A viscoelastic bending moment theory is applied to the subducting slab of the Philippine Sea plate. The slab is steeply bent at the northern end of the Kyushu-Ryukyu subduction zone compared with at the southern part of the arc. The plate boundary on the cross section of the slab beneath Aki-Nada, Iyo-Nada and Bungo Channel, sea regions between Kyushu and Shikoku Islands, represents an arc of a circle, as demonstrated by the microearthquake distribution during over 10 years. The elastic bending moment is evaluated based on this geometry, where the deformation due to viscosity is reduced applying a Maxwell viscoelastic theory. The bending is too large in strength to ascribe it only to the force of negative buoyancy of the slab. We here assume the horizontal pressure generated by the spreading mantle flow accompanying upwelling magmas at the volcanic area in northern Kyushu, where extensive ground movements have been detected by geodetic measurements. The exerted uniform pressure to bend the slab is estimated to amount to (σ=) 72 MPa, when we take the following model constants as standards : the thickness of the slab h = 30 km, Young modulus of the slab (E=) 1.8 x 105 MPa, Poisson's ratio (ν=) 0.27, an average elapsed time since the subduction of the slab initiated (t=) 1.9 × 106 y, the time constant of the viscoelastic slab (τ=) 3.0 × 105y, and the excess density of the slab compared with the surrounding athenosphere (Δρ=) 0.06 × 103 kg/m3. The region of the highest seismic activity is located at the vicinity of the upper locked portion of the plate, where destructive earthquakes have frequently occurred. The 2001 Geiyo earthquake of M6.4 was one of them and took place at the eastern end of this active seismic zone. The focal mechanism of the event shows normal faulting as similarly as those of the majority of small earthquakes in the seismic active region. The main cause of this large earthquake is considered to be the bending force of the plate.
The significant attenuation of seismic motion in the west of the 2001 Geiyo earthquake is inferred from strong motion distributions, observed seismograms and their spectra. Since this attenuation is identified even in the distribution of borehole motions, that is assumed to arise in a deeper part like the mantle wedge. If we assume a low-Q zone in the mantle wedge, a strong motion simulation with Qs = 20-30 can reproduce the observations. This zone may be related to dehydration of the Philippine sea plate.
Many observations have been made in order to clarify the relationship between electromagnetic (EM) phenomena and earthquake (EQ) activities. We have been observing EM waves in east, west, south and north directions by a PLL type synthesized FM tuner from 76 MHz to 90 MHz in VHF band. This frequency band is strictly administrated for FM broadcastings in Japan and hardly influenced by artificial or urban noises compared with other lower frequency bands. In observing EM phenomena at FM broadcasting frequency band, an observation method that distinguishes seismic EM waves from FM broadcasting waves is required. Therefore, we have developed a novel dual frequency observation method in which the observation is made at two different frequencies of fn and fr. During the observation, the Tottori-ken Seibu EQ (M=7.3, d=11 km) occurred on October 6, 2000. Co-seismic broadband EM waves associated with the Tottori-ken Seibu EQ were detected by the observation system located at Hiroshima City University (HCU) 130 km from the epicenter. The observed EM waves rose up to the peak level right after the EQ, then slowly returned to a normal level about three hours later. Following the Tottori-ken Seibu EQ, we had the Geiyo EQ (M=6.4, d=51 km) on March 24, 2001. Our observation system detected two kinds of EM phenomena associated with the Geiyo EQ at HCU 45 km from the epicenter. One EM phenomenon detected in VHF band appeared after the earthquake. The observed EM waves followed similar variation to that of the Tottori-ken Seibu EQ, which had a sharp rise followed by a slower decline lasting about three hours. EM levels of both fn and fr frequency channels were similarly fluctuated. It is therefore suggested that the detected EM waves had broadband spectra. The maximum peak level of -98 dBm was observed in the west direction that was not toward the epicenter. The peak levels of EM waves were detected at 15 : 28 : 20 (JST) after the occurrence time of the Geiyo EQ, at 15 : 27 : 54 (JST). The time when the detected EM waves rose up to the peak level corresponded to the time of seismic swing arrival at the HCU observatory. The other EM phenomenon detected in VHF band disappeared at the time of earthquake. Amplitude fluctuations of about 2 dB were observed since one day before the Geiyo EQ, then, the fluctuations disappeared at the occurrence of the earthquake. These fluctuations were mainly detected from the east and south directions, which coincide with the direction toward the epicenter.
Geological Survey of Japan made a groundwater observation network composed of 21 stations mainly along active faults in and around the Kinki district, which is a range of about 34-35.5°N and 134.5-136.7°E, for earthquake prediction research. These wells were newly made by Geological Survey of Japan after the 1995 Hyogo-ken Nanbu Earthquake (the Kobe Earthquake). Continuous observation at all the wells started in 1998. The 2001 Geiyo Earthquake (M6.7), occurred west off the network on Mar. 24, 2001. The depth of the earthquake is 51 km. The epicentral distance to the nearest observation well is about 170 km and that to the farthest about 380 km. There was no clear preseismic changes in groundwater levels and crustral strains observed at our wells, although coseismic and/or postseismic changes were observed at many of the wells. The analysis of the observational results show that these coseismic and/or postseismic changes were well explained by coseismic volumetric strain changes estimated from the fault model. Usually coseismic and/or postseismic groundwater level changes are caused by two factors. One is coseismic strain changes and the other is strong ground motion, which generates cracks in the aquifer and sometimes causes liquefaction. As the earthquake was relatively deep and far enough from the network, ground motion within the network should have been small. Therefore strain-induced coseismic and/or postseismic groundwater level changes are inferred to be larger than those induced by the other factor.
