Earthquake Prediction Research in Japan Kiyoo Mogi College Of

J. Phys. Earth, 43, 533-561, 1995

Earthquake Prediction Research in Japan

Kiyoo Mogi Collegeof IndustrialTechnology, Nihon University, Narashino275, Japan

The first 5-yearearthquake prediction project in Japan started in 1965.Now, the sixth 5-year project (1989-1993)is in progress. Various kinds of fundamental observationshave been continuedand the density of the observationnetwork has increasedwith time,but there have been no successfulcases of earthquakeprediction. However,reliable data usefulfor long-termand short-termprediction are beingsteadily accumulated,and severalcases have suggestedthe feasibilityof earthquakeprediction. In the Tokai districtof centralHonshu, where a major earthquakeis expectedto occur in the near future, a programfor constant monitoringand assessmentwas formally inauguratedwith the goalof forecastingan M8 classearthquake. The results of research on recent major earthquakesin Japan are reviewedas well as the Tokai earthquake problem.

1. Introduction The Japanese islands and their vicinity constitute one of the most active seismic regions in the world and have experiencedmany major earthquake disasters in the past. This fact has led to great interest in earthquake prediction ever since seismological studies were begun in Japan a century ago. Omori (1909), Imamura (1928) and other Japanese seismologists carried out pioneering research on the periodicity of great earthquakes and on seismic gaps and other subjects. Their research was limited, owing to the level of observations and the inadequacy of data in those days. However, their works contained the beginningsof the important methods used in earthquake prediction today. Earthquakes occur when stress that has been applied to the earth's crust reaches a limit and a sudden fracture (the sudden slip of a fault) occurs in a part of the earth's crust. In general, brittle materials fracture suddenly, and it is difficult to predict this fracture accurately. Earthquakes take place at intervals of 100 or more years, but practical prediction requires that earthquakes be forecast on the order of days. This raises the question of whether such highly accurate prediction is actually possible. According to laboratory experiments, the rupture process strongly dependent on the degree of heterogeneity of materials and ruptures of heterogeneous materials do not occur without warning (e.g., Mogi, 1962). Fortunately, the earth's crust in some seismogeniczones is mechanically complex (for example, in heterogeneous strength

ReceivedAugust 25, 1992;Accepted December 4, 1992

533 534 K. Mogi distribution), and groundwater exists in the pores of the earth's crust. As stress on the earth's crust increases, crustal changes proceed by degrees, and various phenomena may occur immediatelybefore the principal rupture. If researchersfocus on this process and are able to adequately trace it, long-termand short-term prediction may be possible. Severalmajor earthquakeshave occurredsince the start of the earthquake prediction program in Japan. Although there have been successfulexamples of long-termprediction on the basis of seismicgap theory, as for the 1973Nemuro-Hanto-oki earthquake (Utsu, 1972),there have been no cases in which all three elementsof earthquake prediction— place, magnitude and time—have been successfullyforecast. However, reliable data available for prediction are being steadily accumulated, and several cases suggest the feasibility of earthquake prediction. In this paper, the results of research on major recent earthquakes in Japan is reviewed as well as the earthquake prediction problem in the Tokai region where a major earthquake is expectedto occur in the near future.

2. Earthquake PredictionProgram in Japan The earthquake prediction program in Japan that began in 1965is now in its sixth 5-year project. However, this program is supported by observation and measurement data accumulated over almost 100 years. In 1883 the Geographical Survey Institute (GSI) carried out nationwide geodeticsurveys (leveling and triangulation); these surveys have been repeated at regular intervals since then and whenever a major earthquake has occurred. In addition, the Japan MeteorologicalAgency (JMA) has made efforts since its early stages to make uniform seismic observations through Japan and has performed instrumental observationsfor more than 100years. These data and the results obtained from them enable seismologiststoday to discuss earthquake prediction from a fundamental standpoint. In recent years, various observation methods for earthquake prediction have been developedand active research has been promoted coveringa broad range of fields. The Japanese program consists of three categories. The first is observations for long-term prediction, such as geodetic surveys, seismic observations and geoelectric- geomagnetic observations. As an example of these observations, the horizontal deformation of Shikoku Island is shown in Fig. 1 (GSI, 1991b). The broken lines and the solid lines show the maximum contraction and extensionin each area, respectively. The low strain state in eastern Shikoku Island is attributed to the strain release by the 1946.Nankaido Earthquake of M8.0 which occurred off the Kii Peninsula and eastern Shikoku. The high strain accumulation in southwesternShikoku Island is noted. On the basis of the geodetic and seismic data, this region has been marked as a potential region for a future large shallow earthquake (e.g., Shimazaki, 1984). The second is observations for short-term prediction, such as continuous observations of crustal movement, seismic observations,geomagnetic and geoelectricob- servations, geochemicalobservations and groundwater measurements. As an example of these results, an anomalous change in geoelectricpotential prior to an earthquake of M5.6 along the Yamasaki fault in western Honshu is shown in Fig. 2 (Miyakoshi, 1986).

J. Phys. Earth Earthquake Prediction Research in Japan 535

Fig. 1. Direction and magnitude of horizontal strain in Shikoku Island found from the two surveys in 1890 to 1901 and 1989 to 1990. Broken lines, contraction; solid lines, extension (from GSI, 1991b). The high compressive strain in the NW-SE direction in southwestern Shikoku Island is noted .

Fig. 2. Anomalouschange in geoelectricpotential prior to an earthquake of M5.6 along the Yamasaki fault in western Honshu (from Miyakoshi, 1986).

The third is fundamental studies, such as studies of crustal structures of seismogenic zones, historical earthquake studies, active fault surveys, laboratory fracture experiments and theoretical research on earthquake source processes. These observations and research projects are carried out by universities, JMA, GSI,

Vol. 43, No. 5, 1995 536 K. Mogi

Fig. 3. Seismographicstations of JMA, NIED and universities,including array stations in the Tokai and Boso regions having ocean-bottom seismographs.

the National Research Institute for Earth Science and Disaster Prevention (NIED), the Geological Survey of Japan, the Hydrographic Department of the Marine Safety Agency and other governmental institutions. Figure 3 shows seismographic stations of JMA, NIED and universities. They include array stations in the Tokai and Boso regions having ocean-bottom seismographs. Seismic observations for earthquakes of M3 and over have been conducted by JMA, and microearthquakes in the major part of the Japanese islands have been observed by universities and NIED.

