of the things we hope to shed light on is what the origin of these large low-shear-velocity provinces might be. Are they primordial, or are they somehow created and sustained by ongoing processes? What's the research on that, and how might one distinguish these two ideas?

Allen McNamara: There are two very different hypotheses as to the formation; both may be occurring. So one idea is that the large low-shear-velocity provinces are remnants of something more ancient. So perhaps after the Earth formed, either during the differentiation of the Earth or through processes that occurred in those first billion years, when the Earth was undergoing a lot of changes, perhaps a layer, a more dense layer, formed in the Earth. And through time, that layer is slowly being entrained into the upper layer, and as a result, the density contrast between the two layers gets smaller, and that layer ends up eroding. What we see now is just the remnants, or pieces, of what was something much larger. And this is kind of motivated by that trace element geochemistry that we see at hotspots, which indicates that the source material for those basalts is something that's a little bit more primitive, something that looks a little bit more like what we think the original Earth looked like, the original mantle looked like.

Now, another idea, which is also supported by the geochemistry, by the way, is that these large low-shear-velocity provinces are created by the subduction of basaltic oceanic crust through time. So we know that the surface is always subducting into the mantle, sinking into the mantle, and that layer of crust on the surface of the ocean floor, which is about six to seven kilometers thick, we think becomes more dense at mantle pressures. And this is based on mineral physics experiments that have been done in the laboratory: you take a basalt and put it under high pressure, it will change its structure a little bit and become more dense than the rest of the surrounding mantle. And so they could be the accumulations of billions of years of this oceanic crust. So it's a big question for us as to whether these large low-shear-velocity provinces are being created or they're being destroyed.

Oliver Strimpel: One idea that you mentioned when we were talking earlier is that if you have, as a consequence of, say, one of your models, that you find that these large low-shear-velocity regions would be throwing up big plumes to create hotspots, then isn't there a challenge of how that, if it's a primordial feature, how that could be sustained?

Allen McNamara: Absolutely. First, let me back up and say the mantle is a very viscous place. It's very different from a lava lamp that you may have. In a lava lamp, the two materials have a very different viscosity, and they don't stir together very well. The Earth's mantle is very different, because it's so viscous: it stirs extremely well. So it's very hard to have different compositions rising up and not getting stirred with the rest of the mantle at the same time. So if these large low-shear-velocity provinces are rising up through the mantle, as some think that they are, then it's very difficult for that process to occur for four and a half billion years. So that would argue for a hypothesis where these things are accumulating with time.

Oliver Strimpel: For my last question: if you had access to a totally unlimited pot of research funds, how would you spend it?

Allen McNamara: Definitely ocean-bottom seismometers. The Earth's surface is only 30% land, and that's where most of the seismometers are. There are islands around the oceans where there are seismometers as well, so I mean, there is some coverage in the ocean. I think the best thing we can do is just get a better image of what's going on down there. Each new advance from seismology radically changes our ideas of what's going on. If money wasn't an issue, just more of those, to improve our coverage. Let's get better pictures of what's going on in the lower mantle.

Oliver Strimpel: Allen McNamara, thank you very much.

Allen McNamara: Thank you, Oliver. It's been fun.

Oliver Strimpel: For more about Geology Bites, as well as pictures and diagrams that illustrate this podcast, you can go to geologybites.com.