Earthquake prediction and the L'Aquila earthquake

Along with the stories of tragedy in Italy, there are also stories that the earthquake was predicted, and that the predictions were ignored. Was the tragedy made even worse by authorities who wouldn't listen to a scientist who knew what he was talking about?

Now, I'm not a seismologist, but I know a little about predicting earthquakes. I predicted one, you see: the 1989 Loma Prieta earthquake.

There had been a small earthquake just before I arrived at Stanford, so earthquakes were on a lot of people's minds. The San Francisco Examiner had a hilarious cartoon about how to use your pet to determine the size of an earthquake, Stanford was distributing what-to-do-in-an-earthquake pamphlets, and geology grad students were making bad jokes. When I heard that the San Francisco Giants and the Oakland A's might be in the World Series, I said "wouldn't it be great if SF played Oakland, and there was an earthquake?"

Guess what.

And that's how I predicted the 1989 World Series earthquake.

Now, before you read any further, a quick quiz: what was wrong with my prediction?

Now, there aren't any reasonable mechanisms by which baseball can cause earthquakes. On the other hand, Giampaolo Giuliani used measurements of radon gas. Radon is produced as part of the decay series of uranium, and is found trapped in soils and rocks. It's possible that it could be released in the lead-up to an earthquake, according to some models of how ruptures begin. (If ruptures begin with micro-cracks forming, they might increase the permeability of the rock, and allow more radon to move towards the surface.) Radon was one of many earthquake precursors that have been studied since the 1970's, in hopes that some reliable precursor might be found.

None of those precursors have turned out to be reliable. Not radon, not groundwater levels, not ground inflation. Foreshocks, maybe - sometimes there are smaller earthquakes before a larger earthquake. But there are also small earthquakes without larger earthquakes, all the time.

Studies of earthquake behavior, including events that may make an earthquake more likely, continue. Maybe at some point seismologists and structural geologists will understand what happens before an earthquake, and then we will really be able to forecast them. (Or at least say that they have become more likely, as volcanologists do with eruptions. I don't know whether earthquakes warnings will ever be as good as flood warnings or hurricane warnings.)

But even if our understanding does increase, it's worth keeping in mind what would be necessary for an earthquake prediction to be useful. There are several questions that are traditionally asked:

  • Where will the earthquake occur?
  • How big will the earthquake be?
  • When will it happen?

Reliable answers to those three questions are important if emergency planners are going to ask people to change their behavior. Small earthquakes occur every day, somewhere in the world. Earthquakes only affect people who live near them, and only the larger earthquakes are destructive. And if you're going to tell people to get out of their houses, you'd better have some idea of when they will be allowed to go back when nothing happens.

It's that last point, the timing of the earthquake, that is the biggest problem for Guiliani's prediction of the Italian quake. According to some news reports (80 Beats, LA Times), Giuliani's prediction was for a 24-hour period last week (March 29, to be exact). Even if Giuliani's radon measurements were related to this week's quake, his prediction technique got the timing wrong. [Edit: Apparently, he also got the place wrong: his prediction was for an earthquake 30 km away, in Sulmona. See more at olelog and, if you read German, at Amphibol. According to the USGS map of estimated shaking, the shaking 30 km away from L'Aquila was only Mercalli intensity V - unlikely to destroy buildings and kill people. So that's two things wrong. As Ole says, what if people in Sulmona had heeded the warning... and evacuated to L'Aquila?]

If I were teaching The Control of Nature this week, I would turn this into a discussion. Giuliani is working with a technique that hasn't been very successful in the past, and which has mostly been discarded in favor of other earthquake research. But he still thinks it works. So what are the ethical obligations of a scientist in this case? Should one go public with warnings, given the risk that they may be false alarms? Is it worse to create unnecessary panic, or to have people die in the earthquake? (What are the chances that the panic itself would result in deaths, from traffic accidents or from a lack of shelter on cold nights?) In Giuliani's case, Reuters reports that vans with loudspeakers were driving around L'Aquila, telling people to evacuate... a month ago.

