Romain Jolivet on the 2023 Turkey-Syria Earthquakes
Transcript
Note, transcripts are not fully edited for grammar or spelling.
Oliver Strimpel
This is Geology Bites with Oliver Strimpel. This episode and the next one discuss two recent and giant earthquakes. Both caused great loss of life and destruction, but, in almost every other respect, they were very different. In this episode, we talk about the pair of earthquakes that struck Turkey and Syria in February 2023. They occurred on land where the Anatolian Plate, a relatively small continental plate wedged between the African plate and the Eurasian plate, slipped several meters to the west along its strike-slip bounding faults. In the next episode, we focus on the 2011 Tōhoku earthquake, which occurred on the sea floor off the coast of Japan under seven kilometers of water. There, the oceanic crust of the Western Pacific is subducting below Japan. It was sudden movement of the sea floor along the subduction zone thrust fault that triggered the devastating tsunami that led to the Fukushima nuclear disaster.
The two earthquakes that struck near Turkey's border with Syria on February 6, 2023, were the deadliest in the modern history of the region. But the eastern Mediterranean has a long-recorded history of destructive earthquakes. What is the tectonic context of these earthquakes, and do they occur along well-recognized faults? And what do we know about the movement that took place during the February earthquakes? Romain Jolivet studies active faults and how they slip during earthquakes, as well as how they move between earthquakes. He uses satellite radar imagery and particularly a technique called interferometric synthetic aperture radar to reveal the detailed deformation of the ground around active faults. The results from the recent earthquakes show distinctive variations in the movement along known faults, as well as unexpected movement along hitherto inactive faults. Romain Jolivet is professor of geoscience at the École normale supérieure in Paris. Romain Jolivet, welcome to Geology Bites.
Romain Jolivet
Hi, Oliver.
Oliver Strimpel
Let's start with the broad tectonic context. Why does the eastern Mediterranean region have so many earthquakes?
Romain Jolivet
The northward motion of Africa and its subduction beneath Eurasia creates a lot of complicated motion within the eastern Mediterranean. You have slab rollback, where the African slab retreats south while Africa moves north, and this opens a big space in the Aegean Sea. To the east of Turkey, you have Arabia moving north as well, and pushing Turkey. And since there is this space opening in the Aegean Sea, it turns out that Anatolia moves westward, being pushed by Arabia. And that westward motion is accommodated by two major fault zones around Anatolia: the North Anatolian fault, which goes from eastern Turkey to Greece, and goes through the Marmara regions and Istanbul, but also the East Anatolian fault, southeast of Turkey, which is this long plate boundary fault that joins the North Anatolian fault to the Dead Sea fault, this North-South structure that bounds the Arabian plate. And, also, there is the junction with the subduction zone beneath Cyprus.
Oliver Strimpel
So, these major bounding faults that accommodate the westward movement of the Anatolian plate must be lateral or strike-slip faults, I assume.
Romain Jolivet
So, yeah. You have one right lateral fault to the north, which is the North Anatolian fault, and one left lateral fault to the South. And so these two faults accommodate this westward motion. This North Anatolian fault started propagating from east to west about ten million years ago. The East Anatolian fault is a bit younger. It's three to five, I think.
Oliver Strimpel
So, these faults are relatively young, geologically speaking, but are there good human records of earthquakes during the last millennium or two?
Romain Jolivet
In particular, this region of the East Anatolian fault, you're very close to regions that have been populated for a very, very, very long time.
Oliver Strimpel
You mean the banks of the Tigris and Euphrates Rivers?
Romain Jolivet
Yes. So, a very important region in history. Now, in terms of earthquakes, there is some very important earthquakes that have been described. The last ones, in 1893 there was a magnitude 7.1, 1872, 1795 — a bit earlier. So, these are earthquake that ruptured the bits that ruptured about two weeks ago. Back further in time, there is mention of a very large earthquake that took place in 1114. The magnitude of that 1114 earthquake is, of course, very difficult to place, but if you collect all the testimonies, you can draw intensity contours using a scale called the Mercalli Intensity Scale. It's actually a descriptive scale that tells you how much your furniture is going to move or what buildings are going to break, what type of masonry is going to fail. And it turns out that intensity 8, which is very, very high, you could draw this over a quite large region centered on exactly the region where the earthquakes of the last two weeks happened. More recently, there has been a magnitude 6.8. in 2020, the Elazığ earthquake, which ruptured just north of where the earthquake happened about two weeks ago.
