Remarkable video captures fault slip in the Myanmar earthquake
We deep dive the possibilities presented by a witnessed rupture
ဤစာကို မြန်မာဘာသာဖြင့် (Google မှ အလိုအလျောက် ဘာသာပြန်ဆိုရန်) ကိုဖတ်ရန် ဤနေရာကိုနှိပ်ပါ။
Citation: Bradley, K., Hubbard, J., 2025. Remarkable video captures fault slip in the Myanmar earthquake. Earthquake Insights, https://doi.org/10.62481/01cd039c
Earthquake Insights is an ad-free newsletter written by two independent earthquake scientists. Our posts are written for a general audience, with some advanced science thrown in! To get these posts delivered by email, become a free subscriber. If you would like to support our work here, please also consider a paid subscription. If you use our Substack as part of a group, or for teaching, consider a teaching or institutional subscription. Every paid subscription makes a difference!
If you have recently lost your job or are a student or researcher without ability to pay, let us know and we will upgrade your subscription to “paid” at no cost.
We are grateful to the people who made this post possible by uploading the video discussed here to Youtube, and to the authors of a new preprint presenting seismic data from a seismometer near Naypyidaw. We also thank our colleagues Dr. Rishav Mallick and Dr. Harsha Bhat for insightful conversations supporting this post.
This post refers to exciting new seismic data highlighted by a preprint by Ssu-Ting Lai and coauthors, posted on April 29th. This post is independent of, but should be read in parallel with, another preprint by Soumaya Latour and coauthors, which also analyzes the CCTV footage. It is exciting to see people sharing data and analyses rapidly, both with fellow scientists and with the public.
Over the last two weeks, a remarkable video has made the rounds online, which shows for the first time a clear recording of a large fault rupture as it happens. The footage, which captures a dramatic episode of slip on the Sagaing Fault, was uploaded to a Youtube channel focused on recordings following the deadly March 28, 2025 M7.7 earthquake in Myanmar. Please note that the channel also contains unedited records of events following the earthquake that might be distressing to watch for some.
To our knowledge, this is the first video recording of large slip on a fault. A recording from the 2018 Hualien earthquake in Taiwan also showed the surface effects of sudden offset on a fault, but in a dense urban setting where the ground surface was obscured.
Our first reaction to watching the video was: Wow! Our second reaction was: What can this tell us about earthquakes and faulting? That question is the launching point for this post. But to even start answering it, it turns out that we need to discuss a lot of background ideas. Thus, this post will be a lot longer than most of our posts.
Many of our readers will have already seen this video, and some will have already analyzed what is happening. We personally regret that we watched the video before we thought about what we would expect to see. This was a cardinal sin because we were not forced to confront our biases.
Therefore, we will only place a screenshot of the video here. Please don’t try to play it, and if you do, please don’t complain in the comments about the video not playing. After we have thought about what we should expect to see in the video, we will return and analyze the video in some detail.
In this unusually long post, we want to explore some earthquake science ideas that might relate to this video. We would like to clarify that we are not doing “Research”; we are following our own curiosity while trying to highlight some important topics in the science of earthquakes. We might be wrong about some aspects of our analysis, and if so we solicit guidance from our expert readers. We are sure that a lot of great research is currently being done on this earthquake, and probably on the rupture video itself, and we look forward to learning from those studies as they arrive. As with seismic waves, the most interesting studies do not always arrive first. In the meantime, let’s think about what it means to witness an actual fault rupture!
Eyewitness accounts of surface ruptures
Despite the fact that large earthquakes have cut through populated regions many times over human history, eyewitness accounts of the formation of the surface rupture are scarce. However, there are a few good descriptions that stand up to scrutiny. Let’s consider a few of the best examples.
The 1983 Borah Peak earthquake, Idaho
In 1983, Mrs. Lawana Knox witnessed the formation of a normal fault scarp during the M7.3 Borah Peak earthquake in Idaho. She was not located in a densely populated area; she was elk hunting with her husband, Bill, in the remote mountains. Because her husband was driving elk down the mountain at the time of the earthquake, and because given her position seems to have been chosen for a wide field of view of the mountainside, we can only speculate that Mrs. Knox had a rifle and was keen to use it! If so, she would have been intently watching the slope for signs of movement, creating just the right scenario for observing the formation of a fault scarp.
Mrs. Knox was later interviewed by USGS geologist Robert Wallace, but didn’t mention whether she was armed and dangerous at the time, or not.
Some quotes from the report are enlightening:
and
Wallace interprets that there may have been a gap of about ten seconds between when the strong ground motion began, and when the scarp formed (poetically described by Mrs. Knox as “just as though one took a paint brush and painted a line along the hill”).
News reports indicate that Mrs. Knox died in 2019 at age 80. While the obituary rightly focuses on the fruits of a long life and a large family, it makes no mention her unusual role in the history of earthquake science. She saw the Earth paint a masterstroke across the landscape!
The 1990 Luzon earthquake
Other eyewitness accounts come from a M7.7 strike-slip earthquake in the Philippines in 1990. Yomogida and Nakata interviewed many people who were near the trace of the fault, finding thirty reliable witnesses (charmingly? referred to as human seismograms, although we might prefer anthroseismometers) to the formation of the actual surface rupture. As far as we know, this study is the only attempt to date to specifically try to find direct witnesses of surface rupture. In the case of the Myanmar earthquake, which struck in the early afternoon, there should be many such witnesses.
One witness was actually dragged sideways by three meters while holding onto a pole located almost directly on the fault trace. The creators of the earthquake-advice slogan “Drop! Cover! Hold on!” probably did not consider the (remote) possibility of people holding on to something on the other side of the fault.
Another witness also ran to grab onto a post, but by the time she reached it the surface rupture had already formed. Apparently, grabbing onto a post was a common reaction to this earthquake.
From the testimonies of the Borah Peak and Luzon earthquakes, it becomes clear that both surface ruptures formed very quickly, within a few seconds in a given location, and only after the onset of intense ground shaking.
This matches the modern understanding of how fault ruptures usually develop, as pulses of slip that only activate a specific part of the fault for a few seconds. In the early 1990’s that question was still pretty open. A competing model was that the entire fault slips mostly at the same time, which would result in very long slip durations all along the fault during a large earthquake. In 1990, Thomas Heaton described how that model (crack-like ruptures) could not explain the seismic data (or the eyewitness description of Mrs. Knox) recorded during large earthquakes, and proposed instead that the rupture must instead travel along the fault in a narrow pulse of slip. That paper laid new groundwork that is fundamental to how we think about earthquakes today.