I don't have an answer to that question.

(Oh, and for people wondering what was wrong with my 1989 "prediction," apart from the absurdity of relating baseball to earthquakes: if a magnitude 2 earthquake had occurred in Japan during the World Series, could I have claimed a successful prediction?)

A few other discussions of Giuliani, from both blogs and the mainstream media:

LA Times: Did Scientist Predict Italy quake? (Includes good discussion of earthquake prediction history and current direction by seismologists from southern California.)

The Lede (NY Times): Earthquake Warning Was Removed From Internet

Short Sharp Science (New Scientist): Italian authorities dismissed quake warning 2009 L'Aquila Earthquake>

Accidental Remediation: predicting earthquakes

80beats (Discover):
Scientist Smackdown: Did a Seismologist Accurately Predict the Italian Quake?

Other info on earthquake prediction:

Nature Debates (1999): Is the reliable prediction of individual earthquakes a realistic scientific goal?

More like this

In the wake of Monday's earthquake in the L'Aquila region of Italy that killed over 200 people, news emerged that one Italian scientist had predicted the earthquake less than two weeks earlier. Giampaolo Giuliani's predictions were broadcast on March 28 but was dismissed by Italian authorities--and…
Given how many bloggers have already weighed in on the story of an Italian court convicting geologists of manslaughter for failing to issue adequate earthquake warnings before an earthquake that devastated the town of L'Aquila, including Steve Novella, Daniela at Skepchicks, Sharon Hill at Skeptic…
"Prediction is very difficult, especially if it's about the future." -Niels Bohr What's going to happen next? It's perhaps the most important thing to know if we want to be prepared for practically anything in our lives. And without even thinking about it, most of us are actually very good at this…
The Great San Francisco Earthquake(s) On October 8th, 1865, the "Great San Francisco Earthquake" hit south of the city of San Francisco, magnitude 6.3. On October 21st, 1868, the 'Great San Francisco Earthquake" hit near Haywards, east of the city, across the bay, magnitude 6.8. On April 18th, 1906…

I'm somewhat puzzled by your persistent skepticism given that 180 people who would still be alive are now dead because of misplaced skepticism and science denialism.

You say that radon gas emission isn't a reliable earthquake precursor, so my question to you is, is there any science that you consider to be reliable?

I'm somewhat puzzled by your persistent skepticism given that 180 people who would still be alive are now dead because of misplaced skepticism and science denialism.

Which bit of "he got the timing wrong" did you not understand? Or do you think that if everyone had been evacuated on the 29th March, they would have stayed away from their homes until the 6th April?

And you miss the real issue: radon release (or any of the other proposed precursors, like low frequency EM field fluctuations) can precede an earthquake, but just as often you see a signal and nothing happens. No way has been found of distinguishing a 'real' signal from a 'false' one except in a post-hoc analysis - which is useless.

Plus, all the claimed "correlations" I've seen have a large temporal range: sometimes the precursor signal is seen a few hours before an earthquake, sometimes a few days, sometimes weeks. No-one seems to be able to tell the difference between these either. So it boils down to "something might happen, possibly soon. Or maybe not." That's not reliable science by any stretch of the imagination.

OilsMastery: see my edit. According to two European geo-bloggers, Giuliani got both the time and the place wrong. He warned of an earthquake that was 30 km from the fault that slipped - far enough that the estimated shaking at his predicted location was about Mercalli intensity V. (Mercalli intensity V means that the shaking was strong enough to make doors swing and pendulums stop. The damage in L'Aquila was Mercalli intensity VIII - strong enough to do significant damage to stone and brick buildings. Also, note that I haven't seen actual damage estimates from Sulmora, where Giuliani predicted the earthquake - I'm going on the USGS's estimate of shaking intensity.)

It's not clear that 180 (now 205, actually) lives would have been saved if Giuliani's warning had been heeded. (As Ole pointed out, more lives could have been lost if people had evacuated from the town where Giuliani predicted an earthquake, and gone to L'Aquila instead.)

