The Seismological Laboratory of the California Institute of Technology is investigating a broad spectrum of subjects related to earthquake prediction. A selected portion of these investigations that relate to time-dependent geophysical parameters is summarized in the areas of seismic velocities, micro-earthquake levels, crustal distortions as monitored by radio interferometric methods, finite-element strain modeling of the Transverse Ranges region of the San Andreas fault, gravity networks, tensor resistivity as monitored by magnetotelluric methods, laboratory investigations of rock distortion prior to failure, and computer systems for automated handling of the data from large geophysical instrumentation networks.

View full abstract

This paper reviews work at Columbia University on earthquake prediction and related subjects. A series of small earthquakes was triggered by high-pressure fluid injection in western New York near the town of Dale in 1971. The fluid pressures observed at which the earthquakes were turned on and off is similar to that calculated by the Hubbert-Rubey theory. The hydraulic diffusivity was estimated as 1 to 2 times 10⁵ cm²·sec^-1. This value is very similar to that obtained for earthquakes related to fluid injection at Denver and Rangely, Colorado and Matsushiro, Japan. In each one of these four cases of earthquakes related to fluid injection fluids appear to have been injected into a major fault zone. Since the hydraulic diffusivities are nearly the same as those estimated from plots of earthquake precursory times and the length of the rupture zones squared, it seems likely that fluids play an important role in the precursory phase of earthquakes.
The Lamont group has been making a major study of intraplate earthquakes in New York State and vicinity for the past 7 years. This has included installing a network of about 35 short-period seismograph stations that are telemetered to Lamont. The network recorded the earthquakes from western New York related to fluid injection as well as two swarms of earthquakes at Blue Mountain Lake of which the larger events have been preceded by changes in seismic velocity. Thus far very little has been known about intraplate earthquakes in eastern North America even though a number of damaging shocks of that type have occurred in the last 300 years. An understanding of these shocks has become important since most of the nuclear power reactors in the United States are located in the eastern one third of the country. Two northwesterly trending zones of seismic activity in South Carolina and in New England appear to be situated along the landward continuation of two major fracture zones that were active in the early opening of the Atlantic. Similar zones of seismic activity and of relatively young alkalic rocks are found in similar locations in Brazil, western and southern Africa and southern Australia.
Networks of telemetered seismograph stations were installed by the Lamont group about three years ago in the northeast Caribbean, the Alaska Peninsula and eastern Aleutians, northern Pakistan, and around the Nurek Dam in the U.S.S.R In the Aleutians the seismic zone is remarkably thin and is relatively simple. In the region of continent-continent collision in northern Pakistan the downgoing seismic zone is more complicated and appears to consist of at least two zones of activity that extend to depths of about 70km. A major seismic gap for great earthquakes appears to be present along the northern Antillean arc in the Caribbean where underthrusting is of the more typical variety, that is perpendicular to the arc. Another seismic gap appears to exist north of Puerto Rico where the region has not had a great earthquake in at least 400 years. The tectonics of that region is more complicated since slip occurs very nearly parallel to the arc and consists of thrust faulting along nearly horizontal planes. Although the direction of slip is in accord with the overall motion of the Caribbean plate with respect to the Americas plate, the overall mechanics of slip is not well understood. Hence, it is difficult to say with certainty whether the region never has great earthquakes or if the region is a seismic gap in which a great earthquake will eventually occur. Tiltmeters, tide gauges, and other means of monitoring crustal deformation are being employed in two test areas in the Caribbean and on the Alaska Peninsula by the Lamont group.

View full abstract

Data from the continuous observation of microearthquakes which occur along the Yamasaki Fault and in the surrounding regions have been analyzed. There are some active lines of alignment of microearthquakes. Each of these corresponds to an active fault. We can find some properties of the block structure and the block movements of the crust from the temporal variation of the activity.
The fractured zone of the Yamasaki Fault shows some characteristic behaviors of strain change related to the activity of microearthquakes. These phenomena are closely related to the change of water level of the river along the fault and rainfalls at the concerned region.
Heavy rainfalls often triggers the occurrence of shallow earthquakes along the faults. Such characteristics are very useful for the earthquake prediction in very short time range.

