Download this episode
When the Earth’s tectonic plates collide and slide, School of Earth and Atmospheric Sciences Professor Zhigang Peng takes data from seismic sensors and creates “earthquake music.” The results can help scientists learn more about what goes on beneath the planet’s crust.
Renay San Miguel: Hello, I’m Renay San Miguel, and this is ScienceMatters, the podcast of the Georgia Tech College of Sciences.
A massive 2011 earthquake in northern Japan spawned a tsunami that devastated coastlines across southeast Asia.
Here’s what it sounded like above ground:
(Japan earthquake/tsunami TV coverage)
Renay San Miguel: Now listen to what it sounded like below ground:
(Japan seismic sounds)
Renay San Miguel: To Zhigang Peng, professor of geophysics in the School of Earth and Atmospheric Sciences, these earthquake sounds are music to his ears.
Actually, Peng takes that seismic data, and he turns it into music. Earthquake music.
Renay San Miguel: Peng, who’s been at Georgia Tech since 2006, gets seismic data from sources such as the High Sensitivity Seismograph Network in Japan, the University of California at Berkeley’s High Resolution Seismic Network, and the Incorporated Research Institutions for Seismology, sponsored by the National Science Foundation.
He takes that data and sonifies it; that is, he assigns sounds to the data.
For Peng, sonifying seismic data may get the Earth to give up more of its secrets. It may, with computers lending artificial intelligence and machine learning, help us know more about the causes of earthquakes, and their physics.
Renay San Miguel: A desire to learn more about earthquakes is why Peng pivoted from a computer science track to geophysics when he started higher education in his native China. Because even with the great technological advances of the last few decades -- including better, more sensitive seismic sensors -- Peng says we still don’t know a lot about what goes on under our feet.
Zhigang Peng: We can send like satellites to the sky, we can send divers, but the deepest place we can go inside Earth is no more than roughly about 10 kilometers. There’s a lot of things we don’t know.
Renay San Miguel: By the way, 10 kilometers? That’s a little more than six miles deep into the Earth.
Back here above the Earth’s crust, Peng and I sat in his office to talk about his seismic data research, including his earthquake music project, for which he received funding in 2017 from the Creative Curricular Initiative of the Georgia Tech Council on the Arts.
Renay San Miguel: One of them sounded to me like giants throwing things at each other on the Earth. It was just incredible to listen to it—very hypnotic. What does earthquake music, though, teach you about earthquakes?
Zhigang Peng: There are several ways we can learn from it. If we can make an energy to soundwaves, everybody have the experience with sound, as you know. So in that case, you automatically convert something that is a little bit abstract and a little bit complicated to something that people can relate to their daily life. So by doing this mapping or by doing this connection, we can explain some phenomenons to people, sometimes even like elementary school students. They like it very much because by listening to the sound of the seismic event, they can quickly tell the difference between a compression wave with a shear wave. They can tell the difference between a regular earthquake was a slow event or something different. So our human ear is actually very complicated and sophisticated, and it can quickly tell the differences, sometimes even better than the eye. So by adding this extra dimension of sense, we can use that as an effective tool to tell people about what are the differences and why it matters.
Renay San Miguel: Gotcha. Well I have to ask you, on your web it says that these also help you understand the physics of seismic data. How do you mean?
Zhigang Peng: Yes. So I can give you an example. So one of the things you can hear, or tell, is that if I play regular earthquakes to you, it sounds like either fireworks or something that is rapid fire. So another way to think about that is if I'm slapping my finger like this— [fingers snapping] —you did a better job. So the way to do it, you know, why related to earthquake is while I'm pressing the finger and build up stress and eventually slide, so what you hear, this snap sound is basically the disturbance that travels through the air and that’s how you hear.
Renay San Miguel: OK.
Zhigang Peng: So that's a regular earthquake.
Renay San Miguel: Yeah.
