Geoff Abers on Subduction Zones and the Geological Water Cycle
Transcript
Oliver Strimpel
This is Geology Bites with Oliver Strimpel. Subduction zones are places where a slab of oceanic lithosphere plunges down into the mantle below. The slab consists of sediments on top, crustal rocks in the middle, and the lithospheric mantle on the bottom, all plunging down together as a kind of sandwich. In each of these layers is an ingredient that plays a key role in shaping the evolution of the Earth over geological time, and that is water. Geoff Abers has conducted extensive research on water and subduction zones. He uses seismic observations to map the distribution of water in subducting plates and in the overriding mantle. He couples these observations with computer-based models of the physics and chemistry of the subducting plates. He is professor of Geological Sciences at Cornell University. Geoff Abers, welcome to Geology Bites.
Geoff Abers
Thank you; happy to be here.
Oliver Strimpel
I just gave a very high-level summary of what's inside the subducting oceanic slab. Before we talk about water in the slab, can you flesh out the basic structure and composition of subducting oceanic slabs a bit more for us?
Geoff Abers
Sure, almost everything that subducts is oceanic lithosphere formed at a mid-ocean ridge, so that's where the story starts. As two plates pull apart, mantle rises up, rises above the melting temperature, and some of it melts. Those melts are basaltic in composition, and they form the oceanic crust, about 5,6,7 kilometers thick. The top of those basalts erupts on the sea floor makes these pillow basalts. Maybe you see movies of these things bubbling around; they're full of cracks and pores, and they interact at very high temperatures with sea water. The sea water gets in these cracks and pores and reacts to form alteration minerals, minerals that have a lot of water in their structure. Deeper in the oceanic crust, the same basaltic material cools much more slowly without a lot of water present, and we call it a gabbro, and below that is the mantle that’s left over from the melting at the peridotite rock is dominantly olivine. So that's the package that forms at the mid-ocean ridge; as it moves away from the ridge, it cools, and the sediments get deposited. They have a lot of minerals. They have a lot of water in their mineral structure. Things like clays and so forth.
Oliver Strimpel
How do we know what the structure of that oceanic plate is?
Geoff Abers
Well, there's a few different clues. If we go anywhere in the world into any ocean basin, the depth to the sea floor and the heat flow, the rate the heat comes out, is like the same everywhere. And that's telling us that this basic process is very similar globally. The same package is getting delivered the same 5,6,7 kilometers of crust. The shallow parts we've drilled lots of places. The ocean drilling program samples the sediments and the tops of the basalts. One or two places, we've drilled down to the top of the gabbro layers. We have not drilled through the whole crust anywhere, so most of what we know about the deeper parts of the section comes from exposed rocks we call ophiolites. These are slivers of oceanic crust that’s thrust on the continents through various geologic accidents. And all of this is tied together by a tremendous amount of work done over the last several decades measure it the wave speeds of seismic waves, usually through some controlled source. Back in the ‘50s and the ‘60s, people threw dynamite off of boats and recorded on hydrophones or seismometers at various distances. More recently, there are more controlled ways of creating the sound source. But those signals propagate for 10s or hundreds of kilometers, show very distinct changes in wave speed as we go down through the sediments and the top of the basaltic layer in through the gab rows and then the big step in velocities to the faster mantle rocks. We can also tie these wave speeds very carefully to experiments and labs done on minerals and on rock samples to see what kinds of compositions correspond to what kind of wave speeds. And that gives us also some of the first hints of the effects of water. Water, when it reacts with most rocks, will form other minerals that have sort of the same bulk composition, plus some water added. Those minerals typically have slower wave speeds, and so this gives us another tool for calibrating the extent of, say, the alteration in the basalts. Or if any water gets down deeper, we can fingerprint that as well.
Oliver Strimpel
OK, so that's how we can tell something about what's in this lab as it forms the ocean floor and moves away from the mid-ocean ridge towards the subduction zone. But after it enters the subduction zone and it starts to go deeper, can we still get any picture of what the structure is? How do we look at it once it actually starts to dive down to any depth?
Geoff Abers
Well, a lot of the same tools exist for seeing deeper, and, again, some of the main tools come from looking at seismic wave propagation. The speeds that waves travel that oceanic crustal section and see if they change, say, as the plate heats up. And minerals would hold water in their structure breakdown. Those are usually unstable at high temperatures, and as you get deeper, people sample water coming out of seeps in the ground or ultimately get to the volcanic arc of the lavas that are coming out of that to provide hints of what's gone down and come back up again. And it's a very clear fingerprint of seductions of sediment, for example.