Oliver Strimpel
OK, so let's talk about the two February earthquakes. Did they occur on these well-recognized bounding faults?
Romain Jolivet
So, yeah. The main earthquake, the magnitude 7.8, ruptured a big part of the East Anatolian fault, from the former rupture of the Elazığ earthquake, almost down to Antakya. It turns out that the second earthquake, this magnitude 7.5 that occurred about 10 hours later, ruptured the fault that is not identified as a main plate boundary fault. The most recent block model suggests that it is the boundary of a block. However, it wasn't recognized as such before. The actual epicenter of the big 7.8 earthquake did not locate on the East Anatolian Fault. This epicenter is the point where the rupture starts, where the first waves are being radiated, where the break actually starts to propagate within the crust. And it actually occurred on the fault that is parallel to the East Anatolian fault, and the rupture started moving northward toward the city of Pazarcık. And then it reached the East Anatolian fault, where the earthquake propagated both northeast and southwest on the East Anatolian fault, with tremendous amount of slip and huge moment release and a lot of shaking.
Oliver Strimpel
Wow, that's fascinating. So, you say it actually started on a subsidiary, parallel fault, and when that intersected the main fault, that initiated the larger actual slip that was experienced in that first earthquake.
Romain Jolivet
That's what it seems, yes.
Oliver Strimpel
And have we seen motion along that parallel fault in the past?
Romain Jolivet
We haven't. It's actually a region where there's a lot of micro-seismicity and small earthquakes.
Oliver Strimpel
And what about the second earthquake? Did that occur on one of these bounding faults that you talked about?
Romain Jolivet
So, the epicenter of the second earthquake was located on the Sürgü fault, a fault that is left lateral, that has an angle of about 40° with respect to the main East Anatolian fault. It's just north of it. It's not the North Anatolian fault. It is just a few tens of kilometers north of the East Anatolian fault. The question is: why is there a second earthquake on another fault? — which is why we don't call that an aftershock. In general, when we talk about aftershocks, we talk about earthquakes that would occur in the vicinity of the main fault, with a distinct and very specific pattern of occurrence in terms of time distribution. There're more aftershocks in a very short time after the earthquake, and then that number of aftershocks decays exponentially with time. This one is way too big to be an aftershock. It's on another fault, so it's just another earthquake. Why did it happen, and why did it happen 10 hours later? These are two very good questions. In general, we talk about triggering. There is static triggering. There is dynamic triggering. Static triggering is the triggering explained by the change in stress imposed by one earthquake. So, here, the magnitude 7.8 resulted in a stress change in the crust, and that stress change might have brought the Sürgü fault closer to rupture. Hence, it ruptured a few hours later. Dynamic triggering is more about the waves that are being radiated by an earthquake, and these waves passing through a fault will trigger a slip instability in a fault which results in an earthquake. The question that remains… (so it's either static or dynamic) … is why 10 hours later? We believe that friction controls the occurrence of earthquakes, so a fault is roughly a plane that separates two blocks, and friction holds the two blocks together. If friction is just classic Coulomb friction, then there is no time-dependent behavior, and if you change the stress then the earthquake should have occurred simultaneously — as soon as stress changes, the earthquake should have occurred. However, if friction is time-dependent, then you can imagine a mechanism in which your stress change actually triggers a slip instability that then grows bigger and bigger and bigger and results in an earthquake. Or it just changes the way the fault behaves very, very slowly, while it's locked, and it made it more fragile, and it initiated its weakening and it actually led to the occurrence of this magnitude 7.5 earthquake.
Oliver Strimpel
I just want to clarify the mechanisms that are at least theoretically possible, that could have caused this delay, and then this subsequent — not aftershock, but second earthquake in its own right. You mentioned some kind of time dependent behavior of the coefficient of friction. Can you talk a little bit about what that would be?
Romain Jolivet
In order to explain an earthquake, we use a model that suggests that frictional resistance of a fault plane drops while the fault slips, so you have a certain value of friction that holds the blocks together. The fault starts to slip, and while it slips, friction drops. If friction drops, then it's easier to slip, and so that's how you have an earthquake.
Oliver Strimpel
Difference between static friction and dynamic friction…
Romain Jolivet
Yes. And so your friction might evolve very slowly in time, and maybe by shaking the fault gouge, you might have accelerated this weakening of friction. This is purely speculative. There have been papers about this called delayed distant triggering of slip instabilities within fault gouges. So, it's laboratory experiments of fault gouge material being sheared, and you shoot acoustic waves at it. You trigger slip instabilities while the waves are passing, so this is distant triggering, and you would also trigger instabilities a bit later.