View full abstract

The New Madrid seismic zone is one of the most seismically active regions of eastern North America. It is a small region, distant from any plate margin, and surrounded by the relatively uniform crust of central North America. For these reasons it provides an unusual opportunity to pursue earthquake prediction research in an intra-plate environment.
A regional telemetered micro-earthquake network in operation since July, 1974, has yielded much new information on the nature of the seismicity in the New Madrid region. The spatial distribution of hypocenters has indicated the existence of linear seismically active zones, presumably corresponding to active faults.
Teleseismic P-wave residuals recorded by the network have yielded a threedimensional velocity model for the crust and upper mantle beneath the New Madrid region, using an inversion method developed by AKI et al. (1977). The most interesting feature of this model is an anomalous region of low seismic velocities in the upper mantle and crust beneath the area of high seismic activity. The anomalous portion of the upper mantle extends to depths of at least 150km and is characterized by velocities which are about 5% lower than those in surrounding regions.
Observations of possible velocity precursors in the New Madrid region have been inconclusive. Other studies of precursory phenomena are currently being implemented. These include the installation of a network of tiltmeters, the instrumentation of a lake for tilt studies, and the observation of water level fluctuations in wells.

View full abstract

Earthquake prediction research on the recent anomalous crustal activity in the Izu Peninsula is summarized. Swarm activity of shallow microearthquakes in the eastern Izu Peninsula began in August 1975 and became more active in October. The epicenters clustered near Togasa-yama at first but spread over other places in the eastern part of and east off the peninsula in early 1976. Releveling carried out in January-April 1976 has disclosed crustal uplift of 15cm centered at Hiekawatoge, several kilometers north of Togasa-yama. The uplift area, more than 30km in diameter, covered the eastern half of the peninsula. Leveling, gravimetric, and tidal data showed that this uplift had developed only during the preceding 1-1.5 year period. Rapid changes in length of base-lines were also detected by repeated geodimeter measurements.
These phenomena were noted by the Coordinating Committee for Earthquake Prediction and observations were intensified. The information about the crustal activity was made public by the Committee in May 1976.
On August 18, the Kawazu earthquake (M=5.4) occurred in the southern part of the uplift area. Short-term precursors except for foreshocks were not so clear in spite of the intensified observations.

View full abstract

In order to obtain data for conducting research on the earthquake prediction as well as to monitor the seismic activity in and near Japan, seismological networks have been strengthened by many national universities and agencies. Most stations are telemetered to centers where computerized location facilites are installed, and the telemetered records are processed on a real-time or nearly real-time basis. The structure of hypocentral distribution has been made very clear in many parts of Japan by the data obtained from the networks.
Based on the data by the Japan Meteorological Agency, many researches on the seismicity gap and variation in seismic velocity prior to the occurrence of major earthquakes have been made. The seismicity gap appears to be a positive factor to predict the area where a major earthquake Will occur. On the other hand, both positive and negative results were obtained on the temporal variation in seismic velocity. In view of the present accuracy of observations, even the positive cases should be rechecked.

View full abstract

Precise geodetic surveys by the Geographical Survey Institute, in which triangulation stations better than second order would be measured periodically as rapidly as possible mainly by means of trilateration all over Japan, was started in 1973, in order to contribute to the prediction of earthquakes. Of the crustal strains found by comparing with old surveys striking contrast, quiet and violence, can be seen in the southern Kanto area and the central part containing both the Atera and the Neo Valley Faults.

View full abstract

The earthquake off the Izu Peninsula, on May 9, 1974 (M=6.9) was preceded by an anomalous seismic activity which started about 10 years and 4 months before, in July 1963 and can be distinguished from the ordinary seismic activity in the region. This earthquake was not, however, preceded by immediate foreshocks. At the same time there seems to have been some changes in seismic wave velocities in the above region. Similar phenomena were also found for 10 earthquakes which occurred at shallow depths in Japan.
The relation between earthquake magnitude M and the precursor time T in days taken from the above data is given by logT=0.77M-1.65. The relation between earthquake magnitude and precursor time found for other phenomena such as velocity of seismic waves, electric resistivity, crustal deformation, radon concentration in the ground water, etc., is given by respective investigators. These relations as well as my formula are examined. These results seem to suggest that among the above-mentioned precursory phenomena the seismic anomalous activity is the first one to take place. This may be due to the beginning of the formation of many ruptures in the region where the strain energy has been sufficiently stored.

