In the wake of the massive 9.0-magnitude earthquake that has devastated Japan, an obvious question arises... why didn't we see it coming? The fact is that earthquake prediction has long baffled scientists, and the current consensus is that it's actually impossible.
There are very few areas of science where experts are actually willing to make definitive statements that something is impossible. And yet that's pretty much exactly where the field of earthquake prediction stands, at least ever since a 1996 paper in the journal Science in which four seismology experts published a paper simple entitled, "Earthquakes Cannot Be Predicted."
That's a remarkably absolute statement, and it's worth understanding just why seismologists took such a strong stance.
First, we should just quickly review precisely what an earthquake is. Earthquakes occur when energy is suddenly released within the planet's crust, create seismic waves that travel throughout the Earth's interior and causing shaking and ruptures on the surface in the area on or around the fault line between tectonic plates. The release of stored elastic strain energy that builds up along the tectonic plate boundaries is what causes the majority of earthquakes.
So earthquake prediction depends on answering a crucial question: how do you predict the next sudden release of energy? To a certain extent, that's easy enough to answer. After all, there are about 500,000 earthquakes a year, and about 100,000 of these cause tremors that can be felt on the surface. That means there's practically an earthquake every minute, and most of these are clustered in the various regions that find themselves on active fault lines.
Indeed, Brigham Young University geologist Matthew A. Mabey explains in his 2002 piece "The Charlatan Game" that it's dead easy to predict earthquakes. Just tell your friends there will be a magnitude 5 or 6 earthquake in Los Angeles (be sure not to specify when), that there will be a magnitude 6 or greater earthquake in the next two weeks (just don't mention where), and that an earthquake will hit southern California in the next 48 hours (just don't say how strong)... and you will never be wrong.
Of course, what we really want to do is cut past all the minor earthquakes and talk about the ones that can actually do serious damage. There's only about a couple dozen or so earthquakes every year that are 7.0-magnitude or higher, which is roughly when earthquakes start doing serious harm. That means we have to be able to find the twenty or so serious earthquakes out of all 500,000 that happen in a given year. That's a definite needle-in-a-haystack proposition, and we're dealing with a particularly chaotic haystack.
Myths and Failed Explanations
Earthquakes are apparently completely random events that can, without any advance warning, cause utter devastation. When faced with something like that, it's hardly surprising that people come up with various lay theories that link them to more easily understood phenomena, which can include anything from weather patterns, solar cycles, and - perhaps most temptingly - when the last earthquake occurred. Unfortunately, as former US Geological Survey employee and current geology science writer Andrew Alden explains, there's nothing to any of this:
Do earthquakes correlate with the phase of the Moon? No they don't. Do they tend to occur during a particular type of weather? No they don't. What about time of year? No effect. Previous seismic events? Outside of aftershocks, no. Sunspots and solar cycles? Nope, nope. Not even psychics do better than chance. Every prediction method makes lucky hits, but none of them really work.
But it gets worse. Alden goes on to explain that even seemingly scientific theories do no better at predicting when the next big earthquake will hit than the ideas put forward by the general public, something that geologists themselves will readily admit:
It doesn't matter, for instance, how long it has been since the last earthquake along a particular stretch of geologic fault. It doesn't matter what fluid pressures underground are doing-rising, falling, or fluctuating. It doesn't matter what the electrical conductivity of the ground is doing. The behaviour of earthquake faults is a stubborn mystery. Seismologist Yan Kagan of UCLA said a few years ago, "It may require the development of completely new mathematical and theoretical tools. We should not expect significant progress in this direction in the near future."
The Importance Of Chaos
Now, all is not lost. Yes, earthquakes are naturally chaotic, which means there isn't much hope of ever predicting the specific time and location of the next severe earthquake, or when a particular area is going to be hit next. But chaos theory has some pretty rigorous mathematics behind it, and it's possible to model the overall chaotic system with great precision.
We can use the mathematics of chaos to determine the overall energy release for the various tectonic plates, and with that we can know which areas of the world are at the greatest long-term risk from earthquakes. Short-term predictions remain a pipe dream, but this at least gives us long-term forecasts. These can include assessments of how many earthquakes of varying magnitudes a region can expect over a given time period, and it can also provide a probable upper limit for just how severe the earthquakes in that area are likely to get.
We can't know when the big one will hit, but just the knowledge that it will likely happen at some point in the future gives city and building planners and emergency services a chance at being as prepared as possible for the eventual earthquake.
The Failed Success
I can't imagine anyone would dispute the importance of disaster preparedness, but the current situation in Japan makes it painfully clear that there really isn't any level of preparation sufficient to deal with the worst earthquakes. Isn't there anything we can do to get even a little advance warning for those? Well... there is one instance of a "successful" earthquake prediction, but seismologists still aren't sure what to make of it.