I am always amused about how the media can be captivated by a single person's prediction over the volumes of quality, tested scientific research that claims otherwise. Just from what I have read about this "prediction," it does not stand up to scientific muster. Just as Kim mentions, there are a lot of great coincidences that don't necessarily mean you've nailed causation. There are a lot of faults in Italy and a lot of them move, so just because you happen to be in the ballpark (sorry about the pun) with your prediction doesn't mean you have any more of a clue than anyone else. I'm sure there are enough so-called earthquake (and volcano for that matter) predictions out there in the interweb tubes that you could argue we're ignoring thousands of "predictions" - its just stats that some of them are "true".

Ironically, I am working on a research paper on the ethics of natural disaster prediction just when this item comes along. I suggest reading The Politics of Earthquake Prediction by Olsen to see just how badly things can spiral out of control. It is my opinion that radon and other precursors signal a coming quake but, as mentioned, they are unreliable and we simply don't understand how these things might work to be useful enough for valid, specific predictions. Mr. Giuliani will have a nice data set for a paper, I presume. Let him proceed through the appropriate channels to gain credibility and allow for debate and critique.

Have any of you thought of looking further afield for the cause of this quake? precursing this there have been other minor quakes in europe within the past month which could have been reducing enough compressive pressure to cause the situation.

By Dave Major (not verified) on 07 Apr 2009 #permalink

S. Hill - thanks for the book recommendation. I'll look at it.

Dave Major - I don't work on triggered earthquakes (and other ways in which one earthquake changes the risk of others) myself, but it's an active area of research. I don't have enough of a feel for the way triggered earthquakes work to make a reasonable guess about whether the Italy earthquake could have been one or not. (Most of the cases that I know of involve somewhat larger earthquakes, however - M 7 or higher. Larger EQs involve both larger fault areas and larger slip, as well as larger seismic waves, so I would expect them to be more likely to affect distant areas. See Ramon Arrowsmith's blog for a quick estimate of how large the fault rupture is likely to be.)


"Which bit of 'he got the timing wrong' did you not understand?"

Personally I'd rather be early than dead.

"Or do you think that if everyone had been evacuated on the 29th March, they would have stayed away from their homes until the 6th April?"

How much do you want to bet Giuliani was nowhere near the earthquake when it happened?

And in the last few days media in the USA has been talking about the smaller quakes in the Salton Sea area as possible precursors to "a big one" in that area. Of course there have been others who have speculated that the quakes could be precursers to a volcanic eruption.

One thing that was stressed in an engineering geology course I once took was using your knowledge of geologic hazards to support your common sense while not becoming hysterical about it. The trouble is that people expect geology to be predictive on the scale of a human life. But geology deals with processes which generally have a much longer time span.

My cat might be a better predictor of earthquakes than any geologist. But she'll likely only give you about a few hours notice, that is, if you can find her and interpret her behavior correctly. Interpretation of any phenomena is the key and that means integrating more information than just boron.

By Lynn David (not verified) on 07 Apr 2009 #permalink

OilsMastery - here is the problem with your position (at least, one of the problems) - when do you decide to evacuate or not? Do you seriously suggest evacuating entire cities for a week or more each time there is an unproven (or proven unreliable) prediction? Do you know how often he predicts earthquakes? I'm guessing you don't, and I'm speculating that he predicts them fairly frequently and got lucky this time.

You might as well just say that they should permanently evacuate the city - abandon it to the ages because there are earthquakes there.

The solution for now is to be prepared for earthquakes. Reports I've read have indicated that Italy is extremely poorly prepared for earthquakes - they simply don't pay attention to the danger as much as they should. It'd be expensive, but updating building codes and retrofitting the worst of the old buildings would save lives much more reliably than evacuating the city every few months would.