Intriguingly, both of the witnessed ruptures also formed toward the end of the strong ground shaking. If any reader is familiar with other eyewitness accounts of the actual surface rupture forming, please let us know!
To get to where we are going in this post, we probably need to briefly review the concept of seismic wave speeds.
Seismic wave speeds
We now know where earthquake waves come from (a propagating rupture front), but how do they get to where they are going?
Seismic waves move at up to ~6 kilometers per second through the crust, and up to ~11 kilometers per second through the inner core. That makes these waves the fastest macroscopic natural phenomena on Earth, outside of electromagnetic waves and falling meteors.
The speed of a seismic wave through a rock is determined by two properties: first, the stiffness of the rock, and second, its density. A seismic wave is, at an atomic level, a push or pull that is transmitted by atomic bonds. If a material is very stiff, then the applied force is quickly transmitted to the next bond, and from there to the next, and so on. So, stiffer materials — i.e., those that resist deformation — will allow seismic waves to travel more quickly.
In addition, if a material is very dense, then the wave will have to move a lot of mass with each oscillation. Our friend Newton tells us that acceleration trades off directly with mass, so the wave will travel more slowly than if the material were less dense. Our American readers might remember the giant spinning wheel in The Price Is Right. To get that bad boy going required a lot of time!
The fastest seismic wave is the P wave (a.k.a. compressional and dilatational wave, push-pull wave), which is like a sound wave. In this wave, the bonds between atoms are compressed and stretched forward and backward, in addition to being sheared sideways. Most materials resist the change in volume strongly, and this strong resistance causes the P wave speed to be quite high. The speed of P waves in the continental crust (Cp) is about 6 kilometers per second.
The second fastest seismic wave is the S wave (a.k.a. shear wave). Particles caught up in an S wave will move back and forth perpendicular to the P wave direction, but the wave will still propagate outward from the earthquake origin like the P wave. The S wave is slower than the P wave because particles still resist being pulled side-to-side, but do not have to worry about volume changes. Typical S wave speeds (Cs) vary between 2 kilometers per second in young sedimentary deposits, up to about 4 kilometers per second in the deep crust.
It is well known that S waves do not travel through a fluid. This fact, demonstrated by the shadow zone where S waves do not arrive from far-off earthquakes, is actually how the outer core was discovered.
Why don’t S waves travel through fluids? Well, in a fluid, there is no resistance to side-to-side motion; the slippery particles are not bonded together and can simply glide past each other, changing the shape of the fluid with ease. As per Newton, a zero (resistive) force results in zero acceleration. The P wave can easily travel through fluid because the tightly packed fluid particles can’t push through each other, no matter how slippery they are. Fluids are excellent at preserving their volume but not their shape, and at transmitting P waves but not S waves.
The next fastest seismic waves are surface waves, which come in two flavors: Love, and Rayleigh. Surface waves are not radiated directly from fault ruptures. Instead, P or S wave energy becomes trapped along a surface, like Earth’s surface, or an underground fault plane buried underground. These waves were named for the scientists who mathematically predicted their existence: respectively, August Edward Hough Love (in 1911) and John William Strutt (in 1885). In case you are wondering the obvious question: we have Rayleigh waves, rather than Strutt waves, because John William was also the third Baron Rayleigh.
The speed of Love and Rayleigh waves (Cl, Cr) is almost as fast as the S wave, but their effects are confined to the volume of crust around the surface that the waves are moving on, and they decay much more slowly (because they travel in a 2D space rather than a 3D volume). This means that very far-field sites will mostly experience shaking from Rayleigh and Love waves trapped on Earth’s surface — the shaking is carried to those sites along the Earth’s surface by Love and Rayleigh waves. However, it will be hard for far-field sites to sense surface waves trapped on a different surface, like a buried fault plane.
Near-fault seismic recordings
While anthroseismometers can witness a wide range of effects during a surface rupturing event, their memory and timekeeping capability are famously imperfect. In contrast, actualseismometers are indefatigable sentinels that focus all their attention on just one thing: the motion of the small piece of ground they are firmly attached to.
Over the last three decades, an increasing number of large earthquakes have been monitored by at least one near-fault seismometer capable of recording the strong ground motions that occur near the ruptured fault. These fortuitous recordings give us another, more technical, perspective on the phenomena that we might expect to see in the video recording of fault rupture.
Very few seismometers are currently in operation within Myanmar. However, as we have learned from an exciting new preprint by Ssu-Ting Lai and 22 coauthors, a single seismometer located near the Sagaing Fault near Naypyidaw (NPW) was installed in 2016 as part of an international training exercise, and has been running most of the time since then.
The figure below show the recording from the seismometer near Naypyidaw, alongside two other already-published near-fault recordings from two different large earthquakes that are similar to the Myanmar M7.7 strike-slip event.
These seismometers record acceleration. The plots below show the velocity of ground motion over time, calculated by integrating the acceleration over time. The traces are labeled FP (fault-parallel), FN (fault-normal, or fault-perpendicular), and Z (up/down). The SKR seismometer recorded the 1999 Izmit earthquake, but unfortunately had a malfunctioning channel, so we only see FP and Z. We have heavily sliced-and-diced the original published figures, so please refer to the original publications if you want to see the data as it was originally presented.
We tend to avoid showing seismic wiggles in our posts, because we are not actually seismologists, and because the wiggles can be hard to explain. Here, we think it is important enough to try.
You really only need to see a few things to get the picture:
The recordings appear very similar to the naked eye. This means that they are likely documenting the same fundamental processes that are similar between the earthquakes.
Each recording has a large peak in the fault-parallel (FP) component (shaded orange), which is accompanied by a smaller peak in the fault-normal (FN) component. This is the direct signature of pulse of fault slip passing the site. During this pulse, the ground moves at high speeds: up to 0.87 meters per second for the Izmit earthquake, and up to ~1.6 meters per second for the Denali and Myanmar earthquakes. Note that these numbers are about three orders of magnitude lower than the seismic wave speeds we were talking about in the last section — the overall ground movement is not the same thing as the propagating seismic waves.
Each recording shows large oscillations for about 10-15 seconds following the slip pulse, mainly in the fault-normal (FN) component (shaded blue). This means the ground was moving toward and away from the fault, over and over, also at pretty high speeds. We have followed Dunham and Archuleta (2004) by calling these oscillations “Rayleigh pulses.” Their origin will be revealed momentarily.