Zhigang Peng: Now, on the other hand, if I do something slightly different, if I put my hands together and rub against each other, [friction sound of hands rubbing together] so you hear something not as clear here as the slapping finger, but you can hear something, right?
Renay San Miguel: Yeah, absolutely.
Zhigang Peng: Now in addition to that, another thing which we're doing right now is to try to follow what recently people have been doing in image and voice recognition, artificial intelligence, and deep-learning. There are a lot of new tools that’s been developed over the past few years and try to use computer as a way to automatically tell and distinguish between different types of images and different sound.
(Earthquake seismic sounds)
Zhigang Peng: If you have seen a lot of them, then the computer can automatically go and pick new events that almost can match and, sometimes, can outperform human performance.
Renay San Miguel: Six years ago, Peng and his researchers made waves by suggesting that strong hurricanes could trigger earthquakes. That idea was based on the 2011 quake in Virginia recorded shortly before Hurricane Irene slammed into the state. When the storm’s eye reached the quake’s epicenter, a higher rate of aftershocks than usual was recorded. Peng used pattern recognition techniques to separate quake-related seismic data from similar data caused by the storm.
But publication of that study was delayed as questions came up about the single-case nature of the research.
Zhigang Peng: We had a hard time to publish the result because it was a little bit controversial.
So the result did publish earlier this year finally. The main observation is in 2011, there was a magnitude 5.7 earthquake in Virginia. So even though it's not big, but it was widely felt in Eastern U.S. including here in Georgia. People felt the event and actually called us saying “What's going on?” They felt the event. So after—two days after this earthquake, there was a Category 2 hurricane passed by, Hurricane Irene I believe. So it happened to be—the eye center happened to pass through the epicenter of this Virginia—
Renay San Miguel: The eye meets the epicenter.
Zhigang Peng: Exactly. So we don't get this quite often, OK? What we found out is that as the eye center passed through the epicenter, the number of aftershocks, especially the shallow ones—the ones that are very close to surface—increase by a factor of four during about two days as the hurricane center pass through. So our hypothesis is that we suspect that because of the low pressure systems of the eye center, that it effectively unclamped the fault.
Renay San Miguel: Take a little of the weight off.
Zhigang Peng: Yeah. The reason why we had, you know, a hard time to publish is because people came back and, you know, the reviewers came back and say, “You know, it seems reasonable, but it's only one case study. Can you tell us how often it is happening and could this be explained by random chance?” So that's the question that what we're trying to face. So we did actually look at a few other cases. But, unfortunately, if you want to have this sort of perfect condition, we don't get quite often, so we cannot come up with, let’s say, another example or another set that match with this sort of observation we have got.
Renay San Miguel: An earthquake in the area and then a hurricane moving through that exact same area?
Zhigang Peng: Right. Exactly. Now, because of this study and also recently we started another collaboration with a group in Florida International University, we actually managed to get a funding from NASA to support this research.
Renay San Miguel: OK.
Zhigang Peng: So right now—so rather than focus on the Eastern U.S, because we don't get many earthquakes, right—that's the problem—our new study area is Taiwan in East Asia because over there they've got lots of earthquakes, and they also got a lot of so-called typhoons.
Renay San Miguel: Typhoons, yeah.
Zhigang Peng: So our focus there is to understand the relationship between typhoons, especially extreme wet typhoon, those typhoons that will bring a lot of rainfall. And, of course, when you have rainfall, you can trigger a lot of landslide, right? And when you have landslide, you are basically dump and slide a lot of sediments into the river, and the river will carry them away.
And so we're trying to understand whether or not there is a correlational relationship between extreme where typhoon, landslide, and subsequent earthquake or subsequent seismic event.
Renay San Miguel: So tell me, where are we at in 2018 with seismic research?
Zhigang Peng: So I think we're pretty clear now that what is the force and the driving force behind it. You know, we also mostly know where earthquakes would occur. They mostly occur along the so-called plate boundary. That's where, you know, the tectonic plates are moving right into each other, so that's where the stress is built up the most.