Lynn - the media probably is over-blowing it (I live in SoCal but the only local media I pay attention to is NPR, so I don't know) but consider this: the recent Shake Out earthquake simulation/drill modeled a rupture starting in the Salton Sea area as the most likely case for the next "big one". That prediction is based on a lot of science, some of which I've looked at myself, and I think they make a pretty good case for that being the starting point - but what they're leaving out is when it will happen, because they don't know. So if we put a high likelihood of it starting there, even if it happens in 30 years the recent minor events will have been "precursors" ;)

Joking and pedantry aside, this goes right back to the problem of the Italy earthquake prediction - sure we can say that events will probably happen in a region, but we can't say when or exactly where. But once it does happen, if it matches with any part of the prediction, you can bet that ignorant folks will jump on it as being a great prediction that we should have paid more attention to.

I think all of you are missing the point.

If Mr. Guiliani predicts another quake and is
dead wrong, the first prediction will be irrelevant.

If Mr. Guiliani predicts another quake and is
correct within about a week, then all the geologists
who are dissing him now will become despised morons
who will need to embrace the exciting new opportunities
in the food service industry. You don't like it: tough.

By Tim Crouch (not verified) on 07 Apr 2009 #permalink

If he's got a track record (or develops one), I hope he organizes a special session at the December American Geophysical Union meeting (or at the next European Geophysical Union meeting), so that all the various people who are interested in earthquake prediction and the mechanics of earthquake initiation can get together and talk about mechanisms, reliability, and so forth. And I hope he publishes his results in Science or Nature, because a breakthrough in earthquake prediction would belong in one of the general-interest journals.

Believe me, we would all be very excited if an earthquake prediction (or forecasting) method turns out to be right. But the most promising research right now seems to follow other leads - earthquakes that trigger other earthquakes, slow slip, episodic slip and tremor, GPS studies. Radon's been studied for a long time, and seemed like a dead end. Maybe it's due to be revisited, but not on the basis of one earthquake prediction (in which both the location and the timing were wrong).

Also, before I go to bed, a warning about civility. No name calling, no insults, no accusations of massive conspiracies by scientists to suppress the truth. I've seen earthquake prediction discussions turn into flamewars before, and I will not tolerate it in my blog. Be rude to me or to other commenters, and your comment will be deleted. Be rude repeatedly, and you will be banned. (And I will make the final decision about what is rude and what is not.)

This post leaves out the issue of censorship of Mr. Giuliani's work that was posted on the internet. The critics have come out of the woodwork to criticize Mr. Giuliani but because his warning was censored the criticism is based on generalities, not data.

Censorship is incompatible with scientific progress.

We all know that earthquake prediction is a goal we haven't yet reached. For some subset of earthquakes, predictions may be impossible. However, we might have data in this case that represents progress in forecasting earthquakes. Let us not let Mr Giuliani's emotional attempt to warn people as an excuse for criticizing him.

We need to be examining his research and his data not dismissing his work before we have even seen it.

FYI, I have written my own analysis on my diary page on DailyKos.…

I am very disappointed to see what appears to be progress under attack by the scientific community. We should not be attacking something that is good because it is not perfect.

I did a little research and found three scientific papers in UK, Iceland and Taiwan claiming that radon is a good (not perfect!) earthquake precursors: here are the links :
and many more on the archive (just put radon earthquake in the tile field) and, so far, not a single paper saying the contrary.
Of course there are false alarms, but I don't understand why Giuliani's claim is not on the same level as all this positive evidence. Statistically, high radon means 50% or more of probability of a strong earthquake, at least in some regions, and it looks like Abruzzo is one. Now, the only radon detector in that very seismic area is Giuliani's own (that he built and paid himself); who is to blame for this, the outsider Giuliani, or the geologists? I would not mind spending one or two weeks in safe areas every couple of years (the false alarms) if I know 50% of the times this is saving 100's of lifes.

luk says: "I did a little research and found scientific papers ... claiming that radon is a good (not perfect!) earthquake precursors ... and, so far, not a single paper saying the contrary."

That kind of search will likley not result in papers discussing the limitations and uncertainties of the relationship between radon and seismicity ... scientists don't tend to title a paper in such a way. You need to do a much more thorough literature review if you are going to make a claim that there aren't any papers stating something that specific.