3. Coordinating Committee for Earthquake Prediction and Earthquake Assessment Committee In 1969 the Coordinating Committee for Earthquake Prediction (CCEP) was established to gather and examine data on a regular basis and to investigate long-term and intermediate-term prediction. The 30 members of the committee, which meet four times a year, are mostly university staff and government officials who participate in the earthquake prediction program. Two subcommittees meet whenever necessary to examine specific areas and urgent issues. The possibility of a large earthquake in the Tokai region was first pointed out in 1969 (Mogi, 1970). It became a major social issue in 1976, because a disaster of serious consequence is expected if an earthquake occurs. In 1978 the Large-Scale Earthquake Countermeasures Act was enacted. Under this law the prime minister is to issue a public warning, and government and private organizations are to respond immediately by

. Phys. Earth

J Earthquake Prediction Research in Japan 537

adopting effective procedures. As a result of this law, observed data are transmitted to JMA in Tokyo and constantly monitored. If any anomalies are observed, the Earth- quake Assessment Committee, which consists of six university professors, is to meet immediately and assess the data. This is an important national experiment aimed at minimizing the disaster by predicting the earthquake, and it is the first step toward practical short-term earthquake prediction in Japan.

4. Recent Major Topics of Earthquake Prediction

4.1 The 1983 Japan Sea earthquake (M7.7) On May 26, 1983, a large earthquake (M7.7) occurred on the floor of the Japan Sea off the western coast of northern Honshu. This was the largest earthquake to have occurred on the Japan Sea side of the country, and its focal region covered an area 120km long and 40km wide, extending south to north. This focal region was located 100km out to sea, away from the observation network on land. The Japan Sea is far less active than the Pacific side of Japan, so the monitoring networks there were incomplete. Hence it was not possible to forecast this earthquake on either a long-term or short-term basis. If absolutely no precursory phenomena could be recognized, even after the earthquake, the outlook for earthquake prediction would indeed be grim. Thus the results of investigations after earthquakes are an extremely important indication of the possibility of predicting future earthquakes on the Japan Sea side of Japan and inland, where there is low seismic activity. The Japan Sea earthquake was preceded by foreshocks which occurred 12 days

Fig. 4. Epicentral distributions of foreshocks (A) and aftershock (B) of the Japan Sea earthquake. (A) 1-26 May 1983, (B) 26 May-31 July 1983. Large and small stars represent the main shock and the largest aftershock, respectively (from Hasegawa et al., 1985).

Vol. 43, No. 5, 1995 538 K. Mogi

before the main shock, including one of M5.0. Figure 4 shows epicentral locations of the foreshocks (left) and aftershocks (right) (Hasegawa et al., 1985). Large and small stars in the right figure represent the main shock and the largest aftershock, respectively. Figure 5 shows the epicentral distribution of earthquakes that have occurred in recent years in and around northern Honshu (Mogi, 1985). The data are from JMA. These data reveal that seismic activity had declined in quite a wide area including the focal region for about 5 years before the main shock. This indicates that a seismic gap of the second kind (meaning precursory seismic quiescence: a seismic gap of the first kind refers to an unruptured area in a seismic belt (Mogi, 1979)) had appeared. This pattern emerges clearly when the lower limit of the earthquake magnitude is set to 4.0. The appearance of a seismic gap of the second kind is regarded as the most universal long-term precursory phenomenon ever known; the fact that it was also recognized in the case of the Japan Sea earthquake provides us with further valuable data. At the end of 1978 active earthquake swarms occurred at Iwasaki (western coast of Honshu) and Hakodate (in Hokkaido). The heightened earthquake swarm activity in recent years in northern Honshu, including these two areas, had attracted the attention of many researchers (e.g., Sato et al., 1981). A subcommittee of the Coordinating

Fig. 5. Left: Distribution of shallow earthquakes in and around the focal region

of the 1983 Japan Sea earthquake in the 10 years from 1968 to mid-1978 .

Double circle, M•¬6.0; large solid circles, M•¬5.0; small circles, 5.0>M•¬4 .0; small open circles, 4.0>M•¬3.0. Location of eruption of the Usu volcano is

shown by a hatch circle. Right: Distribution of earthquakes in the 5 years

from mid-1978 until the Japan Sea earthquake. The seismic gap of the second

kind is shown by a broken curve (from Mogi, 1985).

. Phys. Earth

J Earthquake Prediction Research in Japan 539

Committee for Earthquake Prediction discussed whether or not the Iwasaki earthquake swarm was an indication of a future large earthquake, but it could not reach a conclusion before the Japan Sea earthquake. Just after the earthquake, the space-time distribution of noticeable earthquake swarms since 1950 was examined. The data in Fig. 6 are taken from the revised Catalog for the list of earthquake swarms in northern Honshu by Ueki and Takagi (1989). To express the space-time distribution, latitude is plotted on the vertical axis and time on the horizontal axis in the right figure in Fig. 6. There was an anomalous increase in earthquake swarm activity at roughly the same time as the appearance of the seismic gap in the adjacent area on the west. This may be viewed as an indication that the earthquake swarm activity was a long-term precursor of the large earthquake. A leveling survey was carried out on the Oga Peninsula, south of the Iwasaki region (GSI, 1982). The Oga Peninsula has a tendency to uplift, tilting steadily in a northwestward direction, but it is noteworthy that the rate of the uplift in the 4 years before 1981 was twice that in the previous 8 years. The temporal changes in the tide level at the tide station at Fukaura and Oga on the west coast of northern Honshu indicate uplift from 1978 (GSI, 1984). The watertube tiltmeters at the Oga crustal movement observatory (Tohoku University, 1984) showed tilt movement rising to the east from 1978. The reason that this east-rising tilt does not coincide in direction with the northwest-rising tilt of the Oga Peninsula as a whole found through leveling surveys may perhaps be that it was caused by a local block-type movement of the tilt observatory. Collating leveling survey, tidal, and tilt observations shows that the area projecting to the west on the Japan Sea coast close to the seismic gap uplifted prior to the earthquake, and that the commencement of this movement roughly corresponds to the appearance of the seismic gap. In light of this temporal and spatial relationship, it is appropriate to consider these movements as long-term precursory phenomena. Ishii et al. (1987)

Fig. 6. Left: Locations of earthquake swarms during the period from 1940 until 1986 in northern Honshu and the aftershock region of the 1983 Japan Sea earthquake. Right: Space-time distribution of earthquake swarms. Earthquake swarms in which the number of earthquakes of M•¬3 or felt earthquakes is 5 and over are shown in these figures (data from Ueki and Takagi, 1989).

