Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
APPENDIX B EARTHQUAKE-PREDICTION RESEARCH OUTSIDE THE UNITED STATES Lynn R. Sykes PROGRAMS OF EARTHQUAKE PREDICTION OUTSIDE THE UNITED STATES General Review of Foreign Programs and Exchanges Japan, the U.S.S.R., and the People's Republic of China all have nation- al programs of earthquake prediction and earthquake-hazards reduction that have been functioning for a number of years. The Japanese program, funded by the Japanese government and now going into its third 5-year segment, has been described in detail in a number of publications. The Japanese program of earthquake prediction has had a great influence upon U.S. seismologists. Until about l0 years ago the subject of earth- quake prediction was greeted with a great deal of skepticism in the United States, was often placed in a category with astrology, and was not considered a proper subject of scientific study. In recent years, however, the research in earthquake prediction in Japan and the Soviet Union has caused a major change in attitudes in the United States. U.S. and Japanese scientists have met four times during the past l0 years for U.S.-Japan Conferences on Earthquake Prediction. These conferences have provided a very important medium for exchanging information and ideas about earthquake prediction and related research. The Soviet Union has had an active program of earthquake prediction and hazards reduction for more than l0 years. Following the destructive Khait earthquake of l949 in the Garm region of the Tadjik Republic, So- viet seismologists in the early l950's put into the field the Complex Seismological Expedition to record and study the spatial distribution of earthquakes, their variations with time, and other statistical prop- erties. The program of scientific research in the Garm region was later expanded to include searches for precursory phenomena associated with earthquakes. In both the U.S.S.R. and Japan, much emphasis has been placed on try- ing to detect precursory phenomena. A wide variety of physical phenom- ena were examined, including changes in elevation of the land, in the velocity of seismic waves, in the frequency of occurrence of small earthquakes, in the electrical properties of rocks, and in earth strains and tilts. Until results emerged from these programs in the U.S.S.R. and Japan, there was considerable skepticism among scientists as to whether many of the early reports of precursory changes represented 5l
52 valid changes associated with earthquakes or were merely interfering ef- fects and noise. It is largely as a result of the massive efforts in Japan and the U.S.S.R. that many geophysicists now accept the view that precursory changes do, in fact, occur before at least certain types of earthquakes. Many of the data used by U.S. and other scientists in sup- port of the dilatancy models of precursory changes preceding earthquakes have come from measurements made in Japan and the U.S.S.R. during the past l0 years. A number of scientists from the United States heard papers on vari- ous aspects of earthquake prediction in the U.S.S.R. at the meeting of the International Union of Geodesy and Geophysics in Moscow in l97l. Several U.S. seismologists also visited Soviet institutes working on earthquake prediction, particularly the Complex Seismological Expedi- tion at Garm, in Central Asia. These papers and visits were instrumen- tal in stimulating U.S. scientists to intensify their search for precur- sory effects of earthquakes. A U.S.âU.S.S.R. program in earthquake prediction and earthquake- hazards reduction was one of several agreements reached at the l972 "summit meeting" between President Nixon and Chairman Brezhnev. A de- tailed plan of cooperative research between scientists of the two coun- tries was drawn up during a visit to the U.S.S.R., in October l973, of a U.S. delegation on earthquake prediction. Several American geophysi- cists visited the Soviet Union, and U.S.S.R. scientists visited the United States under this program in l974. Results of Soviet investiga- tions at Garm were reported at a conference at Aspen, Colorado, in August l974. U.S. seismologists installed a network of seismic instru- ments in the Garm region in l974, and other U.S. projects in Garm and near the Nurek Dam, Tadjikistan, were begun in l975. Earthquake Prediction in the U.S.S.R. Active field programs of earthquake prediction are under way in three areas of the Soviet Union: in the Garm region of central Asia, in the area near Tashkent in central Asia, and in Kamchatka. The first two areas are situated near the belt of moderate-to-high seismic activity that extends from the Mediterranean region across central Asia to the Himalayas and western China. Kamchatka, the most active earthquake re- gion in the U.S.S.R., is a part of the very active seismic zone that borders the northern and western Pacific. Central Asia is generally characterized by north-south compression, which appears to be related to continental collision between India and the rest of Asia. Research in earthquake prediction in Kamchatka is made difficult by the fact that most of the large shallow earthquakes are located offshore beneath the inner wall of the Kurile-Kamchatka trench. The different tectonic environments in these three regions account for the different approaches to earthquake prediction.
