Scientists Work to Predict Earthquakes In the wake of last week's massive and deadly earthquake on the Pakistan border, how likely is it that scientists could predict quakes?
NPR logo

Scientists Work to Predict Earthquakes

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Scientists Work to Predict Earthquakes

Scientists Work to Predict Earthquakes

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


Switching gears now, to the business of predicting earthquakes; from one possibility of a flu outbreak, a disaster, to a disaster that has already happened. People in Pakistan had no advanced warning that a quake was about to strike, no chance to get out of their homes or their office buildings before the buildings crumbled. You know, scientists have been studying earthquakes for years. They've been trying to figure out if there are signs that might tell us when one is on the way. Does the ground move a certain way? You know, we've heard things over the years about water in wells and, you know, maybe these things that happen in advance that people might be warned. Well, have they been making any progress? I know I've been following this story for 25 years, and that's what we're going to be talking about the rest of the hour, the science of earthquake prediction. And if you'd like to join us, our number is 1 (800) 989-8255.

And in fact, earthquake prediction falls into two different kinds of categories. You have this long-term prediction, which tells us what the chances are that a quake might hit, well, in a certain spot in the coming years or decades, right? Then you have the other short-term prediction: How can we know if a quake is going to hit in a few hours or a few days--give us enough warning, you know, to get out of there? One of us--one quake--says, `Hey, you know, maybe we shouldn't be building there.' One says, `Well, we've got buildings there already. How do we know to get out of those buildings and get out of the way?'

Well, my guests for the rest of the hour work on these two different types. Donald Turcotte is a professor of geology at the University of California at Davis. He works on a project called Virtual California, which looks at long-range earthquake predictions in that state, and he joins us by phone.

Thanks for being with us today, Dr. Turcotte.

Dr. DONALD TURCOTTE (University of California, Davis): Thank you very much.

FLATOW: You're welcome.

And William Bakun is at the US Geological Survey in Menlo Park, California. He works on the Parkfield project. This is a project that I've been out--20 years ago I went walking the Parkfield fault line there looking at all these setups they had. Al Lind of the old US Geological Survey took me around there.

And I understand, Dr. Bakun, you worked with Al Lind and published papers with him.

Dr. WILLIAM BAKUN (US Geological Survey): Hi. Yes, nice to be with you.

FLATOW: Thank you.

Dr. BAKUN: Yes, Al and I are old colleagues.

FLATOW: Yes. And that's what we're going to talk about this hour on TALK OF THE NATION/SCIENCE FRIDAY from NPR News.

Let me begin with you, Dr. Bakun. This project that has been going for years at Parkfield--you know, Parkfield had this history of having an earthquake--What?--every couple of decades, 20, 25 years, for almost a century, if I remember correctly. And you thought that would be a prime spot to watch for the precursors of an earthquake. So you put all kinds of instrumentation there, right?

Dr. BAKUN: Yes, we instrumented the Parkfield section with lots of different kinds of instruments in all the places that we thought likely to see possible precursors to the next Parkfield earthquake.

FLATOW: And it didn't happen on--you predicted it would happen when, in '83 or something?

Dr. BAKUN: Oh, yes, we predicted that the earthquake would happen before 1993, and that prediction was wrong, of course. But fortunately, we kept the instrumentation in place and it was all pretty much all working properly when the Parkfield earthquake on 20 September of last year occurred. And we recorded some excellent data that answer some of the questions we were posing for the Parkfield section.

FLATOW: Did it show any precursors before it erupted?

Dr. BAKUN: It really showed very little. We--the earlier Parkfield earthquakes had been characterized by very prominent foreshock activity and there were suggestions of possible strain changes as well. The instruments that we had in place showed very little. In fact, there were no foreshocks down to magnitude 0 and there was only some very problematic possible strain changes in the 24 hours just before the earthquake, but nothing of a level that would allow a public warning.

FLATOW: Yes. That's--you must have been a little bit disappointed.

Dr. BAKUN: Well, in a sense we were, but we did get an answer to the question, is short-term earthquake prediction feasible? And the conclusion is that we don't know how to do it now and it's going to be very difficult. That doesn't mean that someone in the future won't come up with a clever scheme that would allow us to do it. But at the present time, it really is not something that we can rely on.

FLATOW: All right, we have to take a short break. We'll come back and talk lots more with Donald Turcotte and William Bakun after this break. Stay with us.

I'm Ira Flatow. This is TALK OF THE NATION/SCIENCE FRIDAY from NPR News.


FLATOW: You're listening to TALK OF THE NATION/SCIENCE FRIDAY. I'm Ira Flatow.

We're talking this hour about the possibility of predicting earthquakes with my guests, Donald Turcotte, professor of geology, University of California at Davis, and William Bakun at the US Geological Survey in Menlo Park, California. Our number, 1 (800) 989-8255.