Vol. 43, No. 5, 1995 540 K. Mogi discussed quantitatively large scale crustal movements of this region before and after the earthquake. The data in Fig. 7 show a seismic gap appearing in mid-1978 and the various changes that occurred around the focal region (Mogi, 1985). Frequent earthquake swarms, the eruption of active volcanoes, and uplift and tilting of the ground along the Japan Sea coast occurred around the focal region during the same period. Though various long-term precursory phenomena were observed, the only changes that could be described as short-term precursory phenomena were the foreshocks that occurred for 12 days before the main shock. The location of the epicentral region far off the coast could be a major reason for the fact that other marked precursory phenomena were not observed immediately before the main shock. Takagi (1990a) pointed out that three earthquake swarms occurred along the Japan Trench of the Pacific Ocean within one month before the earthquake. They may be a kind of precursory events. As described above, a seismic gap and other precursory phenomena appeared after mid-1978. On 12 June 1978, the Miyagi-ken-oki earthquake (M7.4) occurred off the Ojika Peninsula on the Pacific Ocean side of northern Honshu. A hypothesis was proposed that the Miygagi-ken-oki earthquake may have triggered many long-term precursory phenomena of the Japan Sea earthquake (Mogi, 1985). Though the Japan Sea earthquake occurred far off the coast quite distant from observation networks on land, various kinds of precursory phenomena were observed. This suggests that if careful and appropriate monitoring is carried out, it will be possible to forecast to a certain degree a large earthquake of this magnitude.

Fig. 7. Summary of possible precursors of the 1983 Japan Sea earthquake (from Mogi, 1985).

J. Phys. Earth Earthquake Prediction Research in Japan 541

4.2 Two major earthquake swarms (1987, 1989) along the Japan Trench Two major earthquake swarms occurred in rectangular areas 2 and 3 shown in Fig. 8(a). Solid circles are the largest earthquakes in these earthquake swarms. The first one (S) is the Sanriku-oki earthquake swarm which began on October 27, 1989. The largest shock of M7.1 occurred on November 2 and the second largest of M6.5 occurred on October 29. This earthquake sequence can also be recognized as a foreshock—main shock—aftershock type. The epicentral distribution of the main shock and its foreshocks and aftershocks is shown in Fig. 9 (Tohoku University, 1990). Before this earthquake, an unusual seismic quiescence in Region 2 was noted by Tohoku University (1989). Based on this result, the Coordinating Committee for Earthquake Prediction also suggested the possibility of occurrence of a major earthquake in the near future. The M-T graph in this region, and those in Regions 1 and 3 obtained by Tohoku University (1991) are shown in Fig. 8(b). The decrease in seismic activity before the 1989 Sanriku-oki earthquake in Region 2 is quite clear. The second one (F) which occurred in Region 3 is the Fukushima-ken-oki earthquake swarm which began on February 2, 1987 and continued for a long time. This

(a)(b) Fig. 8. (a) Locations of Regions 1, 2, and 3 along the Japan trench. Solid circles (S and F) show the epicenters of the largest earthquakes of the 1989 Sanriku- oki and the 1987 Fukushima-ken-oki earthquake swarms, respectively. (b) M-T graphs of earthquakes in Regions 1, 2, and 3. Figures simplified from Tohoku University (1991).

Vol. 43, No. 5, 1995 542 K. Mogi

Fig. 9. Epicentraldistributions of shallowearthquakes between November 1989 and January 1990 in and around northern Honshu. A large hexagon shows the largest earthquake (M7.1) of the 1989 Sanriku-oki earthquake swarm (from Tohoku University, 1990). activity is a typical major earthquake swarm with 5 earthquakes of M>6 (M6.7, 6.6, 6.5, 6.4, and 6.1). As seen in Fig. 8(b), this event was not preceded by any seismic quiescence, and the seismic activity increased gradually before the earthquake swarm. In this region, a major earthquake swarm including M7.5 occurred in 1938. After the occurrence of the 1978 Miyagi-ken-oki earthquake, the Coordinating Committee for Earthquake Prediction marked this region as the potential region of a future major earthquake on the basis of the seismic gap theory and the past example of the migration of seismic activity from north to south. Takagi (1990b) pointed out that the major earthquakes of the 1987 earthquake swarm occurred at the places where microseismic activity was especially high, and those of the 1938 earthquake swarm occurred at the same places. He suggested that this relation between the microseismic activity and major earthquakes may give a clue for long-term forecasting of major earthquakes in this region. The difference of the seismic features in the. Fukushima-ken-oki region from that of the Sanriku-oki region may be attributed to the difference in structural heterogeneity in these two regions.

4.3 1984 Western Nagano Prefecture earthquake (M6.8) On September 14, 1984, an earthquake of M6.8 occurred at the southern foot of the Ontake volcano in the western Nagano Prefecture, central Honshu, and caused the collapse of a steep slope of the volcano and disastrous mudflows. Focal mechanism

J. Phys. Earth Earthquake Prediction Research in Japan 543 analyses (e.g., Takeo and Mikami, 1987), aftershock observations (Aoki, 1987), and trilateration surveys before and after the earthquake (Yamashina and Tada, 1985) show that the main fault is a right lateral strike-slip fault with a ENE-WSW direction and the its focal depth is very shallow. Umeda et al. (1986) found that many boulders were thrown out of their former sockets by the main shock, and high acceleration exceeding 1g was estimated in the epicentral area. One day after the main shock, the largest aftershock (M6.2) and its aftershocks occurred near the western end of the main fault. The main shock and the largest aftershock were caused by mutually conjugate fault movements that occurred as the result of compression in the SE-NW direction. Before the 1984 earthquake, an active shallow earthquake swarm began to occur in and around the epicentral region of the 1984 earthquake in 1976 and continued until the occurrence of the earthquake; and a steam eruption of the Ontake volcano which had no record of eruption in historical time up to that time, occurred in 1979 (Aoki, 1990). Sugisaki and Sugiura (1986) observed conspicuous anomalies in gas composition at a fumarole of the Ontake volcano and at mineral springs prior to the earthquake.

Fig. 10. Temporal variation of gas chemistry at the fumarole on Ontake. Arrows indicate the Western Nagano Prefecture earthquake (M6.8) of September 14, which occurred only 9km to the southeast 1 week after the measurement in 1984 (from Sugisaki and Sugiura, 1986).