53 Garm Region Soviet investigators have examined possible precursory changes of a wide number of physical parameters in the Garm region of Central Asia. Among the most important of these, discovered for the first time at Garm, is the so-called ts/tp ratio, the ratio of the travel times of seismic shear (ts) to compressional (tp) waves. Soviet investigators noticed that this ratio, normally about l.75 (a Poisson's ratio of 0.26), decreased by about l0 percent over an interval before moderate- size earthquakes. The ratio was observed then to return to normal im- mediately prior to the occurrence of earthquakes. The duration of this anomaly was found to be longer the larger the size of the earthquake. Soviet investigators postulate that the frequency content of seismic waves is also altered over a period of time preceding earthquakes. Similar changes in ts/tp (or, alternatively, in the ratio of seismic velocities vpvs) have been observed preceding earthquakes in New York State, South Carolina, California, and Japan. The rate of occurrence of very small earthquakes also appears to change preceding moderate and large earthquakes -in Garm. The number of small earthquakes in a fault zone appears to build up very slowly with time and then to undergo a marked reduction (or quiet period) just prior to a larger earthquake. Soviet investigators have reported pre- cursory changes in earthquake mechanisms (which reflect the types of fault motion), which they interpret as related to changes in the tec- tonic stresses prior to earthquakes. An extensive series of measurements of electrical conductivity has been carried out in the Garm region since l967. As measured over base- lines a few kilometers long, electrical conductivity appears to increase by about l0 to 20 percent prior to moderate-size earthquakes. These changes have been attributed to an increase in the water content of the rocks, since dry rocks normally have very low conductivity. A series of geodetic measurements has been made in the Garm region during the past l0 years to detect precursory changes in horizontal and vertical movements. Soviet investigators report anomalous increases in elevation of the land prior to moderate and large earthquakes. Tashkent and Uzbek Republic The damaging Tashkent earthquake of April 26, l966, apparently influ- enced the Soviet government to give increased attention to earthquake prediction and earthquake-hazards reduction. Although this earthquake was only of moderate size (magnitude 5.3), it occurred almost directly under the major city of Tashkent and caused a great deal of damage to older structures. The Tashkent earthquake has become well known among geophysicists since it was in connection with that event that it was recognized that geochemical indicators could be used for earthquake prediction. The content of radon and other gases in well water had been measured on a regular basis for several years prior to the l966 earthquake, primarily
54 in studies of the chemical properties of medicinal waters. The water that was sampled for its gas content was taken from a well that extend- ed into what was to become the hypocentral region of the l966 earth- quake. Prior to the main shock of l966, a marked increase was observed in the amount of an isotope of radon that has a half life of about 4 days. Similar increases were found before several of the larger after- shocks. The increase in radon prior to these earthquakes may be attrib- uted to the openings of small cracks in the rock and to increased flow of fluids near the hypocentral region. More than 20 wells near Tashkent and within the Fergana Basis of Uz- bekistan are now being monitored for precursory changes in the radon and helium content and in temperature and fluid pressure. Marked changes in temperature and fluid flow were observed prior to the l970 Przhevalsk earthquake, near Alma Ata in Central Asia. Changes in temperature and in the content of radon and helium are also now being monitored in sev- eral wells along the major Surkob fault zone in Tadjikistan. Few mea- surements of these kinds have been made in the United States. Hence, it appears that Soviet investigators are perhaps several years ahead of U.S. scientists in studying geochemical indicators that may be useful for earthquake prediction. Kamchatka Since the beginning of l972, Soviet investigators in Kamchatka have been attempting to use changes in the telluric field, in seismic veloc- ities (vp/vs), and in the variations of occurrence of small earthquakes for the routine prediction of earthquakes. The earthquake regime near Kamchatka is very similar to that of southern Alaska and the Aleutians. Both are characterized by rapid un- derthrusting of these island arcs by the Pacific plate. This under- thrust region is characterized by volcanos, very large shallow earth- quakes, a deep-sea trench, and a zone of earthquakes associated with the plunging plate and dipping to great depth under the island arcs. Tec- tonically, the setting is also very similar to that in Japan. Fedotov and Mogi have studied the sizes of rupture zones of very large earthquakes in the western Pacific between Kamchatka and Japan. They conclude that great earthquakes tend to fill in so-called seismic gaps, segments of active faults along which no large earthquakes have occurred for periods of tens to hundreds of years. They find that the rupture zones of these great earthquakes along a single large fault tend to abut without significant overlap. Hence, these great shallow earth- quakes tend to occur with considerable spatial regularity. Fedotov has used this idea of seismic gaps to forecast those parts of the region be- tween Kamchatka and Japan that appear to have the greatest likelihood of future great earthquakes. This technique does not provide more than a very rough estimate of the time of occurrence of future great shocks. Nevertheless, it can be used to guide more detailed studies of earthquake precursors by allowing investigators to concentrate on a few critical areas rather than spreading their attention over the entire 2,000-mile length of the plate boundary stretching from Kamchatka to Japan.