Now what we don't know is A.) Some outliers like, for example, sometimes in the Central and Eastern U.S., we get earthquakes, right? We don't get too many, but sometimes it caught people by surprise. Where does the force come from? In addition, we not only have earthquakes occur in the crust, which is, you know, the brittle, the brittle part of the Earth. We also have earthquakes occur at a few hundred kilometers at depths. So the deepest one is up to about 700 kilometers. So you can imagine at such larger depths, there's high temperature and pressure, so the rock there is probably not going to behave brittle; it's probably more ductile-like flow.
Renay San Miguel: Yeah.
Zhigang Peng: just a few weeks ago there was a magnitude 8.2 earthquake in Tonga, which in southern Pacific. And the hypocenter of depths there is about 600 kilometers. And a few days later there was another magnitude 7.9 earthquake occurred not far from that epicenter. So there are lots of things going on right now at that depths in that region. You know, why what's going on there?
We wanted to better understand what's going on so that we can, of course, explain what's the underlying physics behind that. So that's something we don't know.
Another question is we don't know what else can cause, can trigger, or delay the occurrence of earthquake. We know that, you know, that in the driving force is plate tectonics. We know that those things build up slowly in time, but we don't know what is the straw that broke the camel’s back.
Renay San Miguel: When it breaks, when the stress gives way, something has to happen.
Zhigang Peng: Exactly. Right. And in addition to that, one thing we do know is that when you are close to that threshold, any perturbation could tip it over. So in that case, it could happen this year; it could happen ten years later. It just depends on what the perturbation, the last part of the perturbation matters a lot. And that's actually one of the research we’re studying here.
Renay San Miguel: Let’s take a quick detour back to Peng’s research in the difference between regular earthquakes and what he calls “slow” quakes, which are mostly harmless and hard to detect. Understanding the slow quakes better may point to more information about the physics of earthquakes and whether seismologists will eventually be able to predict them.
Zhigang Peng: They are very similar. They both of them occur along some fault lines, and both of them associate with sliding. The only difference, of course, is one slide very fast, so it can generate significant shaking and damages. Now, on the other hand, slow earthquake because it slides so slowly, that most of time people didn't feel them.
Renay San Miguel: OK.
Zhigang Peng: So they are harmless in some way. And people may ask, you know, like, “Why do you study, why the important because they’re not causing trouble?” Why do we study them, right?
Some of us, including myself, believe that before some large earthquake that there are slow slips that precede them. We believe that someday, by carefully monitoring those slow-slip event, we may, hopefully, come to the time that we can use them to tell when and where big earthquakes will happen.
Renay San Miguel: OK, so to help you with predictions mainly?
Zhigang Peng: Yes.
Renay San Miguel: OK.
Zhigang Peng: Even though, you know, when you talk to seismologists, we normally don't like to use the prediction.
Renay San Miguel: I know. Nobody wants to talk about predictions.
Zhigang Peng: Right, but I think that, you know, if you talk to any other people, the first question people always ask is, “Yeah, what can you tell me and when the next one is going to happen,” right? That's where people care the most. So I think we shouldn't ignore and we shouldn't pretend, we shouldn't say that, you know, that something we never will ever to achieve. I think that with new technology, new observations, and if you put everything together, I think we have a pretty good chance.
Renay San Miguel: In case you’re wondering, Peng has experienced earthquakes several times in his life, both as a child growing up in China, and as an adult researcher.
Also, thanks to his sonification of seismic sounds and his other research, he’s come full circle in his academic life. Peng finally gets to play with computer science, even as he studies the forces that make the Earth move.
My thanks to Zhigang Peng, professor in the School of Earth and Atmospheric Sciences.
Siyan Zhou, formerly a research associate in the School of Psychology, composed our theme music.
If you enjoyed this episode, please subscribe to our podcast from Apple podcasts or SoundCloud.
This is ScienceMatters, the podcast of the Georgia Tech College of Sciences. I’m RSM. Thank you for listening.