Vol. 43, No. 5, 1995 544 K. Mogi

Figure 10 shows the temporal variation of gas chemistry at the fumarole on the Ontake volcano. Arrows indicate the earthquake, which occurred 1 week after the measurement in summer 1984. No other significant short-term precursory phenomena were observed. Around the focal region of the earthquake, there are many active faults including the Atera fault, but no active faults have been found in the focal region (Research Group for Active Faults of Japan, 1991) (Fig. 11) and there are no historical records of strong earthquake in this region (Usami, 1987). After the 1984 earthquake, very accurate aftershock observations, seismic explosion measurements, and extensive gravity surveys were carried out in this region (Aoki, 1990). By these measurements, the detailed crustal structures in the focal region were made clear. Figure 12 shows the epicentral locations of the main shock and aftershocks of the 1984 earthquake and the Bouguer gravity anomaly in and around the focal region (Shichi et al., 1992). The main fault of the 1984 earthquake was located just on the southern side of the steep gradient belt of gravity anomaly dipping towards north and the fault of the largest aftershock was located at the western end of the above-mentioned steep gradient belt. The thickness of the upper low velocity layer changes in the same way (Ikami et al., 1988). Mizoue and Ishiketa (1988).deduced the existence of a molten material at depths of 10-15km in the earthquake swarm region from seismic wave data analysis, and suggested a close relation between the 1984 earthquake and the magmatic activity. Thus, the 1984 Western Nagano Prefecture earthquake which occurred in a structurally very complex volcanic region was preceded by various kinds of precursory phenomena and it was suggested that fundamental research on tectonic structures may give important information for forecasting such inland earthquakes.

4.4 Earthquakes in the Izu Peninsula region No major earthquakes occurred on or in the vicinity of the Izu Peninsula, located approximately 100km southwest of Tokyo, during nearly four decades after a period of activity around 1930. Since 1974, major earthquakes and earthquake swarms have occurred frequently (Mogi, 1990). Figure 13 shows major seismic events in the four successive periods. In 1974, the Izu-Hanto-oki earthquake of M6.9 occurred rather suddenly, and following this event, an anomalous crustal uplift of as much as 15 cm was observed in the eastern Izu Peninsula in early 1975 (A in Fig. 13) (GSI, 1976). After the 1974 Izu-Hanto-oki earthquake, microearthquake swarms were observed also in the eastern part of the Izu Peninsula (Tsumura et al., 1977). In the latter stage of the period (A), the Coordination Committee for Earthquake Prediction discussed frequently whether or not the crustal uplift and the earthquake swarms suggested a future large earthquake. One opinion was that this activity is only an after effect of the 1974 earthquake and that it may decrease with time. Another opinion was that this crustal activity indicates stress build up in this area (the dotted region in A in Fig. 13). In the 1974 Izu-Hanto-oki earthquake, the plate motion in the south-west side of the earthquake fault might have been accelerated via this right-lateral strike-slip fault, and the stress in the dotted region may have increased. Observations were intensified. On January 14, 1978, the Izu-Oshima-kinkai earthquake of M7.0 occurred on the southern boundary of the above-mentioned uplift region, as shown in B in Fig. 13.

J. Phys. Earth Earthquake Prediction Research in Japan 545

Fig. 11. Active faults around the 1984 Western Nagano Prefecture earthquake . Open circle, epicenter of the earthquake; open triangle, Ontake volcano; AF, Atera Fault (from Research Group for Active Faults in Japan, 1991).

Fig. 12. Epicentral distributions of the main shock and aftershocks of the 1984 Western Nagano Prefecture earthquake, and Bouguer anomaly in and around the focal region of the earthquake. Open circle (right), main shock; open circle (left), the largest aftershock; solid circles, aftershocks; open triangle, Ontake volcano; contour interval of the Bouguer anomaly map, 1 mgal (from Shichi et al., 1992).

Vol. 43, No. 5, 1995 546 K. Mogi

Fig. 13. Major crustal activities in the four successive periods A (1974-4975), B (1976-1978, January), C (1978 February-1985), and D (1986-1990) in and around the Izu Peninsula. Solid lines, earthquake faults; shaded areas, earthquake swarm regions; contour curves, crustal uplift; stars, volcanic eruptions.

Remarkable foreshock activity increased several hours before, and then abating temporarily until the main shock occurred (e.g., Tsumura et al., 1978). Several other short-term precursory phenomena were observed (JMA, 1978; Yamaguchi and Odaka, 1978; Wakita et al., 1980, 1988). Some of these precursory changes are shown in Fig. 14. In this case, many of the data were not telemetered to Tokyo, and the monitoring system did not provide complete information on the whole situation, so it was not possible to predict this earthquake. From the end of 1978, marked earthquake swarms began to occur off the east coast of the Izu Peninsula near Ito. On June 29, 1980, a large earthquake of M6.7 occurred when a northward migration was anticipated after the 1978 earthquake, as shown in C in Fig. 13. This earthquake which occurred by left-lateral strike-slip faulting in the N-S direction, was preceded by marked swarm activity. In this stage, models showing that these earthquake swarms may be caused by magma intrusion were presented by several investigators (e.g., Hagiwara, 1977; Mogi, 1982; Ishida, 1984; Sasai and Ishikawa, 1985; Shimazaki, 1987; Tada and Hashimoto, 1988). During the period shown by D in Fig. 13, which started from the 1986 eruption of the Izu-Oshima volcano, marked earthquake swarms occurred successively off the east coast of the Izu Peninsula, migrating from SE to NW after the 1986 eruption of

J. Phys. Earth Earthquake Prediction Research in Japan 547

Fig. 14. Short-term precursory phenomena prior to the 1978 Izu-Oshima-kinkai earthquake (M7.0) (from Mogi, 1982; Wakita et al., 1988).

Vol. 43, No. 5, 1995 548 K. Mogi the Izu-Oshima volcano. On 13 July, 1989, a submarine volcanic eruption occurred just after the peak of an earthquake swarm, at the center of the focal region of the swarm, near Ito (e.g., JMA, 1990). Figure 15 shows the temporal variation of the hourly number of earthquakes observed at the Kamata seismological station of JMA (Mogi, 1990). The ordinate is the logarithm of the number of earthquakes and the abscissa is time. It is noted that a number of earthquake swarms occurred in succession. In this figure, the rate of decreasing earthquake frequency of each earthquake swarm is different between the early stage and the later stage. This suggests a change in mechanical state in the earthquake swarm region. Figure 16 shows the crustal uplift in successive periods determined by leveling surveys (e.g., GSI, 1990) and locations of earthquake swarms obtained from data of

Fig. 15. Temporal variation of hourly number of earthquakes observed at Kamata seismological station of JMA. 1-18, earthquake swarms; A Izu- Oshima-kinkai earthquake; B, Izu-Oshima volcano eruption; No. 18, submarine volcanic eruption off Ito (from Mogi, 1990).

