This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. We hear about earthquakes in the Himalaya, especially when they claim lives and cause damage, and we understand that, broadly speaking, it is the continued northward movement of India plowing into Tibet that causes these earthquakes. But where exactly do the earthquakes occur? How do they occur, and what determines how much damage they inflict? Roger Bilham has conducted a detailed study of the historical record of earthquakes in the Himalaya over the past millennium. He tries to reconcile what we've observed with our current understanding of the physical mechanisms at play. This, in turn, helps us assess future hazard potential. He is Emeritus Professor of Geology at the University of Colorado Boulder. Roger Bilham, welcome to Geology Bites.

Roger Bilham: Hello.

Oliver Strimpel: How far back do earthquake records go, and how much can we learn from the earliest historical records?

Roger Bilham: There are several sources of information, and they change with time. The most recent data comes from seismometers, but they only started operating in about 1900. Prior to that, we have newspaper reports and letters from travelers back to their friends, starting in about 1750. Prior to that, the record becomes more mysterious. There are government records that have occasionally survived. They're of the form, "The king died today, as did a lot of the people in the town." Or they might be in the form of repairs, especially to Tibetan monasteries, when an abbot would write to the Dalai Lama and say, "I lost 45 temples in my district. Could you send me 27 very large lumps of wood, and the money to hire a whole bunch of people to rebuild everything?" Those are valuable if they're spread over a large region, because they tell you that the earthquake was not local but affected a substantial segment of the Himalaya, and that usually means a great earthquake. And sometimes we have records in the form of poems and things like that, which are not so useful, because what we need is quantitative information.

The most important thing is that some buildings have survived for thousands of years, but they have often been damaged. Sometimes the temples, for example in Kashmir, are made out of two-ton lumps of limestone, and the building has survived, but it's clearly damaged, because you can see some of these huge blocks of stone have been shifted out of their original place in the wall. They may have shifted a couple of inches or even a foot, and a few more earthquakes would bring the building down. Sometimes there are records related to the repairs that were made to such structures, but they're very, very rare. So although we have a record of earthquakes going back through history, it's not quantitatively documented in any way that we can easily interpret.

Our biggest task is distinguishing between a short, sharp earthquake that happened locally and a very large earthquake that may have disrupted 300 kilometers of the Himalaya. The reason for that interest is that it's only the very largest of these earthquakes that allow India to slide beneath Tibet. There are thousands of earthquakes; even while we're speaking, several earthquakes will occur. But they're very small, and they play no part in this northward movement.

Oliver Strimpel: What about the geological record, or the geomorphological record, if you like? Can we learn anything from landslides, for example?

Roger Bilham: Yep, people have studied landslides. The great thing about a landslide is that you can often date it from the materials that it covers, or from the start of growth of trees on the landslide, for example. The problem with landslides is that although earthquakes trigger landslides, other things do too, like the monsoon in the Himalaya. So that's not very good. Another source of information comes from lake deposits. There aren't many lakes in the Himalaya, but the study of lake deposits sometimes reveals a steady annual sedimentation rate, from dust and monsoon rains and so on, and then occasionally one will find, interleaved with those annual layers, a layer that has clearly been caused by a shaking event. It's produced what we call a turbidite, some kind of deposit of sand and gravel and sometimes boulders, in the middle of what is otherwise layers and layers and layers of clay. By dating the particles in that coarse sand deposit, one can sometimes hope that it's an earthquake, if you can correlate it with another lake, maybe 100 kilometers away. If you can do that, it usually means it was caused by an earthquake. But again, it's a little bit ambiguous. One gets a date; one gets the idea that something happened. But again, that could just have been a terrible monsoon year, with huge amounts of river water depositing coarse sediments in the lake.

Oliver Strimpel: What about the modern improvements? What kinds of tools have we got now to add to our arsenal?