A decent starting place, although a bit over a decade old now, is Geller's 1997 Geophysical Journal International paper "Earthquake prediction: a critical review". I found it to be an adequate review of the myriad proposed precursor relationships. The paper also discusses several claims of successful predictions in the past as well as predictions that did NOT happen (these 'non-events' are typically not reported in the media).

To those playing the 'censorship' card here ... take a step back. Is Giuliani being treated unfairly in this one specific instance? I don't know ... I don't know enough about him or this specific situation. But, the Earth scientists who are raising skepticism about his claims of prediction are doing so because that is our job! We want to know how accurate and reliable this forecasting method is (as pointed out above, in this case the temporal and spatial accuracy was NOT good; aspects the mainstream media reports are lacking). The best way to evaluate this claim of successful prediction is to be objective and look at within the context of the decades of work leading up to this instance ... this event (and the media portrayal of it) cannot be viewed in isolation.

Actually, the best coverage I've seen of this came from the mainstream media: LA Times. Their reporter did things that I didn't do: go to original sources in Italy and in the local LA seismology community. (And then explain everything well.)

The LA Times has good reasons to have good earthquake reporters on staff. I'm glad that they do, given what's happened to many other newspapers lately.

There's plenty of bad reporting in the world, both in the mainstream media and blogs. But when there's something good, we should recognize it.

Lessons from the great U. of Chicago geology professor, T.C. Chamberlain taught generations of geologists, "The Method of the Multiple Working Hypothesis Theory," in solving complex problems in geology, in addition to discovering large oil and natural gas fields. Finding oil and predicting earthquakes is a combination of both art and science because extensive experience is required. There are no silver bullets in predicting catastrophic earthquakes, but hazards can be significantly reduced by effective, interdisciplinary use of seismology, geophysics and geology. Increases in Radon and grounwater levels alone many not work. However a combination of many subtle indicators, along with a fundanmental geological understanding of local faults in a suspect area can greatly increase the probability of a successful prediction. There
are published case histories that prove this.

By Jack Century (not verified) on 08 Apr 2009 #permalink

False positives are as serious weakness, because the expense and impracticality of emptying a city with every warning. People will quickly stop paying attention.

With every great disaster, you can find at least several people who actually did dream of a tragic event just before it happened... people randomly dream of such things frequently enough to make that a virtual certainty that SOMEONE did. The problem is that to be considered viable he needs a statistically credible track record. Often astrologers try to claim success even if their time is off, or their location is off, the details are a bit off ... they don't seem to realize that such rubberiness of interpretation makes their whole claim statistically invalid.

If Mr. Guiliani predicts another quake and is
correct within about a week, then all the geologists
who are dissing him now will become despised morons
who will need to embrace the exciting new opportunities
in the food service industry.

Wrong. and wrong. You don't understand science much do you. If he actually does prove correct, the majority of 'critics' will gladly accept this new tool of analysis and work to refine it further.

Again, the problem here is the media accepting Guliani's claims at face value. They may have merit and I, for one, don't know enough to judge that mostly because I haven't seen his work. However, the job of science is to take this new work and be highly skeptical until proven otherwise. And no, that is not "ganging up" on someone, that is just how it works. Maybe he sort of came close on one occasion using a method (Radon) that may (or may not) help assess whether an earthquake is more likely - not certain, not schedule - but to then claim that he was stifled from saving all these lives is just media-based sensationalism. Let him prove his case in the scientific channels - and make sure other researchers can reproduce his work and predictions - and then I, for one, would be happy to accept the results. Until then, I am as skeptical as I am about most things like this.

The earthquake that happened in Italy is actually a valuable example on tectonics and their unpredictability. Kids in particular can benefit from the data and research that come from this incident. They can use this information to enhance the accuracy and relevancy of their science fair projects.

Giuliani predicted a quake in SulmoNa (sorry to be pedantic).