Dr. Turcotte, Dr. Bakun says, you know, we so far don't have that short-term prediction scheme nailed down yet. Are we anywhere close to making a long-term prediction scheme?

Dr. TURCOTTE: Well, basically, you would like to know the risk of having earthquakes in various places over the future. And at the present time, mostly this has been done in a rather straightforward, statistical way. And our interest is moving forward towards the kind of modeling that is done for weather forecasting, where you simulate the physics of the problem. And from that, you get a better understanding of when earthquakes are going to occur and their behavior. This is far behind the weather-forecasting-type work, but I believe there is a real hope in the future...

FLATOW: Right.

Dr. TURCOTTE: ...that progress can be made.

FLATOW: But you can predict with certainty, for example, that there are going to be certain places in the world where earthquakes are going to be more likely to occur, and one of them certainly was in Pakistan.

Dr. TURCOTTE: Yeah, that is absolutely correct. And to a first approximation, big earthquakes occur where little earthquakes occur. So if little earthquakes are occurring in a region, you can estimate with moderate accuracy when bigger earthquakes in that region will occur. It doesn't work perfectly, but it's far better than not having any idea where an earthquake will occur.

FLATOW: Tell us about--a little bit--what kind of earthquake--what was so lethal about that earthquake that happened in Pakistan?

Dr. TURCOTTE: Well, by far the most lethal part is the construction. If you have unreinforced masonry walls that are free to collapse, you have very high death rates. And this was the case in Turkey a few years ago and throughout the Middle East, also largely in South America in many places, Mexico. So construction is the number-one problem. It was a shallow earthquake; it was moderately severe, not terribly severe, and it just happened at the wrong place and--very high population density, and the net result was very catastrophic.

FLATOW: And why did that earth--what's going on in the crust of the Earth that caused that earthquake?

Dr. TURCOTTE: Well, it's a part of a collision zone between continents that extends all the way across Asia, creates the Himalayas. That area is basically an extension of the Himalayas. So it's one of the major mountain belts in the Earth. Two continents are running into each other, compressing each other, and that's why you have earthquakes there.

FLATOW: So you have the Indian subcontinent smashing up into Asia and pushing up the Himalayas...

Dr. TURCOTTE: That's right.

FLATOW: ...and causing those earthquakes.

Dr. TURCOTTE: And it's just a little bit further west from that...


Dr. TURCOTTE: ...and same thing. Right.

FLATOW: Yeah, so it's just continued to happen.

Dr. TURCOTTE: Yeah, that's inevitable. Yes.

FLATOW: Yeah. So let's talk about long-term earthquake prediction. You work on something called Virtual California. Tell us what that is.

Dr. TURCOTTE: That--what I refer to it, basically, is a model of the state of California in which you put in as many faults as you know exist, you put in the motions as far as we know them on those faults, run the model forward, and get output as to the earthquake occurrence rate, the differences, the interactions between different faults. And you produce synthetic seismicity of the state of California.

One of the most interesting aspects is, as we move forward, we expect to get very accurate space-based observations of surface deformation. And in, say, the 10-year time frame, this sort of modeling should be able to integrate those observations and hopefully make some real progress in at least understanding earthquakes and possibly even forecasting them.

FLATOW: 1 (800) 989-8255.

John in San Antonio. Hi, John.

JOHN (Caller): Hello. I love your show. I was listening to it earlier.

FLATOW: Thank you.

JOHN: You know how you say that birds are an indication of possibly an earthquake happening? You know, before we start killing them all, maybe we should be a little more cautious. But my question is, magnetic resonance--do you ever look to pick something like that up...


JOHN: ...or fields...


JOHN: ...while an earthquake is happening?

FLATOW: Well, Dr. Bakun, you want to tackle that?

Dr. BAKUN: Yes. The changes in the Earth's magnetic field is something that has been suggested quite often as a possible earthquake precursor. And we had a network of magnetometers installed at Parkfield, and we did not see any significant changes before the 2004 Parkfield earthquake. So we did look for magnetic field changes and we just didn't see them.

FLATOW: What other things did you look for? You know, people used to talk about well water and the levels of water going up and down. Did you look for any of that? Or what other things specifically did you measure?

Dr. BAKUN: Well, we certainly looked for very small earthquakes. We also

looked for changes of the slip of the fault trace in the Parkfield area using creep meters, which are a fairly low-tech instrument that's designed to just look and see if we have some precursory slip along the fault trace. We had arrays of electrical sensors that would detect changes in ground potential. We had magnetometers in the area. We had pretty much the full array of instruments that would detect the kinds of things that have been suggested as possible earthquake precursors.

FLATOW: I remember from touring the fault line, you had laser beams crisscrossing back and forth.

