Read more: 鈥�Instant Expert: Earthquakes鈥�
Many avenues for earthquake forecasting have been explored, from prior changes in animal behaviour to electromagnetic signals. Yet predicting exactly when an earthquake will happen remains impossible today. Still, there is a great deal we do know about the Earth鈥檚 shaking in the future
Forecasting: what we know
When seismologists are asked whether earthquakes can be predicted, they tend to be quick to answer no. Sometimes even we geologists can forget that, in the ways that matter, earthquakes are too predictable. We know where in the world they are likely to happen. For most of these zones, we have quite good estimates of the expected long-term rates of earthquakes (see map). And while we often cannot say that the next Big One will strike in a human lifetime, we can say it is very likely to occur within the lifetime of a building.
Advertisement
We know the largest earthquakes occur along subduction zones, where a tectonic plate dives beneath another into the Earth鈥檚 mantle, with rupture lengths of more than 1000 kilometres and an average slip along a fault of tens of metres. But any active plate boundary is fair game for a big earthquake, at any time. For example, two years before the 2010 earthquake in Haiti, geophysicist Eric Calais and his colleagues published results of GPS data from the region, noting that 鈥渢he Enriquillo fault is capable of a M7.2 earthquake if the entire elastic strain accumulated since the last major earthquake was released in a single event鈥�. While this exact scenario did not play out in 2010, it wasn鈥檛 far off. We can say for sure that people living on plate boundaries will always face risk.
鈥淓arthquakes far from major plate boundaries can often be felt over 1000 kilometres away鈥�
Future large earthquakes are expected in California. Research by James Lienkaemper and his colleagues estimates that sufficient strain is stored on the Hayward fault in the east San Francisco Bay area to produce a M7 earthquake. An earthquake this size is expected, on average, every 150 years. The last one was in 1868. Local anxieties inevitably mount knowing such information, but earthquakes occur by irregular clockwork: if the average repeat time is 150 years, it could vary between 80 to 220 years. So we are left with the same vexing uncertainty: an 鈥渙verdue鈥� earthquake might not occur for another 50 years, or it could happen tomorrow. On a geological timescale there is not much difference between sooner versus later. On a human timescale, sooner versus later seems like all the difference in the world.
Earth scientists have made great strides in forecasting the expected average rates of damaging earthquakes. The far more challenging problem remains finding the political will and resources to prepare for the inevitable.
Why so difficult?
In the 1970s and 1980s, leading scientists were quoted in the media expressing optimism that reliable short-term prediction of earthquakes was around the corner. This was fuelled by promising results from the Soviet Union, and the apparently successful prediction of the 1975 earthquake in Haicheng, China. Since then, this optimism has given way to varying degrees of pessimism. Why are earthquakes so hard to predict?
Any number of possible precursors to earthquakes have been explored: small earthquake patterns, electromagnetic signals and radon or hydrogeochemical changes. Many seemed promising, but none have stood up to rigorous examination.
Consider this example. In March 2009, Italian laboratory technician Giampaolo Giuliani made a public prediction that a large earthquake would occur in the Abruzzo region of central Italy. His evidence? An observed radon anomaly. The prediction was denounced by local seismologists. The M6.3 L鈥橝quila earthquake struck the area on 6 April, killing 308 people.
This gets to the issue of reliable precursors. It is possible that radon was released due to the series of small earthquakes, or foreshocks, that preceded the main earthquake. It is also possible it was coincidence. 快猫短视频s explored radon as a precursor in the 1970s and quickly discovered how unreliable it is. Once in a while radon fluctuations might be associated with an impending earthquake, but usually they are not. Meanwhile big earthquakes hit regions where anomalies were absent. The same story has played out with many other proposed precursors.
That鈥檚 not to say that seismologists have neglected to investigate precursors 鈥� on the contrary they are examining them with increasingly sophisticated methods and data. However, a common bugaboo of prediction research is the difficulty of truly prospective testing. To develop a prediction method based on a particular precursor, researchers compare past earthquakes with available recorded data. One might, for example, identify an apparent pattern of small earthquakes that preceded the last 10 large earthquakes in a given region. Such retrospective analyses are plagued by subtle data selection biases. That is, given the known time of a big earthquake, one can often look back and pick out apparently significant signals or patterns.
This effect is illustrated by the enduring myth that animals can sense impending earthquakes. It is possible that animals respond to weak initial shaking that humans miss, but any pet owner knows that animals behave unusually all the time 鈥� and it鈥檚 soon forgotten. People only ascribe significance with hindsight.
At present most seismologists are pessimistic that prediction will ever be possible. But the jury is still out. One of the big unanswered questions in seismology is: what happens in the earth to set an earthquake in motion? It is possible that some sort of slow nucleation process is involved, and therefore possible that earthquake precursors exist.
For this as well as all earthquake prediction research, the challenge is to move beyond the retrospective and the anecdotal, into the realm of statistically rigorous science.
Megaquake myths
Since the M9.1 Sumatra-Andaman earthquake struck on Boxing Day in 2004, another four earthquakes with magnitudes of 8.5 or greater have occurred on the planet, including the Tohoku, Japan, earthquake in 2011 (see diagram). This apparent spate has led some to wonder if earthquake frequency is increasing. Careful statistical analysis reveals that it is not.
The recent rate of very large earthquakes is unusual, but not a statistically significant increase relative to expected variability. And the overall energy release by earthquakes in the past eight years is still below the combined energy release of the two largest recorded earthquakes: the 1960 Chilean quake and Alaska鈥檚 1964 quake on Good Friday.
Anthropogenic climate change could conceivably influence earthquake rates in certain areas in the future: the process of post-glacial rebound associated with the retreat of large glaciers provides a source of stress that can drive earthquakes. Such earthquakes could have a significant local impact, but their overall energy release will continue to be dwarfed by that of earthquakes caused by plate tectonics.
鈥淐limate change could influence the frequency of earthquakes in the future鈥�
While there is no reason to believe that megaquakes are on the rise, there is little doubt that more and worse megadisasters due to earthquakes lie ahead in our future 鈥� they are the inevitable consequence of explosive population growth and concomitant construction of vulnerable dwellings in the developing world.