Oliver Strimpel
Let's talk now about your own satellite-based study of earthquakes and, of these particular earthquakes. You use radar imagery from satellites. In an earlier podcast, David Sandwell talked about using satellite radar altimetry to measure ocean surface heights from which he derived the shape of the sea floor. But you used an interferometric method or interferometric synthetic aperture radar to be precise, often abbreviated InSAR. How does that work?
Romain Jolivet
It's a technique based on a satellite or an airplane flying over a region and shooting a radar beam to the ground. The radar wave goes from the satellite, bounces on the ground and comes back to the satellites. What we measure is the phase offset between what we've sent and what we receive. A few days later, or few months later, you repeat the same exercise. You shoot your radar wave down to the ground, and you measure the phase offset with what you sent. This phase is a direct measure of the two-way travel time between the satellites and the ground. If you assume that you know what's in the atmosphere, you know the position of your satellites, then you can transfer that into a distance or a pseudo distance. And when you compare both images in the process called interferometry, so just doing simply the conjugate products between these two waves that you have acquired at day one and at day two, then what you have is the change in distance between the satellite and the ground and change in distance is deformation. I'm describing it as if we were doing it with a single point, but actually you do it on a full image, and you can have a full image of ground deformation. So, for instance, on the picture that I provided, you have the interferograms that we managed to construct from the Sentinel-1 constellation from ESA. We're measuring a phase difference between two passes of the satellite. Each fringe cycle that you see corresponds to about 5 centimeters of displacement. You see that the fringes get wider while you go away from the earthquake. Why is that? That’s because each fringe is a measure of displacement, and so the closer the fringes are in space, the larger the displacement gradient at the surface. So, it means that you have more displacement near the fault, less displacements away from the fault, so it's consistent with surface displacements caused by an earthquake. On this butterfly-shaped image, you have actually the combination of two earthquakes. You have one earthquake on the Eastern Anatolian fault, the one that's roughly southwest, northeast oriented, and you have other fringes that go to the northwest, and these correspond to motion on the Sürgü fault, the second magnitude 7.5 earthquake. Here, we're measuring meters of displacement. If we're doing time-series analysis using a lot of images, you can actually go down to displacements on the order of a millimeter.
Oliver Strimpel
Wow, that's an incredible resolution, and this is a wonderfully dramatic picture, as you say, with all kinds of butterfly shapes radiating out from the faults. We have these images on the episode Web page. So, you generated InSAR interferograms of the earthquake region by comparing radar images taken just over a week before the earthquakes to those captured a few days afterwards. Can you describe what the interferograms look like and what they reveal?
Romain Jolivet
The first thing that we can see is the length of the rupture. We see these fringes over a length of about 300, 350 kilometers, and they're closing in on the fault, on the edges of the region where you have these high-density fringes. This tells you how long the rupture was on the fault. This is for the main 7.8 rupture. You see, these fringes are also closing in on the fault that ruptured during the second earthquake, and that really shows you the extent of that rupture. You also see that we're losing signal. We can't really see anything when we're close to the fault. What does it tell us? It tells us that there has been a tremendous amount of displacement — that the surface displacement gradient is actually very high near the fault and much higher than the pixel spacing. We lose what we call coherence. We're unable to count fringes, we're unable to relate to one pixel to another because there has been too much motion.
Oliver Strimpel
What is the spatial resolution of each pixel on the ground?
Romain Jolivet
2.3 meters roughly in the east-west direction and 14 meters in the north-south direction.
Oliver Strimpel
And how small a deformation can you detect?
Romain Jolivet
With a single interferogram, we can detect a centimeter of deformation.
Oliver Strimpel
That's incredible… from space… so these are very highly registered images or highly processed images, to be able to compare the before image with the after image.
Romain Jolivet
The real challenge is from the space agency, who is able to fly the satellite within a tube of less than 50 meters radius, which means the satellite comes back every 12 days following a flight path that is less than 50 meters away from the previous one 12 days ago.
Oliver Strimpel
It is incredible.
Romain Jolivet
It is incredible.
Oliver Strimpel
Do they have to do little burns? Rocket burns to keep correcting it?
Romain Jolivet
There is a huge ground segment doing a tremendous job. The observation program was designed to last for 20 years.
Oliver Strimpel
This is the European Space Agency's Sentinel-1 series of radar satellites that you're referring to.
