Deep-sea turbidite has potential for paleoseismicity analysis. Many deep-sea turbidites were intercalated in two sediment cores collected from two slope basins off Tokai area along the eastern Nankai Trough. Geological and topographic setting of the basins suggests that these turbidites were of earthquake origin. Depositional age of each turbidite layer was determined by radiocarbon dating using planktonic foraminifera in hemipelagic mud. The results indicated that large earthquakes along the eastern Nankai Trough might have occur periodically every 100-150 years during the last 3000 years.
Many great earthquakes occurring along the Nankai Trough, where the Philippine Sea Plate subducts under the Eurasian Plate, have been recorded in historical documents such as ancient Japanese government's and private diaries in Kyoto, Edo (Tokyo) or elsewhere when severe damage was experienced. The Pacific coast of western Japan along the Nankai Trough has been deformed significantly by great earthquakes, which are recorded in geomorphological or geological evidence such as emerged marine terraces and tsunami deposits. There are many fossil calcareous assemblages of raised sessile organisms along the coast of Cape Muroto up to a height of about 9 m. Calcareous assemblages had been grown up to 1 m thick in some places. Sessile organisms in this area consist of Pomatoleios kraussii (phylum ANNELIDA), which lives in the midst of the intertidal zone. The growth process of the assemblages of sessile organisms could suggest the history of relative sea-level change ; that is tectonic uplift since the late Holocene around Cape Muroto. The author collected some cores from these assemblages using a hand-coring system with a single core tube and a diamond bit, and dated them using the AMS14C method. Dating results suggest that relative sea-level was fairly stable, at least during 4500-2800 yrsBP, 2700-1000 yrsBP, and after 900 yrsBP. There should be coseismic uplifts of more than 2 m between these stable phases. Historical records in Japan, however, suggest that inter plate earthquakes along the Nankai Trough have recurred at intervals of 100 to 200 years. The author concluded that there were two types of earthquake that have made this area uplifted. One type is the inter-plate earthquake, which recurred at intervals of 100 to 200 years, and which had scarcely contributed to cumulative uplift due to inter-seismic subsidence. The other type is the intra-plate earthquake, which originated in shallow sea active faults near Shikoku island, and resulted in cumulative uplifts of the cape. The recurrence intervals of intra-plate earthquakes were longer than those of the inter plate type.
The slip distributions of the 1944 Tonankai earthquake (Tanioka and Satake, 2001b) and the 1946 Nankai earthquake (Tanioka and Satake, 2001a) estimated using tsunami waveforms were reviewed. In the source region off Kii peninsula, the large slip extended over the entire locked zone. However, in the source region off Shikoku, Shima peninsula or Atsumi peninsula, the slip on the up-dip or shallow part was limited. Parts of these may be explained by the existence of splay fault (up-dip subfault) systems in the accretionally prism. The tsunami waves from those earthquakes were observed not only at tide gauges in Japan, but also at Honolulu, Hawaii. The tsunami at Honolulu from the 1946 Nankai earthquake was smaller than that from the 1944 Tonankai earthquake, although the seismic moment of the 1946 Nankai earthquake is much larger than the 1944 Tonankai earthquake. The numerical tsunami computation was carried out for the 1944 and 1946 events. The largest slip region near Shikoku for the 1946 Nankai earthquake was close to the shore, so the large parts of tsunami energy were trapped in the shallow sea and were not transmitted into the deep ocean. That is one of the reasons why the tsunami observed at Honolulu from the 1944 event was larger than that of the 1946 event.
By carefully analyzing the source process of the 1946 Nankai earthquake and its correlation with plate boundary structure, we attempt to explain the occurrence pattern of historical earthquakes in the Nankai Trough, in which great earthquakes tend to rupture separately either the western or eastern portions of the Nankai Trough. The source process of the 1946 earthquake consists of two major subevents, each corresponding to segments A and B, defined by Ando (1975), which have long been thought to correspond to units of earthquake rupture in the western Nankai Trough. Furthermore, rupture in each subevent begins near the eastern edge of the respective segment, where there are pronounced contortions of the plate boundary : a subducting seamount chain off Cape Muroto and a rapid change in subduction angle beneath the Kii Peninsula. We suggest that these seismotectonic features of the plate boundary shape control to some extent the pattern of great earthquake occurrence in the Nankai Trough.
