This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites. I'm Oliver Strimpel, a former astrophysicist and museum director with an endless curiosity about the geology of our extraordinary planet. We've all seen images of volcanoes erupting, mostly the ones that erupt frequently, such as those in Hawaii or Indonesia. But others lie dormant for centuries, millennia or even longer before erupting, often cataclysmically. So what lies behind this enormous disparity? Steve Sparks is Professor of Earth Sciences at Bristol University. His work applies the physics of fluid motion to the magma and rocks below volcanoes. In recent years, this work has fundamentally changed our understanding of how and when volcanoes erupt. Steve Sparks, welcome to Geology Bites.

Steve Sparks: Well, thank you very much, Oliver, for inviting me. I'm very pleased to be here and to talk about my favorite subject, volcanoes.

Oliver Strimpel: We know that as you go down below the surface of the Earth, things get hotter. So I have a naive picture in my mind that at some depth it becomes hot enough to melt the rocks, so there's a pool or layer of liquid magma sitting below, and a volcano is a place where the magma somehow manages to force its way up to the surface.

Steve Sparks: That, Oliver, is a very good first-order picture of a volcano. It's a place where melts from deep within the Earth accumulate and then erupt. If you go down beneath the volcano, what you're likely to encounter is a lot of very hot rock. In some places, there will be regions of rock which have been infiltrated by liquid rock, melt, and in some places there will even be pools of almost pure melt around. So that's a pretty good picture. And the volcano is really the story of how that melt gets from deep within the Earth, typically from around 100 kilometers depth or even more, and how that melt manages to work its way up to the surface and erupt.

Oliver Strimpel: So does that naive picture of a vat of liquid magma sitting down there and then bursting up to the surface from time to time explain what we see?

Steve Sparks: To some extent, but unfortunately not entirely. We now know that that's not a really accurate picture of how a volcano works and what the insides of a volcano look like. Probably around the end of the 19th century, people did indeed start to think that volcanoes consisted of vats of almost pure molten rock. And as the 20th century moved on, it became apparent that these vats of molten rock might be at only a few kilometers depth. Unfortunately, we now know from really detailed studies of minerals, and of the structure underneath volcanoes using geophysical methods, that that really can't be the whole story. In fact, those vats probably only exist for relatively short periods of time. Most of the time, if you went beneath a volcano, you wouldn't actually find very much except very hot rock, with melt infiltrated within the pore spaces of the rock.

Oliver Strimpel: So is that infiltrated melt part of the new picture of what's going on?

Steve Sparks: Exactly. Geologists have a name for this: mush. I think a good analogy is if you go to a beach on a dry, sunny day, and you have your sand, and you pour some seawater onto the top of it. Of course, the water soaks into the sand and infiltrates between the grains of sand. That's a pretty good picture of what most of it will look like underneath a volcano. Most of the time, there will be rock with crystals, but there will be little spaces between the crystals, and that's where the liquid rock spends most of its time.

Oliver Strimpel: So where does this liquid, the analog to the seawater being poured onto the sand, if you like, come from?

Steve Sparks: Well, that comes from processes going on much, much deeper down. As I mentioned earlier, the sources of many volcanoes are probably something like 100 kilometers below the surface, and that's where the interior rocks of the Earth melt. They melt to form, again, little tiny films of liquid in between crystals, and with time, that liquid will work its way out. It's lighter than the surrounding crystals, and just like a balloon going up in the air, that melt wants to rise and separate itself from the crystals. When it does that, it does form bodies of pure liquid, but those are extremely unstable, and they will move up through the rocks rather rapidly and accumulate at a higher level. So if we go down beneath the volcano, say to the region between the surface and about 30 kilometers, that liquid rock, which we call magma, will have come up from 100 kilometers depth, but then it will accumulate in that shallower region and pond. As it does so, because it's relatively cold up near the surface of the Earth, it'll cool. It starts to freeze, and it will start forming crystals. And as it crystallizes, there will be less and less melt present, but it won't get rid of all the melt. So there will be little pockets of melt left behind as this magma freezes in the interior of the Earth.