Dr. BAKUN: Yes, we did have that, and it turned out that, over several years of study, we found out that there were sources of noise in those instruments. So we have replaced that instrumentation with global positioning satellite measurements that provide us very good measurements of the deformation in the Parkfield area. And those instrumentations provided a very good image of the slip on the fault before, during, and after the earthquake.

FLATOW: You mean, your GPS is so accurate, you can do fractions of a centimeter or things like that?

Dr. BAKUN: Well, we have many of these instruments installed around, and when you look at them over periods of time, you can see the rate of slip and image that into various sections of the fault and determine changes in the slip rate of, yes, a few centimeters of slip per year.

FLATOW: You've got the military grade, I think, and my car doesn't do that--Is the monitoring at Parkfield continuing?

Dr. BAKUN: Yes, it is. The afterslip in the Parkfield earthquake is ongoing and it's very important. We often think that most of the action occurs during the earthquake, but at Parkfield we've learned that a good part of the slip is occurring afterward, which we call afterslip. And that's important in being able to understand the matchup of the driving motion of the plates and the release of that motion during earthquakes and in aseismic processes like afterslip.

FLATOW: So do you think now because you haven't been able to detect one in advance that you're not as hopeful about the possibility in the future of finding a way?

Dr. BAKUN: Well, I think that most seismologists would have thought Parkfield was an ideal place to be able to predict earthquakes. And the lesson there is that since we really didn't see anything before this earthquake, it's going to be very hard and we can't rely on that as way to address earthquake hazards. We have to look toward building codes and enforcement of those building codes. That's not to say that someone clever won't come up with a scheme in the future that would allow the prediction of earthquakes. And so I'm sure that research...


Dr. BAKUN: ...will continue to try and look for those things.

FLATOW: Dr. Turcotte, in studying San Francisco and Virtual California--and next year is the 100th anniversary of the 1906 earthquake--what did you find? What is the likelihood of a quake in San Francisco region occurring?

Dr. TURCOTTE: Well, the sort of numbers that we have said here is that roughly it could be 10 percent in 30 years for a magnitude 7 and above. I mean, you sort of have to specify magnitudes, you have to decide where the earthquake is going to occur on the San Andreas fault, but roughly that order of magnitude.

FLATOW: That's pretty big, 7 and above, is it not?

Dr. TURCOTTE: Yes. Although probably I can't say--we don't know for sure, but the 1906 earthquake was probably about a 7.7 or something like that.

FLATOW: Yeah. Are there enough newer buildings to, you know, not create a havoc in San Francisco like it did a hundred years ago?

Dr. TURCOTTE: Well, certainly, things are much, much better. The big question in California concerns the older buildings. Whereas the newer buildings have built to very much improved codes, there's been relatively little money spent on improving many of the older buildings in San Francisco and elsewhere. And that's just an allocation of priorities, which is an economic and political decision.

FLATOW: Yes, go ahead.

Dr. BAKUN: Ira, I should say that the hazards of San Francisco and the region certainly is due to earthquakes on the San Andreas fault, but there are many other faults in the San Francisco Bay region that would cause serious damage in San Francisco. And when you consider the regional hazard, earthquakes on the southern faults, you would come up with probabilities that are somewhat higher. And this has been put forward in a working group report on California earthquake probabilities.

FLATOW: Yeah, we conc--here, back east, we think that there's just one long giant fault, but there's a whole bunch of them...

Dr. BAKUN: Yeah.

FLATOW: ...out there that could go at any time. Yeah. And you think that something--there's a higher probability on these other faults?

Dr. BAKUN: Well, one has to consider the probability on the San Andreas fault as well as on the Hayward fault and the Rodgers Creek fault and also on faults that we may not have recognized yet. Many of the earthquakes in California in recent years have occurred on blind faults that geologists did not recognize in advance.

FLATOW: Talking about earthquake prediction this hour on TALK OF THE NATION/SCIENCE FRIDAY from NPR News. Talking with Donald Turcotte and William Bakun. Our number, 1 (800) 989-8255.

Let's see if we can take a phone call or two. Let's go to Jean in Cleveland. Hi.

JOAN (Caller): Hi, this is Joan in Cleveland.

FLATOW: I'm sorry, Joan. Jean...

JOAN: No problem. You know, I have a dear friend that lives out in the Bay area, and also a nephew, and I'm worried about California. I was talking to a friend in the military the other night and he said that older people in Japan keep a jar of cockroaches, and when they get real frenzied, that clues them in that there's going to be an earthquake. Has Dr.--I'm sorry I forget your guest's name. But has he heard of anything like that? And I'll hang up and take my answer off the air.

FLATOW: All right, thank you.

JOAN: Thank you very much.

FLATOW: All right.

