Patrick Fulton on the 2011 Tōhoku Earthquake
Transcript
Note, transcripts are not fully edited for grammar or spelling.
Oliver Strimpel
This is Geology Bites, with Oliver Strimpel.
We all know about the massive earthquake that occurred in March 2011 off the east coast of Japan. It had a registered magnitude greater than 9.0, making it the most powerful earthquake ever recorded in Japan and the fourth most powerful earthquake in the world since modern record-keeping began in 1900. It triggered a tsunami with waves up to 40 meters that swept ashore, killing nearly 20,000 people and causing the Fukushima nuclear disaster. The earthquake occurred where the Pacific Plate is subducting under the plate that lies beneath the Japanese island of Honshu. During the earthquake, the eastern edge of the overriding plate jumped over 50 meters to the east over a length of about 500 kilometers. It was this unexpectedly large slip that displaced enough water to set the giant tsunami in motion. Why was the slip so large? Was it something to do with the nature of the fault zone on the boundary between the two plates as it nears the Japan trench?
Patrick Fulton led a team that probed the Tōhoku-Oki fault, which is the fault along which the main rupture occurred. His team overcame enormous engineering challenges to install what he calls a temperature observatory right through the fault zone. The observations revealed a temperature peak across the fault, which in turn enabled him to estimate the coefficient of dynamic friction at the plate boundary. And it turned out to be extremely low.
Patrick Fulton is an Assistant Professor in the Department of Earth and Atmospheric Sciences at Cornell University.
Patrick Fulton, welcome to Geology Bites.
Patrick Fulton
Thank you. Thank you for having me.
Oliver Strimpel
Can we start with a bit of tectonic context for the subduction zone off the east coast of Japan? Does Japan sit on its own plate, or is it on the eastern edge of the Eurasian plate?
Patrick Fulton
Japan is actually in a confluence of a number of different plates. The northern part of Japan, where this earthquake happened, actually sits on the North American Plate. And some people will see that the plate goes all the way around from the United States, up through Canada, around from Alaska, and reaches down into the northern part of Japan. Some people call that little part of it in Japan, the Okhotsk plate, kind of a micro plate, and then when it hits Mount Fuji, South of Tokyo, that's where it intersects with the Eurasian Plate.
Oliver Strimpel
And how fast are the plates converging in the region where the earthquake occurred?
Patrick Fulton
Somewhere around 8 centimeters per year.
Oliver Strimpel
I think that must be one of the faster convergence speeds of plates around the planet.
Patrick Fulton
That's right, it's fast, and some of it's because the rocks are old and cold, roughly 130 million years, and the old cold rocks are fairly dense, and it's trying to sink down into the asthenosphere.
Oliver Strimpel
So, the slab pull is especially effective for that reason. So, what exactly happens when an earthquake occurs along a subduction zone?
Patrick Fulton
So, when the downgoing oceanic plates are pulled down from their own weight into the asthenosphere, they get stuck to the overlying plate, and they build up elastic energy, and every once in a while they relieve that elastic energy, and that's what we observe as earthquakes.
Oliver Strimpel
And how deep down does the earthquake normally start — does that stress get released?
Patrick Fulton
So, in subduction zones, about 30 kilometers depth is where we see a lot of earthquake starts. That's kind of the middle of the seismogenic zone.
Oliver Strimpel
And was the Tōhoku earthquake at that depth?
Patrick Fulton
It was. It was about 30 kilometers deep.
Oliver Strimpel
As I said in the introduction, there was a 50-meter slip recorded at the surface, and that was unexpectedly large. What would one normally expect for such an earthquake, and what can we infer from this anomalously large slip?