Roger Bilham: Well, it all really started with seismometers in 1900. Then in about 1960 we got a global network of seismometers that really permitted us to look at earthquakes everywhere, and that told us that there are huge numbers of smaller earthquakes. But the really significant advances occurred in 1985 to 1990, when GPS satellites became available and it was possible to actually measure the movement, the northward movement of India, and the slowing down, the deceleration, of materials as they approach the Himalaya. That really became the number one tool. We suddenly realized that it wasn't just the Himalaya moving, but that India was moving northward at between 14 and 18 millimeters a year. And that means that, knowing the rate of convergence of the Himalaya, the squeezing of the rocks, one could make an assessment of how frequently we needed great big earthquakes to occur.

And then a new tool came on the horizon, and that was the illumination of the Earth's surface with radar. An image taken before an earthquake and one taken after the earthquake could be compared, and one could make interference fringes between them, showing the displacement of particles of rock. It's not so good with forests and so on, but this technique has proved invaluable for the precise documentation, not so much of the slow convergence, as of the displacements that actually occur during earthquakes. For example, the Nepal earthquake in 2015 told us an enormous amount about how earthquakes occur, and that now requires us to reinterpret the historical record.

Oliver Strimpel: Let's talk a little bit about the geology of these earthquakes. There are a number of major faults that run along the Himalaya that have names such as the Main Central Thrust, the Main Boundary Thrust and the Main Frontal Thrust. Can you describe these briefly and tell us which, if any, of these is currently active in accommodating the ongoing collision between India and Asia?

Roger Bilham: Yeah. The main thrust in this collision process is called the Main Himalayan Thrust, and it is a surface which you can consider as lying above the Indian plate, which is going northward, and below the base of the Himalaya, which is going south relative to it. But for much of the time, that fault is locked to the Indian plate. The Main Himalayan Thrust is the thing that moves in these great earthquakes, and those are the ones that are of particular interest to us, because they accommodate India's northward movement. Now, the Main Himalayan Thrust lies at a depth, for most of its path, of between two kilometers and about 20 kilometers below the surface, but it emerges at the surface along the southern edge of the Himalaya, and where it emerges, you can dig it up. You can actually see the process of India's descent at work. One can see the offset of layers of mud, and one can sometimes pick up bits of carbon that have been incorporated from trees and fires, from the people that lived there or from natural fires, and those carbon products enable us to date the time of movement.

So the Main Himalayan Thrust comes to the surface at a place that we call the Main Frontal Thrust. It's a bit of a misnomer, because it's one of many splay faults arriving at the surface, only one of which may be active for a few hundred years. The interesting thing about the Main Frontal Thrust is that it has had to advance southward over the last 50 million years, because the Indian plate is going north and the Himalaya is going south relative to the Indian plate, and that means the Main Frontal Thrust has to keep jumping forward occasionally. Currently, we can see part of that process through bulges that have occurred in occasional earthquakes in history. We've never seen this bulge appearing in the plains of the Ganges, but we can see distorted rocks and uplift along riverbeds and so on that clearly tell us that the Main Frontal Thrust is trying to advance southwards.

When an earthquake actually occurs, one can actually see the movement at the surface, where the Main Frontal Thrust hits the surface in the Ganges plain or wherever. You can walk across it. You can say, this is where it slipped. In Taiwan, a man woke up in the night and watched the frontal thrust of the Chi-Chi earthquake slip before his very eyes. The house next door rose about 10 feet in front of him, slowly. Quite a remarkable sight. So the earthquake, occurring at depth, somehow translates the whole of the Himalayan carapace southward, the sediments of the Himalaya and the cities and villages and houses on it, and it offsets the sediments at the front edge of the Himalaya. So it's a very real surface. It does get offset. You can dig a trench through it and find when it slipped. But nobody's actually seen the Main Himalayan Thrust move and written it down.

Oliver Strimpel: As I understand it, there are three possible situations at a fault plane. First, it may be completely locked, with no motion across the plane. Second, there may be slow, steady motion, which we call creep. And third, we have the sudden, very brief but rapid movement during an earthquake. Can we talk a bit about the physics of these states, and how they can transition from one to the other?