View full abstract

The most recent data obtained from the high-gain seismograph network of Tohoku University are used to make an investigation on seismic activity and tectonics in the northeastern Japan arc. Seismic activity is closely related to the tectonic development of island arcs, and detailed analyses of the spatial distribution of microearthquakes in the northeastern Japan arc have led to conclusive evidence that almost all the shallow microearthquakes in the land area occur only in the layer with a P-wave velocity of 5.9 km/sec which was determined from explosion seismic observations.
The configuration of the deep seismic zone in this region is discussed in relation to the plate subduction beneath the arc. The accurate determination of the hypocenters and focal mechanisms of microearthquakes has made it possible to investigate the subduction process beneath the arc.

View full abstract

A field research on precise measurement of arrival times of seismic waves from man-made sources have been carried out, for the purpose of detecting precursory change in P-velocities. Seismic waves from detonations of about 500kg of explosives in bore-holes at a fixed site on Oshima Island, have been observed at about ten stations distributed over the South Kanto district. Observations were performed ten times from 1968 to 1976, approximately once a year. The reproducibility of wave-form is fairly good, and overall accuracy of travel-time including he errors in correction of shot location and the time correction of reading is about ±10 msec in maximum. Though the travel-time at every station seems to have been slightly delayed as a general tendency, this delay may be attributed to the change in source conditions due to repeated explosions. Thus it is likely that no significant changes in P-velocity have occurred in this period.
As to the anomalous crustal activity in the eastern Izu Peninsula, the dilatancy-diffusion model (DD model) does not explain sufficiently the apparent absence of P-delay in our experiment. Two hypothetical models other than DD model were proposed. One is an expansion of magmatic body, and other is a creep dislocation in the aseismic lower crust. These models seem to explain the present anomalous activity better than the DD model, so far as it has not been accompanied by significant decrease in P-velocity in the shallower part of the crust.

View full abstract

To get suggestions for the researches of earthquake prediction in Tokai district, short-term precursors prior to the recent great earthquakes along the Nankai trough are investigated here.
The Tonankai earthquake. The anomalous tilt of the ground had been observed several hours before the occurrence of that earthquake by a leveling party surveying in Tokai district. The party said also that they recognized with a level unusual quakes of the ground just before the earthquake. It is interesting that the similar phenomena were reported by two survey parties in China.
The Nankai earthquake. From the analysis of hourly value of tidal records, precursory land uplift prior to the earthquake was recognized in the southwest coast of Shikoku. Although the reliability of that uplift was not so high, another precursory phenomena, abnormal changes in the ground water, had occurred in several wells in the area. It is, therefore, probable that there occurred significant changes in the crustal stress and strains prior to the earthquake.
It is noteworthy that those precursors were observed in the areas near the boundary of the focal regions of the earthquakes.

View full abstract

Earth tidal strains observed concurrently at two stations in the western Chubu region have been analyzed to see if there are any temporal variations of their amplitude in relation to major earthquakes in the region.
The tidal strain records obtained at Kamitakara during 7 years and at Inuyama during 2 years are bandpass-filtered, and correlated with the corresponding theoretical tidal function predicted for a solid earth, using least squares mainly in the time domain, to estimate the tidal admittance over the periods. The admittances determined for the two-component Kamitakara strains appear to show a consistent and gradual increase during 10-11 months prior to the central Gifu earthquake of 1969 (M=6.6), with a maximum variation reaching about 15%, and after the earthquake they seem to decrease and recover their normal value a half year later. No such significant variations as in this period can be found over the entire period. The resolved amplitudes of the semidiurnal M2 and diurnal O1 waves show similar trends. The results from the Inuyama observations involve somewhat larger fluctuations, but appear to reach a peak 3-4 months before the earthquake and then decrease to the time of its occurrence.
Secular strains and tilts observed at the above two stations also indicate appreciable changes during the corresponding periods. If these temporal variations of the tidal strain amplitudes are not spurious but real phenomena, they might be interpreted as a manifestation of some precursory changes in crustal rigidity around the seismic source region.