Oliver Strimpel
Let's talk about our theme today, which is the geological water cycle.
What do you mean when you talk about the geological water cycle as distinct from the water cycle of the atmosphere and oceans that most people are much more familiar with?
Geoff Abers
Yeah, that's a great question. Most of the water that we know about sits at the Earth's surface in the oceans and rivers and the atmosphere in ice, and it moves around, you know, fairly quickly between these different reservoirs. But there's this other cycle where water is exchanged between the surface reservoir and the mantle, the deep interior. Water comes out in tiny, tiny quantities and measured in, you know, 10s of hundreds of parts per million through mid-ocean ridges and hot spots and other sources of volcanoes that tap deep in the earth, and then returns to the mantle primarily in subduction zones. And so this deep water cycle and is regulated at plate tectonic speed so, you know, centimeters, 10s of centimeters a year, is a cycle between the surface water that we see and the water that's in the mantle.
Oliver Strimpel
OK, so can you step through what happens to the water in the subducting oceanic slab after it enters the trench?
Geoff Abers
Some of the water comes out pretty quickly, you know; water that’s sitting in cracks and pores in the shallow part of the crust gets squeezed in the upper few kilometers as it's going down, and you can see that coming out in seeps and trenches. But then a lot of the water is bound in minerals that are attached to the ongoing plate and stays in them until these minerals reach conditions where they're unstable. And this is true of nearly all hydrous minerals. You heat them up enough and you squeeze them enough, water is released. The big flux of that water coming out that we know about is into the mantle sitting underneath volcanic arcs, and so that water will come out in volcanoes. And volcanoes have a few weight percent water in the magmas that are coming out of the mantle that feed them. Those two things, what gets squeezed out of the trench that comes on the arc probably counts about 2/3 of the water coming into the trench. The rest of it seems to be able to stay in the down-going slab to go into the deeper parts of the mantle, where it can feed much longer-scale geologic processes, long-term water cycle.
Oliver Strimpel
Can you connect the dots a bit between the release of water from this downgoing subducting slab and the water that's spewing out into the atmosphere from a volcanic eruption?
Geoff Abers
Yeah, so the plate that's descending, the top of it is the stuff that has the sediments and the oceanic basalts that have a lot of water bound in it. That top surface at some point is going to encounter the hot flowing part of the mantle, they call the mantle wedge, where sort of hot material is coming in and then getting dragged down by the plate. As it starts to heat up, the minerals that hold the water in it will start to break down. That water is released. That water, then, because it's much more buoyant, rises and, we call, flexes the mantle up above it. Adding a little bit of water to a hot rock will lower its melting temperature, and so the idea is that if it's hot enough, you get enough water, and that'll actually trigger melting in the mantle above the downgoing plate. Then those melts then rise more or less vertically to form the volcanic arcs that we see all around the world, and this is a key piece that ties subduction, oceanic plates going into the trench, with this parallel chain of volcanoes that we see everywhere around the world.
Oliver Strimpel
In my introduction, I said that you couple your seismic observations with computer-based models of the physics and chemistry of subducting plates. What exactly do the models try to replicate?
Geoff Abers
My main colleague in this, Peter Vankin at Carnegie Institution, is doing all the computer modeling. What we've been working on for the last 10 or 20 years has been doing largely 2-dimensional models of the thermal structure and flow fields of different sections of subduction zones around the world, where we try to pattern inputs and geometries as closely as we can to real places, and then there’s a few different things we can do. One is is to make them predictions of the thermal structure of the arc, the temperatures. But then we put in a model of the rock composition of the mineralogy and petrology. Look at what minerals are stable where given those temperatures and pressures, how much water they hold, and then make predictions from that of where we would expect water to be released at different depths, both going down the surface of the slab and then within the plate. So the models then give us predictions of water storage and water release as well, and different subduction zones.
Oliver Strimpel
And what do the results predict?