Each recording shows a smaller period of movement before the fault slip arrives (shaded red). We have labeled this the “precursor” phase, and we will also discuss it further in a moment.
These recordings are very different from the human observations that we described earlier. Seismometers can’t actually see fault ruptures, and they don’t try to grab onto poles. But these recordings do document fundamental motions of the Earth that can tell us a lot about what is happening as the rupture burrows past. In particular, it is surprising (to us) that there is so much motion toward and away from the fault (the fault normal direction), given that these are all recordings of strike-slip earthquakes, where we usually think mostly about fault-parallel motions.
There are some clear differences between these recordings and some of the eyewitness testimonies. For the Borah Peak earthquake, the slip happened after the strong ground motion subsided, about ten seconds after the shaking started. In contrast, for the three seismometer recordings, the slip happened within two to three seconds of the first sensible shaking, and strong ground motions continued after the rupture passed.
To cut to the chase and keep this post as short as possible, this is because each of the three earthquake recordings is for an event that experienced supershear rupture, while the Borah Peak earthquake was a subshear rupture.
Subshear and supershear earthquakes
We’ve now talked about the speed of seismic waves, and the speed at which a fault actually slips at a given location. However, there is a third speed that we have so far neglected, which is actually the main topic of this post: the speed at which a rupture propagates along a fault.
A fault rupture begins its life as a tiny crack in an over-stressed rock. As the friction that previously held that crack stationary is overcome by stress, the sides of the crack begin to move slightly past each other. This motion transfers energy that was previously stored in the strained-but-stuck rock away from the slipping area and toward the tip of the crack, forcing the crack to grow a bit more. This absorbs a lot of energy but also releases even more stored energy, further growing the crack. This is a potentially run-away situation: as long as the expanding crack produces enough energy to overcome the resistance, and as long as the driving tectonic stress is large enough in the newly-propagated-into area, the rupture can continue to grow.
In the area behind the crack tip, the fault is actually slipping, causing a lot of dramatic effects. Because the rocks are sliding so quickly, there is a lot of frictional heating along the fault, which is thought to be critical to how the fault weakens and how the rupture evolves. We can actually find evidence of this heating along some fault zones: pseudotachylites are glassy veins around and along fault zones that tell us that faulted rock actually melted during an earthquake. The name comes from the similarity to tachylites, a type of basaltic glass.
The rupture propagation speed can change over space and time due to factors like the level of stress on the fault, the properties of the rocks in or around the fault zone, the amount of fluid trapped along the fault, and many other factors. This speed is a very important factor for the resulting earthquake phenomena.
If we are thinking about rupture speed, there are basically two types of ruptures that occur in nature, of which one is common, and the other is rare.
Subshear, or sub-Rayleigh, ruptures
Sub-Rayleigh ruptures are common. This type of rupture propagates more slowly than S waves (and Rayleigh waves). Rayleigh waves travel at about 90% of the shear wave speed. Apparently ruptures cannot stably propagate at speeds between the Rayleigh and S wave speed — some very complicated math explains how ruptures at these speeds do not generate sufficient stresses over a large enough area in front of the advancing crack tip to continue to propagate. So, while sub-Rayleigh means something different than subshear, these two terms are functionally the same.
As a sub-Rayleigh rupture propagates, the waves it generates race away in all directions — both the P and S waves move faster than the rupture itself. Due to the Doppler effect, the waves that are released in the same direction that the rupture is moving will pile up in front of the rupture, while the waves released in the opposite direction will be diluted.
If we are a seismic station located near the fault trace, and a subshear rupture passes by, we will experience several effects. Here, we are using as examples some numerical models of strike-slip ruptures by Mello et al. (2016).
The image above is a map view of a right-lateral strike-slip rupture that is propagating from top to bottom along a fault (dashed black line). The rupture tip has just reached the center point of the image. The S waves, generated along the length of the rupture, have piled up, causing large motions ahead of the rupture tip. The arrows show the direction of this motion. The boxes at bottom show a time series for a site located near, but not on, the fault trace: t1 first, then t2, etc.
An interesting point is that, for a sub-Rayleigh rupture, the largest motions are actually across the fault, and occur before the rupture itself passes by. In addition to these large motions, a lot of strong ground shaking will precede the rupture. This timing is similar to what Mrs. Knox observed in Idaho.
Laboratory experiments have recorded these kinds of sub-Rayleigh ruptures in miniature, as shown in the figure below. The left panel is a view of the rupture in progress; we can see what is going on due to the transparent plastic that is used instead of rock, and an extremely clever experimental setup that can capture the details of the high-speed ruptures As a great scholar once said: “Life in plastic, it’s fantastic.” The rupture began (nucleated) at the lower right, and is moving upward to the left with strike-slip offset. The P wave front has already exited the building, but we can still see most of the S wave front, and the rupture tip itself (marked “SR rupture”). The experimental setup can actually detect the motion of specific points; the graph on the right shows the fault-parallel (FP) and fault-normal (FN) motions for a sensed point near the fault. The fault-normal velocities (green line) grow larger than the fault-parallel velocities (red line), as predicted by the numerical model.
The discussion above is for sub-Rayleigh ruptures, which travel slower than the S wave speed (and which in practice must also travel slower than Rayleigh waves).
One of the most remarkable features of the Myanmar earthquake is that there is pretty clear evidence that it ruptured southward from the epicenter for a very long distance (about 430 kilometers) in about 90 seconds. That means that on average, the rupture was traveling at about 4.8 kilometers per second — well above the S wave velocity. (See our previous posts for details.) There is a good chance that the video records a supershear rupture passing by.
So, what should we expect in a supershear rupture?
Supershear ruptures
Supershear ruptures are relatively rare. As we noted above, ruptures cannot stably propagate between ~90% and 100% of the shear wave speed. For a rupture to go supershear, it has to therefore leapfrog this Forbidden Zone. These ruptures occur when a subshear rupture accelerates to near the Rayleigh wave speed, and becomes unstable. At that point, if conditions are right, the stresses acting ahead of the advancing crack tip can actually cause a new crack to develop that can propagate at supershear speed.
The speedometer from Spaceballs seems apt:
When a supershear rupture nucleates, the original subshear rupture can actually continue to propagate along the fault, following in the wake of its precocious child.