Roger Bilham: Yes. A fault like the Main Himalayan Thrust is doing nothing most of the time. It's stressed by compressional forces from the south and the north, but it can't slip. And that's the fate of most seismogenic faults, faults that slip in earthquakes. There's a special class of conditions that can conspire to enable a fault to slip slowly. For example, the San Andreas Fault in central California is creeping at up to 20 millimeters a year, almost the full slip rate. In those special conditions, we know that there are rocks, like serpentinite, that are slippery and allow this slow slip. Sometimes it's continuous slow slip; sometimes it's episodic slow slip. It doesn't do anything for a few years, and then suddenly it slips a few millimeters, but it does so aseismically. Now, of course, it would be great if the Main Himalayan Thrust were a creeping fault everywhere, but it's not. We've been able to demonstrate that it's locked throughout most of its 2,000-kilometer length and its 100-kilometer width.

However, we found a part of that fault that is creeping, and it's creeping in a rather special way. The region is at the intersection between the locked Indian plate, that's the Himalaya locked to the Indian plate being driven northwards, and the part of the Indian plate that descends under Tibet. It descends under Tibet gracefully, through a creeping process, where the rocks are simply too hot to have any strength to hold the fault closed. But there's a transition region between this descent, which is slow and has no earthquakes, pure creep, and the purely locked region. That transition zone, we've realized in the last 10 years, is somewhere between 10 kilometers and 50 kilometers wide.

What's going on in that zone is not entirely clear. It could be a pressure-temperature transition that is slowly occurring and enabling some minerals to slip through a process of creep, and other minerals to fracture. It could be a series of bumps on the surface, separated by regions that are stressless because the bumps are taking all the strain. These bumps, which we call asperities, fail in microearthquakes, and almost certainly something of the sort is going on, because we see these microearthquakes. In fact, they define the width of this transition zone. It could also be that the materials of the zone are highly pressurized by fluids, through a metamorphic process where one mineral changes to another and releases fluids, like carbon dioxide, which at that depth is a liquid, or water. Those high-pressure fluids have the ability to kind of float the surface locally. So there are several things going on.

But the important thing is that this transition zone has a very important property: its width, its north-south width, determines the amount of elastic strain that can be stored before an earthquake occurs. So imagine the Indian plate moving northward. It's locked to the base of the Himalaya until it hits this transition zone, and beyond the transition zone it's disappearing under Tibet gracefully. Nothing's happening, no earthquakes; it's creeping. But in the transition zone, elastic strain is building up, and it does so for tens of years, hundreds of years or even thousands of years. And when it goes, the longer you've waited, the larger the earthquake.

Oliver Strimpel: What happens during an earthquake? Does the locked portion temporarily unlock?

Roger Bilham: Yeah. At the moment the earthquake goes, we hit a critical point in the amount of strain that the rock can sustain. We call this the failure strain, and it's about one part in 10,000. That strain level varies a little bit, but it's common to most rocks: one part in 10,000. When the earthquake initiates, it does so very rapidly, and it sets up a sequence of actions that we are beginning to understand. One of the important things about an earthquake on a thrust fault is that theory says that it should not be able to slip.

Let me explain with an analogy. Imagine a brick sitting on a concrete road. You push the brick from behind, and at a certain level of push, a very small amount of force, it starts to move. Now scale that brick up to have dimensions of many kilometers. Let's increase its height to two or three kilometers and its length to 5 or 7 kilometers. It turns out that if you increase its length to about 2 kilometers, you can push it, but if the length is increased to about 7 kilometers, the force you need to push that brick along that concrete road is so big that instead of the brick sliding, it fractures. The back end just crumbles up into small pieces and continues to crumble. So the friction in this case is much too high to permit the rock to slide.

This is the first paradox of thrust faults in the Himalaya: in principle, they shouldn't slide at all, but they obviously do. We believe the process is dynamic. For a brief second, or microsecond, the surface of contact between the base of the Himalaya and the top edge of the Indian plate becomes detached. Somehow friction goes to zero. How it does that could simply be a dynamic motion, where there's a sudden jolt and a separation of the rocks through their momentum. Or it could be that the friction caused by the movement of that small patch is enough to liquefy the rock and turn it into molten granite or whatever, in which case you've lowered the friction by turning it from a solid rock interface into a liquid interface. Or some people believe it explosively fails, in the form of a gas transition, which levers the rocks apart. We just don't know what's going on down there. But there are various physical models that have been proposed that seem to explain how you might be able to do it. Whatever happens, at the moment the earthquake occurs, the rocks start to move, and the part that's detached is maybe only 100 meters long. It propagates along the fault and allows the top surface to move southward.