J. Phys. Earth Earthquake Prediction Research in Japan 549

Fig. 16. Crustaluplifts in the Izu Peninsula(contour lines) and the focal regions of earthquakeswarms off the eastcoast of the Izu Peninsula(dotted areas) in successiveperiods (data from GSI and EarthquakeResearch Institute; Mogi, 1990). the Earthquake Research Institute. There is a close relation between the marked uplift and strong earthquake swarms (Mogi, 1988;Ishii, 1989). When the east coast region of the Izu Peninsula was uplifted, marked earthquake swarms off Ito occurred. This result was explained by the above-mentioned magma intrusion model. From 1986 to 1989, both the rate of crustal uplift and earthquake swarm activity increased, and in 1989 a submarine volcanic eruption occurred off Ito. Figure 17 shows the horizontal and vertical distributions of earthquakes in the three earthquake swarms in 1988 and 1989 (JMA, 1989, 1990). In this figure, the northwestward migration of swarm activity and the decreasing focal depths are notable. A submarine volcanic eruption occurred on July 13 nearly at the center of the focal region of the third of the three above-mentioned swarms, and formed a new crater (Oshima et al., 1991). When the earthquake swarm began in July, just before the eruption, the Coordinating Committee for Earthquake Prediction remarked that the focal depths of earthquakes in the swarm were extremely shallow. During the active earthquake swarm preceding the volcanic eruption, very

Vol. 43, No. 5, 1995 550 K. Mogi

Fig. 17. Horizontal and vertical distributions of earthquakes in the three (A, B, C) earthquake swarms off the east coast of the Izu Peninsula in 1988 and 1989. Figures compiled from JMA (1989, 1990). marked ground deformations—the domed uplift by leveling survey (GSI, 1990), the increase in the horizontal distance by geodimeter measurement (Tsuneishi, 1991) and GPS observation (Fujinawa et al., 1991), the ground tilt by a borehole tiltmeter (Yamamoto et al., 1991)—were observed in the focal region of the earthquake swarm. Figure 18 shows the tilt curves observed by a borehole tiltmeter at Ito. Okada and Yamamoto (1991) and Tada and Hashimoto (1991) explained these ground deformations quantitatively by a magma intrusion model. Since the 1989 event, the crustal activity in this region has been nearly quiescent.

J. Phys. Earth Earthquake Prediction Research in Japan 551

Fig. 18. Tilt curvesobserved by a borehole tiltmeter at Ito and temporal variation of number of earthquakes in the earthquake swarm of 1989 July. Figure simplifiedYamamoto et al. (1991).

Collating the northward migration tendency of large earthquakes since 1974, the subsequent seismic gap, the state of the submarine topography, active faults, and the periodical occurrence of strong earthquakes (Ishibashi, 1985), it can be said that future earthquakes may occur in the western part of Sagami Bay, so care is required. However, we must consider that the great Kanto earthquake in 1923 (M7.9) occurred along the Sagami trough and released crustal strain in this region. Consequently, it is difficult to judge the strain accumulation in the western part of Sagami Bay. However, in view of the situation described above and the fact that several tens of years have passed since the great Kanto earthquake, it is necessary to maintain an intensive observation system and to continue monitoring.

4.5 Earthquake prediction problem in the Tokai region The possibility of a large shallow earthquake occurring in the Tokai region of central Honshu, Japan was first suggested in 1969 (Mogi, 1970), and the Coordinating Committee for Earthquake Prediction designated this area as a region for special observation. In 1973 the area was specified as a region for intensified observation, and observations of crustal movements, seismic activity etc. were increased. In 1976 Ishibashi (1976, 1981) presented a hypothetical fault model of the expected Tokai earthquake and emphasized that its focal region would occur behind Suruga Bay, possibly inflicting great damage on the coastal region. This report had an enormous social impact on the district centered on Shizuoka, and there were increased calls for the establishment of a system for predicting this earthquake. Observations in this area were reinforced, and currently 133 data sets are telemetered to JMA in Tokyo, where they are monitored 24 h a day. Figure 19 is the distribution of observation stations located from the Tokai region to the southern Kanto district, and Fig. 20 shows the locations of recent earthquakes observed by this dense network. If any anomalies are discovered, the Earthquake Assessment Committee will be convened. If the Committee decides that

Vol. 43, No. 5, 1995 552 K. Mogi

Fig. 19. Distribution of observation stations for full-time monitoring of precursory phenomena of earthquakes in the Tokai region. Data are telemetered to the center (JMA) at Tokyo.

Fig. 20. Locations of recent earthquakes (1989-1991) observed by the seismic observation system in and around the Tokai region.

J. Phys. Earth Earthquake Prediction Research in Japan 553 the anomalies are the immediate precursors of a great earthquake, the Director General of JMA is to report these to the Prime Minister, who will immediately announce them to the public. There are two reasons why a major earthquake is expected in the Tokai region in the near future, as follows. (1) Large low-angle thrust type shallow earthquakes have occurred repeatedly at quite regular intervals of 100 to 150 years along the Nankai trough—Suruga trough. These earthquakes are caused by the subduction of the Philippine Sea plate beneath the Eurasian plate. Figure 21 shows the focal regions of large earthquakes in this area since 1700. In both the 1707 and 1854 earthquakes the rupture zone extended over the whole region from the Nankai trough to the Suruga trough (Ando, 1975; Ishibashi, 1976; Utsu, 1977). In the 20th century the 1944 Tonankai earthquake (M7.9) and the 1946 Nankaido earthquake (M8.0) occurred along the Nankai trough, but the zone along the Suruga trough, which is the eastward extension of the former, remains unruptured. The Tokai region along the Suruga trough is therefore a seismic gap of the first kind, and has a high potential as the site of the next large earthquake. Since there are no records of a large earthquake having occurred independently in the Suruga trough, however, there are problems in making this simple inference. (2) The results of leveling surveys, tide-gauge observations and triangulation surveys (trilateration surveys nowadays) by GSI over several decades have revealed a marked subsidence of the west coast of Suruga Bay and a reduction of the horizontal distance across the Suruga trough as much as one meter (e.g., GSI, 1977, 1991a). The marked subsidence along the trough, where the Philippine Sea plate is sub-

Fig. 21. Focal regions of great shallow earthquakes along the Nankai-Suruga and Sagami troughs in successive periods. The Tokai region along the Suruga trough is a seismic gap of the first kind.