Because a supershear rupture moves more quickly than the S waves it radiates, those S waves will tend to pile up in the wake of the rupture front, forming a type of Mach cone. This is often compared to a “sonic boom” caused by a jet fighter, but it is important to remember that the sonic boom is caused when the jet exceeds the P wave speed of air (the speed of sound — a mere 340 meters per second); as we know, S waves do not propagate in a fluid or gas (air).
Because supershear ruptures travel faster than S waves, they will outpace them. However, there is still a kind of ground motion that precedes the rupture, called the “dilatational precursor” — this can be loosely translated into “P waves before the rupture.”
What do the Mach cone and dilatational precursor look like? Luckily, we have more lab quakes and numerical models to help us visualize all of these effects. The following image shows a supershear lab quake, with all of its attendant phenomena. We can again see the S wave front, led by the S wave Mach cone, along with a dilatational lobe in front of the rupture tip, and a trailing sub-Rayleigh rupture lagging behind the rupture tip.
Here’s the important part: the velocity traces in the right panel, which are created by tracking the motion of a specific point near the fault, look just like the recordings from the near-fault seismometer recordings of actual supershear earthquakes. We see a huge fault-parallel pulse combined with fault normal motion (orange), followed by mostly fault-normal oscillations (blue). We also see movement before the slip pulse itself arrives (red).
This agreement between field observations and lab experiments has convinced us (personally) that supershear ruptures, when recorded by instruments close to the fault, should have a predictable sequence of phenomena.
And let us not forget that supershear and subshear ruptures are actually expected to coexist. This happens when the original subshear rupture does not die out after creating the leading supershear rupture. That means that we might expect to see supershear-like motions, followed by subshear-like motions, in quick succession.
OK, OK, we are almost ready to watch the video, we promise. But first (second? third?), we have to ask ourselves whether we think the rupture was supershear at the video location, or not. It is clear from the above discussion that we might expect to see different behavior in each case, so it’s worth thinking about.
Was the Myanmar rupture supershear at the video and the seismometer locations?
As we speculated in several earlier posts, it now seems pretty clear that the Myanmar earthquake was a significant supershear event, at least for a large part of its rupture length — it traveled too far, too fast to have propagated entirely at sub-Rayleigh speeds.
However, the video was taken only about 124 kilometers south of the epicenter. The seismometer near Naypyidaw was located 246 kilometers south of the epicenter (and about 2.6 kilometers to the west of the rupture trace). The overall timing of the rupture indicates that large parts of it went supershear. But was it supershear at those two locations? Many ruptures only nucleate a supershear rupture for part of their length.
Let’s look at the NPW station first.
Very interestingly, the brand-new seismic data from the NPW station (left) seem to clearly match the velocity pattern of a dilatational precursor (right). Recall that dilatational precursors are a feature of supershear ruptures. Near-field seismic recordings of supershear ruptures are very rare, and so the recordings that do exist have gotten a lot of attention, like the seismogram at PS10 from the 2002 Denali earthquake in Alaska. We suspect that (a lot) more will be said about the NPW data as well.
The prior research discussed above suggests that we can fingerprint supershear ruptures by looking at the fault-parallel versus fault-normal velocities: a fault-parallel velocity that is higher than the fault-normal velocity is diagnostic of a supershear rupture. The higher, the better. At NPW, we see exactly that: a ratio above 1.5.
So, we do think that the rupture near NPW was moving at supershear velocity.
We don’t have the same kind of information from the video. However, we can use the timing from the NPW station to estimate the average rupture speed between the origin and the seismometer. We can also estimate the P wave, S wave, and Rayleigh wave speeds, using seismic recordings of aftershocks.
Furthermore, we know that the rupture itself was detected near Naypyidaw about 51 seconds after the origin. And, we have some speed limits to work with: the rupture can’t propagate between the Rayleigh (Vr) and shear wave (Vs) speeds. There might be some other limits — for instance, supershear ruptures below a special speed known as the Eshelby speed (Ve = (√2)*Vs) are thought to be unstable, so supershear ruptures should mostly travel between the Eshelby and P wave speeds.
First, let’s make a distance-vs-time plot showing these various speeds, along with the estimated rupture arrival time at the NPW station. Our goal is to figure out how we can connect the rupture at the origin (time = 0, distance = 0) to the rupture at near NPW (time = 51 s, distance = 246 km). We have two dots, and some allowable slopes.
To get this plot, we estimated the P and S wave speeds using measured P and S wave arrival times of the M6.7 aftershock at NPW. Our S wave speed (3.41 km/s) compares well with the average S wave speed estimated from regional tomography (Fadil et al., 2023). We calculated the Rayleigh speed Vr = 0.9 * Vs, and the Eshelby speed Ve = sqrt(2)*Vs. The rupture time comes from the onset of the fault-parallel slip pulse at NPW, and does not account for the distance of the station from the fault, which might change the time by about 1 second.
The first thing to notice is that the black square (rupture arrival at NPW) is almost the same as the red dot, representing the Eshelby speed. This means that the average rupture velocity that we observe matches the slowest limit of stable supershear rupture. So, in an average sense, it seems likely that the movie location saw supershear rupture. And, it is possible that the rupture moved at supershear velocity the entire way — although that seems a little unlikely.
We can examine a few other scenarios. What if the rupture started at the Rayleigh speed (the fastest possible “slow” rupture), and then accelerated to the P wave speed (the fastest possible supershear rupture)? In that case, the movie location could record a very fast supershear rupture. Note that these are the speed limits — the sub-Rayleigh rupture can travel no faster than the Rayleigh speed, and the supershear rupture can travel no faster than the P wave speed — allowing us to explore some end-member cases.
What if there was an early episode of supershear rupture, but that rupture somehow slowed down, passing the movie site at Rayleigh speed, and then for some reason accelerated again? In that case, the movie site could witness a pretty typical subshear rupture that had already travelled at that speed for up to ~70 kilometers.
Actually, we can put that up-to-70 kilometers of sub-Rayleigh rupture in a lot of different places, and still fit the data. (Note: we cannot put it too close to the NPW, because then S waves from the sub-Rayleigh rupture would have reached NPW before the rupture itself.)
This line of reasoning suggests to us that we can’t actually predict what we should see in the movie, using only the rupture arrival time at NPW. The movie may have documented a very fast supershear rupture passing by, or a slow subshear rupture, or anything in between. We will have to keep our minds open. Ugh.
Preparing to watch the video, finally
We are really almost very close to being ready to watch the video. But first (fourth?), some quick forensics are in order.
Where are we exactly, and what are we looking at?
