Earthquake prediction

Predicting the location and timing of earthquakes is an active area of research in many countries throughout the world. Although significant progress has been made in understanding the causes and consequences of earthquakes, scientists are still unable to predict with sufficient accuracy the occurrence of major temblors.

Earthquake Occurrence

Chinese scientists pioneered the study of earthquakes hundreds of years ago. Since that time, predicting the location and time of major earthquakes has been an important part of seismology. Earthquakes occur, with varying frequency, in diffuse belts in nearly every region of the globe. The distribution of earthquakes is explained easily by the modern theory of plate tectonics, which holds that the surface of the earth is composed of a mosaic of interlocking rigid plates that move relative to one another at speeds up to 12 centimeters per year. Motions along the boundaries of the plates produce earthquakes; if the plates are not able to accommodate the motions easily, then large earthquakes may accompany the relative motion between the plates. The widespread occurrence of earthquakes makes them important to everyone, and for people living near plate boundaries, earthquakes play an even more potentially destructive role in shaping the environment.

Earthquakes are generated when some portion of the earth's rigid outer layer, called the lithosphere, ruptures catastrophically along a sharp discontinuity or fault. This creates significant ground motion near the source of the rupture. Earthquakes occur most commonly at the three types of boundaries of lithospheric plates, which are known as convergent or destructive, divergent or constructive, and transcurrent or transform. The largest number of earthquakes are at divergent plate boundaries located along mountain ridges in the ocean basins. Because these earthquakes are small and far from population centers on the continents, little effort is expended to predict ruptures along divergent plate boundaries. In contrast, earthquakes at convergent or transcurrent plate boundaries, although fewer in number, are larger. Most convergent and transcurrent boundaries coincide with continental margins along which the majority of the global population lives. For this reason, earthquake prediction research focuses on convergent and transcurrent plate boundaries such as those in Japan and California.

Slip on fault planes of large earthquakes is on the order of 10-20 meters, and the forces responsible for faulting are simply the result of the relative motions of the plates at the plate boundaries. Rocks near a region of an impending earthquake may accumulate motion and change volume and shape for hundreds of years prior to causing a rupture. When the lithosphere does finally break, energy stored by the rocks is released suddenly as seismic waves that travel through and around the surface of the earth. These waves generate the intense vibrations associated with an earthquake. For great earthquakes, the rupture may extend for as much as 1,000 kilometers, and it may propagate at speeds in excess of 10,000 kilometers per hour.

Earthquake Categorization

Seismologists categorize earthquakes by several different features, but the two most important for earthquake prediction are an earthquake's location and its magnitude or size. The location is described by an epicenter, which is the projection of the earthquake's focus within the lithosphere onto the earth's surface.

The magnitude is a number from 1 to 10 on a scale devised by Charles Richter that describes the relative changes in ground motion recorded on a seismometer. The so-called Richter scale is logarithmic; an increase from a value of 1 to 2 corresponds to a tenfold increase in ground motion and to an approximately thirtyfold increase in the amount of energy released during the rupture. Contrary to popular belief, there is no upper limit to the Richter scale. The Richter scale is based on a “standard seismometer” placed at a “standard distance” from the epicenter of the earthquake. The traditional Richter scale magnitudes are denoted by “M” to distinguish them from other more recent magnitude scales. Richter originally devised his magnitude scale to be most appropriate for describing moderate local earthquakes in California. Unfortunately, despite its widespread use, the scale is not a good measure of the energy released from very small or very large earthquakes.

Potential Triggers

Seismologists have learned much about the rupture process that causes earthquakes by studying the ground motion close to and far away from the source. The development of modern seismological instruments and procedures in the early twentieth century led seismologists to the discovery that different rupture mechanisms are at work in different plate tectonic settings. Early attempts at earthquake prediction used analysis of the frequency of major earthquakes in specific regions of the globe to determine whether any significant pattern was apparent. This approach proved to be fruitless. With the acceptance of elastic rebound theory to describe the rupture mechanism for earthquakes, seismologists shifted their attention away from statistical analysis of earthquake occurrences toward developing methods to understand the “trigger” of major earthquakes. Most seismologists agree that the energy necessary to produce a major earthquake is accumulated slowly relative to the time it takes for a rupture to occur. If no trigger were involved in the rupture process, prediction of earthquakes would be extremely difficult, if not impossible. Modern seismologists interested in earthquake prediction primarily rely on developing methods to understand any precursory phenomena that would enable them to predict at least several days or weeks in advance the location of large (M > 6.0) to great (M < 8.0) earthquakes. Of course, if an impending earthquake is far removed—for example, more than 1,000 kilometers—from any population center, the need to alert the public is minimal.