The authors have been trying to simulate the conspicuously concentrated area of damaged buildings, the so-called damage belt, caused by the Hyogo-ken Nanbu earthquake of 1995. To reproduce the damage belt, we need to have a realistic 3-D basin structure model and we have constructed an initial model, with which we succeed to reproduce a high amplitude area with belt-like shape (Matsushima and Kawase, 2000). In this paper, we calibrate this basin model using aftershock recordings that we observed right after the main shock. First we use source parameters derived by Nakagawa (1997) to calculate synthetic motions using the 3-D finite difference method developed by Graves (1996). The results show a fairly good match between the synthetics and the data overall, and from these results it seems that the initial basin structure model is acceptable. As we look closely, there are slight differences in amplitude and timing of the major peaks in the synthetics compared to the data, and as the differences vary from one aftershock to another, it would be better to tune up the source mechanisms to account for these differences. The reason is that to calibrate the basin structure, we need to have a well-calibrated source mechanism. So we estimate source mechanism, seismic moment, and source depth of aftershocks by a grid-search technique using body waveform data for the aftershocks. By using these estimated source parameters, the synthetic motions at rock sites become closer to the data, but there is still a systematic time difference of peaks at basin sites. At basin sites the SP-wave, i.e. P-wave transformed from S-wave at the boundary between the basin and the upper crust, can be observed quite clearly, especially in the vertical component. The time differences between the SP-wave and direct P-wave (SP-P time) of the synthetics are constantly 0.4 second longer than that of the data. This time difference is independent of the source parameters. In order to fill the difference of the SP-P times between the data and synthetics, we modify the velocity model of the upper crust of the initial model. We change the layer of the upper crust with an S-wave velocity of 3.20 km/s to the layer with a velocity of 3.46 km/s in the basin side and to the layer with a velocity of 2.85 km/s in the rock side. The synthetics using the modified model and the newly estimated source parameters show that it gives better results in simulating the observed seismograms.
As a result of 1995 Hyogo-ken Nanbu Earthquake (also called the Kobe Earthquake), a distinctive fault ruptures appeared along the northwestern coast of Awaji Island, which are called the Hokudan Fault System. This system consists of two fault strands, i.e., the Nojima Fault and the Ogura Fault. The southern portion of the Nojima Fault, which is about 1.5 km long from the branch point (NSM), shows a little displacement. On the other hand, the Ogura Fault and northern portion of the Nojima Fault show a distinctive displacement. In order to delineate shallow resistivity structure around the southern portion of the Nojima Fault, 2-D resistivity surveys were made along the three transects across the fault. Our further aim was to reveal relationship between resistivity structure and difference of fault activity. Clear resistivity contrast was detected when crossing the surface trace of the fault. Resistivity value of larger than 500 Ωm was observed for granitic rocks, whereas the value of less than 200 Ωm were measured for Tertiary or Quaternary sediments. Surface positions of all large resistivity boundaries are consistent with surface trace of the fault. Furthermore, dip angles of two resistivity boundaries are in fair agreement with other geological evidence reported from the survey area. A distinctive low resistive zone which has been previously observed along the Ogura Fault and the northern part of the Nojima Fault, is not detected along the southern portion of the Nojima Fault. This variation of the shallow resistivity structure around the different fault zone may indicate a difference in fault activity. The southern part of the Nojima Fault is reported to be less active during the latter half of the Quaternary, on the other hand the Ogura Fault and the northern Nojima Fault have been active until present. Our result thus shows that a conductivity structure around a fault zone can be used as indicator for recent fault activities along faults.
The Oubaku fault is an active fault bounding the south-eastern edge of the Kyoto Basin. Although the structure of the fault was inferred to be reverse by the past several reports, the vertical displacement of it has remained unknown and the inferred strikes have been different author by author especially in and around Gokasho, Uji. In order to know more precise subsurface structure of the Oubaku fault, we carried out a seismic reflection exploration and drilled two boreholes (130 m and 400 m) with several geophysical explorations at Gokasho, Uji. The result of the seismic reflection exploration showed the existence of a reverse fault, as had been inferred before. The vertical displacement of the fault is about 170 m and the orientation of the strike is NNE-SSW. The other several faults are also recognized but the vertical displacements of them are much smaller. The subsurface structures of them mainly show the characters of reverse faults but they are more complicated than expected before.