We performed an amplitude variation with offset (AVO) analysis using 2D multichannel seismic reflection data across the Nankai Trough to determine the physical properties along the plate boundary decollement. With angle gather data obtained after a series of preprocessing, including prestack time migration, we calculated two major AVO attributes for the decollement the range from the deformation front to 45 km landward : zero-offset P-wave reflection amplitude (P) and gradient of amplitude versus sin2 θ (G). Attribute P corresponding to an approximation of the reflection coefficient for normal incidence, depends on Pwave acoustic impedance contrast above and below the decollement. Attribute G mostly depends on the Poisson's ratio contrast above and below the decollement. Combining variations of these attributes (P and G) along the decollement and structural features of the accretionary wedge, the Nankai Trough decollement zone can be divided into three different physical property zones landward from the deformation front : large physical property change (LPC) zone, small physical property change zone (SPC), and moderate physical property change (MPC) zone. For these three zones, both absolute values of P and G decrease by stages landward along the decollement, suggesting landward decreases of both contrasts of P-wave acoustic impedance and Poisson's ratio above and below the decollement. A significant, landward decrease of these AVO attributes may be related to interplate coupling process during the interseismic period.
We have developed an observation system for ocean bottom crustal deformation using an acoustic ranging-GPS link. A seafloor-positioning test at Suruga Bay revealed that our system can obtain the horizontal location of a sea-bottom station within 5 cm (95 % confidence interval) using accurate GPS positioning data. GPS positioning error is a major source of error in our whole system. We carried out experiments on kinematic GPS positioning with a long base line and a moving antenna to investigate its accuracy. The results of the experiments suggest that data with PDOP (Position Dilution of Precision) of 2 or less can reduce the GPS positioning error to 7 cm. The network of the observation system for ocean bottom crustal deformation may lead to advances in tectonic studies along subduction zones and contribute to predictions of interplate large earthquakes in the future.
With a view toward understanding the mechanism of interplate earthquakes at subduction plate boundaries around Japan, remarkable progress was made in the long-term observation system of seafloor crustal deformation. The method for the seafloor geodesy has been exploited in order to overcome several difficulties. The position of the sea-bottom transponders on the Kumano Trough was repeatedly observed using a GPS/Acoustic measurement system in May and August 2000. Early results from processed data show about 2 cm differences between May and August with a scatter of 4 to 10 cm in horizontal positioning. This paper describes techniques developed to realize precise acoustic ranging, automatically detecting and correcting the Doppler shift of acoustic propagation signal due to ship's drifting and attitude motion, and theoretically forming a pulse of one wavelength in cross correlation between transmission and reception of long-coded signals. Furthermore, by using GPS surface positioning, acoustic ranging and sound speed profiling data, the temporal changes of sound speed profile were estimated with sufficient accuracy for acoustic positioning.
Crustal stress and strain measurements in a deep borehole are essential for studying plate subduction earthquakes. Especially in the case of the Nankai Trough earthquake, it is expected that crustal movements are detectable even on land, because the focal region caused by the subduction of the Philippine Sea plate is close to land. We have tried to search for a suitable field and a method for observations and investigations based on some hypotheses. Special emphasis is given to better understanding the hinge line as an adequate related field. The hinge line is located on the surface just above the boundary between creep and seismogeniclocked zones on the upper surface of the descending slab. If soft materials such as serpentine exist on the surface of the slab, deformation and flow here will become initiate an earthquake. The basic data of crustal stress and the rock physical properties were acquired by the scientific drilling of a borehole. This borehole was drilled to a depth of 550 m at Shingu city, Wakayama prefecture, on the Kii Peninsula. In-situ experiments such as crustal stress measurements by the hydraulic fracturing method were conducted in the well. Furthermore, we have developed an intelligent strain meter for measuring in-situ rock stress deeper than 500 m using the overcoring method. The final goal of this study is to monitor temporal changes of in-situ stress and strain, and to reveal their storage processes in relation to a great earthquake.
Recently, Tada (1996) and Matsumura (2000) reported that crustal deformation rate and seismicity pattern have interestingly changed in the Tokai area, central Japan. We formulate recurring earthquake model in the Tokai area based on the rate- and state dependent frictional law derived from laboratory experiments of rock friction and on the location of a seismically locked region. Using the model parameters based on previous studies, we estimate the time when the next Tokai Earthquake will occur and the location of plate interface by fitting simulated calculation to observed data. Some of the results explain the recent decrease in the rate of uplift change especially at the three observational points. We find out that the mechanism of microearthquakes is well explained by the rate change of stress field. The present model shows the vertical rotation of the axes of principal stresses and predicts the precursory change of seismic activity in the region, where the distance from the assumed trench along the relative plate motion direction is about 150 km, in the depth range of 10-20 km. The present seismicity with P axis almost horizontal will become more inactive several years before the occurrence of a great earthquake, and/or the direction of P axis rotates abruptly about 80 degrees so as to be orthogonal to plate interface.