55 Soviet geophysicists have been carrying out a series of large under- water explosions for seismic research off the east coast of Kamchatka since l965 to search for possible precursory changes in the velocity of compressional waves before large earthquakes. Seismic records for several explosions have been obtained each year since l965 at several stations near the east coast of Kamchatka. The Soviet investigators reported a change in travel time of about 0.l second over a period of a few years prior to a magnitude 7.2 earthquake that occurred at a depth of about l00 kilometers beneath the region of their experiment. It is not clear whether the change in travel time they observed was actually a precursor of the earthquake or related to uncertainties in the loca- tions of the explosions relative to the rough topography of the ocean floor. Soviet scientists have also been studying the natural telluric field of the earth near the east coast of Kamchatka. However, since most of the large earthquakes are located in the ocean off the east coast, it is very difficult in this region to use an active measuring technique in which a controlled current is fed into the ground and changes in the electric and magnetic fields are observed at various sites. Neverthe- less, some studies have been successful, and a number of short-term vari- ations in the telluric field observed by Soviet investigators are thought to be earthquake precursors, despite the fact that the natural telluric field is characterized by a great deal of noise related to variations in the ionosphere and changes in rainfall. Japanese Program in Earthquake Prediction A national program in earthquake prediction was initially launched in Japan in l965. The funding and planning of the Japanese program have been in 5-year increments, and the program is now mid-way through the third 5-year increment. During the ll-year period l965-l975, a total of $28 million was allocated for the program (excluding salaries), and for l975 the amount was about $4 million. While this budget is not as large as that for the Japanese space project, it is strikingly larger than the budgets for other projects in solid-earth science in Japan. The Japanese program of earthquake prediction places great emphasis on the collection of a wide variety of data that are seen as essential to progress in predicting earthquakes. Three new centers were set up to promote data collection and processing; the Crustal Activity Monitoring Center, for geodetic data and tide-gage data, was set up in the Geograph- ical Survey Institute; the Seismicity Monitoring Center, under the Japan Meteorological Agency, for analyzing and recording earthquakes of magni- tudes larger than 3; and the Earthquake Prediction Observation Center, for analysis of microearthquakes (of magnitudes smaller than 3), crust- al deformation, magnetic data, and other data from university sources. Data from these three centers are presented to the Coordinating Commit- tee for Earthquake Prediction (CCEP), which consists of specialists from universities and government institutions. The CCEP is actually a headquarters for earthquake prediction.
56 The Japanese program has established several stages of earthquake alert for scientists, government agencies, and the public. In many ways, this alert sequence is similar to the alerts issued by the Tsuna- mi Warning System, based in Hawaii. (A tsunami is a seismic sea wave.) Whenever anomalous phenomena, such as land deformation, are observed by either nation-wide routine observations in special areas such as active faults and densely populated areas, the region of the anomaly is desig- nated as an "area of intensified observation." If the anomaly is later suspected to be precursory to a major earthquake, the designation changes to "area of concentrated observation" and all types of observations are concentrated there. If precursory phenomena are eventually more posi- tively identified, a warning, or prediction, of an earthquake may be issued to the public. In his review article, Rikitake mentions that the CCEP is responsible for the judgments involved in the above procedures. However, the mem- bers of the CCEP work on a part-time basis only, and it is therefore difficult for them to examine the relevant data on a continuous basis. In addition, with respect to program activity, the CCEP is limited to coordination only. Though the CCEP cannot assign study projects to other organizations when anomalies have been observed, it can provide the available background information and suggest the kinds of observa- tions needed. Geodetic Surveys Anomalous crustal uplift has been noted before several large historic earthquakes in Japan, some of them in this century, by geodetic observa- tions. As a result, the Japanese prediction program places great empha- sis on both vertical and horizontal geodetic monitoring. About 20,000 kilometers of first-order leveling lines have been established all over Japan. Plans call for the leveling to be repeated at 5-year intervals in order to maximize the probability of detecting anomalous crustal movements and to monitor strain buildup. Some of the best observations of precursory crustal movements were made prior to the Niigata earthquake of l964. Several leveling lines established before the earthquake were resurveyed at intervals before the earthquake and again afterward. These surveys showed that precur- sory uplift began about l0 years before the earthquake. These data have played a central role in developing models of the physical process in precursory phenomena, such as the dilatancy/fluid diffusion model. A leveling survey of the Boso Peninsula, southeast of Tokyo, indi- cated a possible anomalous uplift there in l969. This peninsula was uplifted in the great l923 Kanto earthquake and had been subsiding since then. Thus, the apparent uplift in l969 was feared to be a precursor of a large future earthquake. Therefore, the CCEP, in l969, designated the south Kanto area as an "area of intensified observation." Leveling surveys have been repeated over the Boso and Miura Peninsulas nearly every year since l970. It is now thought, however, that the apparent anomalous uplift in l969 was a result of noise or errors in the geodetic observations. The late l960's were the first time that precise leveling