Patrick Fulton
Yeah, this was a very interesting and confusing earthquake. Generally, we thought that most of the slip would happen where they start. So, at that 30 kilometers depth, you mapped out on a map where the contours of slip were, you would see maybe at the beginning of the earthquake it was somewhere around 5 or 10 meters of slip. And then it would decay out. As the rupture tried to expand, it would essentially hit the brakes of the surrounding rocks and die out. But that was not the case in this earthquake. In this earthquake, most of the slip actually happened at the very shallowest parts. So, where it was roughly 30 meters, a huge amount of slip, down where it started – at the shallowest part, at the sea floor, it jumped as much as 50, 60 meters.
Oliver Strimpel
That reminds me of the podcast I did with Roger Bilham, in which he talked about the earthquake in 2015 in Goroka, in Nepal. And, in that case, the rupture did not propagate to the surface -- the brakes, if you like, were effective. Thus, the stress was built up further along the fault, and everyone's expecting another big quake sometime soon there. So, why do we think this one was so different?
Patrick Fulton
This is really kind of the big question and kind of motivated some of our work. On land, like in the Gorkha earthquake, the rocks are strong, and they are brittle, and they break. But sometimes they don't go all the way to the surface, because they intersect areas where there's a bunch of soft sediments where the rupture kind of will dissipate into different areas or may deform more plastically some of the rocks and sediments above it. Here, we think in subduction zones, the rocks are often softer, especially when they get to the shallow parts, that we almost never have really seen a surface break like this, or at least of this magnitude. And, so, that led to a lot of questions about why the shallow part slipped so much. Was it because the shallow part was really strong and built up a lot of elastic energy? Or was it really weak, and the rupture came through, and it just essentially hydroplaned — didn't have any brakes on the shallow part?
Oliver Strimpel
That actually takes us to your work, which was indeed to probe the fault zone to see if you could gather data that would help us figure out what the friction was along the surface. So, how did you go about this?
Patrick Fulton
One way that we can figure out how much shear resistance or how much brakes were on the fault when it slipped is by measuring how much heat is generated during the earthquake. So, just like when you rub your hands together, it generates heat, and it's a function of how much you move them past each other — the displacement — but also how much force and friction is between your hands as you're shearing it. So, if we could measure how much heat was generated on the fault, we could figure out this big question. And, to do that meant that we would have to drill down, at depth, where the earthquake happened, and measure any temperature anomaly: measure the excess heat across the fault after an earthquake, which is really challenging because that heat could generate really hot temperatures, maybe up to 1,000 degrees on the fault, in a very thin zone, maybe a centimeter thick during the earthquake, but then it diffuses into the surroundings. And so, we figured that we would have to be there within about a year or two afterwards to be able to do that. And this earthquake, with a lot of slip, would give a very big signature. And, because so much of that slip was at shallow depth, that meant that perhaps we could actually drill into it and put sensors down there to actually measure the temperature anomaly across the fault.
Oliver Strimpel
Has anyone successfully measured a temperature anomaly across a fault during or right after an earthquake?
Patrick Fulton
Not quite. There has been a few attempts, particularly after the Jiji earthquake in Taiwan — a large earthquake — but there it took many years before they could drill across the faults, and they could only get one profile of temperature before the hole collapses. So, if you see something across the fault, you don't know whether that little temperature blip is really due to frictional heat. Does it diffuse over time as you would expect? Or, is it due to water flowing through a fracture? Or, is it maybe just due to differences in rock type as you go down? So, we learned a lot from that experience, and we learned that you really need to monitor the temperature over time. And, you really need to get a good understanding of the rocks themselves — take core samples and rock samples. There's also been a long-term study of trying to look for frictional heat, not from a single earthquake, but over millions of years of shearing along the San Andreas Fault. And there you would expect a large increase over many kilometers of heat coming out around the fault zone. And there, it was not observed, despite a lot of attempts, including some of my own, not during my PhD. Which has led us to think that maybe rocks aren't as strong as we think they are, at least, how they behave in the lab. Maybe there is something different during an earthquake or something different about mature plate boundary fault zones.
Oliver Strimpel
That makes it all the more interesting that you actually were successful in determining a temperature anomaly. So, as I mentioned earlier, you went about this by installing what you call a temperature observatory. Tell us about what that is and how you did that.