Think of a caterpillar. Its back end is locked to Tibet. Its front end is locked to the base of the Himalaya, about 10 kilometers to the south. And the region in between suddenly separates, or, should we say, the friction goes to zero. That could be the arching of the back of the caterpillar. Now the caterpillar moves itself forward by moving that hump along, opening a surface and allowing itself to move, very slowly, of course. In an earthquake, this happens in milliseconds. We know that this pulse of zero friction, or very low friction, propagates at about 2 kilometers per second. We saw it happening in the Nepal earthquake. The seismometers recorded the sequential fracturing of rock from west to east and from north to south across the base of the Himalaya.

Oliver Strimpel: So does this ripple, this caterpillar-type process, start under the Himalaya, where it was formerly creeping, or does it start in the transition zone? And does it extend into the locked zone and unlock that for a microsecond?

Roger Bilham: Yeah, I mean, that is the absolute real question, and that brings us to the second paradox. The earthquake starts where the elastic strain has been stored for tens or hundreds or even a thousand years. The elastic strain has a limited potential to move the rock southward. In the case of the Nepal earthquake, we know that it shifted only about half of the Himalaya in a north-south sense, and it shifted the compressional strain that initiated the earthquake southward, to some midway point close to Kathmandu. We know it's there because radar interferometry showed a huge bulge centered on Kathmandu. In fact, measurements after the earthquake showed that the airport runway had risen about a meter and had been tilted backwards by the elastic strain being translated to that point.

Now, the paradox is this. The amount of elastic strain you can store in that transition zone is limited by the failure strain of the rock, and the failure strain is about one part in 10,000, or 1 meter in 10 kilometers. So if the zone is 10 kilometers wide, you only have one meter of elastic strain available to drive an earthquake, and that apparently is not enough to drive an earthquake entirely across the Himalaya. You can get halfway, or a third of the way. In the case of the Nepal earthquake, we know that the slip available at the moment of failure was two or three meters, and that was enough to drive the elastic strain halfway across the Himalaya. So only half of the Himalaya moved southward; everything south of Kathmandu remained locked.

Of course, in the immediate aftermath of the earthquake, people were worried. Well, goodness me, we've got all this elastic strain sitting on Kathmandu; it's going to produce another earthquake. We calculated that it would have been about a magnitude 7.1 if it occurred, and that it could occur at any moment. People were in fact incredibly worried. But we were able to find a parallel in Nepal's history. An earthquake happened in 1833 that looked almost identical to the 2015 earthquake. It had the same magnitude and the same sort of rupture area. We can tell that from the damage to monasteries and from felt reports in Nepal and in Bihar province in northern India, so we know those two earthquakes were almost the same. And yet there were no earthquakes south of Kathmandu following that earthquake in 1833. So this was a kind of reassuring piece of good news from history.

The bad news, however, is that we now believe there are two of these increases of strain in the Kathmandu region. You can think of them as reservoirs of elastic energy that are completely invisible. If we hadn't seen the one that exists right now, centered on Kathmandu, appear in the geodetic data, we wouldn't know anything about it. We would assume they had all fizzled away. And I think we would also have expected it to disperse in the form of creep or something. But in the last seven years, we've seen nothing of the sort. It's just sitting there. It hasn't evolved. It hasn't produced more earthquakes, not even microearthquakes. It's sitting there dormant. And the question you might ask at this point is, well, what's going to happen? It's quite possible that in 180 years' time, which is the time to produce yet another earthquake in the same place, that earthquake will dump a little bit more elastic energy in the Kathmandu region, which would be sufficient to drive that patch southward. And if it does so, it will basically have been refueled with ancestral elastic energy, and it will move in a much larger earthquake. In fact, it may release not just the earthquake energy that's been left by the last two earthquakes; it may release a lot of energy that's been stored along strike. So the next event may actually result in a magnitude 8 earthquake. That shouldn't surprise us. The geologists have dug up the Main Frontal Thrust south of Kathmandu and found evidence for slip of the order of 10 to 15 meters, and that means the earthquakes that slipped that amount probably went through the same sequential process that we believe is occurring now, of which the 2015 earthquake was a part.