57 surveys had been carried out in this region a few years apart, but thi experience suggests that such a short period of time does not appear tv be enough to obtain a reliable indication of precursory changes in ver- tical motion. The changes for the period l965 to l973 indicate a pat- tern that is very similar to that which has been going on since l923. Also, the amount of strain buildup since l923 is about one-third that released during the l923 earthquake. Hence, the threat of a repeat of the 1923 earthquake does not now appear as imminent as it did in l969. Geodimeters are now being used in Japan on a regular basis for moni- toring changes in horizontal distances, and the geodimeter surveys are now being integrated with previous triangulation surveys. Under their present program of earthquake prediction, the Japanese plan to estab- lish a nation-wide network of geodimeter lines consisting of about 6,000 triangles, which they plan to remeasure every 5 years. As of l973, Japan had established l7 observatories for monitoring crustal movements with tiltmeters and strain meters, and some of the best data on precursory changes in tilt have come from Japan. It is planned to set up many bore-hold tiltmeters, along with other instru- ments, in holes several tens of meters deep, in a project to increase the number of tiltmeters and strainmeters in Japan. Matsushiro Earthquake Swarm The Matsushiro earthquake swarm of l965-l977 occurred soon after the Japanese had started their first program of earthquake prediction. A wide variety of field measurements were made at Matsushiro that bear upon searches for precursory phenomena. This is one of the few times when changes both in gravity and in vertical motion were monitored. Although there is considerable debate about the scatter of the data, these observations appear to be in accord with fluid flow into a dilat- ant region. One of the outstanding features of the Matsushiro swarm was the grad- ual enlargement with time of the fault region experiencing earthquakes. It was found that microearthquakes tended to migrate into a new region along the fault zone prior to the occurrence of moderate-size earth- quakes a few months later. Based on such observations, warnings were issued that moderate-size earthquakes could be expected within a few months. One of several earthquakes predicted successfully in this way was filmed by cameramen who set up their equipment in advance. Seismic-Activity Monitoring Programs The Japan Meteorological Agency (JMA) now records seismic data on mag- netic tape at 67 of its weather stations. The JMA network is used to locate earthquakes of magnitude larger than 3. It is very difficult to locate precisely earthquakes that occur along the inner wall of the Japan trench off the east coast of Japan. Since the largest Japanese earthquakes occur in this offshore area, the third 5-year plan for earthquake prediction calls for the installation
58 of several ocean-bottom seismographs for continuous monitoring of this offshore activity. This work is also under way with the JMA. Several university groups in Japan have active programs of monitor- ing earthquakes smaller than magnitude 3. A great deal of research is currently under way to search for changes in the relative numbers of large and small earthquakes (the so-called b value) and for changes in the frequency of occurrence of small earthquakes. Plans are under way for telemetering of much of the data on small earthquakes to several regional centers for analysis. To overcome problems of man-made noise in the Tokyo area, the Na- tional Research Center for Disaster Prevention installed borehole seis- mometers in a 3,500-meter (ll,500-foot) well a few tens of kilometers north of Tokyo. Seismic Wave Velocities Explosions have been detonated on Oshima Island, south of Tokyo, ap- proximately once a year since l968 to monitor the seismic waves thus generated. No change in the velocity of the compressional (p) wave exceeding 0.l% appears to have taken place. Ohtake has reported changes in vp/vs prior to 3 earthquakes in Japan, including one of the Matsushiro swarm. Magnetic and Electrical Fields Proton-precession magnetometers with digital recording have been set up at l2 stations in Japan. However, noise caused by local variations in the magnetic field of as much as 2 gammas between stations a few hun- dred kilometers apart tends to mask possible magnetic precursory sig- nals. Also, stray electric currents from the extensive electric railways in Japan lead to large local magnetic disturbances. Seismic Gaps Extensive work has been done by Mogi and Utsu in identifying seismic gaps in Japan. The June l7, l973, Nemuro-Oki earthquake, off the coast of Hokkaido occurred in one such seismic gap that had been cited as a likely place for a future earthquake. Utsu has identified five sites in central and southwestern Japan at which earthquakes with magnitudes estimated at 7 or greater occurred between 7l5 AD and l325 AD as places that should be watched carefully as possible sites of future large earthquakes.
