Thursday, April 13, 2006

Earthquake Prediction: Ready for Prime Time?

Earthquakes and the tsunamis that sometimes accompany them are one of the most frightening and fatal types of natural disasters. The December 26, 2004 earthquake and tsunami that struck in and around the Indian Ocean killed more than 200,000 people, and millions more have died in similar disasters. One of the main ways people die in an earthquake is in collapsing buildings, and over the years civil engineers have developed building codes and other techniques that reduce (but do not eliminate) the danger of structural collapse during an earthquake. Unfortunately for billions of people who live in developing countries, these measures are expensive. If the choice is between living in shaky but affordable housing on the one hand, and going without shelter on the other, most people take their chances with a house that may fall down in an earthquake. The poor of this world have more pressing things to worry about than earthquake safety, but that doesn't make their lives any less valuable.

Viewed as an engineering problem, the question of how to save lives in earthquakes and tsunamis has several possible solutions. The only one we have pursued to any great extent up to now is to make sure that structures will withstand the likely force of an earthquake. (As far as tsunamis go, there is little one can do except run for higher ground.) If—and this is a big "if"—earthquakes could be predicted with good accuracy, the problem becomes simpler. A few hours before an earthquake strikes, simply clear everyone out of dangerous buildings until the danger is past. This second solution is not without its own problems, but if it could be implemented, the cost of an early-warning system would be much less than earthquake-proof buildings for everybody, and the potential to save lives would therefore be much greater. The only problem is, how do you predict earthquakes?

Historically, earthquake prediction has been regarded as a pseudo-science. The abundance of post-earthquake "premonition" stories such as animals acting strangely, unusual sounds, and lights in the sky is a set of data that few scientists take seriously, and with some justification. Human beings are not emotionless recording machines, and memory is a highly subjective thing. Perfectly ordinary and random incidents that happen just before a frightening event take on an ominous cast when recalled later. But the shady neighborhood that earthquake prediction has resided in up to now should not prevent scientists and engineers from exploring ideas about how to do it.

The December 2005 issue of IEEE Spectrum, a highly regarded magazine for professional electrical and electronic engineers, carried an article on recent efforts to develop technical means of predicting earthquakes. (The article can be found at The lead author. Tom Bleier, described how ELF waves (extremely-low-frequency electromagnetic waves) and other measures such as satellite-sensed electromagnetic waves and surface temperatures have appeared at times to be correlated with certain large earthquake events. He made what to this author sounded like a good case that there is something to the idea that such correlations are real. However, a good physical explanation for why such correlations should occur is presently lacking.

The article inspired three geophysicists to write a letter to the editors of IEEE Spectrum protesting the publication of claims that they said should be rejected (the letter can be viewed at Robert J. Geller, Alex I. Braginski, and Wallace H. Campbell argued that there is no scientific basis for the kind of earthquake prediction that Bleier and his colleagues are doing. They claim there is so much noise from other natural and man-made sources at the frequencies in question that any exercise in earthquake prediction amounts to sophisticated tea-leaf reading. Their opinion is that the scientific community has examined the methods of Bleier and company and found them wanting.

This controversy reminds me of the early days of tornado prediction. From the late 19th century until 1938, forecasters at the U. S. Weather Bureau were forbidden even to use the word "tornado" in a forecast. The prevailing opinion was that there was no reliable way to predict tornadoes and such a forecast was likely only to cause needless panic. It wasn't until 1948 when some U. S. Air Force weathermen at Tinker Air Force Base in Oklahoma had their airfield trashed by a tornado that anyone began to apply serious scientific effort toward the problem of tornado forecasting. They came up with a combination of conditions that looked like it would work. Five days later, they noted the same conditions prevailed, and, not being under the restrictions of the civilian Weather Bureau, took it upon themselves to issue a tornado forecast to Air Force personnel. Later that same evening, probably the only tornado in history that was greeted with jubilation struck Tinker Air Force Base again! The weathermen published their findings in 1950 and 1951, but for several years afterwards tornado forecasts were restricted to military facilities unless they were leaked to the media. Other researchers attempting to publish research papers relating to tornado forecasting were blocked by skeptical reviewers. It took the better part of a decade to overcome the attitude that forecasting tornadoes was so chancy as to not be worth upsetting the public. But in combination with radar-based early warning systems for tornadoes that were put in place in the 1950s, annual tornado fatalities in the Midwest plummeted. (The story of tornado prediction is told in Marlene Bradford's Scanning the Skies: A History of Tornado Forecasting.)

Time will tell whether the new techniques of earthquake forecasting will bear fruit in the form of reliable, specific predictions. In the meantime, its proponents should prepare themselves for a long battle with skeptics. We can hope that if there is anything to it, engineers, scientists, and the public will be open-minded enough to welcome the practice and take it seriously enough to save lives with it in the future.

Sources: See URLs above referring to items in IEEE Spectrum. Marlene Bradford's Scanning the Skies: A History of Tornado Forecasting was published in 2001 by the University of Oklahoma Press, Norman.

1 comment:

  1. Tom Bleier's company, quakefinder, has an interesting website that I found: It explains all about their research on earthquakes and electromagnetic signals, including data from various stations in California.