Earthquake observatories

Worldwide during the late 1950s, there were only about 700 seismographic stations, which were equipped with seismographs of various types and frequency responses. Few instruments were calibrated; actual ground motions could not be measured, and timing errors of several seconds were common. The World-Wide Standardized Seismographic Network (WWSSN), the first modern worldwide standardized system, was established to help remedy this situation. Each station of the WWSSN had six seismographs—three short-period and three long-period seismographs. Timing and accuracy were maintained by crystal clocks, and a calibration pulse was placed daily on each record. By 1967 the WWSSN consisted of about 120 stations distributed over 60 countries. The resulting data provided the basis for significant advances in research on earthquake mechanisms, global tectonics, and the structure of the Earth’s interior.

By the 1980s a further upgrading of permanent seismographic stations began with the installation of digital equipment by a number of organizations. Among the global networks of digital seismographic stations now in operation are the Seismic Research Observatories in boreholes 100 metres (330 feet) deep and modified high-gain, long-period surface observatories. The Global Digital Seismographic Network in particular has remarkable capability, recording all motions from Earth tides to microscopic ground motions at the level of local ground noise. At present there are about 128 sites. With this system the long-term seismological goal will have been accomplished to equip global observatories with seismographs that can record every small earthquake anywhere over a broad band of frequencies.

Locating earthquake epicentres

Many observatories make provisional estimates of the epicentres of important earthquakes. These estimates provide preliminary information locally about particular earthquakes and serve as first approximations for the calculations subsequently made by large coordinating centres.

If an earthquake’s epicentre is less than 105° away from an observatory, the epicentre’s position can often be estimated from the readings of three seismograms recording perpendicular components of the ground motion. For a shallow earthquake the epicentral distance is indicated by the interval between the arrival times of P and S waves; the azimuth and angle of wave emergence at the surface are indicated by a comparison of the sizes and directions of the first movements shown in the seismograms and by the relative sizes of later waves, particularly surface waves. It should be noted, however, that in certain regions the first wave movement at a station arrives from a direction differing from the azimuth toward the epicentre. This anomaly is usually explained by strong variations in geologic structures.

When data from more than one observatory are available, an earthquake’s epicentre may be estimated from the times of travel of the P and S waves from source to recorder. In many seismically active regions, networks of seismographs with telemetry transmission and centralized timing and recording are common. Whether analog or digital recording is used, such integrated systems greatly simplify observatory work: multichannel signal displays make identification and timing of phase onsets easier and more reliable. Moreover, online microprocessors can be programmed to pick automatically, with some degree of confidence, the onset of a significant common phase, such as P, by correlation of waveforms from parallel network channels. With the aid of specially designed computer programs, seismologists can then locate distant earthquakes to within about 10 km (6 miles) and the epicentre of a local earthquake to within a few kilometres.

Catalogs of earthquakes felt by humans and of earthquake observations have appeared intermittently for many centuries. The earliest known list of instrumentally recorded earthquakes with computed times of origin and epicentres is for the period 1899–1903. In subsequent years, cataloging of earthquakes has become more uniform and complete. Especially valuable is the service provided by the International Seismological Centre (ISC) at Newbury, Eng. Each month it receives more than 1,000,000 readings from more than 2,000 stations worldwide and preliminary estimates of the locations of approximately 1,600 earthquakes from national and regional agencies and observatories. The ISC publishes a monthly bulletin—with about a two-year delay—that provides all available information on each of more than 5,000 earthquakes.

Various national and regional centres control networks of stations and act as intermediaries between individual stations and the international organizations. Examples of long-standing national centres include the Japan Meteorological Agency and United States National Earthquake Information Center in Colorado (a subdivision of the United States Geological Survey). These centres normally make estimates of the magnitudes, epicentres, origin times, and focal depths of local earthquakes. On the Internet, data on global seismicity is continually accessible through the Web site of the Incorporated Research Institutions for Seismology (IRIS).

An important research technique is to infer the character of faulting in an earthquake from the recorded seismograms. For example, observed distributions of the directions of the first onsets in waves arriving at the Earth’s surface have been effectively used. Onsets are called “compressional” or “dilatational” according to whether the direction is away from or toward the focus, respectively. A polarity pattern becomes recognizable when the directions of the P-wave onsets are plotted on a map—there are broad areas in which the first onsets are predominantly compressions, separated from predominantly dilatational areas by nodal curves near which the P-wave amplitudes are abnormally small.

In 1926 the American geophysicist Perry E. Byerly used patterns of P onsets over the entire globe to infer the orientation of the fault plane in a large earthquake. The polarity method yields two P-nodal curves at the Earth’s surface; one curve is in the plane containing the assumed fault, and the other is in the plane (called the auxiliary plane) that passes through the focus and is perpendicular to the forces of the plane. The recent availability of worldwide broad-based digital recording has enabled computer programs to be written that estimate the fault mechanism and seismic moment from the complete pattern of seismic wave arrivals. Given a well-determined pattern at a number of earthquake observatories, it is possible to locate two planes, one of which is the plane containing the fault.