This is 30km from L'Aquila. And his prediction was wrong by a week.
So, assume we interpret his statement as "there will be a quake within ten days of March 29, within 50km from Sulmona. What does the italian government do?
1) Nothing. 200+deaths.
2) Organizes people to move out of the area. Which means, you know, stopping them from working, moving hospitals, children don't go to schools, they all live in tents for two weeks... easy? Who is moving the old people who don't want to leave their house? How expensive is that anyway? How many deaths does the moving cause? We're talking abput moving several hundreds of thousands of people. Remember, you must keep police in the evacuated area, to prevent crime.
3) just tell people and let them take their choice. Result: those with a second home at the seaside (the richest 5% of L'Aquila's population) just go there. Everybody else spends a night or two in a tent and then goes back home.

Bonus point: imagine the government does 2). Then it must do it whenever there's a quake alert. How many would that be, in Italy? Italy has had 4 major earthquakes in the last 25 years, and many, MANY more smaller quakes. How can we afford moving everybody every time?

Bonus point 2: the main problem is that building quake-resistant housing is mandatory but not enforced in Italy. Not many years ago a number of schoolchildren died in a minor quake because their school was very lousily built.

Thanks for the spelling correction estraven.

And yes, about the scenario for evacuating people. It's clearly better to be alive and inconvenienced than dead, but if a prediction is wrong, there's a huge psychological and economic toll on the people who are evacuated.

Retrofitting buildings for earthquake resistance can be expensive (especially if you want to preserve the appearance of easily damaged stone buildings - it was bad enough for Stanford University in the 1990's; I imagine it would be far worse in a country with beautiful medieval architecture). But in the long run, it seems like the wisest choice. (And if modern buildings aren't built to be quake-resistant - that's a huge tragedy.)

I am geologist myself and I m gutted at the way self-styled professors and other specialists try to diss their fellow geologists.

At least Giuliani tried to save some lives. He may have been wrong by 30 km and by a week but at least he tried to save some lives.

Most of you here should leave your ego and envy aside and should give him a call and work with him instead.

Hey Kim, i dont care if you delete this message....

In my italian blog "scienzeedintorni" I have posted an article about a Giuliani's work wich is full of uncorrelated data: he writes the forecasted earthquake without a diagram in wich radon emissions and time of quakes are related....
and in this period he did not see neither the strongest events, nor more than 15 events that occurred near his station.... (I crossed his data with the ones of the iris earthquake browser).
In a recent happening in Italy he said that firstly, working in the gran sasso phisics laboratory he observed a lot of radon after a turkey quake and that he saw also the sumatra earthquake.
I'm sad, because he his from my country...

Ever hear of geologist Jim Berkland? He is the man who predicted and coined the Loma Prieta quake and it was published on Fri. 13th in the Gilroy Dispatch, 1989. He was even suspended from his job as Santa Clara County Geologist--for 2 months because of it all. He had tons of data to base his prediction...It's all in my book The Man Who Predicts Earthquakes. I didn't know you predicted it...He got credit big-time for his accurate forecast.

What's kind of amusing to me is that when the Tom Jordon's of the world say "People send out predictions based on various stuff. It's always hard to evaluate" (from the LA Times article), and even my favorite bloggers pile on, I also dismiss Guiliani out of hand. But I have been working with a respected drilling scientist who kind of likes Guiliani's 'prediction'. I started to think about it, and you bet your sweet you know what that if Tom Jordon had put together some stress measurement and detected some creep event that fed into a 'prediction' that worked, you'd see it on the cover of the New York Times and the National Academy would be writing memos. But because some mud logger spouts off its another kook.
Look, 30 km is within the radius of influence for any reasonably large earthquake (see the study of well-heights during the Nisqually earthquake of Seattle, Montgomery et al. Geology article?) and it is a rock-mechanics 'fact' that any weakening is going to be highly non-linear with time. So why don't we give Guiliani a break and think a bit about what he did. Nobody's asking for a billion dollar 'prediction' program out of this.