View full abstract

Repeated gravity surveys have been carried out in several places for the purpose of detecting small changes in gravity accompanied with seismic land-deformations and faulting motions. Statistically significant gravity changes were found during the 1964 Niigata earthquake and the 1963-67 volcanic activities in Oshima Island. On the basis of the gravity-elevation relation, many authors estimated crustal density change induced by the Matsushiro earthquake swarm.
A recent highlight in gravimetric work is a gravity decrease with a free-air rate over the earthquake swarm area in the eastern part of Izu Peninsula. The rounded dome-like gravity-decreasing area has been gradually extended in a N-S direction. Another topic is a tidal gravity change recorded by an Askania gravimeter in the epicentral area during the 1974 Izu Hanto-oki earthquake. The postseismic change in daily-mean tidal factor, ranging from 1.1 to 1.6, was correlated with daily maximum amplitude of volcanic tremor observed beside the crater of Oshima Island, 40km distant from the tidal station.

View full abstract

A correlation between the locally anomalous secular variation in the geomagnetic total intensity and some large earthquakes which occurred in and near Japan since 1969 is examined. By combining the data obtained at 11 magnetic observatories with the data of the first order magnetic surveys at about forty stations, an equation for the most probable distribution of the normal secular variation in time and space is determined. Some anomalous changes in the geomagnetic field are separated from observed secular variations at each station by using the equation.
The result shows that the seismomagnetic effect may be detectable with a certain reliability in Japan in spite of small amount of the local anomaly theoretically expected. This suggests that the uppermost layer of the earth's crust in Japan may be highly sensitive to the tectonomagnetic effect. This inference may be also deduced from the fact that an interesting distribution of the local anomalies of the secular variation closely relating to the recent crustal movements was found in southwestern Japan.

View full abstract

A resistivity variometer of unusually high sensitivity has been in operation at a crustal movement observatory about 60km south of Tokyo since 1968. The variometer records changes in ground resistivity caused by crustal strain in association with tidal loading. When a large earthquake occurs, the variometer often records a step-wise change. Steps are observed for earthquakes having a magnitude of 8 or thereabout even if the epicentral distance exceeds 1, 000km. It is also the case for earthquakes of magnitude about 5 or more in the Tokyo area. Detailed investigations indicate that the variometer seems likely to detect extremely small coseismic strains in the form of resistivity steps.
The resistivity steps are often preceded by a precursory change having a precursor time of a few hours. A Weibull distribution analysis of 21 examples brings out that the characteristic parameter controlling the distribution of precursor time is more or less the same as that for a group of short-term precursors such as land deformation, tilt and strain, and underground water. This kind of resistivity change, that may probably reflect an occurrence of premonitory rupture in the focal region, seems to have an important bearing on short-term earthquake prediction.

View full abstract

An outline of geochemical studies for earthquake prediction starting from the end of 1973 in Japan is presented. In these studies most emphasis has been placed on the measurement of the radon concentration in groundwater, but variations in abundances and isotopic ratios of other elements, water temperature, water level of groundwater have also been measured.
At the present time a total of 30 observation sites for the periodical radon measurement are operated mainly by University of Tokyo and the Geological Survey of Japan. Among them the continuous measurement technique by using a recently developed ZnS(Ag) scintillation detection system has been adopted at five sites operated by University of Tokyo and Nagoya University. Tokyo Metropolitan Government runs the project of radon monitoring in groundwater at over 400 sites in and around the Tokyo area.
Two examples of continuous measurements of the radon concentration in groundwater obtained by using the above-mentioned system are given. Temporal changes of the radon concentration in groundwater in Izu and Tokai districts are compared with the occurrence of three earthquakes (M=5.5, 5.4, and 4.5) with the epicenter distances of some 20 to 130km.