In 1974, China's National Earthquake Bureau made two crucial decisions: that a major earthquake would hit the country within the next two years, and that strange animal behaviour and other weird phenomena could be a useful predictor a coming earthquake. Over 100,000 "honorary observers" were trained to be on the lookout for animals behaving as though scared or anxious, for well water that suddenly clouded and bubbled, and for random lightning strikes in clear skies, among other things.
Shortly after a minor earthquake in December 1974, reports started flooding in reporting all these events, and after a string of more small tremors in February 1975, Chinese officials made the decision to evacuate the people of Haicheng, a northern city with a population of over a million. The day after the evacuation, a 7.3-magnitude earthquake struck, and experts have predicted that over 150,000 people would have been killed or injured if the government had not stepped in at precisely the right moment.
It's considered the only successful earthquake prediction in history, although it's still debated whether the officials were acting on the unusual signs, the swarm of minor quakes, or a combination of the two. This kicked off a minor buzz of renewed excitement for earthquake prediction among seismologists, and the US National Research Council even issued a report in 1976 that short-term earthquake prediction would be a reality within 10 years.
But even this doesn't have a happy ending. On July 28, 1976, a pair of 7.8-magnitude earthquake hit the northern Chinese city of Tangshan, and this time there was little to no advanced warning. The earthquake left about 250,000 people dead and another 164,000 seriously injured. For all the good their prediction methods had done in Haicheng, they had been useless in Tangshan, leaving it an open question whether the initial "success" was just a spectacularly lucky coincidence.
What It Would Take To Go Further
Part of the problem is that, even if we can confidently identify the limits of seismology, we're still not even particularly close to that. Even after a hundred years, the science remains in relative infancy. Andrew Alden explains by comparing seismology to predicting the weather:
Before weather predictions became widely accepted, we had to spend decades learning about climatology, fluid dynamics and the physical laws that govern the ocean and the atmosphere at all scales. We also had to collect immense amounts of historical data and set up satellites to monitor the ocean-atmosphere system. If earthquake predictions are to become as effective as weather predictions are today, we must take seismology to a comparable level. Today's research is aimed at that more fundamental goal, and the nearest that scientists will come to issuing predictions is making long-term and medium-term forecasts on a strictly experimental basis.
Right now, seismology is still only taking baby steps in that direction. A lot of research is built around the mechanical concept of stress and trying to apply the behaviour of objects like stressed beams to seismic activity. Researchers Chung-Han Chan and Ross S. Stein demonstrated how a transfer of stress caused by one earthquake to another region precipitated the next earthquake, which might allow us to zero in on the location of the next big earthquake, even if its timing still eludes us. But these methods have only very weak forecasting ability, and it's too early to really know what, if anything, these methods will actually amount to.
The Dangers of False Prediction
There's no real way to get around the fact that earthquakes are fundamentally chaotic. Sure, there's some mild evidence that there are certain signs that can predict when an earthquake is imminent, but there's no guarantee that signs like weird animal behaviour or even a swarm of minor earthquakes will consistently predict when a big earthquake is imminent, or even if a decent number or major earthquakes are actually preceded by these signs. And false alarms, while obviously not nearly as bad as a truly devastating earthquake, can take their own toll, and a string of bad predictions can lead to a lessened emphasis on the one constructive thing we actually can do, which is preparation.
This is a realisation that geologists have come to only after more than a century of failed predictions, some of which destroyed once stellar careers. Renowned California geologist Bailey Willis created a mass panic in 1925 when he predicted a massive earthquake would hit in the next five to ten years, and the blowback from his doomsday predictions forced him out of seismology ever, and it took a relatively minor earthquake in 1933 to remind people that he wasn't completely wrong when it came to improving building safety.
In 1976, US Bureau of Mines researcher Brian Brady claimed his laboratory experiments predicted devastating earthquakes would hit the waters just off the coast of Peru in either 1981 or 1982, likely creating a deadly tsunami. While seismologists were rightly doubtful that any laboratory experiment could make such a prediction, people throughout South America grew understandably anxious over the announcement, and even after Brady retracted his prediction a team of American scientists had to visit South America to resolve what had become a minor diplomatic incident.
What We're Left With
It's not fun to admit that science has fundamental limits on what it can do, particularly when so many human lives potentially hang in the balance. But it's crucial that we understand those limitations and recognise all the things that we can do. Earthquake prediction is, in all likelihood, a scientific dead end, a distraction that keeps our focus off something that's actually tangible and important, and that's preparedness.
Earthquake preparedness - particularly when aided by long-term forecasting of likely problem areas - can at least give us the best possible chance of minimizing the extent of the destruction in the event of disaster. That may seem like rather hollow consolation, particularly in the wake of the current crisis in Japan. And yes, frankly, it is. But it's still the preferable alternative to the false promise of pseudoscience.