Romain Jolivet
Yes. So, as you said in the beginning, I also study what happens between earthquakes, and what happens between earthquakes is very, very, very slow. So, we must accumulate deformation over very long periods of time so that we're able to measure it. For instance, this East Anatolian fault. It's accommodating a relative motion of one centimeter per year on its northeastern section and five millimeters per year on its southern section, the one that ruptured in February. And so if you want to capture that centimeter per year or that five millimeters per year of motion, you better have several years of observations, and the more the better.
Oliver Strimpel
So, the interferogram presumably shows the amount of displacement and the length of the displacement that takes place along the fault. But these so-called butterfly-wing patterns that you see and how they're spaced also tell you about how the displacement decays as you move away from the fault. What do you see in terms of the deformation change as you move away from the fault and what can one infer from that?
Romain Jolivet
That gives us information on how deep the fault ruptured within the crust. Typically, the brittle-ductile transition is somewhere between 15 and 20 kilometers, so we can't expect earthquakes to happen anywhere between the surface and this brittle-ductile transition. If you look at the gradient of deformation away from the fault that tells you about the distribution of slip — of seismic slip — at depth. This fault here, the East Anatolian fault, is a vertical strike-slip fault, OK, and its slip extended all the way down to the brittle ductile transition.
Oliver Strimpel
And how far down is that?
Romain Jolivet
So USGS places it at 20 kilometers.
Oliver Strimpel
So, just qualitatively speaking, the deeper the rupture goes, the further away you expect to see deformation?
Romain Jolivet
Yes, the deeper it is, the longer the characteristic wavelength at the surface, and so the further away you're expecting to see deformation. Here you have a lot of slip. We're talking about 8 to 10 meters of slip, and if it extends down to 20 kilometers, then yes, you will see centimeters of displacement hundreds of kilometers away from the fault.
Oliver Strimpel
You said 8 to 10 meters of slip, which is enormous. Do we see that vary dramatically along the fault?
Romain Jolivet
So, the little parallel fault where the earthquake started, where the epicenter is located, doesn't seem to have slipped that much — about a meter of slip. But then it branched on the second one, which triggered a massive earthquake where you have 8-10 meters of slip. But then, along this large fault, you also have variations of the amount of slip, especially in the southern portion of the fault. You can see patterns of fringes are making circular patterns around this southern section. That means that there is along strike of variations of the amount of slip along the fault. For instance, near the main slip region, the fringes are parallel to the fault. That means there is no variation, it just slipped a lot and that's it. If you go to the south, you see fringes making wiggles around the fault — that tells you that there is a lot of variation in the slip distribution.
Oliver Strimpel
Is it too early to say if the amount of slip along the fault decreases monotonically as you move away from the epicenter?
Romain Jolivet
Looking at the interferograms, I wouldn't say that it's monotonic patterns in the interferograms. It shows lots of different segments of different size that ruptured and there’s variations, and it definitely did not decrease monotonically. There's bumps and humps. It changed along the strike.
Oliver Strimpel
If it changes, along strike, does that suggest that stress has accumulated differentially along the fault, so that there are places where this earthquake has in fact increased the probability of a future earthquake?
Romain Jolivet
Yes, on the edges of that main rupture, stress has been increased, especially in the southern tip of that rupture. For instance, there has been a big earthquake of 6.4 a few days ago. That means that stress has been transferred on the edges of the main rupture, and the probability of having an earthquake on the edges of the main rupture definitely increased.
Oliver Strimpel
We discussed earlier why that second magnitude 7.5 earthquake was not an aftershock. But there were in fact thousands of actual aftershocks. Are any of those visible in the InSAR data?
Romain Jolivet
Near the city of Adana, you have a very distinct small pattern of fringes. That really, really looks like very shallow small earthquakes, potentially a magnitude 5. So yes, we can see some of the aftershocks. The problem then is that these aftershocks, compared to the main shocks, are so small that they are completely lost within the fringe pattern corresponding to the main shocks. The second problem is that these aftershocks mostly occur where we lost coherence, where the ground has moved so much that we can't reconstruct the InSAR signal, and so we can't see them. But if the aftershocks are far away and big enough, then yes, we can see them. Furthermore, there will be another pass of the satellite, and so we could do the interferograms and try to capture aftershocks within these new interferograms.
Oliver Strimpel
Is that pattern that you described earlier for the first earthquake — where it actually started on a parallel subsidiary fault with less displacement, and then when it hit the main fault along the eastern edge of the Anatolian plate, it suddenly accelerated and kind of unzippered, if you like, both to the northeast and the southwest — is that something that was a surprise, or is that a pattern you often see in larger earthquakes?