JOAN: Yeah. Bye.

FLATOW: Bye-bye.

Well, what do you think?

Dr. BAKUN: You'll have to repeat that question. If you see something in a jar--I...

FLATOW: A jar of cockroaches.

Dr. BAKUN: Oh, cockroaches. Well, there's been a great deal of speculation about animals behaving strangely before earthquakes. And unfortunately, people don't really remember the comparable strange behavior of animals at ordinary times. So I don't know of any study that demonstrates that animals do react before earthquakes in a systematic way.

FLATOW: Let's see if we can get another call on. Let's go to Edward in San Diego. Hi, Ed.

Mr. EDWARD SIEGEL (Caller): Yeah, hi. I used to work in the field of acoustic emission and Don knows who I am. We're old adversaries. It is well known that even in regular breaking or placketfall(ph) material there are sub-audio, what's called one-over-f-noise precursors. There's a difference between prediction and early warning alarms. If a burglar comes to the door of your house and, you know, something senses him on the lawn, that's not a prediction. He's outside the door. He's one minute from breaking in. And that's also what animals feel--low-frequency sub-audio infrasound. The expert on it is a guy named Alfred Bedard or Albert Bedard at NOAA. You'll be reading about creepy feelings in haunted houses and so on. People like Clement Tatreau(ph) back--I'm sure you people know who he is, possibly be deceased by now, from Lawrence Livermore; someone who worked at USGS, Baxter Armstrong, myself, have been talking about that, Anthony Frasier-Smith...

FLATOW: So you're saying there were these...

EDWARD: Low-frequency sub-audio signals...

FLATOW: Below your hearing level, you feel them...

EDWARD: Very far below, sometimes a thousand times...

FLATOW: And you think that's what animals may be feeling, things like that?

EDWARD: Exactly.


EDWARD: Exactly. And if I mention my--well, can I mention my name? People can do a Google search...


EDWARD: Edward Siegel. Ira knows who I am, the guy who does Waterwork(ph). But the point is, this has been known for 30 or 40 or 50 years. It's very hard to talk to geophysicists--I started out in geophysics, and my experience was a lot of geo, very little physics. So sub-audio infrasound, Alfred Bedard at NOAA...


EDWARD: I think there are 30,000 pages about his work, including spooky feelings in haunted houses. He can predict tornadoes, he can predict avalanches, earthquakes--it's not prediction, it's early warning.

FLATOW: Right.

EDWARD: Perhaps a few days or weeks--yeah. That's what the cockroaches feel. I used to raise birds in the Bay area. I was injured in Loma Pireta. They have large transduces on their legs called baretootsies(ph), and the ones that are alive have survived dogs trying to attack them. When a dog runs at them, they feel it; they get right into the water.

FLATOW: All right, Ed, let me get a reaction because we're running out of time.

Dr. BAKUN: Well, early warning systems are very useful and, of course, they really aren't earthquake prediction except in the sense that they predict--that they can predict strong incoming ground motions after an earthquake has already occurred. And they aren't very useful at sites that are right near the earthquake but at more distant sites they can provide warnings of incoming ground shaking, 10, 20, 30 seconds in advance. That's enough time for some useful actions to be taken.

FLATOW: Dr. Bakun, did you pick up any of those things at Parkfield before or did you not listen for those?

Dr. BAKUN: You're talking about the...

FLATOW: These sub-audio waves that he's talking about.

Dr. BAKUN: Well, I don't think he was talking about predicting earthquakes, but early warning. And I--no, we haven't looked for that at Parkfield.

FLATOW: Do you not think it's--you know, it's there?

Dr. BAKUN: I think it's something that we could look at, but I have not seen anything to indicate that it's a significant effect.

FLATOW: Dr. Turcotte, anything there?

Dr. TURCOTTE: I could be corrected, but as I remember, a number of years ago, the USGS sort of set up a zoo on the top of a fault and watched the animals, and they had a nice earthquake, and they didn't even notice it.

FLATOW: All right. That's where it'll end today. We'll pick it up some other time. Gentlemen, I want to thank you for taking time to talk with us. Donald Turcotte, professor of geology, University of California, Davis. William Bakun is at the US Geological Survey in Menlo, Park, California. Thanks again for being with us.

Dr. TURCOTTE: Thank you.

Dr. BAKUN: Thank you.


FLATOW: If you'd like to write us, send your letters to SCIENCE FRIDAY, 55 West 45th Street, fourth floor, New York, New York 10036. You want to podcast us? Surf over to our Web site at, click on the `podcast' button. Also, SCIENCE FRIDAY's Kids' Connection is there. We take this material and make free curricula for you to teach in your schools. Just click on the `teachers' button and you can also send us e-mail at

I'm Ira Flatow in New York.

Copyright © 2005 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.