Oliver Strimpel: That doesn't necessarily bode well for Kathmandu. But I wanted to ask you, coming back to our analogy of the brick moving: if you lubricate the bottom of the brick, it's going to slip more easily. And I know that water acts as a lubricant in the lithosphere. Does the monsoon rain penetrate far enough to introduce a kind of enhanced seasonal risk of earthquakes?

Roger Bilham: It does, but it does so in a different way from the one you're imagining. Water takes a long time to filter through 10 kilometers of rock. It's possible to imagine scenarios where sufficient water loading could force water into cracks and make a continuous hydraulic column down to the place where earthquakes occur, but that is not what we see. On an annual basis, the weight of the water of the monsoon loading the Ganges plain is enough to stretch the Indian plate and to produce a tiny increase of stress in the transition zone. We see its manifestation in the form of greater numbers of microearthquakes in the monsoon than in the dry season. So earthquakes that were about to occur in this transition zone are just brought forward a little bit in time by the weight of the water in the monsoon.

In fact, there's another possibility, other than that. There are huge numbers of glaciers in the Himalaya, and they've been melting very rapidly recently, and you could think of those glaciers as a load. Going back to the brick analogy, you can think of the brick as being the weight of the rocks of the Himalaya, and you could add a few pebbles for the weight of the ice. If you remove those pebbles, you'd need to push a little bit less hard to get an earthquake to occur. So the unloading of the Himalaya by ice melting is likely to produce more earthquakes, or bring forward in time earthquakes that would otherwise have occurred. We think there might be a little evidence for that. It looks like between the 12th century and the 14th century, the whole of the Himalaya may have slipped in great earthquakes. That's information that's been pieced together by looking at trench excavations of these gigantic events. And it's possible that the Little Ice Age may have had something to do with that. In other words, the ice that was there started melting, and the sudden decrease in the loading on the transition zone may have initiated a propagating pulse of earthquakes.

Oliver Strimpel: We've already touched on the earthquakes and the danger posed by them in Kathmandu. Let's talk a little bit more about the damage inflicted by Himalayan earthquakes. Obviously, the magnitude of an earthquake will increase its damaging impact, and you've already mentioned this new idea that there may be reservoirs of elastic energy stored underneath the ground. But do other geological factors, perhaps local factors, affect the impact of an earthquake?

Roger Bilham: Well, the damage from an earthquake is due to the vigor with which buildings are shaken. In fact, if there were no people around, earthquakes wouldn't matter at all. But most of our problems are caused by buildings, and so slight changes in the physical conditions where these buildings are constructed can in fact enhance their propensity to be damaged. For example, large sedimentary basins amplify the shaking from an earthquake, and ridges and mountains do indeed enhance shaking. In the Kathmandu Valley, for example, you have both occurring. The valley is full of sediments that have much lower strength than the surrounding rock. So you can think of it as a bowl of Jell-O in an iron container. If you shake the iron container, the iron container just moves as you shake it, but the Jell-O inside the iron container can slop about. Its surface may move two or three, or maybe four or five, times more than the actual shaking that you're imparting to the basin itself, if it hits the resonance of the valley, and the resonance of the valley sediments is usually triggered by these earthquakes. So buildings in sedimentary basins experience a lot more acceleration and damage than those built on solid rock.

On solid rock itself, the top of a mountain tends to flop around by a kind of whiplash process, in much the same way that if you held a stick and wiggled the bottom end of it around, the top end would flip backwards and forwards a lot faster; the accelerations at the end of the whip will be larger. So there are places where you can expect worse damage and places where you can expect lesser damage. In the Kathmandu Valley, in a very specific village, you can walk from the base of a small ridge up into the village itself, only a few hundred meters above the river floor, and you find twice as much damage occurring to the structures at the top of the hill as at the base of the hill. It's that specific.