Vol. 43, No. 5, 1995 554 K. Mogi

ducting, and the simultaneous reduction in the horizontal distance across the trough indicate that a considerable amount of elastic strain energy is accumulating in this area. The leveling route between Omaezaki and Kakegawa (see Fig. 22) is approximately

perpendicular to the trough axis, and leveling surveys have been repeated here at frequent intervals in recent years. By repetition of surveys, it was found that the leveling data contained some seasonal variation, as seen in curve A in Fig. 22. Curve B in which a seasonal variation is eliminated is nearly linear. These have revealed that subsidence at Omaezaki, which is closer to the trough, is proceeding nearly constantly (GSI, 1991a), and that strain energy is steadily building up in this area. The main issue is when this earthquake will occur. Attempts have been made to forecast the occurrence of the anticipated earthquake on the basis of the average recurrence interval of large earthquakes and the amount of strain accumulated (e.g., Rikitake, 1977), but these attempts have only provided a rough guide-line. The most effective method of forecasting the time of occurrence of earthquakes is by observing precursory phenomena. The main purpose of the various observations in the Tokai region is to detect any long-term or short-term precursory phenomena that are thought to be precursors of the anticipated Tokai earthquake. Here, some recent observations of crustal activity in the Tokai region are presented. Figure 23 shows maps of vertical displacements of the Tokai region in successive

periods, which Ishii (1990) obtained from leveling data by GSI. He noted that the pattern have changed significantly with time. Such temporal changes in pattern were noted also by Mizuno (1985). The tectonic significance of such changes should be discussed further. Before a major earthquake occurs, the focal region of the coming earthquake tends

to become seismically inactive and, simultaneously, the wide-ranging area surrounding this region tends to become active. Recent seismic activity in the Tokai region was examined from this viewpoint (Mogi, 1992). The distributions of shallow earthquakes of M4.2 and over in successive periods show that there was some activity before 1972 along the Suruga trough and the Nankai trough, but no earthquakes of M•¬4.2 have occurred recently. This quiescence is continuing. On the other hand, many large earthquakes have occurred in the surrounding area in recent years. Figure 24 shows the epicentral distributions of earthquakes of M•¬5.5 during the two successive periods

(1950-1972) and (1973-1991). It should be noted that a number of recent large earthquakes have occurred in the area surrounding the focal region of the anticipated Tokai earthquake. Therefore, it is conceivable that the changes described above are long-term precursors of a large earthquake. However, there are occasions in which this pattern is not directly related to the occurrence of a large earthquake. Future developments should be watched closely. At this stage, no other changes that could be regarded as long-term precursory phenomena have been observed. As mentioned above, observation stations are densely located in the Tokai region. The possibility of short-term earthquake prediction (on the order of several days) depends on whether precursory phenomena are observed immediately before the earthquake. Since precursory phenomena normally have a marked regional character, the type of phenomena that preceded the Tokai earthquake of 1854 should be expected before the occurrence of the next one. Unfortunately, there are no reliable data available

J. Phys. Earth Earthquake Prediction Research in Japan 555

Fig. 22. Vertical movement of Omaezaki in relation to Kakegawa. A, original curve; B, corrected curve in which a seasonal effect was eliminated (from GSI, 1991a).

Fig. 23. Vertical displacements of the Tokai region in successive periods obtained from leveling data by GSI (from Ishii, 1990).

Vol. 43, No. 5, 1995 556 K. Mogi

Fig. 24. Epicentral distributions of earthquakes (M•¬5.5) during the two successive periods (1950-1972) and (1973-1991). Recent large earthquakes have occurred in the area surrounding the focal region of the anticipated Tokai earthquake (from Mogi, 1992).

Fig. 25. Temporal variations of crustal deformation before the 1944 Tonankai (A) and the 1946 Nankaido (B) earthquakes along the Nankai trough (from Sato, 1977; Mogi, 1984/1985). Locations of Kakegawa and Tosashimizu are shown in Figs. 22 and 26, respectively.

J. Phys. Earth Earthquake Prediction Research in Japan 557

Fig. 26. Anomalouschanges of groundwater beforethe 1946Nankaido earthquake of M8.0.These changes began to occur one day—oneweek beforethe earthquake.Figure modified from Komukai(1948). from 1854.However, the 1944Tonankai earthquake and the 1946Nankaido earthquake occurred adjacent to the Tokai region. They could be described as tectonically related to the future Tokai earthquake and thus should provide valuable information. As shown in Fig. 25, in both of these earthquakes the trough side began to uplift a day or two before they occurred (Sato, 1970, 1977; Mogi, 1984/1985). Before the Nankaido earthquake, marked anomalies were also observed in the wells and hot springs along the coast (Komukai, 1948) as shown in Fig. 26. Consequently, if similar phenomena occur before the Tokai earthquake, the changes should be sufficientlymarked to be recorded by the present network of highly sensitive observation stations in that region.

5. Conclusion An outline of the earthquake prediction program which started in Japan in 1965 has been presented, and research on recent major earthquakes from the viewpoint of earthquake prediction has been reviewed. Examining and reconsidering what kind of response we made prior to these earthquakes and what kind of observational data were obtained after the earthquakes are necessary processes that must be gone through when considering the possibility of earthquake prediction in the future and when moving earthquake prediction forward toward practical application. In Japan, there has not been a single example of successful earthquake prediction in practical terms. This demonstrates directly the current state of earthquake prediction. In the Tokai region, however,since the possibilityof prediction exists, active efforts must be made to minimize the expected disaster by forecasting the earthquake. For earthquake prediction, the promotion of modern and multidisciplinary observations are essential, and various kinds of fundamental studies, such as the overall

Vol. 43, No. 5, 1995 558 K. Mogi understanding of process of earthquake occurrence and tectonics of seismogenic zones, are also important. These subjects are discussed in other articles in this volume.