Oliver Strimpel: Is that ponding, or pooling, that you just talked about a bit like the classical notion of a magma chamber?

Steve Sparks: It isn't a classical magma chamber, because a classical magma chamber is one where you've got a large volume of liquid and really very few crystals; in fact, it's largely liquid rock, in the classical terms. Most of the time, the melt which is eventually going to erupt will be just like those grains of sand and water at the seaside: the melt will be in between the tiny crystals, on a very small scale. But those regions which contain melt are vast. There are people in the United States who have used a process called tomography, rather like a CT scan in medicine, where you use seismic waves to find out what's underneath. And what they find is that there are absolutely gigantic amounts of such rock, with tiny amounts of melt in it, underneath volcanoes like Yellowstone or Mount St. Helens, on a really vast scale, right from near the surface down to the bottom of the crust, at 30 kilometers or more. So there's a lot of this liquid, but it's all infiltrated in this mush.

Oliver Strimpel: It comes out somehow.

Steve Sparks: Yes, it does. It likes to come out; it likes to separate. It's not really an exact analogy, but if you have a pint of beer with tiny bubbles in it, then when you first pour the beer, the bubbles will be dispersed through the liquid, but the bubbles will rise up because they're lighter. It's somewhat similar to that: you've got liquid which is lighter than its surroundings, and it wants to move upwards and leave the dense crystals behind. That's essentially what happens. But liquid rock is very, very viscous. Of course, you can imagine putting some honey in your freezer overnight: it'll become very stiff and difficult to flow. We call that high viscosity. And so that process of the melt getting out of the rock is extremely slow. It may take hundreds of years or thousands of years, in some cases tens of thousands or even hundreds of thousands of years, for the melt to make its way out of the rock and separate from it.

Oliver Strimpel: So after that separation, which, going back to your two analogies, would correspond to the water separating out from the sand, or the bubbles rising to the surface of the beer, something really interesting happens, which I think is a big piece of the explanation of what we see at the surface. You did some experiments to try and determine what that is.

Steve Sparks: We did, with a colleague of mine, Alison Rust, and a very bright master's student, Guillaume Seropian. We did some experiments which looked at what happens after that liquid has extracted itself from the rock. If you imagine a column of this rock with dispersed liquid in it, over very long periods of time that liquid will squeeze out, move up to the top of the column and start forming a pure layer of liquid. And that's becoming much more like what people imagined in the 19th century, the layer of pure liquid. Now, that layer of pure liquid is lighter than the rock above it, and that's an unstable situation. We did some experiments which looked at how long it would take that liquid layer to essentially separate and move on upwards. The analogy which, I guess, a lot of people will be familiar with is a lava lamp, with two layers of liquid. You've got a thin layer at the bottom of the lava lamp, and when you switch the light on, it gets heated, and the effect of the heating is to make the layer at the bottom of the lava lamp light. Everyone will have seen what happens: the lower layer gets hot and light, and it forms a blob, which moves up to the top. We're thinking of something quite like that, except we're not heating the layer. We're simply looking at separating the layer, which is light, and as it separates and thickens, the forces due to the density contrast, the fact that it's lighter, become greater and greater. At some point it goes unstable, and it then moves up to higher levels, towards the volcano, pretty quickly.

Oliver Strimpel: It must be different from the lava lamp as well, in that these rocks are very dense. So we're not talking about the kind of density contrast that we see in the lava lamp. It's much more extreme, isn't it?

Steve Sparks: It's larger, but it turns out that's not really the important part of the physics. The important part of the physics is that the layer of molten rock that separates has a viscosity, and we've already said that that's pretty high by most standards. It's like honey in the freezer, or pitch, you know, asphalt when you're making a road. It's really sticky stuff; it doesn't flow very fast. But the crust itself, the rocks around it, although they will flow, are extremely viscous. They can be billions of times greater in viscosity, even trillions of times greater in viscosity, than the layer of melt. It's pretty counterintuitive, but crustal rock that seems totally solid to us does in fact behave like a liquid and will flow. But it's a super-stiff, extremely high-viscosity liquid, very, very different from the melt below, and it's that difference that governs how fast this process happens. If you try to get a blob of light liquid to rise through another liquid which is much, much stiffer and more viscous, it's that viscous liquid which controls how fast you can do it. So that's the difference. What we did in the experiments was essentially to test out a theoretical model which predicted how big the blob would be and how often a blob would form. And the main factor is, in fact, the difference in viscosity, as well as, as you've already said, the difference in density between the liquid and the overlying rock.