Oliver Strimpel: And would some of the sediments in your sedimentary basin even be prone to liquefaction?

Roger Bilham: Yes. Liquefaction is a very special process. It occurs where you have a shallow water table, so the water is only a few feet below the surface, and the worst case is where that's capped by a layer of clay. When you shake that up, the pressure in the sand-water mix below the clay layer tends to rise, and that tends to move the particles of sand apart slightly. As a result, it loses its strength. So if a building relied on the frictional properties of that sand layer always being strong, and you shake that layer and it loses its strength, the building will actually sink. In the Bihar earthquake of 1934, for example, a lot of bridge parapets sank into the river, and the decks between them were crumpled and broken. Interestingly, there were a few gasoline stations with underground tanks, and the reverse occurred: the tanks rose to the surface, came out, and had to be reburied after the earthquake.

Oliver Strimpel: There must also be non-geological, human factors that can make a huge difference to an earthquake's lethality.

Roger Bilham: Yes, unfortunately. If you want to build a house, if you've got enough money to build a concrete-frame house, and you ask a contractor to build it for you, you first of all ask him what his price is going to be. You may ask around, and you'll take the lowest price. Now, the lowest price is usually the one where the contractor has to cut corners in order to survive, and that's putting it mildly. Many contractors are sufficiently corrupt not to care about the strength of the buildings they put up. They'll just put them up in the cheapest possible way. Pure corruption is in fact the cause of many, many unsafe buildings in developing nations. It's cheaper to use brittle steel rather than compliant steel, or a sand-cement mix that's much more dilute than it should be, so you lose strength. That's the kind of problem that one finds in many Indian villages and cities that are going from traditional design to concrete-frame design, which is necessary largely because we've run out of wood and it's not possible to make lightweight buildings anymore.

However, in the villages, it's usually ignorance that conspires to make buildings that are unsafe. To give you an example, most of the Nepal structures that fell down in the last earthquake are made of boulders held together with wet clay, which dries into a brittle, soil-type material that doesn't have any strength; you can easily break it. In an earthquake, this is the worst possible kind of condition, because horizontal accelerations tend to make the walls sway backwards and forwards, fracturing the clay-rubble bonding. Very often the boulders are rounded; they're not even angular, so there's no cohesion between the rocks themselves. So you have a situation where probably 90% of all the structures in the villages, the distant, remote villages where most of the rural population lives, are vulnerable to shaking. The government has encouraged retrofits to these buildings, but even the small cost of these retrofits, like using chicken wire to hold the clay together in layers, is too expensive for the villagers to afford. So even though retrofits are known and advised, they're not enforced in most of the villages, and the villagers just don't know anything about it. They simply need to build a house. They go down to the river and get some rocks, they go to the fields and get some clay, and they build their house. And I'm afraid that we will see many earthquakes in the future that will bring these houses down.

It's unfortunate. It's so easy to fix, but we can't prevent it. For example, had the 2015 earthquake occurred at midnight rather than at midday on a sunny Saturday, when everyone was out picnicking and so on, there would probably have been four or five, maybe ten, times as many fatalities. As it was, 10,000 people were killed by that earthquake. There could quite easily have been a hundred thousand.

Oliver Strimpel: What are you working on at the moment?

Roger Bilham: We're looking at earthquakes in the central part of India. It's often forgotten that earthquakes can occur in the middle of the Indian plate. The reason is that the Indian plate is being stressed because of its northward collision with Tibet. So every part of the Indian plate is on the verge, on the brink, of failure. If you do anything like construct a large reservoir there, you can trigger an earthquake, and earthquakes can also occur quite spontaneously throughout the Indian plate. This is important if you want to site things like nuclear power stations: you have to avoid the kinds of faults that could occasionally fail.

Oliver Strimpel: Roger Bilham, thank you very much.

Roger Bilham: Thank you.

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