View full abstract

We review recent laboratory observations relevant to earthquake mechanics and prediction. These continue to be largely from triaxial experiments, although new techniques have gained prominence, such as direct shear, holography, and scanning electron microscopy.
Two processes have been identified which produce stick-slip: (i) fracture of asperities, and (ii) indentation creep of asperities. Rupture velocities along sawcuts in granite and dunite were 2-3km/sec. Stick-slip was observed in serpentine, chlorite, and certain clays, when thin layers were sheared at high pressure. Fault creep has been studied systematically to 700°C and 5 kbars; some brittle cracking persisted, although plastic flow of quartz, mica, and possibly amphibole, was detected.
Dilatancy was found to persist through many stress cycles, although the detailed effect of cycling is still not clear. Dilatancy caused marked anisotropy in elastic moduli, compressional and shear velocity, and attenuation. The degree of localization before peak stress is a point of some disagreement. Microfracturing and optical studies show no localized dilatancy prior to fracture, whereas holography and slit diffraction show a narrow zone along the eventual fault direction. Small saturated laboratory samples showed an increase in the V_p/V₈ ratio with stress; a decrease occurred under very restricted conditions. Permeability of granite increased about fourfold prior to fracture. Dilatancy was observed in gabbro at temperatures up to 400°C, 2 kbars.
Several investigators described the properties of actual or simulated fault gouge. Based on equilibrium studies, some clays could be stable to depths of 12km in fault zones. Although gouge tends to reduce the amplitude of stick-slip over that on clean faults, stick-slip has been observed in thin layers of both clays and crushed silicate rocks. Large changes in compressional velocity with shear and normal stress were reported for a granite gouge.
Preseismic creep has been widely observed, on sawcuts, faults, and in both triaxial and direct shear experiments at room temperature, at pressures from 1.5 to 6 kbars. The amount of creep was appreciable at intermediate pressures and with more ductile materials or thicker gouge.
In a direct shear apparatus both resistivity and pore pressure changes preceded each stick-slip event, although velocity changes could not be detected. Resistivity decreased with stress for intact rocks, even with partial saturation. Dilatant volume changes prior to frictional slip were less than 10^-4, which represented a porosity increase of about 10%. Dilatancy altered the normal stress-remnant magnetism relation.

View full abstract

By development of a new triaxial compression technique, all components of principal stresses and strains of rocks were measured under general triaxial stress states in which the intermediate compression is not always equal to the minimum or maximum compression. The result shows that dilatancy is highly anisotropic under the general triaxial stress states. Previous experimental results on intact rocks and faulted rocks including the results reported in this paper suggest different rupture processes in the earth's crust which may be due to differences of the state of the earth's crust with respect to such factors as its mechanical structures, stress states, and so on. In some regions, marked precursors can be observed, but in other regions, it may be difficult to observe any appreciable precursors.

View full abstract

The Tokai district has been considered to be one of the probable sites of great interplate earthquakes in the near future. This idea is based on the time series of historical great earthquakes and the fairly high rate of crustal movement observed in this region for more than 80 years. The last great earthquake was the Ansei earthquake of 1854 (M=8.4). The western half of the source region of this earthquake ruptured again in the 1944 Tonankai earthquake, but the eastern half, i.e., the Tokai district, remain unruptured for 122 years. The recurrence time interval of great earthquakes in this region estimated from historical earthquakes, topographical and geological evidences, observed strain rates, the rate of plate convergence, etc. varies from 120 years to more than 600 years. There are several observations which may suggest that this region has already been in the long-term precursory stage. However, it is questionable that these observations represent real precursors to the future Tokai earthquake. It seems possible that the next great earthquake will occur south of the Izu Peninsula and will release the strain in the Tokai district. In any case, the Tokai district should be watched carefully for every conceivable precursors in order to make a successful prediction.

View full abstract

Dilatancy, the opening of cracks in rock under applied shear stress, has been proposed as a mechanism to explain seismic velocity anomalies prior to earthquakes. In recent years many studies have been made in search of precursory changes in V_p/V_s, but positive results have been extremely limited. Since dilatancy induced velocity anomalies require undersaturation of rock pores, we look carefully at porosity and fluid compressibility within the crust in an effort to determine the conditions that are most favorable for induced velocity anomalies. Simple porosity-temperature-pore pressure models of the crust are examined. Various dilatant strain distributions are imposed on the models and in. every case the most significant potential for velocity anomalies occurs within the upper 3 to 5km of the crust. Low porosity and high temperature at depth have the net effect of giving smaller velocity anomalies than in near surface regions. Actual crustal porosity, pore pressure and strain distributions prior to an earthquake are not well known. Nevertheless, our results suggest that searches for velocity anomalies should be concentrated in the upper few kilometers of the crust where the most significant anomalies are favored to occur.