Geoff Abers
There's a few different things. At shallow depths, because you're bringing very cold oceanic lithosphere fairly quickly underneath some overriding upper plate, you're effectively cooling it. It's sort of refrigerating upper plate, and so temperatures inside the down-going plate can stay pretty cold for several 10s of kilometers’ depth at least. Probably at around 70 or 80 kilometers, this plate starts to encounter the hot flowing part of the wedge as it goes down. In the deeper parts, it drags down through viscous flow of the mantle, and then hot mantle as it comes in to replace it. So the first thing that we see is that at this depth of around 70-80 kilometers, the top surface of the plate undergoes a fairly rapid heating pulse. That heating pulse is big enough that we think in most cases the water that's bound in minerals in subducting sediment, and probably the top of the oceanic crust, is released there. And that's really the water that's available to feed the oceanic volcanic arc. The volcanic arc is just, you know, the slabs are little bit deeper, maybe 100, 110, 120 kilometers deep when the volcanoes appear, so we think this is there’s a tight connection between this thermal structure and where the volcanic arc is. That's the other thing that we see, is that if water were to get deeper into the downgoing plate, and there’s some hints, just seaward of trenches, that this happens, water gets through the oceanic crust into the oceanic mantle. That water can stay bound in minerals like serpentines for quite a ways deeper than the plate at those depths doesn't necessarily heat up to the point where serpentines break down. That’s temperatures like 600 and 650 degrees C. And so that's a pathway for water to bypass the volcanic arc and go into the deeper part of the earth where it can be stored for much longer.
Oliver Strimpel
OK, so the models help you predict how much of it gets recycled at a fairly shallow depth and then goes to feed the volcanoes and spews out in the atmosphere again. And how much of it might actually just stick around in the slab for long enough in the form of minerals like Serpentine and then actually wind up going into the mantle. .
Geoff Abers
Yeah. One thing that we see, though, is that given the range of ages of incoming plate, which have to do with sort of how thick the cold part is and how long it takes to heat up, how fast they're descending, the geometries. There’s actually an incredible variability in some subduction zones. The water boils off very quickly. There's really not much room for water to stay in the subducting plate, past the volcanic arc. This would include places like, say, where the Wanda Fuca plate off of Oregon and Washington, British Columbia, is subducting underneath the Cascades.. There, that plate is maybe only 8,000,000 years old when it hits the trench, and it's descending very slowly. So we think there water goes off very quickly. Other places, like Tonga, like northern Japan, like the Marianas, these very old places subduction very quickly. The interiors of those places can stay cold for hundreds of kilometers. And so those become the highways for water to go to the deep Earth. When we add this up globally, right now at present rates, probably about 1/3 of the water that's hitting trenches is making it past the volcanic arc into the deep Earth through these kinds of processes.
Oliver Strimpel
And to get that average, 2/3 boils off, if you like, then goes back up, and one-third goes deeper, you actually put real-world parameters for the temperatures and the thicknesses. How many subduction zones did you put in parameters for?
Geoff Abers
We've gone around the world to sort of 55 or 60 segments of subduction zones trying to look at every 500-ish kilometers of Arc segments around the world where we think we know the geometry and the seduction rate, and something about how much sediment is actually going in. I should say the one-third, two-third’s number, there’s a lot of uncertainty. We're assuming a certain amount of water is coming in the mantle of the incoming plate. We’ve seen in a few places, the first place was off of Nicaragua, and we've seen since off of parts of Alaska and some other subduction zones, where it looks like, as the plate starts to bend before it reaches the trench or 50-100 kilometers seaward of the trench, that bending makes faults that seem to act as conduits for water to get into the mantle, and so we see seismic wave speeds at the very top of the mantle should start to slow down before the plate hits the trench, and we think that's a fingerprint of water getting into the mantle, reacting to form serpentines probably. But we don't really know how deep that goes, how many places that goes. It's clearly places that does not happen, and because the mantle part of the system is very big. Also, because those minerals hold a lot of water as Serpentine or something like 14 weight percent water by mass. Not really knowing how much water is in that subducting mantle gives a really big uncertainty into the system.
Oliver Strimpel
Now that you've got this model, how can you so-called ‘ground truth’ it?