Observation and interpretation of precursory phenomena

The search for periodic cycles in earthquake occurrence is an old one. Generally, periodicities in time and space for major earthquakes have not been widely detected or accepted. One problem is that long-term earthquake catalogs are not homogeneous in their selection and reporting. The most extensive catalog of this kind comes from China and begins about 700 bce. The catalog contains some information on about 1,000 destructive earthquakes. The sizes of these earthquakes have been assessed from the reports of damage, intensity, and shaking.

Another approach to the statistical occurrence of earthquakes involves the postulation of trigger forces that initiate the rupture. Such forces have been attributed to severe weather conditions, volcanic activity, and tidal forces, for example. Usually correlations are made between the physical phenomena assumed to provide the trigger and the repetition of earthquakes. Inquiry must always be made to discover whether a causative link is actually present, but in no cases to the present has a trigger mechanism, at least for moderate to large earthquakes, been unequivocally found that satisfies the various necessary criteria.

Statistical methods also have been tried with populations of regional earthquakes. It has been suggested, but never established generally, that the slope b of the regression line between the logarithm of the number of earthquakes and the magnitude for a region may change characteristically with time. Specifically, the claim is that the b value for the population of foreshocks of a major earthquake may be significantly smaller than the mean b value for the region averaged over a long interval of time.

The elastic rebound theory of earthquake sources allows rough prediction of the occurrence of large shallow earthquakes. Harry F. Reid gave, for example, a crude forecast of the next great earthquake near San Francisco. (The theory also predicted, of course, that the place would be along the San Andreas or an associated fault.) The geodetic data indicated that during an interval of 50 years relative displacements of 3.2 metres (10.5 feet) had occurred at distant points across the fault. The maximum elastic-rebound offset along the fault in the 1906 earthquake was 6.5 metres. Therefore, (6.5 ÷ 3.2) × 50, or about 100, years would again elapse before sufficient strain accumulated for the occurrence of an earthquake comparable to that of 1906. The premises are that the regional strain will grow uniformly and that various constraints have not been altered by the great 1906 rupture itself (such as by the onset of slow fault slip). Such strain rates are now being more adequately measured along a number of active faults such as the San Andreas, using networks of GPS sensors.

For many years prediction research has been influenced by the basic argument that strain accumulates in the rock masses in the vicinity of a fault and results in crustal deformation. Deformations have been measured in the horizontal direction along active faults (by trilateration and triangulation) and in the vertical direction by precise leveling and tiltmeters. Some investigators believe that changes in groundwater level occur prior to earthquakes; variations of this sort have been reported mainly from China. Because water levels in wells respond to a complex array of factors such as rainfall, such factors will have to be removed if changes in water level are to be studied in relation to earthquakes.

The theory of dilatancy (that is, an increase in volume) of rock prior to rupture once occupied a central position in discussions of premonitory phenomena of earthquakes, but it now receives less support. It is based on the observation that many solids exhibit dilatancy during deformation. For earthquake prediction the significance of dilatancy, if real, is in its effects on various measurable quantities of the Earth’s crust, such as seismic velocities, electric resistivity, and ground and water levels. The consequences of dilatancy for earthquake prediction are summarized in the table. The best-studied consequence is the effect on seismic velocities. The influence of internal cracks and pores on the elastic properties of rocks can be clearly demonstrated in laboratory measurements of those properties as a function of hydrostatic pressure. In the case of saturated rocks, experiments predict—for shallow earthquakes—that dilatancy occurs as a portion of the crust is stressed to failure, causing a decrease in the velocities of seismic waves. Recovery of velocity is brought about by subsequent rise of the pore pressure of water, which also has the effect of weakening the rock and enhancing fault slip.

Strain buildup in the focal region may have measurable effects on other observable properties, including electrical conductivity and gas concentration. Because the electrical conductivity of rocks depends largely on interconnected water channels within the rocks, resistivity may increase before the cracks become saturated. As pore fluid is expelled from the closing cracks, the local water table would rise and concentrations of gases such as radioactive radon would increase. No unequivocal confirming measurements have yet been published.

Geologic methods of extending the seismicity record back from the present also are being explored. Field studies indicate that the sequence of surface ruptures along major active faults associated with large earthquakes can sometimes be constructed. An example is the series of large earthquakes in Turkey in the 20th century, which were caused mainly by successive westward ruptures of the North Anatolian Fault. Liquefaction effects preserved in beds of sand and peat have provided evidence—when radiometric dating methods are used—for large paleoearthquakes extending back for more than 1,000 years in many seismically active zones, including the Pacific Northwest coast of the United States.

Less well-grounded precursory phenomena, particularly earthquake lights and animal behaviour, sometimes draw more public attention than the precursors discussed above. Many reports of unusual lights in the sky and abnormal animal behaviour preceding earthquakes are known to seismologists, mostly in anecdotal form. Both these phenomena are usually explained in terms of a release of gases prior to earthquakes and electric and acoustic stimuli of various types. At present there is no definitive experimental evidence to support claims that animals sometimes sense the coming of an earthquake.