Patrick Fulton
So, one of the big challenges with this project is that perhaps we could drill across the fault in a place where it had a lot of slip and perhaps measure the temperature across it. But to do that, we were just at the limits of where technologically we could actually drill. We're working in a water depth of seven kilometers, and then the fault zone beneath that was another 820 meters below the sea floor. So, no one had ever drilled that deep within the ocean before, let alone trying to put instruments under there.
So, to do that, we first had to characterize the hole. We drilled and found out exactly where the fault was first. We took core samples out. We did geophysical logging. But now we drill underneath the sea floor, and we put a protective casing at the very shallow part, just so that we could reenter the hole. Then we built the observatory. So, we built a 855-meter-long steel pipe inside of it. We built our temperature observatory, which was essentially a static climbing rope, or a rope that does not stretch much, where we had woven in temperature sensors, each of them with their own data logger, and each of them encased in titanium so all the electronics would not implode due to the huge pressures at those depths. And then the temperature sensors themselves have an accuracy of a thousandth of a degree Celsius. We want to be able to see just the smallest little details and the smallest little increases of what the temperature is, so we then lowered that down to the sea floor and then reentered the hole and installed that bit of casing all the way down, and then left it there and went away for several months and then came back to collect the data. Those temperature sensors, 55 of them at different depths, were measuring the temperature at those depths over time. And, once the hole re-equilibrated, we could see if we saw a temperature anomaly.
Oliver Strimpel
Just before we get to your results, I'm completely boggled by the idea that you can return to a hole that — I don't know, what is it? About a meter across or something — at the depth of seven kilometers of water from a ship that's like floating on the waves above. How on earth is that possible?
Patrick Fulton
Well, it's just barely possible. So, essentially, we have a 850-meter-long observatory of steel pipe connected to another 7 kilometers of drill pipe that bring it down to the sea floor, and we're trying to reenter that into the hole. There is no motor on the bottom of this. All we have is the ship, which can use dynamic positioning to stay in one position on the surface of the water, but down below we have a TV camera that can see only a few meters ahead of it. At the bottom of the whole assembly. And so, with the light on it, we could see where we are. We tell the captain: “Well, we think it might be over to the north,” and he or she will move the ship over there and then the whole thing — all that steel pipe below us is like a wet spaghetti noodle —It's just hanging there. And it slowly starts to wiggle over in that direction in response. And we did that for many hours until we see the Observatory, or the hole, rather, on the sea floor, and a little bit of tubing that sits above the sea floor. And once we're above it, someone says: “Now! Now! Now!”, and then the drillers drop everything into it. And, luckily, it was a hole in one. We got straight in and then continued to lower it down.
Oliver Strimpel
That's absolutely incredible. I want to also know you selected the precise location for the hole, because it seems like you had 850 meters of observatory that you could feed in below the level of the sea floor. The earthquake itself was at 30 kilometers depth, so obviously the zone of the fault plane makes it all the way up gradually to the surface at some point. So, how did you pick a point where the fault plane was at about, it said, 820 meters down?
Patrick Fulton
We needed to find a place where we could drill across the fault at somewhere less than a kilometer deep. If it's too shallow, maybe we start to get to less weight acting on things, and so we wanted to find very clear position of the fault less than a kilometer deep underneath the ground. We also want to find a place where we knew the amount of slip fairly well and that was large. Here is a position where there was clear evidence that it was roughly 50 meters of slip. And, the other kind of constraining thing was we needed to find a place where the water depth was less than 7 kilometers. And the reason for that is: we're going to put our temperature sensors down there, and they're going to record, but we need to get them back at some point in time. There is only one remotely operated vehicle, essentially, an unmanned submarine, that could go to those depths and pull the sensors out, to get our data back out, and its maximum operating depth was seven kilometers.