Geoff Abers
Two main ways we've been doing this. One observation that we see is a layer that looks like oceanic crust that has wave speeds that look similar to those that we see on the sea floor, but underneath about 100 kilometers down, and then this goes away, and we think where it's going away is where these dehydration reactions occur, where the crust reacts to form much denser, higher-speed minerals. And we made these observations using a few different kinds of seismic waves in thin layers, so think we can see the depth distribution of water release. And, for instance, in Cascadia, which is this very hot place, we think we see this happening in much shallower depths than we do, say, under Alaska, where the incoming plate is colder, subducting faster, and able to keep the water in the plate down deeper. The second test that we've been able to do is looking not at the speed of the waves, but sort of the amplitudes of the waves are more precisely how the energy in the waves is being absorbed as it travels through the rock. The absorption, we call seismic attenuation, is a fairly good function of temperature, as the temperatures especially approach the melting temperature of the rocks. And what we observe is waves propagate very efficiently. Very little absorption in the downgoing plate and in the upper plate when the downgoing plate is less than about 70 or 80 kilometers deep, and very quickly underneath the volcanic arcs enters this very high attenuation zone where we attribute that to much higher temperatures. So this gives us a little bit more of a direct handle of testing the thermal models themselves that are underlying the mineralogy and the water release. We also look at the depths of earthquakes. It turns out there’s a lot of earthquakes inside these downgoing plates. In fact, this is how they're first recognized. There's a lot of arguments to suggest that earthquakes are probably in some way due to the water release, either directly or maybe indirectly through increasingly poor pressure because the water is suddenly releasing the system. And we've also seen good correlations between the depths and the geometry of where earthquakes are and where the thermal models predict the water really should happen.
Oliver Strimpel
So when you make this comparison, then, for your population of 50-odd subduction zones, generally speaking, you're able to reproduce the seismic wave speeds and the attenuation that's observed by using reasonable parameters. The reason I'm pushing on this is there are those people who are very skeptical of models, that you have enough parameters. So whatever you see you can put together a parameter set that will predict that, and then the question is, have you learnt anything?
Geoff Abers
Yeah, it's a great point. Our philosophy has been to try to keep the modeling no more complicated than it absolutely has to be to explain these observations. So keeping this in two dimensions, we're tying the geometry so things we know about we can see from the seismicity, about where the plates go. We know how fast the plates go. We know a lot about these upper plates. There are these complicated transitions, though. The shallow part of the system is essentially a fault with one rigid plate going underneath another one. As you get deeper, as things warm up, that fault transitions into probably a ductile shear zone. And we see these things exposed on Earth everywhere we see deep rocks exposed, and then it transitions to this hot flowing viscous mantle wedge. These kinds of transitions are very hard to replicate without at least a little bit of numerical equation-solving. You could make different decisions, but I don't think you could replicate this whole suite of observations that we’re talking about with purely simple analytical models to get there. You can certainly make the model more complicated and there are people who do. There are people make up all sorts of other processes of blobs that fly around, and those, I suppose, help you imagine what might be down there. But I feel like we're at a point where these models are just complicated enough to capture the essence of the first order of things that are happening here.
Oliver Strimpel
Let's come back to the geological water cycle. What are we talking about in terms of quantities of water that we can subduct over geological time?
Geoff Abers
Yeah, well, of course this is a slow process. This isn't something that happens overnight, so our numbers suggest that at seeing about an ocean’s worth of water subducting in 1 billion, maybe 1.3 billion years. Assuming everything stays steady with today's rates, which is a big if. If 2/3 of this is coming out in arcs and four arcs, that means over the age of the earth, we've probably subducted one ocean’s worth of water into the mantle.
Oliver Strimpel
Wow, so is there more water in the mantle or more water in the oceans?
Geoff Abers
There's probably more water in the mantle than in the oceans right now. There are things that we know from the very small amounts of water that do come out at mid-ocean ridges and tell us about the water in the uppermost mantle. There's more water that comes out in hot spots like Hawaii. These give us some clues, and there's a bunch of other geochemical arguments, but numbers that geochemists who look at these direct or indirect proxies for melting are somewhere between one and five or ten oceans’ worth of water are sitting somewhere in the mantle. Where exactly that sits is a good question. A lot of water probably sits not in these hydrous minerals, things like micas and clays and serpentines, but as hydrogen as defects in nominally anhydrous minerals, you can hold at high pressure has a fair bit of hydrogen, say in an olivine crystalline structure. The transition zone between about 410 and 670 kilometers depth has minerals that seem to be very good at holding a lot of water in their structures, so a lot of people think that’s probably the biggest sponge in the mantle to hold a lot of this water.