59 Summary A great variety of research and data collection programs are under way in many areas in Japan to search for and monitor precursory phenomena that can be used to predict earthquakes. In addition to the geophysi- cal disciplines mentioned above, geologists and geographers have been examining active faults, folding, and other crustal movements in rocks of the Quaternary Period (approximately the past 2 million years). Long-term rates of movement have been estimated for many faults. Also, an intensive program of laboratory studies in rock mechanics is under way to increase understanding of the physical basis of earthquake pre- diction. It is quite possible that Japan will become the first country to achieve routine prediction of earthquakes. The Japanese have the ex- perience gained from an ll-year national program of earthquake predic- tion, a vast number of trained scientists and technicians active in earthquake studies, and a relatively small geographic area to monitor, compared with the size of earthquake regions in the United States, the U.S.S.R., and China, for example. Also, most damaging earthquakes in Japan occur at shallow depths within the islands. These shocks gener- ally tend to be more damaging even though they are smaller than the great earthquakes located off the east coast of Japan. Shallow earth- quakes within lithospheric plates, such as those within the Japanese islands, are the sources of many of the precursory effects detected thus far. The shallow nature of the sources, and the fact that the source regions can be readily surrounded by instruments, make it much easier to monitor possible precursory changes than if the earthquakes were located off the coast. Earthquake Prediction in China A massive effort to detect a wide variety of precursory phenomena asso- ciated with earthquakes is under way in China. The scale of this pro- gram, and most of its results over the last few years, became known to foreign scientists when a seismology delegation from the United States visited China in October and November l974. There has been heightened interest in the United States in earthquake prediction in China follow- ing announcements by the Chinese that they had successfully predicted a major earthquake that occurred in northeastern China on February 4, l975. If that event was indeed predicted using acceptable scientific criteria, it is a major milestone in the history of seismology and in the forecasting of natural disasters. Field programs to detect possible precursory phenomena have been under way for the past few years at the provincial centers of Kunming (Yunnan), Chengtu (Szechuan), and Lanchou (Kansu), within the zone of high seismic activity that crosses China from south to north, as well as in the Peking area, in the region of the damaging Hsingtai earth- quakes of l966, and near the Hsinfengking dam in Kwangtung province. The very high level of seismic activity and the relative accessibil- ity of many of the active regions of China to monitoring (the most
60 active zone for earthquakes affecting the United States, by comparison, is located almost entirely offshore, along the Aleutian trench) are two factors that have led to the rapid accumulation of precursory data and can be expected to remain important for prediction of large earthquakes in China in the future. A wide variety of precursory phenomenaâthe existence of which have been reported previously in Japan, the U.S.S.R., and the United Statesâare being studied in China, largely empirically. Relatively little effort appears to have been made thus far in designing field or laboratory experiments around theories or models, such as the dilatancy/fluid-diffusion hypothesis. Most Chinese scientists appear to be convinced that earthquakes are preceded by a wide range of precursory phenomena and that these phenom- ena occur in a variety of tectonic environments. Changes in the follow- ing variables have been reported prior to earthquakes and are being studied actively in extensive field programs: crustal movements, seis- mic velocities, frequency of occurrence of small earthquakes, seismic attenuation, the flux of radon from water in deep wells and springs, water level and temperature in wells, geomagnetic field, natural geo- electric field using telluric currents, and earth resistivity using an applied potential. Chinese scientists generally distinguish precursory phenomena as either long term (years), intermediate term (months to weeks), or short term (days to minutes). Visiting U.S. scientists generally believe that several of the re- ported changes in radon flux, seismic velocity, and frequency of occur- rence of small earthquakes appear to represent valid precusory phenomena. Some of the other reported anomalies may result from other causes and may have been associated in time with the earthquakes only by chance. Although precursory phenomena were observed hundreds of kilometers from the epicenters of magnitude 4-to-5 shocks, most of the anomalies with high signal-to-noise ratios did not