View full abstract

On February 22, 1976, a magnitude 5 thrust earthquake occurred at a depth of 25km in the subduction zone near the Adak microearthquake network in the central Aleutians. The earthquake was located in a previously identified source volume with maximum dimensions of about 10km. This source volume falls within a larger seismotectonic feature characterized by a continuous band of seismicity along the arc in the depth range 17-27km. Within this band there was a period of decreased seismicity over a region 50km wide, centered on the main shock, beginning about 41/2 months prior to the earthquake. During a 5 week period before the earthquake, six small foreshocks occurred in the smaller source volume. High-resolution hypocenter locations showed a clear alignment of the foreshocks, progressing toward the location of the main event. The line of progression was roughly parallel to the direction of slip in the mainshock. The first and last foreshocks were characterized by thrust focal mechanisms common to earthquakes occurring during 'normal' periods. The remaining foreshocks were also thrust mechanisms, but with the principal axes of stress rotated nearly 90° on the basis of observed first motions and S/P amplitude ratios. For these same foreshocks there was also a suggestion of a very small P delay to a station with a ray going through the hypocentral region of the mainshock.

View full abstract

Most of active faults (=Quaternary faults) known on land in Japan are ranked to class A(1<S≤10mm/year), class B(0.1<S≤1mm/year) or class C(0.01<S≤0.1mm/year), according to a long-term average slip rate (S) during late Quaternary. A map of active faults which was recently compiled in scale of 1:4, 000, 000 describes locations of most faults of class A and class B. Statistics of recent historic on-land destructive earthquakes in Japan imply that the class C faults are much greater in number than the higher classes faults and that one third of future destructive earthquakes greater than magnitude 6.5 would occur from faults shown in the map.
Probable magnitude and recurrence time of earthquakes from active faults on land in Japan are calculated based on a long-term slip rate (S), length of fault (L_t) and length of time (t) elapsed without earthquakes up to the present. Recurrence time (R) of destructive earthquakes from an active fault on land is shown to be about 1, 000 years or longer.
Active faults with a elapse ratio E(=t/R)≥0.5 are marked in the active fault map as 'precaution faults.'

View full abstract

The latest results conceming the prediction of the June 17, 1973, Nemuro-oki earthquake (magnitude 7.4), off the Pacific coast of Hokkaido are discussed with special emphasis on the mode of strain release. This earthquake filled in the seismic gap that had been cited as a likely place for a future earthquake, but the preseismic land subsidence extending over the Eastern Hokkaido was not recovered. These facts suggest that the seismic slip at the plate boundary was deficent. Presently available evidence indicates that such a slip deficiency is unlikely to be taken up in the near future, by either aseismic faulting or occurrence of a great shock. More likely case proposed here is that much of the stored strain energy may remain unreleased during the interseismic quiescent period of seismic activity. In this case, it is suggested that the area of faulting and amount of slip differs greatly each time inter-plate earthquakes recur in at least a certain area, in view of the idea that plate motions at the subduction zone are not regular in size on a time scale of several hundreds of years or more.

View full abstract

The Coordinating Committee for Earthquake Prediction released information about anomalous land upheaval around Kawasaki City, Kanagawa Prefecture in 1974. This was taken by common people as a kind of earthquake prediction. Social consequences of the "prediction" were investigated in central and peripheral areas of the city through sampled citizens' evaluations and responses to it. The results were analysed to find what should be prepared for the proper delivery of necessary information and what assessment should be made for the model of prediction.
Main items of the questionnaire survey were designed to determine the citizens' cognitive accuracy for information content and its source, emotional impacts or aroused anxiety, actual behavior taken after the announcement and evaluations towards prediction and its announcing style.
Some difference in response was seen between subjects living near and far from the predicted epicenter, the former being likely to respond emotionally, whereas the latter tending to be solely in the cognitive dimension. Those sampled generally accepted the present prediction with favor, but more accurate authorized statement in various dimensions may be necessary next time.

View full abstract