REFERENCES

Ando, M., Source mechanisms and tectonic significance of historical earthquakes along the Nankai trough, Japan, Tectonophysics, 27, 119-140, 1975. Aoki, H., The Western Nagano Prefecture earthquake, Proc. Earthq. Predict. Res. Symp. (1987), 109-114, 1987 (in Japanese). Aoki, H., The 1984 Western Nagano Prefecture earthquake-an observational research of the seismogenic field-, in Second Decade Coordinating Committee Earthquake Prediction, ed. CCEP, pp. 164-174, Tokyo, 1990 (in Japanese). Fujinawa, Y., S. Shimada, S. Ohmi, S. Sekiguchi, T. Eguchi, and Y. Okada, Fixed-point GPS observation of crustal movement associated with the 1989 seismic swarm and submarine volcanic activities offIto, central Japan, J. Geophys. Res., 39, 141-153, 1991. Geographical Survey Institute, Crustal deformation in the central Izu Peninsula, Rep. Coord. Comm. Earthq. Predict., 16, 82-87, 1976 (in Japanese). Geographical Survey Institute, Crustal deformation in the Tokai district, Rep. Coord. Comm. Earthq. Predict., 17, 113-115, 1977 (in Japanese). Geographical Survey Institute, Vertical movements in the Tokai district, Rep. Coord. Comm. Earthq. Predict., 28, 53-58, 1982 (in Japanese). Geographical Survey Institute, Vertical movements in the Tokai district, Rep. Coord. Comm. Earthq. Predict., 31, 60-68, 1984 (in Japanese). Geographical Survey Institute, Crustal movements in the Izu Peninsula, Rep. Coord. Comm. Earthq. Predict., 43, 227-252, 1990 (in Japanese). Geographical Survey Institute, Crustal movements in the Tokai district, Rep. Coord. Comm. Earthq. Predict., 46, 272-289, 1991a (in Japanese). Geographical Survey Institute, Horizontal strains in the Shikoku district, Rep. Coord. Comm. Earthq. Predict., 46, 394-401, 1991b (in Japanese). Hagiwara, Y., The Mogi model as a possible cause of the crustal uplift in the eastern part of Izu Peninsula and related gravity change. Bull. Earthq. Res. Inst., Univ. Tokyo, 52, 301-309, 1977 (in Japanese). Hasegawa, A., N. Umino, and A. Takagi, Seismicity in the northeastern Japan arc and seismicity patterns before large earthquakes, Earthq. Predict. Res., 3, 607-626, 1985. Ikami, A., T. Yoshii, R. S. Matsu'ura, S. Kubota, Y. Sasaki, A. Hasemi, T. Moriya, H. Miyamachi, and K. Wada, A seismic-refraction profile in and around Nagano Prefecture, central Japan, J. Phys. Earth, 34, 457-474, 1986. Imamura, A., On the seismic activity of central Japan, Jpn. J. Astron. Geophys., 6, 119-137, 1928. Ishibashi, K., Re-examination of a great earthquake expected in the Tokai district: possibility of the Suruga Bay earthquake, Abstracts, Seismal. Soc. Japan, 1976, No. 2, 30-34, 1976 (in Japanese). Ishibashi, K., Specification of a soon-to-occur seismic faulting in the Tokai district, central Japan, based upon seismotectonics, in Earthquake Prediction, ed. D. W. Simpson and P. G. Richards, pp. 297-332, AGU, Washington, D.C., 1981. Ishibashi, K., Possibility of a large earthquake near Odawara, central Japan, preceding the Tokai earthquake, Earthq. Predict. Res., 3, 319-344, 1985.

J. Phys. Earth Earthquake Prediction Research in Japan 559

Ishida, M., Spatial temporal variation of seismicity and spectrum of the 1980 earthquake swarm near the Izu Peninsula, Japan, Bull. Seismal. Soc. Am., 74, 199-221, 1984. Ishii, H., Recent abnormal uplift on the Izu Peninsula (1980-1988), Bull. Earthq. Res. Inst., Univ. Tokyo, 64, 313-324, 1989 (in Japanese). Ishii, H., Vertical movements in the Tokai district (1962-1989), Tohoku Geophys. J., 33, 107-113, 1990. Ishii, H., S. Miura, and A. Takagi, Large-scale crustal movements before and after the 1983 Japan Sea earthquake, J. Phys. Earth, 34, Suppl., S159-S174, 1987. Japan Meteorological Agency, Continuous observations of crustal deformation by the borehole strainmeters, Rep. Coord. Comm. Earthq. Predict., 20, 34-35, 1978 (in Japanese). Japan Meteorological Agency, Seismic activity in and around the Izu Peninsula (May-October, 1988), Rep. Coord. Comm. Earthq. Predict., 41, 227-242, 1989 (in Japanese). Japan Meteorological Agency, Seismic activity in and around the Izu Peninsula (May-October, 1989), Rep. Coord. Comm. Earthq. Predict., 43, 140-156, 1990 (in Japanese). Komukai, R., Report of the Nankaido earthquake in the 21th year of Showa, Earthquakes and Damages, Special Volume of Hydrographic Bulletin, Hydrographic Department of Japan, 12-52, 1948 (in Japanese). Miyakoshi, J., Anomalous time variation of the self-potential in the fractured zone of an active fault preceding the earthquake occurrence, J. Geomag. Geoelc., 38, 1015-1030, 1986. Mizoue, M. and Y. Ishiketa, Unusual crustal discontinuity underlain by molton material detected by microearthquake observations in the southeastern foot of the Ontake volcano, the eastern part of Nagano Prefecture, Earth Monthly, 10, 700-704, 1988 (in Japanese). Mizuno, H., Leveling results in the Tokai region and the estimation of the accuracy of leveling based on them, Earthq. Predict. Res., 3, 425-440, 1985. Mogi, K., Study of elastic shocks caused by the fracture of heterogeneous materials and its relation to earthquake phenomena, Bull. Earthq. Res. Inst., Univ. Tokyo, 40, 125-173, 1962. Mogi, K., Recent horizontal deformation of the earth's crust and tectonic activity in Japan (1), Bull. Earthq. Res. Inst., Univ. Tokyo, 48, 413-430, 1970. Mogi, K., Two kinds of seismic gaps, Pageoph, 117, 1172-1186, 1979. Mogi, K., Earthquake Prediction in Japan, Saiensusha, Tokyo, 351 pp., 1982 (in Japanese). Mogi, K., Temporal variation of crustal deformation during the days preceding a thrust-type great earthquake-the 1944 Tonankai earthquake of magnitude 8.1, Japan, Pageoph, 122, 765-780, 1984/1985. Mogi, K., Precursors of the 1983 Japan Sea earthquake, Earthq. Predict. Res., 3, 493-517, 1985. Mogi, K., Relationship between the crustal anomalous uplift and the earthquake swarm activity in the Izu Peninsula, Earth Monthly, 11, 195-198, 1988 (in Japanese). Mogi, K., Earthquake swarms off the east coast of the Izu Peninsula, in Second Decade Coordinating Committee Earthquake Prediction, ed. CCEP, pp. 203-217, Tokyo, 1990 (in Japanese). Mogi, K., Recent seismic activity in and around the Tokai region, Rep. Coord. Comm. Earthq. Predict., 47, 254-256, 1992 (in Japanese). Okada, Y. and E. Yamamoto, A model for the 1989 seismo-volcanic activity off Ito, central Japan, derived from crustal movement data, J. Phys. Earth, 39, 177-195, 1991. Omori, F., Outline of the large Messina earthquake, Toyo Gakugei Zassi, 26, 377-389, 435-445, 491-496, 1909 (in Japanese). Oshima, S., M. Tsuchide, S. Kato, S. Okubo, K. Watanabe, K. Kudo, and J. Ossaka, Birth of a submarine volcano "Teisi Knoll," J. Phys. Earth, 39, 1-19, 1991.