Luckily, the video uploader provided coordinates for the video recording in the Youtube comments,: 20°52'55.4"N 96°02'07.0"E. In Google Earth imagery from 2023, this is a large construction zone without any of the buildings we can see in the video. Fortunately, the site is tagged with the company name.
We looked up the website of the Great Success Energy Company, which shows that the company has developed a large solar array at this site over the last several years. The website also has a nice drone view of the newly built power station where the rupture was filmed. We can also clearly see the gate that features in the video, at top left.
This gives us a great starting point, but we still don’t know which side of the fault the power plant is on! A strike-slip fault looks the same from either side; we don’t know if the camera is pointed south or north.
We examined Planet Labs satellite imagery collected in the last few months. While the resolution is low, we can see the power station and its two gates, as well as the two blue-roofed outbuildings visible in the video. The video was filmed looking toward the south. The rupture crosses the solar array, the camera is on the east side of the fault, and in the movie we see the west side of the fault moving northward past the camera.
At this point, we know some of the things we should be looking for when we watch the movie:
Strong motions before the rupture arrives that could (conceivably) discriminate between subshear and supershear rupture.
The amplitude of the fault-parallel motions vs. fault-normal motions.
The total duration and amount of slip at the surface.
Possible direct measurement of the rupture propagation speed
Movements that happen after the rupture arrives.
With that in mind…
Finally, the actual video
We made it!!!
What did you see?
We are lucky that the video captures many objects responding to (and documenting) the ground movement, both before and during the actual rupture. Let’s introduce our cast of characters:
The Gate
Access to the power plant is by two large concrete entryways, of which we see only one in the video. These concrete portals are equipped with two-part rolling gates made of black metal. It is critical to notice that the gates sit on wheels and do not seem to be attached to hinges; they are floating. We will focus on the left-hand part of the gate in the video.
The plants
At left, two landscaping plants sit in a small area of grass, soaking up the sun and providing a gentle touch to the otherwise austere setting. To their right, three-tiered plant in a small pot stands in shadow.
The Upper Water Container and shed #1
This enigmatic character starts out completely hidden from view from behind second story of the building at top left (shed 1). It emerges dramatically from hiding part way through the scene. A similar container sits on the ground below. We suspect that this elevated tank provided pressurized water for the use of workers living in a dormitory.
The Lower Water Container
Less reticent than its more elevated partner, this water container sits comfortable next to shed #1.
The Concrete Driveway
In the foreground, a pristine concrete driveway basks quietly in the sun. The original expansion joints are the only breaks in the surface.
Shed #2
In the background at top left, a small house or shed faces us directly with its door propped partly open.
The Transmission Tower
At upper right, a tall power transmission tower, presumably carrying electricity generated by the solar array, stands quietly.
To make things a little easier, we used an online pixel tracking app (Co-tracker 3) to mark and follow specific pixels in the movie. Our modified movie, trimmed in time, is here. The tracking is imperfect, but we think it at least helps us see the motions all at once.
The video timeline
The timestamp on the top right of the video provides a measurement of time, but unfortunately the clock in the camera is several minutes early — if we took that time (12:46:24) at face value, then the entire sequence would have happened before the earthquake even started (12:50:52). So, we can only use the relative timing of the video frames.
We opened the video in software that gives us more accurate timing for each video frame. We will refer to the time of events in this reference frame, starting at t=0 seconds on the first frame of the video. Notably, there is a freeze-frame at the start of the video that lasts for the first 4.3 seconds (at least in our downloaded file). We then scrolled through the video and noted down the times of various events. The difference between our noted times is the only thing of importance. Any real research product would have to be more rigorous; we are just skimming the cream off the top.
Let’s write down our observations first, and then we will try to interpret them later.
The first action occurs at 9.6 seconds, when the camera starts to shake slightly and the plants jiggle a little.
At 11.7 seconds, the left side of the gate and both of the plants make a slight move to the left. Recall that we are facing south, so left in this case is east. This motion of the gate and plants toward the left (east) indicates movement of the ground toward the west: the ground is moving, and the plants and gate are being left behind due to their inertia.
In the following three figures, the red arrows show the direction of loosely attached objects, and the green arrow shows the inferred direction of the ground motion.
At 12.4 seconds, the gate and plants switch direction, now moving to the right (west). Where previously only the left side of the gate moved (since the right side was already fully extended), now both sides roll to the right. The unsecured upper water container perched behind shed 1 (presumably holding several tons of water) emerges from its hiding spot, tilting to the right and starting to spill water. Because the water tank is probably a meter across or so, the fault-normal motion causing it to emerge is probably about a meter. At this point, we see no evidence of surface cracks anywhere.
At 13.5 seconds, the gate starts to move toward the left again. The plants are now swaying strongly. The upper water container has tilted too far to recover and continues to fall. This might mean that the leftward (eastward) motion was not immediately compensated by a westward motion sufficient to “catch” the water tank.
The first cracks start to appear in the concrete driveway at 13.8 seconds. This is the first visible sign of ground breaking. The cracks will grow visibly over the next several seconds. One expansion joint will open up and then close again.
At 14.2 seconds, we notice the first sign of the large surface rupture that will grow into a scarp, crossing the dirt road area at top left. The water tank is spilling and is about to plummet from its perch, and the door to the house behind has opened.
At 15.0 seconds, the falling water tank hits the ground. The resulting explosion of water blasts a small black object to the right, where it hits the scarp and bounces back. It is remarkable that the entire scarp grows during the time that the water container is airborne. This is notably similar to the eyewitness testimony of Witness #12 of the Luzon earthquake — blink, and you will miss the scarp formation!
As the fault scarp grows, the roof of the house behind (shed #2), which seems to be built on the fault trace, starts to partially collapse. We see the scarp rise upward and slip sideways all at once. As far as we can tell, there is no separation of upward and sideways motion.
At this point, the slip on the fault is well underway. The dramatic sideways motion will continue until 15.9 seconds. During the slip, several areas directly along the fault trace experience significant upward motion (see the potted plants and road just above the gate). This is not necessarily due to actual uplift; strike-slip faults often develop small pressure ridges (areas of uplift) and sags (areas of downlift) along the fault trace. These do not necessarily indicate actual tectonic displacement along the fault, but can be due to local geometric imperfections of the fault trace, especially where the fault has to cut across young surface sediments for the first time. We cannot detect any fault-normal motions during the period of fault slip.
At 16.0 seconds, just as the slip on the fault ends, we see the start of the collapse of the transmission tower at top right. Also at this time, the cracks in the concrete stop growing.