Oliver Strimpel
So, inside this pipe, and you have your sensors, I think you said 55 of them, how far apart are they down inside that ball?
Patrick Fulton
We want to make sure that we can really see the profile of an anomaly, and so we made our sensors 1 1/2 meters apart from each other, anticipating that the full anomaly may be over some tens of meters wide. And so, most of them are about 1 1/2 meters apart, and then at shallower depths we have some other ones that have further spacing across that allow us to get some information about the background temperature gradient through the Earth.
Oliver Strimpel
So, you then sealed the hole and left these sensors to their own devices for about 9 months. And then you return. So, how did you pick that time interval, and what were you doing during that time with the sensors?
Patrick Fulton
When you first drill the hole, you get all those broken-up pieces of rock out of the hole by circulating cold ocean water, so that drilling process actually cooled off the hole right around the edge of the bore hole for a while, and it would take some time, usually a few weeks to months, to re-equilibrate with the formation temperatures. So, then we went away. We were anxious — wanted to know the answer. But this is still a very active plate-boundary fault zone with lots of huge aftershocks, including a magnitude 7.3 aftershock, pretty much right at this site on a deeper fault below our observatory. And so, we were afraid that, with all these sensors hanging on a rope, if we left it too long, maybe the fault was moving, and it would shear through the observatory and grab hold of the sensors, and no matter how much we tugged on them, we wouldn't get them back. So, when we first went back, we couldn't find the observatory. Some of that is just due to the challenges of navigating in seven kilometers of water depth. You can use GPS in the air, on the ship, but communicating to the ROV underneath you, just the smallest little bits of imprecision in your communication can lead to you being several tens of meters away from where you think. But luckily, some of our Japanese colleagues offered us to use some of their time with the ROV. A few weeks later, we went back, and we were able to find it, and I grabbed hold of it with the ROV, and we pulled it all the way back up to the sea floor and we recovered all of it. The fault had not moved. It had not sheared through everything — which was actually an important observation in itself. It meant that a lot of the observations on land where GPS sensors were showing that Japan was still moving to the east, maybe as much as a millimeter per day, It wasn't because of slip on the fault. It's probably due to movement within the mantle; the mantle recovering, relaxing after having such a big earthquake.
Oliver Strimpel
Well, that's fascinating. So, you're saying the actual landmass of Honshu was moving, but the sea floor was not?
Patrick Fulton
Or part of the sea floor was, but that movement was not between the overlying plate and the downgoing plate. It was between both of those things, and the mantle underneath was churning a little bit after the earthquake.
Oliver Strimpel
So, was the GPS location of the ball actually slightly different when you came back after 9 months?
Patrick Fulton
Well, we don't have a very good GPS location of it. The main issue was that the position was correct, but our communication for where we thought we were was slightly miscalibrated, because of the wavelengths that you use to communicate from the ship to the ROV. It's just a very long path — any little bit of imprecision just adds up. And, if you're 10 … 20 meters away from where you think you are, down there, you can't see very far ahead of you with an ROV, and light doesn't travel that well through the water. And at these depths, this is the deepest, darkest trenches. It is very dark down there.
Oliver Strimpel
OK. Well, let's talk about what the temperature sensors had logged during those 9 months in the borehole and what you found out.
Patrick Fulton
As is the case pretty much everywhere on Earth, the temperature increases as you go deeper. It's roughly about 3 degrees on the sea floor. And then the temperature seems to increase down to about 25 degrees Celsius, kind of room temperature down at the bottom of the observatory. And if we remove that background gradient, we see the anomalous temperatures. At first, it was cold because of that drilling disturbance. But then, we started to see a relatively big temperature anomaly of about 0.3 degrees Celsius, centered right on the fault zone, the same position where we had logged and have all these geophysical images of the fault. We've actually taken a core sample out from the fault itself, right there is where we see a very large temperature anomaly that we interpret to be the frictional heat signature from the magnitude 9 earthquake.