Vol. 43, No. 5, 1995 560 K. Mogi

Research Group for Active Faults of Japan, Active Faults in Japan-Sheet Maps and Inventories, Univ. Tokyo Press, Tokyo, 437 pp., 1991. Rikitake, T., Probability of a great earthquake to recur off the Pacific coast of central Japan, Tectonophysics, 42, T43-T51, 1977. Sasai, Y. and Y. Ishikawa, Changes in the geomagnetic total force intensity associated with the anomalous crustal activity in the eastern part of the Izu Peninsula (5)-1982-1983 Off-Kawanazaki swarm earthquakes and crustal uplift-, Bull. Earthq. Res. Inst., Univ. Tokyo, 60, 147-177, 1985 (in Japanese). Sato, H., Crustal movements accompanying the Tonankai earthquake in 1944, J. Geod. Soc. Japan, 15, 177-180, 1970 (in Japanese). Sato, H., Some precursors prior to recent great earthquakes along the Nakai trough, J. Phys. Earth, 25, Suppl., S111-S121, 1977. Sato, T., K. Tanaka, and H. Sato, Some features of recent seismic activity in the northern part of Tohoku district, Sci. Rep. Hirosaki Univ., 27, 33-45, 1981 (in Japanese). Shichi, R., A. Yamamoto, A. Kimura, and H. Aoki, Gravimetric evidences for active faults around Mt. Ontake, central Japan: specifically for the hidden faulting of the 1984 Western Nagano Prefecture Earthquake, J. Phys. Earth, 40, 459-478, 1992. Shimazaki, K., Seismic gap near the western end of the Nankai trough, Abstracts, Seismol. Soc. Japan, 1984, No. 2, 262, 1984 (in Japanese). Shimazaki, K., Reported intermediate-term earthquake precursors in the Izu Peninsula, Japan, USGS Open-File Rept., 87-591, 26-53, 1987. Sugisaki, R. and T. Sugiura, Gas anomalies at three mineral springs and a fumarole before and inland earthquake, central Japan, J. Geophys. Res., 91, 12,296-12,304, 1986. Tada, T. and M. Hashimoto, Origin of the anomalous crustal activity in the northeastern Izu Peninsula (2)-tension crack model-, Abstracts, Seismol. Soc. Japan, 1988, No. 1, 139, 1988 (in Japanese). Tada, T. and M. Hashimoto, Anomalous crustal deformation in the northeastern Izu Peninsula and its tectonic significance-tension crack model-, J. Phys. Earth, 39, 197-218, 1991. Takagi, A., Japan Sea earthquake, in Second Decade Coordinating Committee for Earthquake Prediction, ed. CCEP, pp. 136-163, Tokyo, 1990 a (in Japanese). Takagi, A., Fukushima-ken-oki earthquakes, in Second Decade Coordinating Committee for Earthquake Prediction, ed. CCEP, pp. 175-190, Tokyo, 1990 b (in Japanese). Takeo, M. and M. Mikami, Inversion of strong motion seismograms for the source process of the Naganoken-seibu earthquake of 1984, Tectonophysics, 144, 271-285, 1987. Tohoku University, Continuous observation of crustal movements before and after the Central Japan Sea earthquake, Rep. Coord. Comm. Earthq. Predict., 31, 69-83, 1984 (in Japanese). Tohoku University, Microearthquake activity in and around the Tohoku district (November, 1988-April, 1989), Rep. Coord. Comm. Earthq. Predict., 42, 20-42, 1989 (in Japanese). Tohoku University, Microearthquake activity in and around the Tohoku district (November, 1989-April, 1990), Rep. Coord. Comm. Earthq. Predict., 44, 19-35, 1990 (in Japanese). Tohoku University, Microearthquake activity in and around the Tohoku district (November, 1990-April, 1991), Rep. Coord. Comm. Earthq. Predict., 46, 11-30, 1991 (in Japanese). Tsumura, K., I. Karakama, I. Ogino, and M. Takahashi, Observation of earthquake swarm in the Izu Peninsula (1975-1977), Bull. Earthq. Res. Inst., Univ. Tokyo, 52, 113-140, 1977 (in Japanese). Tsumura, K., I. Karakama, I. Ogino, and M. Takahashi, Seismic activities before and after the

J. Phys. Earth Earthquake Prediction Research in Japan 561

Izu-Oshima-kinkai earthquake of 1978, Bull. Earthq. Res. Inst., Univ. Tokyo, 53, 675-706, 1978 (in Japanese). Tsuneishi, Y., Continuous monitoring of crustal activity in east Izu Peninsula by automatic distance measurement, J. Phys. Earth, 39, 131-139, 1991. Ueki, S. and A. Takagi, Catalog of swarm-type seismic activities in the inland area of the Tohoku district, Earth Monthly, 11, 190-194, 1989 (in Japanese). Umeda, Y., A. Kuroiso, K. Ito, Y. Iio, and T. Saeki, High accelerations in the epicentral area of the Western Nagano Prefecture earthquake, 1984, J. Seismol. Soc. Jpn., 39, 217-228, 1986 (in Japanese). Usami, T., Materials for Comprehensive List of Destructive Earthquakes in Japan, Univ. Tokyo Press, Tokyo, 434 pp., 1987 (in Japanese). Utsu, T., Large earthquakes near Hokkaido and the expectancy of the occurrence of a large earthquake off Nemuro, Rep. Coord. Comm. Earthq. Predict., 7, 7-13, 1972 (in Japanese). Utsu, T., Possibility of a great earthquake in the Tokai district, central Japan, J. Phys. Earth, 25, Suppl., S219-S230, 1977. Wakita, H., Y. Nakamura, and Y. Sano, Short-term and intermediate-term geochemical precursors, Pageoph, 126, 267-278, 1988. Wakita, H., Y. Nakamura, K. Notsu, M. Noguchi, and T. Asada, Radon anomaly: a possible precursor of the 1978 Izu-Oshima-kinkai earthquake, Science, 207, 882-883, 1980. Yamaguchi, R. and T. Odaka, Precursory changes in water level at Funabara and Kakigi before the Izu-Oshima kinkai earthquake of 1978, Bull. Earthq. Res. Inst., Univ. Tokyo, 53, 841-854, 1978 (in Japanese). Yamamoto, E., Y. Okada, and T. Ohkubo, Ground tilt changes preceding the 1989 submarine eruption off Ito, Izu Peninsula, J. Phys. Earth, 39, 165-176, 1991. Yamashina, K. and T. Tada, A fault model of the 1984 Western Nagano Prefecture earthquake based on the distance change of trilateration points, Bull. Earthq. Res. Inst., Univ. Tokyo, 60, 221-230, 1985.(in Japanese).

Vol. 43, No. 5, 1995