At the same time, our tall plant performs a deep final bow and a spray of some kind of material (water?) appears at the very left side of the image. It appears that the end of slip was associated with a strong deceleration on both sides of the fault.
At 16.6 seconds, the lower water container (finally!) gets a chance to participate, by falling towards the camera.
This movement coincides with a final pulse of west-directed fault-normal motion, moving the plants and gate to the left once again. This is our final noted event. The video continues to 27 seconds.
We are left with a shifted landscape, but there is no evidence of particularly extreme damage in this location. How the solar panel array fared is another question.
The timeline, in short
We noted down our observations on a simple timeline. We would like to reiterate that this is not a research product, it is just a record of our own observations. Other people might see things differently, or interpret the same observations differently. We would love to hear what other people see!
Interpretations
So, how do these interpretations relate to our discussion of seismic waves?
The P wave
First, we cannot tell from the video when the hypocentral P wave arrives, because at this distance it would have been too small to detect by video, and it would likely have been cut out of the cropped video. The video starts with the plants already shaking, which could be due to wind or due to a rising P wave train from the approaching rupture.
We do not expect a large P wave arrival here, because P waves are not radiated along the fault plane itself, or along the fault-perpendicular plane passing through the origin:
It’s too bad that the video is incorrectly timed and truncated at the beginning, because we can guess the P-wave arrival time (~21.5 seconds after the origin), which would have given us a pinpoint to work with. Oh well.
The time between the onset of shaking and the rupture
What we can detect is the time between the first sensible shaking and the onset of surface deformation: ~4.2 seconds. This is quite similar to the three seismic records we showed before (Denali, Izmit, and Myanmar).
Could this shaking be the oncoming S waves? This seems possible: if the rupture was moving at the Rayleigh velocity for 71 kilometers prior to the movie site (the slow to fast transition model discussed above), then the S waves would have outdistanced the rupture, and we would expect about 3.1 seconds of S wave shaking before the rupture arrives. In this case, the rupture at the video site would be a sub-Rayleigh event. The timing isn’t quite right, but it’s pretty close.
In some other cases where seismometers have recorded nearby supershear ruptures, the earthquake only went supershear after a period of sub-Rayleigh rupture, and that early dawdling allowed the S waves from earlier parts of the earthquake to reach seismometers before the rupture itself — much like the tortoise outracing the hare.
Alternatively, could this shaking be the dilatational precursor in front of a supershear rupture? The timing, at least, seems to be consistent with that, too: ~3 seconds of early motion were recorded in the NPW supershear seismic record, and also at PS10. But as we will discuss in a moment, the fault-normal motions in the video seem to be in the wrong direction.
We think it might even be possible for a rupture propagating at supershear velocities to still have S waves arrive at the video location before the rupture does. This seems like a contradiction! How do we think this might happen? Well, if the rupture front is not vertical but instead tilted backward, then the rupture might arrive below the video site first, with the surface rupture trailing behind. If the tilt is great enough, then there might be enough time for the S waves from that deeper rupture to travel upward to the surface, before the shallow rupture catches up. The following figure is a back-of-the-envelope sketch. The red line indicates the part of the rupture for which the S waves would first arrive at the surface; the P waves would arrive about 1 second prior.
This kind of tilted rupture front is actually not that unexpected. That is because seismic velocities tend to increase strongly downward into the crust. When there is a layer of sedimentary deposits near the surface, Rayleigh wave speeds can be especially low in the upper several kilometers. This could slow the propagation of the shallow rupture versus the deep rupture.
Abdelmeguid et al. (2025) modeled exactly this scenario, for both subshear and supershear ruptures. Their models do show this kind of shallow setback in the rupture front. It would be extremely interesting to see synthetic seismograms for near-fault sites generated from these models, to test whether this effect could explain the short-duration shaking in advance of the rupture arrival in the video.
The direction of shaking
The first strong motions appear to be fault-normal pulses, first to the west, and then most strongly directed toward the east (toppling the water tank). This is quite consistent with a subshear rupture, in which the main fault-normal motion is expected to first go slightly westward, then strongly eastward. A supershear rupture in which the fault-normal motion due to the dilatational precursor should be only westward, according to the models we showed earlier.
This idea was first proposed by Tom Gabrieli on Bluesky quite soon after the video came to light, based on a similar approach to our own, and also using figures from Mello et al. (2010). His early analysis seems to assume the video is looking north, and that the motions of the objects reflect the motion of the ground directly, not in an opposite sense. Fortunately, the two effects cancel each other out, so we find ourselves agreeing with the conclusions.
The surface rupture timing
The period of visible surface deformation lasts about 2.1 seconds, counting the concrete cracking, but we estimate that the real surface slip involving the fault trace only lasts for about 1.7 seconds. The preprint by Latour et al. estimates a shorter slip duration of about 1.3 seconds, estimated in a different way. These values are very similar to the eyewitness accounts from Idaho and Luzon, and the records from near-fault seismometers, including the seismometer at NPW.
This short slip duration basically proves the validity of the rupture pulse model for these particular kinds of events: long, strike-slip, and occasionally supershear ruptures.
The rate of rupture propagation
We attempted to measure the actual rupture propagation speed using frame-by-frame examination of the first motion in the near-field (to the right of the video) and the far-field (at the top left), a distance of about 100 meters. We saw motion start during two consecutive frames. That constraint still allows speeds between infinity and about 0.7 km/s; a useless measurement. If the video had a much faster frame rate, we would have probably been able to measure the surface rupture propagation speed directly! This is an important consideration for future camera installations along fault traces.
The fault slip
How much surface slip actually occurred? This is a surprisingly tricky thing to estimate, because all of the objects on the far side of the rupture are different distances away, and there aren’t a lot of easy ways to estimate distance.
We think the fault slipped a little more than two meters during the video. Here is our thought process:
During the slip, small posts in the background move relative to the fixed fence. If we can estimate the distance of the posts from the camera, and compare their motion to the scale provided by the fence, we can estimate the slip on the fault accounting for parallax effects.
How far apart are the fence posts? We estimated the width of the fence panels by dividing the full length of the fence from Planet imagery (~64 meters) by the number of fence panels visible in the aerial shot from the Great Success Energy company (24), we get 2.7 meters per panel. This is consistent with the width of the shipping container structures inside the compound.