Over the years, investigators have suggested a variety of potential triggers to earthquakes. These include rapidly changing or severe weather conditions; variations in the gravitational forces among the moon, sun, Earth, and other planets in the solar system; and volcanic activity. Scientists have searched historical seismicity records, including extensive catalogs for California, for relationships between the suggested triggers and the occurrence of earthquakes, without much success. For example, every 179 years, the planets of the solar system align. Some researchers suggested that this alignment would increase the gravitational forces acting upon earth and thereby trigger an increase in seismicity. The last such planetary alignment was in 1982, and no significant increase in earth seismicity occurred.

Prior Changes in Physical Properties

Because the research on earthquake triggers has been largely unsuccessful, seismologists have turned their attention away from potential triggers of major earthquakes toward the role of changing physical properties prior to an earthquake. Some promising properties include shifts in ground elevation near the site of an impending earthquake; variations in the velocity of certain types of seismic waves as they traverse regions that may produce a major quake; increased escape of radon, helium, and other gases from vents and cracks in the earth's surface prior to the earthquake; changes in the electrical conductivity of rocks near the region of impending rupture; and fluctuations in pore fluid pressure in the rocks near major fault zones. In addition, seismologists have focused on recognizing certain premonitory swarms of smaller quakes, called foreshocks, that may foreshadow a major earthquake.

Another technique is assessing the time between major earthquakes in a specific region. If an area that is expected to produce earthquakes is seismically quiet—that is, a gap exists in its seismic activity—then the area may be more likely to experience an earthquake in the near future. This is referred to as “seismic gap” theory.

Finally, strange animal behavior has been linked to periods of several days to several hours prior to an earthquake. Some researchers have claimed that cats and dogs tend to run away from home or exhibit unusual behavior, such as seeking out special hiding places, before the onset of an earthquake. Individual reports of odd animal behavior are substantiated by the increase in advertisements for lost pets in local newspapers during the days before a major earthquake. Scientists have suggested that some animals are sensitive to minute changes in their environment, which allows them to “sense” an earthquake prior to onset of severe ground shaking. Research in this area is actively pursued in China and Italy. Most workers, however, are interested in developing instruments that would be able to measure the same effects that disturb animals. Although many of these effects are known to occur prior to a major earthquake, scientists still must develop highly sensitive devices that will alert them in enough time to evacuate or prepare the region near an impending earthquake.

Seismogram Analysis

Several methods are used by seismologists to study earthquake prediction. The techniques include analysis of seismograms to identify either foreshock or aftershock patterns that signal an impending large earthquake, examination of active fault zones in the field to determine the frequency of great earthquakes over the last tens of thousands of years, and investigation of deep boreholes to characterize the orientations and magnitudes of stresses associated with active faults. In addition, elaborate arrays of sophisticated instruments are frequently deployed near active faults to collect geophysical data that may shed light on earthquakes.

The energy carried by seismic waves is recorded on seismometers or seismographs, which are instruments that monitor ground motion. Seismometers are composed of a mass attached to a pendulum. During an earthquake, the mass remains still, and the amount that the earth moves around it is measured. Ground motion is recorded on a chart as a series of sharp peaks and valleys that deviate from the background value, measured during times of no earthquake activity. The arrival of the waves at different times at different places allows seismologists to calculate the epicenter of an earthquake. The height or amplitude relative to the background noise of the first peak in a long series of peaks associated with a particular earthquake is an estimate of the magnitude of that earthquake.

Since the early twentieth century, seismometers have recorded hundreds of thousands of earthquakes per year worldwide. Seismologists have carefully cataloged many seismograms, the actual paper records of ground motion from a particular location, so that they may be easily compared. Examining these records in detail has allowed seismologists to deduce certain characteristics of major earthquakes. They have noted, for example, that most large earthquakes are followed by a series of smaller earthquakes in the same region. These smaller earthquakes are called aftershocks, and they allow seismologists to constrain the orientation and dimensions of the rupture or fault plane that produced the main earthquake. With the development of modern seismometers and digital recording networks in the 1970s and 1980s, seismologists began to recognize certain precursory seismicity patterns in addition to aftershock sequences. These precursory phenomena are referred to as foreshock sequences and, as yet, are poorly understood. Seismologists hope that with enough data on the overall seismicity of an area, they will be able to note deviations from normal patterns that would signal the onset of a major earthquake.