Romain Jolivet
The more earthquakes we see, the more complex earthquakes we see. For instance, the 2016 Kaikoura earthquake in New Zealand had an incredibly complicated shape. The faults were crosscutting each other at angles that were completely unexpected and ruptured, and it resulted in an earthquake of comparable magnitude. Same thing in the Indian Ocean in 2012. We had three magnitude 8 strike-slip earthquakes with rupture at depth that are incredible considering the thickness of the oceanic crust. So, this one, an earthquake that starts on the little fault, joins a big one, and then goes backwards on a second fault, while unzipping the main Eastern Anatolian fault — the rupture that went southwest just did a U-turn. It started going north, and then it went south! Yeah, that's kind of unexpected.
Oliver Strimpel
It does seem surprising in view of the overall big picture of southwestward movement of the Anatolian Plate, that you would have slip along two faults that are on the eastern edge. You'd have one moving northward and then the other one moving southward, and then going both directions. Is that all consistent with that movement of the Anatolian Plate or have we seen some local deviation from that movement of the plate as a whole.
Romain Jolivet
The slip is left lateral, so that is consistent with the general motion. What is a bit awkward is that you have a fault that's turning. It veers from almost an east-west direction to a more north-south. How do you get a rupture through an elastic plate to turn around, and how is stress accommodated, but that it actually fuels that kind of earthquake. There are some very intriguing earthquakes, for instance, in 2013, this earthquake that occurred in Baluchistan in Pakistan. An earthquake that started a strike-slip earthquake on the fault that was almost north-south, and then it propagated south and ended on the thrust fault. But it stayed as a strike-slip earthquake. So, it's a strike-slip earthquake that ruptured this trust fault, which is awkward. In the topography in the long term geology that Balochistan earthquake looks like it should be thrusting. It should be a dip-slip fault, and we should expect dip-slip earthquakes, and we have a very strong strike-slip earthquake. Here, we have to explain that U-turn. We have to explain the position of the epicenter. We have to explain the occurrence of the two earthquakes. We have to explain the complexity of the rupture in the South, but in terms of general motion that is consistent with the motion of Anatolia, and there is no real question here.
Oliver Strimpel
I know you're continuing to monitor the InSAR data as the satellite makes successive passes about every 12 days, I think you said. Will any of that additional data help throw light on these questions, or will it be more directed towards the aseismic slip, if you like, that happens in between these major earthquakes?
Romain Jolivet
I like to study, in particular, aseismic slip, because it can give us hints about how a fault weakens, how it transitions from slowly moving to rapid dynamic earthquakes. There's already a slowly slipping section identified near where the 2020 earthquake happened. Now, another region that people should focus on following this earthquake is the Laval Fault.
Oliver Strimpel
Is that called the Dead Sea fault also?
Romain Jolivet
Yes, it's the Dead Sea fault. It’s this gigantic strike-slip fault that goes roughly from Antakya, from where the earthquake stopped down to Sinai, across Israel, across northern Syria. And there is a big history of earthquakes along this Dead Sea fault based on mostly testimonies and paleoseismology studies.
Oliver Strimpel
Is there any evidence that the aseismic slip along these faults slowly speeds up before you get a catastrophic slip, or is it the other way round? Or do you not actually have the resolution to see that kind of behavior?
Romain Jolivet
Can we capture a seismic slip before an earthquake that would be the signature of an impending major rupture? We don't know how to do that. But we are completely unable yet to differentiate between an aseismic slip event, harmless, that just starts, accelerates, does a few centimeters of slip over a couple of days or weeks, and then just slowly stops. And we can't differentiate that kind of harmless event from a slow slip event that would accelerate and then trigger or result in a big earthquake. There is also evidence that slow slip actually triggers small earthquakes and swarms of earthquakes, and sometimes they never trigger the big one. We'd like to find something occurring before the earthquake, but then we'd like also to be able to relate it physically to the occurrence of a big earthquake. There is access of research. There is models for the nucleation of earthquakes. There is observations mostly in subduction zones, but there is no causal link between most of the fans and big earthquakes. Of course, that's a very important research direction.
Oliver Strimpel
Well, Romain Jolivet, thank you very much.
Romain Jolivet
Thank you.
Oliver Strimpel
For more about Geology Bites, as well as pictures and illustrations that support this podcast, go to geologybites.com.