Oliver Strimpel
How wide is the fault zone down there?
Patrick Fulton
Very interesting. It's very thin compared to a lot of other fault zones. So, the fault zone has its main core — it's roughly a couple meters or less than 4 meters thick, and it's filled with dense clay. And then around it there are other subsidiary faults and fractures.
Oliver Strimpel
That's really like a knife edge on these scales that we're talking about. So, you had your temperature signal, and you already said you had to subtract out the thermal gradient associated with the steady state heat flow that's coming up from the mantle to the surface. But were there any other effects that you had to account for in order to really understand the contribution of the earthquake to the temperature anomaly?
Patrick Fulton
We didn't know if that anomaly was due to fluid flow or to thermal properties. We did lots of measurements on those core samples that we got to make sure that that anomaly that we saw could not be due to differences in thermal conductivity or other rock property. We also looked at the data very closely to see if it’s very smooth and does it look like the temperature is diffusing as we would expect, or are there lots of signatures of water moving through open faults and fractures. And, in the plate boundary fault itself, we did not see any indications of that. But we did see water flowing through fractures away from the main fault in other faults and fractures, and that gave us a whole other insight into earthquake mechanics.
Oliver Strimpel
During the nine months that you were logging the temperature down in that borehole, did you see this temperature normally start to diffuse away through thermal conductivity? Did it peter out?
Patrick Fulton
It was hard in that time because we were still seeing a lot of the effects of the drilling disturbance, but generally the shape of it and the relative stable behavior of it through time suggested to us that it was a very conductive signature.
Oliver Strimpel
Just a moment ago, you were saying it was actually quite significant that you managed to pull your whole sensor string out of the hole intact without any of it snagging or getting caught. I assume you meant the possibility that the hole was deformed during those nine months.
Patrick Fulton
The hole could have been sheared across. The fault could have still been moving, and we didn't see that. Some of the other things that we measured while we were down there, before we installed the observatory, was to take the resistivity image 3-D 360-degree image of the walls, and we could see how the tectonic stresses were deforming that hole after we drilled it. So, when we drilled it, it was a cylinder. But we could see from those images that, in some places, the tectonic stresses were forcing it, a little bit, to collapse in certain places, where the hole was becoming more oval-shaped. And that told us about the magnitude of the stresses down near the fault, beyond just what we see in the temperature. That told us what the stresses were during the earthquake. So, after the earthquake, we could see that the amount of shear stress on the fault was essentially zero. So, not only did this fault slip and have no brakes on it, and that the stress on the fault went to a value that was very small, that, after the earthquake, at least in the year afterwards, had not built up a lot of stress.
Oliver Strimpel
That's interesting. So, the coefficient of friction was very low and at the same time, all the stress was relieved, perhaps because of that in part. So, did you manage to get some quantitative results from this that you were able to put in perspective with the parameters that we've observed in other faults around the world?
Patrick Fulton
Yeah, I guess I jumped ahead of myself there. And so, I told you that we saw this 0.3 degrees Celsius temperature anomaly, but I didn't tell you what it meant. And so, we could use that heat anomaly and the width of it to constrain — and with the thermal properties of the rocks down there — to constrain how much heat must have been deposited on the fault during the earthquake. And I had said that the fault zone was about two meters thick, but there are, in that two meters thick, there are lots, thousands, tens of thousands of little sip surfaces from probably tens of thousands of past earthquakes across that fault, And so each of them are as thin as even a few millimeters. And so, most of the slip can be in a very thin zone, or it could be distributed across those two meters. It's unknown… And it's not really that relevant for the calculation we do. But what we found, by figuring out how much heat energy was on the fault, that heat energy per unit area related to the shear stress or the shear resistance, times displacement — we take that, divide it by 50 meters of slip, we get the average shear resistance on the fault. We know that the average shear resistance is related to the friction coefficient times the normal stress, which is related to the weight of the rocks. We calculate the weight of the rocks based on their density. And we end up with a friction coefficient that is 0.08, which is outrageously small. Most rocks have a friction coefficient of 0.6 to 0.8. So, it has told us that the fault had no brakes, for very little resistance to slip once the rupture got into this shallow part.