Field Analysis

Information about prehistoric seismic activity must be obtained by examining ancient fault zones exposed at the surface of the earth. Large motions between two rock masses produce characteristic features that may be identified in the field. Geologists are now examining the recent rock record near the San Andreas fault in California. Careful mapping of areas that have been excavated across the fault zone has yielded evidence for large earthquakes prior to historical and seismological records. The now well-established technique of carbon-14 dating was applied to organic material trapped in the fault zone to determine the approximate age, location, and intensity of several ancient earthquakes. The data, although sparse, seem to suggest that great earthquakes occur every fifty to three hundred years. In addition, there is some indication that great earthquakes may occur closely spaced in time with significant periods of quiescence between them. This type of analysis is similar to the seismic gap theory, where catalogs of seismograms are examined to determine which known faults or regions that have been active previously are currently inactive and perhaps are ready for renewed activity.

Many countries are carrying out elaborate experiments in areas of repeated seismic activity. One example is in central California on the San Andreas fault near the town of Parkfield. There, geophysicists arrayed a variety of instruments, including seismometers, tiltmeters, gravimeters, and laser surveying equipment, to measure ground motion, elevation changes, gravity variations, and minute amounts of slip on the fault. Based on seismic records, scientists discovered that earthquakes with magnitudes of approximately 5.0 M occur with predictable frequency. Since 2004 Parkfield has been the site of the San Andreas Fault Observatory at Depth (SAFOD), with extensive instrumentation in a 3-kilometer borehole angled into the fault. The experiments are designed to learn as much as possible about the changes that occur in the region prior to, during, and after an earthquake of moderate size. Scientists hope that these data will allow them to know what features to monitor to predict much larger earthquakes in other areas.

Investigation of Deep Boreholes

Another method used by scientists to understand precursory phenomena associated with earthquakes is drilling deep boreholes into the earth's crust near major fault zones. One such hole is in Fort Cajon, California. At this site, researchers examined changes in pore fluid pressure and electrical conductivity in the borehole. In addition, instruments measured the orientation of fractures in the borehole's walls. These fractures are related to the forces acting on the rocks deep in the crust, and some researchers attempted to relate these manifestations of stress to earthquake fault orientation. Geophysicists hope that the newest techniques will measure these stresses in real time, so they will be able to compare these data to those obtained from studies of seismicity. Understanding the behavior of a major fault zone at depth may prove useful in predicting earthquakes in the future.

Progress in Prediction

Many countries are actively involved in earthquake prediction research. Since the early 1960s, these efforts have been particularly active in Japan, the People's Republic of China, and the United States. The overall goal of these research efforts is to attain the same level of reliability in earthquake prediction as in weather prediction. Although the majority of effort has been focused on predicting the exact time and place of a major earthquake, an equally important, though often overlooked, aspect of earthquake prediction is an assessment of the severity of ground shaking for a specific site. This information is crucial for public policy discussions on the location of dams, hospitals, schools, and nuclear reactors, all of which may be at significant risk during a major earthquake.

Despite the numerous uncertainties that enter into forecasting an earthquake, some have been successfully predicted. The most spectacular was the Haicheng earthquake of northeast China in February 1975. Five hours before the earthquake, warnings were issued and several million people from towns in the vicinity of the predicted epicenter were evacuated. Devastation was widespread, but loss of human life was minimal. Scientists who later visited the area estimated that hundreds of thousands of lives were saved. Unfortunately, the Chinese were only able to predict that a great earthquake was to strike the Tangshan region within five years. In August 1976, a very strong earthquake struck this area without warning, leaving 700,000 people dead.

By the early twenty-first century, little progress had been made on predicting earthquakes. Though scientists could utilize more advanced tools to study the effects of potential earthquakes, they were still unable to accurately assess the likelihood that an earthquake would occur within a given range of dates. However, during the early 2020s, scientists began experimenting with using machine learning algorithms to help warn the public before an earthquake occurs. Research conducted at the University of Alaska Fairbanks suggested that properly trained machine learning programs could analyze low-level tectonic activity across large areas, providing days or months notice before an earthquake occurs.