occur at such great distances but were limited to within a few fault-rupture lengths from the epicenters. Several of the better examples of precursory phenomena have a time dura- tion, t, that falls close to the line log t = a + cM, discussed by Scholz et al. (l973) and Whitcomb et al. (l973) where a and c are constants and M is the magnitude. In addition, various short- term effects that fall well below the line defined by the equation were also observed before moderate and large earthquakes. Among various Chinese scientists, attitudes range from acceptance of nearly all reported anomalies as valid precursors to earthquakes to the view that a large body of data should be accumulated before certain types of anomalies are chosen as the more reliable precursors. The more senior scientists generally hold the latter view. Particularly at the provincial and county levels, where a great amount of the monitoring and data analysis is now being done, attempts are being made to use a variety of types of anomalies to predict earth- quakes routinely. In addition to the February 4, l975, earthquake, Chinese scientists claim to have predicted several other earthquakes successfully. They emphasize, however, that they have failed in some other predictions, that they have not estimated the time, place, or
6l size very accurately in still others, and that they have made some pre- dictions that turned out to be false alarms. They are not yet prepared to tabulate these results so as to produce a false-alarm ratio or a success-failure percentage. Most Chinese scientists believe that obser- vations using a multiplicity of techniques, rather than reliance on a single method, are necessary for reliable prediction. The scale of the Chinese effort to predict earthquakes is very large indeed. Some l0,000 trained workers are involved and, although figures are not available, the number of untrained volunteers working part-time is probably as large. Although the cost of this effort is difficult to convert into a U.S. monetary equivalent, it might be in the range of $50 million to $l00 million a year. Although the reasons for investing such a large amount of human and economic resources in this program were not discussed during the American visit in l974, there appear to be several fairly obvious explanations. From a practical standpoint, the prediction of a forthcoming great earthquake is at present the only way to avoid repetition of some of the terrible calamities of the past few hundred years. The loss of 820,000 lives in the Huahsin earthquake of l556 AD, near Sian, was due almost entirely to the collapse of houses in the densely populated valley of the Wei and Huang rivers. Housing construction in that area today, as in much of rural China in general, is undoubtedly not greatly different from that of l556. New construction, though of brick, is generally not designed for earthquake resistance and is likewise liable to major damage in a great earthquake. To bring rural construction, which houses over 600 million Chinese, up to earthquake-resistant de- sign standards would require an effort so immense that China's goal of industrial self-sufficiency would be drastically retarded. On the other hand, following an earthquake, reconstruction in the limited regions that will suffer great earthquakes in the next l00 years would not be a burden on the economy, provided the inhabitants survive. If, therefore, the great earthquakes can be predicted, the inhabitants' lives can be saved without having to resort to the draconian solution of seriously modifying or rebuilding the 30 percent or so of China's total existing housing that occupies its earthquake-prone regions. Given the practical need for reliable prediction, then, the Chinese believe, further, that earthquakes are predictable. In their very well- documented history of large earthquakes, examples of precursory phenom- ena are numerous and, in particular, certain of these recur. Peculiar- ities in the behavior of animals and anomalous changes in the water level or water quality in wells have been reported repeatedly. There- fore, at least empirically if not theoretically, the problem seems to be a soluble one. An extensive description of Chinese work in earthquake prediction was published in l975 by the U.S. Delegation on Seismology that visited China in l974.
62 SELECTED REFERENCES Rikitake, T. (l974). Japanese National Program on Earthquake Predic- tion. Tectonophysics, v. 23, n. 3, pp. 225-236. Sadovsky, M.A., et al., eds. (l973). Earthquake Precursors. Moscow. Anonymous (l97l). The Tashkent Earthquake of 26 April l966. FAN. Tashkent. Sadovsky, M.A., ed. (l972). Physical Bases of Seeking Methods of Pre- dicting Earthquakes. Moscow. Tectonophysics (l972). Special issue entitled: Forerunners of Strong Earthquakes. Various papers by Soviet authors, v. l4, n. 3 and 4. Transactions, American Geophysical Union (l975). Earthquake Research in China by Members of the American Seismology Delegation to the People's Republic of China, v. 56, pp. 838-88l.