Oliver Strimpel
How might we explain such a very low coefficient of friction?
Patrick Fulton
Those sediments in the fault zone that are really slippery — they already have a low friction coefficient, lots smaller than most rocks in general. And perhaps a combination of the rupture going through them, and other effects associated with the geometry of the fault being very shallow or very horizontally-dipping, together, may have allowed the rupture to find a weak plane. Instead of diffusing out or rupturing into hard rocks, it had found this easy pathway to just continue to slip all the way to the sea floor.
Oliver Strimpel
So, you mention clay. I guess clay is a sort of lubricant as compared to, say, other materials like more silica-rich materials that many sediments have formed; is that right?
Patrick Fulton
Yeah. In the lab, granular materials, sands, or crystalline materials, have very high friction coefficients of around 0.6 — 0.8 when you put them under stress. But clays are interesting. Their material structure is different. They're platy minerals. They tend to be a bit weaker in general, and these particular clays are pure smectite clay, which is largely ash — not a lot of sediments from rivers. I said these are some of the oldest oceanic rocks, 130 million years old or older. They were probably deposited out in the middle of the ocean, far away from any land, at such great depths that all the animal life has dissolved, when they get to these great depths, where they're deposited. So, pretty much all of this material, this sediment, is ash from volcanoes that have gone through the atmosphere and found their way to the bottom of the ocean.
Oliver Strimpel
So, ash from volcanoes is not in itself necessarily clay-rich, right? It's what happens once it goes down at depth. And you're below the depth at which carbon will redissolve. So, no carbonates are intact down there. Is that what happens?
Patrick Fulton
That's right. These soft particles and ash, when they go down, and they are exposed to water, and they can breakdown into these clay minerals.
Oliver Strimpel
So, obviously an earthquake like this one has enormous destructive potential… Do these studies have practical implications for earthquake preparedness, or maybe even for prospecting for economically important minerals that are deposited by the kinds of transient hydrothermal fluid flows that you saw further up in your borehole observatory?
Patrick Fulton
So, we saw that other faults and fractures would respond to aftershocks. Pulses of fluid flow through them. They were opening up, and that's a very similar process to how we think a lot of ore-minerals are precipitated. That's also another process that can change the fluid pressure or change the stress state as the fluids move through. They can redistribute fluid pressures which affect stress state, and those changes in stresses can trigger subsequent earthquakes. So, that tells us a lot about the processes that can affect aftershocks. But just in general, understanding something about our initial question: was the shallow part locked and building up a lot of stress, really strong? Or was it really weak, and just sit-along for the ride? That can have direct and important implications for earthquake and tsunami hazard preparedness.
Oliver Strimpel
Perhaps one of the more analogous parts of the world would be the Pacific Northwest coast of the United States and Cascadia, where people talk about a huge earthquake with tsunami potential waiting to happen. Is that a similar situation there?
Patrick Fulton
It is a similar situation. This is a place that also, like the northern part of Japan, has not had a lot of earthquakes in the modern era. People had thought that Cascadia doesn't have big earthquakes. It's just quiet. Maybe it's just always gradually sliding, not building up any stress. But Native American stories and also now paleo seismological records, evidence in the rock record of past tsunamis, that it does have big earthquakes, including a magnitude 9 earthquake in January 1700. There, the geology is a little bit different. Instead of having lots of ash and clay, there are a lot of sediments coming off the Columbia river. It's not as easy to drill into the fault, although we could perhaps do some observations to learn more about the seismic potential there as well.
Oliver Strimpel
Patrick Fulton, thank you very much.
Patrick Fulton
Thank you.
Oliver Strimpel
For more about Geology Bites, as well as pictures and illustrations that support this podcast, go to geologybites.com.