In spite of the fact that we know now exactly what causes earthquakes, their sudden, repetitive appearance is as disturbing as the former ignorance of causes, so the quest to be able to predict their arrival has become a central interest of geologists. In fact, from earliest times, alongside the fanciful reasons for the occurrence of earthquakes, there were theories that pointed in the direction of predictions. Aristotle, for example, tried to explain the existence of earthquakes by theorizing that air was trapped in cavities below the surface of the earth and the rumbling of earthquakes was the result of that as the blocked cavities sought to free themselves. Pliny the Elder of Roman times, the man who died at the time of the eruption of Vesuvius, thought similarly to Aristotle, that there was a blockage of air inside the earth and caves and wells were necessary to release this trapped air. Often he would propose a solution that would involve drilling holes to allow the trapped air to escape.
A variety of attempts have been made over the years to predict earthquakes by looking at the minor seismic events that always occur around the time of an earthquake, hoping that the way these appear and their strength might indicate when the earthquake would occur. These have been successful to a point. For example, at the time of the Mount St. Helen’s earthquake and volcanic eruption in 1980, the small shakings that preceded the earthquake and the major eruption were an accurate forecast of what was going to happen. But the exact time of the explosion was not predictable and so the nearest that experts could come to a prediction was to say that there would be an explosion within a few weeks. That level of accuracy has been proved successful, not only in the case of Mount St. Helen’s, but also in the many earthquakes that recur in Alaska, where patterns are frequently repeated over long periods of time. The Tangshan earthquake of 1976 was kept secret for three years because the Chinese felt they had developed similar methods for predicting earthquakes but, unfortunately, they were totally wrong in regard to the Tangshan quake
Many reports, some dating back hundreds of years, have pointed out that animals of all sizes and kinds seem to detect earthquakes hours or even days before humans. At the time of the tsunami that followed the Indonesian quake of 2004 it was observed that almost no animals were caught in the event. It seems that they moved away from danger areas in good time. This behavior of animals was also noted and recorded in 1755 at the time of the Lisbon earthquake and tsunami. In the year 2004, a book by a Japanese geologist, Motoji Ikeye, was published in Singapore by World Scientific Publishing Company and titled Earthquakes and Animals. In this book he describes in detail one possible explanation for animals being able to sense an oncoming earthquake. Ikeye was aware that, in recent years, in major earthquakes in Japan, China, and India, there were numerous reports of animals fleeing from the scene twelve to twenty-four hours before the earthquakes struck. Ikeye also knew, from his own experiments, that flashes of lightning or lights of some kind accompany earthquakes, perhaps due to collisions or friction among rocks as they are shaken. The thesis that he developed related to these lights. He concluded that they frightened animals and so they moved away from them. It seems to be a very persuasive argument, one that is likely to be investigated further in the future.
Ikeye’s book included list of animals that were seen to be running away from a place that later experienced an earthquake. This list is more detailed than any other similar record. His list includes reports of avoidance of earthquakes by dogs, cats, sea lions, hippopotami, squirrels, rats, seagulls, snakes, turtles, fish, dolphins, octopus, crocodiles, and rabbits. Each had its own pattern of behavior and the following details concerning dog and cat activities will indicate the varieties within each category of animal.
Some dogs howled like wolves while others refused to be separated from their owners, either insisting on staying inside or trying to get the owner outside. Some dogs left their homes before the earthquake and returned several days later while others barked continually up to thirty minutes before the earthquake. With cats, some tried to get into bed with their owners, waking them up and even biting them. Forty-five minutes before an earthquake other cats meowed to be let out of the house. The places selected by Ikeye for the collection of his data about animal behavior prior to earthquakes were taken from Japan, Turkey, Taiwan, and India.
Most research on prediction still remains focused on the characteristics of rocks and terrain in areas that have a history of earthquakes. Tilting of the ground, changes in elevation, even rising water levels in wells can all be indicators of stresses deeper down. Appearance of small cracks may be the surface manifestation of a fault hundreds of miles below ground. In places that were hit by very powerful earthquakes in the past, it is possible to predict the frequency of recurrence by examining soil and rock layers below ground and calculating the likelihood of another quake. This is a very rough method of prediction, making it possible to say no more than that an earthquake of such and such magnitude will recur sometime within the next so many years. In California, at Pallet Creek on the San Andreas Fault a little over fifty miles north of Los Angeles, a location where slippages on the San Andreas Fault have been repeatedly noted in the form of disrupted sedimentary layers, an expert from the California Institute of Technology decided to dig down at this location and make accurate assessments of the soil layers to discover when past quakes had occurred. His purpose was earthquake prediction. He thought that if he could date older slippages he might be able to draw up a timetable of recurrences and thus an average frequency of earthquakes for this part of the San Andreas Fault. It was already well known that most California earthquakes occur on or near this fault. A history of 1,400 years of earthquakes was identified and from these the average recurrence interval was calculated at about 150 years. Since the last great earthquake on the San Andreas near Pallet Creek was in 1857, the finding from this investigation and prediction came uncomfortably close to the year 2007.
Earthquake prediction is always an inexact science. No one can claim certainty about the future, yet scientists are always at work seeking to gain as much predictability as possible. Occasionally a successful forecast occurs and then efforts are redoubled to capitalize on the event. In China, in 1969 on a particular morning, zookeepers noticed unusual animal behavior—swans avoided water, pandas screamed, and snakes refused to go into their holes. About noon on that same day a 7.4 magnitude earthquake struck the city. Ever since then, and especially now in the light of Ikeye’s 2004 report, scientists take careful note of any relevant animal behavior. Although probability seems to be a very weak method of prediction at first glance, it is turning out to be the best of all. When dealing with a very large number of variables in a situation where most of the variables operate independently of one another, the ordinary predictive methods of science do not apply. Scientific prediction requires stability in several variables so that the behavior of one can be evaluated.
Weather forecasters were among the first to recognize the value of probability prediction but the idea did not originate with them. It came from scientists working with very small things, like cells or atoms, where all the common laws of physics give way to random behaviors. Only probability is predictable in those domains. Meteorologists trying to cope with increases in the frequency and power of both hurricanes and tornadoes decided to settle for probability methods in giving storm warnings. It is now the same in geology. The basis of the method is the record of past events. If there are extensive records of say earthquakes in a given region, including magnitudes and dates, it can be said with a stated degree of probability that an earthquake of a certain size will hit within a given time period. This approach has become standard practice over the past twenty or thirty years with the Global Seismograph Network (GSN) that was set up in the early 1960s. In 1990 it rated the probability as high, of a magnitude 6 quake striking the eastern United States before 2010.