Minimizing

Earthquake prediction remains critical to modern society because most of the world's population lives along convergent plate boundaries and, therefore, within the destructive reach of a great earthquake. The purpose of earthquake prediction, then, is to prepare a society for any earthquakes with magnitudes capable of disturbing normal life. This may include warning and evacuation of the local population or assessing the risks of severe ground motion on current or future structures. In addition to strong ground motion, earthquakes may be responsible for hazards such as tsunamis, avalanches, and fires. In both the great San Francisco earthquake of 1906 and the Tokyo earthquake of 1923, many of the fatalities attributed to the earthquakes were actually the result of the subsequent fires that consumed the cities. Another danger of earthquakes is soil liquefaction, which occurs when the seismic waves cause the soil to lose rigidity and slide away. When this happens, the soil can no longer support structures. Although the structures may be strong enough to withstand the shaking associated with an earthquake, their foundations may be undermined, causing the buildings to topple.

Perhaps the most promising aspect of earthquake prediction is the development of stringent building codes. After each major earthquake in Southern California, for example, municipal, county, and state statutes are changed to reflect new data concerning the behavior of building materials during strong ground motion. Engineers now know that unreinforced masonry buildings are likely to be destroyed in even a moderate earthquake. Because the seismic risk is high in Japan and California, these areas now have the most stringent building codes in the world. That these codes can prevent much unnecessary loss of property and human life unfortunately can be demonstrated by a comparison of the 1971 San Fernando and 1988 Armenian earthquakes. The earthquakes were of similar magnitude—slightly greater than 6.0 M—yet the San Fernando earthquake resulted in about fifty deaths, most of which were in one wing of an old hospital building, while tens of thousands perished in the Armenian earthquake because of the collapse of poorly constructed masonry buildings.

Humans will never be able to prevent earthquakes. Understanding how, why, when, and where earthquakes occur, therefore, is extremely important to society. Earthquake prediction, like weather prediction, is one way that society seeks to minimize harmful effects of these complex natural phenomena.

Principal Terms

crust: the outermost layer of the earth, which consists of materials that are relatively light

elastic rebound theory: the theory that states that rocks across a fault remain attached while accumulating energy and deforming; the energy is released in a sudden slip, which produces an earthquake

faulting: the process of fracturing the earth such that rocks on opposite sides of the fracture move relative to each other; faults are the structures produced during the process

lithosphere: the earth's rigid outer layer, which is composed of the crust and uppermost mantle

seismicity: the temporal and spatial distribution of earthquakes

seismology: the study of earthquakes and their causes

stress: a force acting in a specified direction over a given area

Bibliography

Berlin, G. Lennis. Earthquakes and the Urban Environment. Vols. 2 and 3. CRC Press, 1980.

Bolt, Bruce A. Earthquakes and Geological Discovery. Scientific American Library, 1993.

Doyle, Hugh A. Seismology. John Wiley, 1995.

Eiby, G. A. Earthquakes. Van Nostrand Reinhold, 1980.

Farley, John E. Earthquake Fears, Predictions, and Preparations in Mid-America. Southern Illinois UP, 1998.

Grotzinger, John, et al. Understanding Earth. 5th ed. W. H. Freeman, 2006.

Hough, Susan. Predicting the Unpredictable: The Tumultuous Science of Earthquake Prediction. Princeton UP. 2010.

Iacopi, Robert. Earthquake Country. 4th ed. Lane Books, 1996.

Lomnitz, Cinna. Fundamentals of Earthquake Prediction. John Wiley & Sons, 1994.

Mogi, Kiyoo. Earthquake Prediction. Academic Press, 1985.

National Research Council (U.S.), Panel on the Public Policy Implications of Earthquake Prediction. Earthquake Prediction and Public Policy. Government Printing Office, 1975.

"New AI Model Could Predict Major Earthquakes Months Before They Happen." SciTechDaily, 7 Sept. 2024, scitechdaily.com/new-ai-model-could-predict-major-earthquakes-months-before-they-happen/. Accessed 10 Feb. 2025.

Rikitake, Tsuneji. Earthquake Forecasting and Warning. Kluwer Academic Publishers, 1982.

Tarbuck, Edward J. et al. Earth: An Introduction to Physical Geology. 10th ed. Prentice Hall, 2010.

Uyeda, Seiya. The New View of the Earth. Translated by Masako Ohnuki. W. H. Freeman, 1978.