The JMA seismic intensity scale has been playing an important role in earthquake engineering such as seismic risk analyses and planning of disaster response. Since 1996, the seismic intensity has been measured automatically with seismic intensity meters for its objectivity and rapid announcement to the public and officials. The seismic intensity meters have been densely installed. However, it has been pointed out that the scatter of the instrumental seismic intensity at neighboring sites is not small. Therefore, the spatial stochastic properties of the instrumental seismic intensity should be examined for proper utilization of the seismic intensity scale as the regional intensity of seismic ground motions. In this study, seismic intensity difference (the difference between instrumental seismic intensity observed at the two sites during each eathquake) is introduced and investigated to examine the statistical characteristics of the instrumental seismic intensity. First, the probability distribution of the seismic intensity differences is analyzed, and then the probability density functions, averages, standard deviations and percentiles are derived. Second, above-mentioned statistics are estimated using seismometer arrays of the Chiba and SMART-1 databases. Then, the relationships between these statistics and the station separations are analyzed. We have found that the means and standard deviations monotonously increase with increasing logarithm of the station separation distances ranging from about several meters up to six kilometers. The 50th, 80th and 95th percentiles are calculated to examine the seismic intensity difference. Finally, the probability of occurrence of the same seismic intensity scale at two sites is considered. This consideration reveals that the probability can be given by the analytical expression using the probability density function of the seismic intensity differences.
A dense GPS network was established in 1997 around Ou Backbone Range (OBR), northeastern Japan, by deploying 9 new continuous GPS stations so as to complement the sparse portion of GEONET (GPS Earth Observation Network) operated by Geographical Survey Institute (GSI), Japan. This new observation network is aimed to investigate the present deformation and to understand the relationship between the occurrence of intraplate earthquakes and the deformation process of an island-arc crust. GPS data are analyzed using a precise point positioning strategy of GIPSY/OASIS-II. Coordinates of the GEONET stations in daily SINEX files have been supplied by GSI. A map of the intraplate deformation field derived by combining the two data sets and by subtracting the motion of the North American Plate demonstrates displacements toward WNW along the Pacific coast. This is caused by the interplate coupling between the Pacific and North American Plates. Displacements along the Japan Sea coast, on the other hand, are considerably small in magnitude, and directed to NW and N in the southern and northern area, respectively. OBR lies halfway between the two coasts and shows the transitional pattern of both deformation characteristics. By producing grid data of displacement velocities and taking spatial derivatives, we obtained a map of strain rate distribution. We found that the region between 38.8°N and 39.8°N in OBR demonstrates notable concentration of EW contraction. The region coincides with the area of active seismicity and includes the focal areas of 1896 (M7.2), 1962 (M6.5), 1970 (M6.2), and 1998 (M6.1) events. The observed strain is larger than can be explained by the total moment release of earthquakes that occurred in the same period of this study. Possible sources of the strain concentration may be viscoelastic deformation due to the large earthquakes and/or aseismic slip along the deeper extension of active faults.
We analyze seismic waveform data from three artificial chemical explosions conducted before and after the 1998 northern Iwate Prefecture earthquake (M 6.1), on September 3, 1998. These explosions were carried out at the same location ; about one month before, two months after, and two years after the earthquake. The waveforms of the three explosions are quite similar at three stations deployed on and around the hypocentral area of the M 6.1 earthquake. In order to detect small differences of travel time and waveform of coda, we calculate their coherence and phases by using the cross-spectral analysis. The result of phase spectral analysis shows that the seismic-wave velocity decreased 0.1-0.9% during the period including the earthquake occurrence. On the other hand, seismic velocity change is not found between the waveform data for the two explosions conducted after the earthquake. This observation strongly suggests that the velocity change took place associated with the occurrence of the M 6.1 earthquake.
In the final stages of the Matsushiro earthquake swarm (August 1965-October 1967), a large volume of groundwater was generated that continues to flow as of this writing. We studied the spring water origin by measuring oxygen/hydrogen isotope ratios and concentrations of Na+ and Cl, the main dissolved ions. We took water samples from June 1999 to October 2000. Data plots for δ18O vs. δD are distributed along a well determined linear regression line having an endpoint, i.e., river water, at the ordinary value of rainwater. The regression line can be extrapolated toward estimated δ18O and δD of “andesitic magmatic water” originating from magma in subduction zones. This implies that the Matsushiro groundwater can be regarded simply as a mixture of 2 fluids, i.e., surface water and andesitic magmatic water. We obtained the carbon isotope ratio of CO2, the main component of free gas in spring water. δ13C ranges from -7.1‰ to -3.1‰, suggesting that the source of CO2 is also magmatic. The ratio 3He/4He shows that He in the free gas is from the mantle or magma. These 3 pieces of evidence - (1) δ18O and δD values, (2) δ13C of CO2, and (3) 3He/4He - suggest that the origin of Matsushiro water is magmatic. Considering the presence of an electric conductive layer and seismic reflective layers 15 km beneath the Matsushiro area, we presume that this andesitic magmatic water accumulates as a thin layer at this depth. An impermeable sheet presumably lying just above the water layer was formed by precipitates from magmatic water. We present the following model of the relation between groundwater and earthquakes : When the impermeable sheet broke and high-pressure water with CO2 rose into the upper crust, the crust was weakened, causing the Matsushiro earthquake.