Oliver Strimpel: I must say, it's really amazing to think that hard crustal rocks can behave like the upper layer of a lava lamp and permit a blob of melt to rise up through them. So what materials did you actually use in these physical experiments to simulate these blobs?

Steve Sparks: We used very simple liquids which are quite viscous, like corn syrup. Sugar syrups are extremely good for this kind of thing, because we can add a bit of water and change the viscosity very easily. And we managed to get very high viscosity ratios by doing the experiments in a relatively narrow tube. Imagine a tube a bit like a lava lamp, really. We just had a cylinder full of a very viscous liquid, we slowly put in the light liquid layer at the bottom, we observed these blobs, and we compared them with the physics theory, and we found it worked extremely well. And we had two bits of luck. One was that we realized that by having a narrow container, we could do experiments with much greater differences between the viscosities of the layers. And the other thing that we realized was that that's what happens in nature underneath a volcano. If you imagine going underneath a volcano from the surface to, say, 30 or 40 kilometers, it's actually a bit like a fat cylinder. So, perhaps more by luck than judgment, we were able to simulate the volcanic situation as well by using a narrow cylinder.

Oliver Strimpel: Did you supplement these real-world physical experiments with computer-based simulations?

Steve Sparks: We did that, and we found extremely good agreement between the predictions of the computer code and our experiments. And because it agreed so well with the experiments that we'd done, we had a lot more confidence that the model was telling us something real.

Oliver Strimpel: Once the experiments, both physical and computer-based, gave you confidence that the mathematical model worked well, you could then apply the model to the real-world situation, by plugging in the enormous viscosity contrasts of a billion or even a trillion.

Steve Sparks: Just enormous. Imagine it again; you can see how it might work. If you've got your honey, and you put a marble or a steel ball on top of the honey, and the honey was warm and quite low in viscosity, then the ball would perhaps sink down to the bottom very quickly. But if you put the same honey in your freezer overnight, and it's now much colder and more viscous, and you put a marble on top again, it would take an awful lot longer to sink down. And so that's telling us that the timescale of volcanoes is governed by this process. If we think about the many volcanoes, let's take Vesuvius as an example, back in the days of Pompeii. When we look at the geological record at Pompeii, what we see, both before and after the famous eruption, is that every several hundred years, the volcano would have an enormous eruption, and those eruptions would last perhaps a day or two. So what we see at many volcanoes is enormously long times in between eruptions, and then suddenly huge amounts of molten rock being erupted all in one go. That's a very typical observation that we want to explain.

Oliver Strimpel: So what accounts for the very long periods of time?

Steve Sparks: There are two things. One, as I've already mentioned, is that it takes a very long time to squeeze that melt out of the mush and form a pure liquid layer. And it takes a long time for the layer of liquid, once it's formed, to go unstable, because it's overlain by this extraordinarily high-viscosity crust. So those are two very long times. But then, once it goes unstable, we find that the magma travels extremely fast to the surface, perhaps in a few weeks or a few days, and once it gets to the surface, it erupts extremely quickly. So we have the elements in this physical model of what we see at most volcanoes, which is very short eruptive events alternating with very long periods of the volcano doing pretty well nothing.

Oliver Strimpel: So does this model also explain why some volcanoes erupt much more often than Vesuvius, maybe every year or so?