Using the movie and the aerial photo provided by GSE, we estimated the lines of sight from the camera to the landscape. We identified specific fence posts, along with specific smaller posts located on both sides of a fault-parallel road in the background. The small posts were offset from the fence during the slip. Some trigonometry later, we calculated independent estimates of the total offset: 2.12, 2.09, 2.40, 2.26, 2.08, 2.41 meters, for a combined estimate of 2.23 ± 0.3 meters.
This value is very surprising, because satellite imagery for this area indicates that the blocks on either side of the fault moved ~3.5-4.5 meters in the earthquake! In order to verify this, we ran our own pixel correlation using satellite imagery from Planet Labs around the camera site, with the program MicMac. The pixels are 3 meters wide, but the correlation process allows estimation of much smaller offsets, quite accurately.
As expected, we see a pretty sharp fault trace near the video location (profile D), with about 4 meters of slip across it.
So, where is the missing slip?
Option 1 is that our measurement of slip from the video is somehow wrong. There are always uncertainties in these measurements, and we could potentially at least close the gap that way. However, the factor of 2 difference between our estimate and the satellite displacements is hard to explain.
Option 2 is that the rest of the offset happened somewhere else — as slip on other fault strands, or as bulk deformation around the fault. We do not believe that there is another equally large surface offset nearby. However, we do know that some off-fault deformation happened near the video site (the concrete cracks are examples), and there is room for significant off-fault deformation in the pixel correlation data. It is pretty common for faults to slip more at depth than at the surface, with the remainder of the offset accommodated in a zone around the fault in the near-surface sediments. For instance, in the 1930 M7.3 Idu, Japan, earthquake, a tunnel about 150 meters below the surface (in volcanic rock) was offset about two and half meters, while at the surface the fault slipped less than a meter (in sandy clay lake deposits) (Bonilla, 1970). So, it is possible that the slip below the video site was greater at depth (explaining the shift observed by satellite images), but that the shallow sediments absorbed a significant share of the offset as distributed deformation, resulting in less offset along the fault trace itself.
Option 3 is that the rest of the slip happened at a different time. We only see about half a minute of time in the video, and it’s possible that further slip happened later. Fault slip estimates from satellite images involve differencing two images taken days apart. In this case, the first post-event Sentinel-2 flyover occurred on March 30th, about two days after the earthquake. It is fairly common for faults to continue to slip slowly after the main fast event, without radiating much seismic energy: this is known as post-seismic creep, or afterslip. Usually (but not always), this adds up to significantly less than a third of the slip in the mainshock, and it more typically occurs around the edges of the slip patch. It would be surprising if this part of the fault aseismically slipped ~1.5-2.5 meters, effectively doubling the coseismic slip.
It seems that we will have to await more information, or a more convincing analysis that demonstrates a much larger coseismic slip in the video than our own attempt. On-site measurements of surface displacement at the fault trace would be of great help!
The end, at last. What does it all mean?
Well, it’s been a long ride, and we didn’t emerge with many certainties.
Writing this post took a lot of time (we started on the day we first saw the video), mostly because we had to read through a large literature about fault rupture physics that we had previously ignored. We had that luxury because most of the processes that we normally write about are somewhat divorced from the details of the rupture process. However, those details are fundamental to the science of earthquakes. The fact that we felt the need to dig into these details illustrates the great importance of truly new observations.
Supershear ruptures have certainly caught the public attention: the word itself is extremely exciting and a bit hyperbolic. Yes, there are certainly implications of supershear vs. subshear rupture for damage. However, it isn’t exactly clear how the various effects balance out — so much so, that determining whether a rupture was supershear or not often involves quite a lot of pretty technical detective work!
We think there is a better reason to care about rupture velocity. In a way, supershear ruptures “lift the veil” on how faults actually work, by separating out the rupture tip from the otherwise all-encompassing radiated wave field. This allows seismologists to ask more pointed questions directly addressing the physics of faulting. The fact that supershear ruptures were first predicted by theory, and then produced in the laboratory, well before they were reliably observed in nature, is actually one of the great predictive successes of physics (in our opinion).
Three decades of experience have now confirmed that large, straight, strike-slip faults are pretty likely to produce supershear ruptures during large magnitude events. It seems like we can expect a significant rupture with at least some component of supershear slip about once per year, globally. The trick, of course, is that there are a lot of faults that can possibly host one of these ruptures, and we don’t know which one will rupture next. Can we reasonably improve our chances of getting more data like this video?
Many instruments are used to record earthquakes, but cameras have not historically been included in the official science package, or at least not very often. There is an underwhelming video of an earthquake on the Parkfield section of the San Andreas fault, which shows some swaying trees with no surface rupture. A set of instruments including cameras are included in a recent proposal to densely instrument a section of the Elsinore Fault southeast of Los Angeles (the RuFZO initiative), with additions of “standard discs of known mass and size for benchmarks” and “camera angles that allow for image differencing.” Indeed, a well-designed camera network that allows quantitative measurements and views from multiple angles could certainly be extremely interesting — especially if it had a high enough frame rate to actually record the rupture moving across the landscape!
Over the last decade, there’s also been an explosion of footage from CCTV cameras and cell phones. Most of that footage shows shaking, landslides, building collapse, and other earthquake effects — these processes can occur in a wide zone around the fault. Capturing the rupture itself requires being in the right place, looking in the right direction, with enough lighting to see what’s going on.
That said, it seems very likely that more videos of the Myanmar fault rupture exist, especially near Mandalay/Sagaing and the capital Naypyidaw (and their airports, both of which lie directly next to the fault trace). How can we find these videos, and incentivize their owners to upload them online? Similarly, how can we collect the eyewitness accounts of the fault rupture process? It seems like this would be a useful endeavor.
To maximize the chances of recording the next big earthquake rupture that happens, we could try to make sure that already-installed cameras are aimed in the right direction (and, hopefully, have accurate clocks). If any of our readers are living near a fault trace, you might consider maintaining an accurately timed camera pointed at the fault. Who knows: you might be the next Great Success Energy Company!
On that note, we would like to thank Great Success Energy for installing their security cameras with a view of the Sagaing Fault, and then for uploading the video so that the world can finally see a fault move! They have made an important contribution to earthquake science — a truly Great Success!
As always, we are eager to hear from readers with comments, criticisms, and corrections. Tell us what we missed, or got wrong, in the comments!