Oliver Strimpel
Just to clarify what you said earlier about observational sampling of the mantle, you said at mid-ocean ridges, we see the water that might be coming from relatively shallow in the mantle because that's where we think the material comes from. But at hot spots where sampling much deeper mantle, potentially, according to some people, all the way down to the core mantle boundary. So was that the point you were making, that we can actually look at these two different geological structures and we can therefore get a handle on what water might be in the shallow and the deep mantle?
Geoff Abers
That's right. In mid-ocean-ridge basalts, around the global ridge system, somewhere between 50 and 200 parts per million by weight of those rocks are probably water, and the lavas that come out, and that tells us about the water in the source region. There seems to be more in hot spots. And this is one of the lines of evidence that people have used to suggest that the hot-spot source region may be over geologic time fed by subduction, because we're bringing all this water down in subduction zones. It seems to be one of the main conveyors. There is, of course, debate about this because these are very deep processes that are very hard to fingerprint. But that's at least a self-consistent model for this very long-term geological water cycle and the return flow.
Oliver Strimpel
I'm intrigued by the fact that over geological time you're talking about subducting and essentially locking up in the mantle reservoir of the order of the same amount of water that we have in the oceans today. So would it be the case that if plate tectonics was just a bit more vigorous, we might have subducted away the whole ocean and we might have a totally dry planet?
Geoff Abers
That's an interesting question, what happens as you go back in geologic time, or even, you know, slightly different scenarios in terms of what plates are subducting when. Right now we're in the setting where plates are on average older than they were 100 million years ago when they were subducting, just because of how continents have moved around and how ocean basins have formed. So even over a couple 100 million years, there may be some pretty big variations in how much water is getting into the mantle. There are a bunch of competing effects, though, so if you subduct things faster, you of course are bringing your water into the system, but you're also then spreading at ridges faster, and the plates that hit the trenches are younger, so they're going to be hotter. And if they're hotter, there's less room in them to hold the water, so they may start with less water going in, So it's not really clear if just speeding up plate tectonics is going to increase or decrease this net water flux. Secular cooling of the Earth over geologic time is also another important factor that's coupled to how fast plate tectonics works. Some people who have done this modeling suggest that the ability of the subducting mantle lithosphere to hold water and serpentines, so again, has to be colder than about 600 degrees C before it subducts. This is a relatively recent phenomenon over the four and a half billion years of Earth history, and so it wasn't really until that started you started to drag down a lot of water. The key observation here, though, is, from what we can tell, sea level hasn't really varied a lot. You know, maybe tens or maybe a couple 100 meters over as much of geologic time as we can tell anything. Ocean basins are about four kilometers deep, so on average this balance hasn't radically changed.
Oliver Strimpel
Interesting question is whether we actually have a stable equilibrium there.
Geoff Abers
Yeah, I mean, it's hard to think of what a proper feedback mechanism would be. There's a lot of feedback mechanisms in the surface water cycle that help regulate various parts of it. The very slow rates of things going into the mantle and what happens at subduction zones versus what's happening at the outgassing at ridges and plumes. It's not so obvious why or how these things could talk to each other and there could well be an imbalance over time that is changing. We just don't know this, so I think I'd be a little bit cautious thinking that there is actually a feedback mechanism in place, although maybe it's there and we just haven't figured out how that might work.
Oliver Strimpel
What are you working on at the moment?
Geoff Abers
Lately, I've been very interested in taking the same kinds of approaches we've been using to look at this sort of larger-scale water cycle to look more closely at the volcanoes themselves, and thinking about, can we see better the magmas where they actually form in the mantle that flux through the crust to create the volcanic eruptions? Does this process happen steadily or episodically? Are there observations that you can make in the deeper parts of the crust or of the mantle that more directly tie to eruptive cycles in volcanos? Can we use these kinds of methods and tools, deeper sizing, imaging, to better predict volcano behavior?
Oliver Strimpel
That could be extremely useful if it gets as far as even hazard prediction.
Geoff Abers
Yeah, that's a lot of the motivation, as we're understanding how these bigger systems work. Then can we use that understanding in some more useful ways to worry about volcanic hazards? Understand also better things like core formation, other aspects of things that volcanoes do that are maybe a little bit more directly applicable to human existence.
Oliver Strimpel
Geoff Abers, thank you very much.
Geoff Abers
Thank you, it's been wonderful chatting.
Oliver Strimpel
For more about Geology Bites, as well as pictures and illustrations that support this podcast, go to geologybites.com.