We investigated deviative components from average crustal deformation in southwestern Japan with the annual component and the average velocity removed from the raw-time series data of the GSI's GPS network (GEONET), during the period between April 1996 and October 1999. The result shows anomalous crustal deformation at GPS sites around the Bungo channel after the 1996 Hyuganada earthquakes with the following three characteristic periods : (1) increased displacement rate associated with the Mw=6.8 Hyuga-nada earthquake of October 19, 1996 and with the Mw=6.7 Hyuga-nada earthquake of December 3, 1996 ; (2) an increa-singly rapid anomalous period of deformation from June 1997 to around the end of 1997 ; and, (3) a slowdown period and return to normal rates of deformation after the end of 1997. During these periods, no earthquakes larger than Mw=5 occurred in the Bungo channel area. Linearized least squares, using a rectangular fault, locate the slip area of this slow event beneath the Bungo channel, Kyushu, southwestern Japan; an area dipping west with the lower edge at a depth of 50 km, very close to the plate boundary between the Philippine sea and the overriding continental plates in this region. In addition to its proximity to the plate boundary, this estimated slip of the hanging side is opposite to the motion of the Philippine Sea plate. These results support the hypothesis that the observed anomalous crustal deformation is due to a rebound process of the overriding continental plate against the Philippine Sea plate, as is pointed out by Hirose et al. (1999). In order to estimate the rupture propagation of this slow earthquake in detail over time, we apply a time-dependent inversion scheme for the data at GPS sites on the Shikoku and Kyushu islands. The result shows that a reverse slip occurred in an area beneath the Bungo channel between the Shikoku and Kyushu islands with a time span ranging from several months to a year, depending on the location estimated. During the initial period of slowly rising displacement rates, the visible slip area is resolved by our inversions beneath the coastal area of southwestern Shikoku. In the next stage of accelerated displacements, the slip area expanded southwestward to the Bungo channel beginning around June 1997, when a postseismic slip almost ended in the area near the epicentral region of the 1996 Hyuga-nada earthquake. After this accelerated period, the estimated slip area remained almost the same in size and location over time, with the time transient anomalous deformation signals ending.
Seismicity characteristics in and around Shikoku, southwest Japan are investigated using recent data from microearthquake observations. Earthquakes occurring in the regions are distinctly divided into crustal and mantle earthquakes according to about 22 km in focal depth. Crustal and mantle earthquakes have predominantly the strike-slip type with P axes in the east-west and north-south directions, respectively. Although the Philippine Sea plate is subducting under Shikoku in the northwest direction from the Nankai trough, the P axis of mantle earthquakes in eastern Shikoku is inferred to be stable in the N-S direction during the period before the 1946 Nankai earthquake until today. On the other hand, earthquakes with normal fault type occur infrequently in the aseismic portion deeper than the north dipping mantle earthquake zone. A gradual lowering of crustal earthquake activity has continued since 1968 in which our microearthquake observation started in central Shikoku. Taking the abrupt decreasing of events in Shikoku before the 1946 Nankai earthquake by JMA data into consideration, the present activity is expected to become lower extremely just before the next Nankai earthquake.
The shape of the Philippine Sea slab beneath southwest Japan was investigated using the hypocenter distribution and the fault plane solutions obtained by Japan Meteorological Agency. Two seismic belts were found in an equal depth section of the slab and these did not relate to the double seismic zone in the subducting slab. Then, the existence of a broken off piece of slab was proposed beneath Kii channel. This piece collided with Philippine Sea slab and created a high seismicity area at each side. Three events that were not located at a high seismic area in the slab were found. Two were considered to be the events that occurred at the near aseismic part of the slab. The other may have occurred at another broken off piece of the slab.
After a renewal process is estimated for data on repeated occurrences of large-scale earthquakes, we predict the hazard of the next rupture as a function of time using the estimated parameters. However, when the number of events in the data is small, the hazard function based on the maximum likelihood estimates (MLE) has not only a large uncertainty in the estimation but also sometimes has a serious bias. To see the uncertainty of estimated hazard functions, we use a Bayesian inference. In this manuscript I discuss the uncertainty of the estimated hazard functions of the forthcoming Nankaido earthquake based on historical data consisting of ten large events. Moreover, assuming the same distribution of the ratio of the time-interval length between the successive events to the slip size of the first event of the pair, an extended renewal process for the Shimazaki-Nakata's Time-predictable model can be considered. Thus we estimate not only the alternative hazard function of the next Nankaido event but also its uncertainty based on occurrence time data for the last three large events associated with records of coseismic uplift sizes. The same extended renewal model is applied to the same ten large events noted above associated with the estimated slip sizes that are calculated from Usami's magnitude estimates. The three datasets lead to significantly different predictions, even if the estimation uncertainty is taken into consideration.