Steve Sparks: It does, in general terms. There are some volcanoes, like Hawaii, where the liquid rock is extremely hot down where it's formed, at 100 kilometers depth, and it separates out from the crystals quite easily. Also, Hawaii is a volcano above an enormous source of melted material, so the supply, the flow from depth, is very high. When we have those two ingredients, we get a volcano like Hawaii, which erupts more or less all the time, or practically all the time, and it doesn't take very long to get through the crust from 100 kilometers depth. If we go to a volcano like Mount St. Helens, it's quite a different story. Firstly, the magma underneath Mount St. Helens is produced deeper in the Earth at a much slower rate. So the flow from depth, or the supply, is much, much less than at a volcano like Hawaii, perhaps by a factor of 100. And this magma finds it much more difficult to get up through the crust, and, as I mentioned before, it often gets stuck and intrudes and crystallizes and forms a melt which becomes even more viscous. So the timescales slow right down, because this more viscous liquid has to get out of the mush, and then it has to go unstable, as we've been describing, and that takes much, much longer. So with Mount St. Helens, we're dealing with a volcano which erupts every few decades; the frequency is much less. And the scale of the eruptions when they happen is still quite large, because there's been time to accumulate a lot of liquid rock.

Oliver Strimpel: So what about the volcanoes that erupt only every millennium or so? Are they just an extreme case of what's going on with Mount St. Helens?

Steve Sparks: Well, that would be like Vesuvius. But I think even more intriguing is a volcano like Yellowstone, or some in the Andes, where the time interval between the really big eruptions is of the order of tens or hundreds of thousands of years. In those cases, the magma gets stalled, and it just takes an enormously long time to separate and an enormously long time for the layer of liquid to become unstable, and so we end up with gigantic eruptions. Now, the old theory simply doesn't work to explain what happens at a volcano like Yellowstone, because of what scientists have found by examining tiny crystals. We can tell something about how long that liquid took to get up and form a giant magma chamber before the eruption. And so even though the gaps between Yellowstone eruptions can be hundreds of thousands of years, we know that the magma chamber probably took only a few decades, perhaps a century or two, to form. So we have the same picture: the melt accumulates at depth over a huge length of time, and then, when it does go unstable, there is very rapid behavior, and we get a shallow magma chamber and a huge eruption. So we can't really explain that by the old theory.

Oliver Strimpel: I see. So really, the role of the magma chamber has been demoted a bit, to just a temporary storage area on the way up, once an instability has already developed. But the long-term storage happens in this mush that is much further down.

Steve Sparks: Much further down, exactly. And this was one of the puzzles for quite a while. There was a kind of argument with the geophysicists, who use seismic waves and magnetics and gravity and other methods of imaging the inside of volcanoes, what was going on down at depth, and they had a great deal of difficulty finding classic magma chambers. So the people on the magma chamber side said, well, the geophysicists' images aren't high enough resolution, or they're not taking other things into account. But now we know that these giant magma chambers are actually very ephemeral. That sort of solves the problem, because most of the time, underneath a volcano like Yellowstone, there isn't a giant magma chamber, but there is an enormous volume of rock with liquid dispersed in it and slowly squeezing out. And that's the case at most volcanoes, most of the time.

Oliver Strimpel: For my final question, I'd like to ask you whether this new understanding of the mechanism of volcanic eruptions can help us predict when eruptions will occur and how big they'll be.

Steve Sparks: In principle, it should. We now have a theory. I would contend that before this idea of the mush and the destabilization of the mush came along, we didn't really have a theory of why and when volcanoes erupt, or to predict what the volumes would be. At least, we had theories which explained some observations but not others, so they were partially successful theories. These new ideas, I think, are giving us the basis for doing exactly what you say. But I think we're quite a way from getting to the point where we can really do good forecasting. It's extremely difficult to make measurements down beneath a volcano, for obvious reasons. We usually have to make indirect measurements of one kind or another, and they're not direct measurements of the things we'd like to know. So we've probably got a long way to go before we can start making really good forecasts.

Oliver Strimpel: Steve Sparks, thank you very much.

Steve Sparks: OK, thank you.

Oliver Strimpel: For more about Geology Bites, go to geologybites.com, where you can also find images of the models and experiments described in the podcast.