Several Italian scientists may be charged later this year with manslaughter over the deaths of 308 people who died in and around l'Aquila in 2009.  Is this reasonable?

I wanted to write about something new in this post, but as Google News failed to inform me of any interesting geology-related happenings (unless you include this BBC article which is just one big rock pun) I will have to make do with something almost-current I have wanted to write about for a while.

Santa Maria Church in Paganica, damaged by 2009 L'Aquila Earthquake. CC-BY-NC by Flickr user pablo72.Back at the beginning of June, seven Italian scientists and officials were accused by the l'Aquila public prosecutor of manslaughter relating to the 2009 L'Aquila earthquake, and it appears they will go before a judge in December for a preliminary hearing.  Most of them were members of the Commissione Grandi Rischi (the Committee for Major Hazards), a committee made up of scientists and members of the Italian Civil Protection force designed to prepare for large disasters such as this.  What are they charged with?  Failing to predict the earthquake.  (you can read the indictment document here; although it is 224 pages long and in Italian, so I haven't been able to read it properly)

Most seismologist agree that we cannot currently predict the timing earthquakes, at least not accurately than on the scale of decades (the location of earthquakes, on the other hand, is much better understood; I'll come back to this later).  Many also believe that it may never be possible.  Both the American Geophysical Union and the Italian Instituto Nazionale di Geofisica e Vulcanologia have underlined this by responding to the charges, with the open letter on the latter's website collecting over 5,000 signatures.  So why how could anyone sue the scientists for failing to predict this earthquake, something that is not scientifically possible to do?

In the months leading up to the 6 April earthquake there were many, small tremors in the area.  This is nothing unusual, most active faults in the world go through periods of increased seismic activity.  Occasionally this is a prelude to a major quake; but more often or not they lead to nothing, and peter out without doing too much damage.  Indeed, as the Commissione Grandi Rischi pointed out at the time, these seismic swarms are often good news: they allow built-up energy to be dissipated fairly harmlessly.  The important thing is, swarms can happen without leading to large earthquakes and large earthquakes can happen without precursory seismic swarms.  Increased frequency of small earthquakes is, therefore, not a reliable predictor of large earthquakes.

If that had been all, then it's unlikely that anything more would have come of this.  However, there was also
Giampaolo Giuliani.  Although often described in the media as an "expert", he worked as a lab technician and his research on predicting earthquakes was done in his spare time.  This alone is obviously not a reason to discount it, and the work he was doing was far from the lunatic fringe that often is attracted to predictions of this sort.  What he was looking at was radon gas, which is formed naturally as part of the uranium decay series as uranium naturally found in rocks breaks down.  Back in the 1970s, increased radon emissions were recorded before several large earthquakes.  The mechanism behind these emissions was poorly understood, it was suggested that when the rocks around a fault were stressed almost to breaking point there would be changes in the pore spaces squeezing fluids out or that micro-fissures would open up.  Radon emissions were linked to other phenomena such as changes in ground water levels that had also been observed.  Researchers were disappointed, however, when further investigation revealed that none of these indicators were consistent.  A 1995 New York Times article demonstrates that radon had been discounted by that time.

It is in that context that the the Commissione di Grandi Rischi announced on March 31, six days before the earthquake, that there was "
no scientific basis for any forecast".  They never ruled out the possibility of a quake, emphasising that as l'Aquila sat on an active fault it was only a matter of time, but they did try to allay fears that Giuliani had raised of an imminent quake, and they did silence him.  In hindsight, was this the right thing to do?  Did the failure to issue a warning, based partly on Giuliani's research and partly on the increased seismicity, amount to criminal negligence?  In short, no.  As Giuliani's method had been shown to be inconsistent in the past, and increased seismicity often occurs without leading to a large earthquake, there was no scientific reason to suspect the quake.  Without reliable evidence, no warning could be issued.

However, 6 days later there was a large earthquake, which Giuliani claims vindicates him.  This brings me on to a more general point about predictions.  Just because Giuliani's prediction turned out to be correct, does that make his method correct?  The short answer, again, is no.  There is a good chance this was simply coincidence.  The AGU has now shown an interest in his work, and he is now has funding to continue his research, and I wish him luck.  If radon gas emissions are linked with earthquakes, at least at l'Aquila, further research should show this.  If not, radon is still hazardous, and so greater understanding of how it reaches the surface could still prove useful.

How can I say Giuliani's prediction is likely to be coincidence?  The chance of him correctly predicting an earthquake may be low, but that is the wrong way of looking at it.  Giuliani's prediction has to be seen in the context of all the predictions of earthquakes that are made all the time.  They range from plausible if unlikely, like Giuliani's, to downright wacky, like this guy on  The sheer amount of predictions means that every time there is an earthquake, someone will be able to claim to have seen it coming.  Just as a large earthquake in California later this year would not be proof that revelation13's bible codes worked, neither is the l'Aquila earthquake on its own proof that radon gas works.

This problem is intensified by people starting with an earthquake and looking back, trying to find signs that could have predicted that earthquake.  A nice example of this from l'Aquila is this study on toads,  who all packed up and left their breeding pool shortly before the quake only to return after it was over.  What exactly the toads were trying to achieve on their jaunt -- they were pretty safe in the pool (remember, earthquakes don't kill people falling buildings do) -- is unclear, as is how often this happens when there are no earthquakes.  After all, who would find this toad trip interesting if it wasn't for the earthquake?  People always underestimate the chances of a coincidence like this happening.  When you have something as vague as some animals behaving "strangely", then there's a pretty good chance that you will find that anywhere.  If it wasn't toads it could be bees, snakes, or cows; in everyday life this would usually be quickly forgotten.  

Returning to the matter of criminal negligence, should the Commissione di Grandi Rischi have issued a warning anyway, despite the lack of evidence, just to be on the safe side?  While it may have made a difference this time, in answering this question you have to consider what would happen if you always gave a warning in this sort of situation.  This time, Giuliani was out by a week and 55 km (not including the numerous false alarms he gave in the weeks preceding the main shock).  If the authorities had acted, it would have involved evacuating a large area for a long period of time.  This disruption itself would obviously be very costly, but if this was carried out in all similar circumstances the scientists' credibility would soon disappear.  This would lead to easily preventable death in the future, when people ignore real warnings.  A single false alarm may not be as devastating as a false assurance, but multiple false alarms soon add up.

In the end, this is all rather beside the point.  While predicting the timing of earthquakes may be difficult, we have a pretty good idea of where they will strike: they are concentrated along areas of active tectonic strain.  Most earthquake deaths are preventable.  Compare the magnitude 6.9 1989 Loma Prieta earthquake that struck San Francisco with the magnitude 7.0 2010 Haiti earthquake (l'Aquila was magnitude 6.3, for comparison).  In San Francisco, 63 people died.  In Haiti one estimate puts the number of dead at 230,000.  The difference?  Building regulations.  Dull though it sounds, having well built buildings saves lives.  The l'Aquila public prosecutor is also pressing charges against several builders, whose work it appears did not comply with the regulations.  They would be better off concentrating fully on those, the real criminals in this story, not the scientists who failed to warn about an event that was impossible to predict.

Thanks to Ben Goldacre on Bad Science for originally bringing this story to my attention.