References:
Abdelmeguid, M., Elbanna, A. and Rosakis, A., 2025. Ground motion characteristics of subshear and supershear ruptures in the presence of sediment layers. Geophysical Journal International, 240(2), pp.967-987. https://doi.org/10.1093/gji/ggae422
Benz, H., Furlong, K., Herman, M., McKenzie, K., Schmitt, R., Reitman, N., Thompson, E., Barnhart, W., Yeck, W., and Pananont, P. (2025). 2025 M 7.7 Mandalay, Burma (Myanmar) Earthquake, U.S. Geological Survey StoryMap, https://storymaps.arcgis.com/stories/5f3e33e35c5247c9bf5204fa0d6e56e5
Bonilla, M.G., 1970. Surface faulting and related effect. In: Earthquake Engineering, ed. R. Wiegel, Prentice-Hall, Inc., p. 58.
Bouchon, M., Toksöz, N., Karabulut, H., Bouin, M.P., Dietrich, M., Aktar, M. and Edie, M., 2000. Seismic imaging of the 1999 Izmit (Turkey) rupture inferred from the near‐fault recordings. Geophysical Research Letters, 27(18), pp.3013-3016. https://doi.org/10.1029/2000GL011761
Bradley, K., Hubbard, J., 2025. Surface ruptures of the Myanmar M7.7 earthquake mapped from space. Earthquake Insights, https://doi.org/10.62481/51b7df8c
Bradley, K., Hubbard, J., 2025. Updates on the M7.7 Myanmar earthquake. Earthquake Insights, https://doi.org/10.62481/9e49eb4a
Burtin, A., Hovius, N. and Turowski, J.M., 2016. Seismic monitoring of torrential and fluvial processes. Earth Surface Dynamics, 4(2), pp.285-307. https://doi.org/10.5194/esurf-4-285-2016
Churchill, R.M., Werner, M.J., Biggs, J. and Fagereng, Å., 2022. Afterslip moment scaling and variability from a global compilation of estimates. Journal of Geophysical Research: Solid Earth, 127(4), p.e2021JB023897. https://doi.org/10.1029/2021JB023897
Dunham, E.M. and Archuleta, R.J., 2004. Evidence for a supershear transient during the 2002 Denali fault earthquake. Bulletin of the Seismological Society of America, 94(6B), pp.S256-S268. https://doi.org/10.1785/0120040616
Ellsworth, W.L., Celebi, M., Evans, J.R., Jensen, E.G., Kayen, R., Metz, M.C., Nyman, D.J., Roddick, J.W., Spudich, P. and Stephens, C.D., 2004. Near-field ground motion of the 2002 Denali fault, Alaska, earthquake recorded at pump station 10. Earthquake spectra, 20(3), pp.597-615. https://doi.org/10.1193/1.1778172
Fadil, W., Wei, S., Bradley, K., Wang, Y., He, Y., Sandvol, E., Huang, B.S., Hubbard, J., Thant, M. and Htwe, Y.M.M., 2023. Active faults revealed and new constraints on their seismogenic depth from a high‐resolution regional focal mechanism catalog in Myanmar (2016–2021). Bulletin of the Seismological Society of America, 113(2), pp.613-635. https://doi.org/10.1785/0120220195
Heaton, T.H., 1990. Evidence for and implications of self-healing pulses of slip in earthquake rupture. Physics of the Earth and Planetary Interiors, 64(1), pp.1-20. https://doi.org/10.1016/0031-9201(90)90002-F
Lai, S.T., Oo, K.M., Htwe, Y.M.M., Yi, T., Than, H.H., Than, O., Min, Z., Oo, T.M., Maung, P.M., Bindi, D. and Cotton, F., 2025 preprint. Capacity Building Enables Unique Near-Fault Observations of the destructive 2025 M w 7.7 Myanmar Earthquake. Earth System Science Data Discussions, 2025, pp.1-23. https://doi.org/10.5194/essd-2025-216
Latour, S., Lebihain, M., Bhat, H.S., Twardzik, C., Bletery, Q., Hudnut, K.W., Passelègue, F., 2025 preprint. Direct estimation of earthquake source properties from a single CCTV camera. arXiv Geophysics, https://doi.org/10.48550/arXiv.2505.15461
Mello, M., Bhat, H.S. and Rosakis, A.J., 2016. Spatiotemporal properties of Sub-Rayleigh and supershear rupture velocity fields: Theory and experiments. Journal of the Mechanics and Physics of Solids, 93, pp.153-181. https://doi.org/10.1016/j.jmps.2016.02.031
Mello, M., Bhat, H.S., Rosakis, A.J. and Kanamori, H., 2010. Identifying the unique ground motion signatures of supershear earthquakes: Theory and experiments. Tectonophysics, 493(3-4), pp.297-326. https://doi.org/10.1016/j.tecto.2010.07.003
Reitman, N.G., Wang, Y., Kuo, Y.T., Hanagan, C., Hatem, A.E., DuRoss, C.B., Barnhart, W.D., Yin, H.Z., Briggs, R.W., Thompson-Jobe, J.A., Nicovich, S.R., Lynch, E.M., Powell, J., and Schmitt, R.G. (2025). Remote Surface Rupture Observations for the M7.7 2025 Mandalay, Burma (Myanmar) Earthquake, https://doi.org/10.5066/P1RYMWCK
Wallace, R.E., 1984. Eyewitness account of surface faulting during the earthquake of 28 October 1983, Borah Peak, Idaho. Bulletin of the Seismological Society of America, 74(3), pp.1091-1094. https://doi.org/10.1785/BSSA0740031091
Yao, L., Ma, S. and Di Toro, G., 2023. Coseismic fault sealing and fluid pressurization during earthquakes. Nature Communications, 14(1), p.1136. https://doi.org/10.1038/s41467-023-36839-9
Yomogida, K. and Nakata, T., 1994. Large slip velocity of the surface rupture associated with the 1990 Luzon earthquake. Geophysical Research Letters, 21(17), pp.1799-1802. https://doi.org/10.1029/94GL00515
Zhang, Z., Xu, J., Huang, H. and Chen, X., 2017. Seismic characteristics of supershear and sub‐Rayleigh earthquakes: Implication from simple cases. Geophysical Research Letters, 44(13), pp.6712-6717. https://doi.org/10.1002/2017GL074158
Judith, I’ve kept this post open in Safari since you posted it so that I can refer to it when time allows. I contend that those of us “scientist-wanna-bes” wh find seismology compulsively engaging and wade through this particular material should be given an opportunity to be awarded a minor in geology just for “stick-to-itiveness” 😏. Thanks for this. Completely fascinating and informative. What a lab! What a lecture!
This is an extraordinarily interesting post. Thank you so much for all the time and effort that went into it.