Explosion seismic experiments were carried out in 1980 for the investigation of the crustal structure in the northeastern part of Shikoku. Sixty-three seismograms were obtained at 42 temporary stations aligned on the two N-S trending parallel profiles crossing the Median Tectonic Line which is the most prominent geotectonic line and the greatest active fault in Japan. A P-wave velocity structure of the surface and granitic layers was obtained from the first arrival time data. A structure of the deeper part of the crust was also roughly estimated from the later arrivals. The Median Tectonic Line was estimated to be a north-depressed fault with a vertical offset of up to 2km in the shallow velocity profile. A possibility of the extension of the offset further downward to the deeper crust is also suggested.

View full abstract

In a southeastern part of Kyoto basin, we observed the reflected waves from the basement some hundred meter deep by using air gun as a source and signal-enhancement technique by stacking. Both the velocity up to and the depth of the basement were greater than those previously given by KITSUNEZAKI et al. (1971). It is useful to use reflected waves for exploring sharp boundary, such as a basement, because it is relatively easy to observe them at far distances where surface waves are not disturbing. Synthetic seismograms also support the prospect to detect reflected phases from a sharp boundary.

View full abstract

Preceding the eastern Yamanashi earthquake of magnitude 6.0, an anomalous tilt change up to 0.4 micro-radian was detected by the borehole-type tiltmeter at the Enzan station locating 31km north-west of the epicenter. The anomaly began 18 days preceding the earthquake and continued 10 days. The tiltmeter was installed at the bottom of the well of 89m depth. The resolution is 0.006μ radian and the long-term stability is better than a few μ radian per year. This tiltmeter had recorded similar tilt changes of the order of 0.4μ radian twice since it was installed in 1979. Both the changes had been caused by a large amount of precipitation accompanied by typhoons. During these 10 days, however, the maximum daily precipitation was 20mm, temperature change was smaller than 0.05°C at the bottom of the well, and there was no construction near the station. Therefore, we may interpret this anomalous tilt change as a precursor of the earthquake. The station is located in the fracture zone associated with the active fault, the Daibosatsurei-nishigawa fault; core samples obtained from the well were fractured and their P wave velocity was extremely low. The direction of the anomalous tilt vector change is nearly orthogonal to the strike of this active fault. This suggests that the observed upheaval of the epicentral block is the consequence of the hinge movement at the fault.

View full abstract

The Fukui-earthquake fault which generated the Fukui earthquake in 1948 has been investigated by measuring microtremors with periods of a few seconds to determine its accurate position and vertical offset. The fault cannot be observed in geological surface survey, being masked by thick sedimentary layers. Eight horizontal-component seismographs with a proper period of 2 sec have been set up at intervals of 200m along a profile which may cross the fault.
The amplitudes of microtremors with a period of 2 sec change rapidly around the fissure zone generated during the 1948 Fukui earthquake. The amplitudes at the western side of the fissure zone are twice as large as those at its eastern side. The predominant periods also change around the fissure zone, being 0.6 sec at its eastern side and 1.5-2.0 sec at its western side. Therefore the position of the Fukui-earthquake fault can be determined within an accuracy of 200m. The depth to the basement estimated from the predominant periods of microtremors are about 100m at the eastern side of the fissure zone and about 300m at its western side. From this difference it can be concluded that the Fukui-earthquake fault is an ‘active fault’ which has moved repeatedly.

View full abstract

Three-dimensional anelasticity (Q_s^-1) structure of the upper mantle beneath the northeastern Japan arc is obtained using spectral amplitude ratio of P and S waves observed by the seismic network of Tohoku University. An S-to-P spectral ratio technique is adopted to avoid dependance on the source spectra of the earthquakes used here, assumingQ_p/Q_s to be 2.25. The crust and upper mantle are divided into many rectangular blocks, and an anelasticity parameter of each block is deduced by using the damped least squares method.
The anelasticity structure of the crust and upper mantle beneath the northeastern Japan arc consists of an extremely high-Q (Q_s≅1500) inclined oceanic plate and a relatively low-Q land-side region which can be divided into two zones: an intermediate-Q (Q_s≅500-800) land-side mantle between the Pacific coast and the volcanic front, and a low-Q (Q_s≅150) land-side mantle between the volcanic front and the coast of the Japan Sea. Moreover extremely low-Q zones exist in the crust beneath the active volcanoes.

View full abstract

An attempt is made to restore the original pulse forms from dispersed wave trains by using group velocity data in place of phase velocity data. For a given period, the group velocity is obtained, eventually, by usual method to interpolate group velocities of adjacent peaks of dispersed train, but the validity of the method is interpreted in slightly different way from the usual one.
Calculations are made for fictious dipersed train whose original form is known and for actually recorded data whose original form is unknown. The estimated pulse is of a form not completely free from uncertainties, but may be utilized to check an assumed source mechanism.

View full abstract

The viscous relaxation spectra and the marginal stability curves of Rayleigh-Bénard convection in a transversely isotropic fluid were computed. An incompressible, transversely isotropic fluid is expressed in terms of two material constants, L and δ. L is the viscosity pertinent to the shear in the horizontal plane and δ is the anisotrpy factor. δ is likely to be very large in the mantle if thin less viscous (molten) layers are aligned in the horizontal plane. At large δ the relaxation spectra become nearly constant over a wide range of wavenumber and its magnitude is determined essentially by the product L·δ. The flattness is consistent to the observed relaxation spectra. A similar effect is found in the Rayleigh-Bénard convection; the marginal stability curve flattens out to small wavenumber at large δ. This implies a possibility of thin convection cells in the earth's mantle.

View full abstract

The Adak seismograph network has located 8123 local earthquakes in the central Aleutian island are between August, 1974 and October, 1982. The seismically most active area monitored by the Adak network is the main thrust zone at a depth of 15km to 50km. This zone is characterized by thrust focal mechanisms with the greatest principal stress axis parallel to the relative motion of the Pacific and North American plates. In the Adak Canyon region, there is a zone of east-west extensional normal faulting overlying the thrust zone. Oblique subduction of the Pacific plate may be responsible for this normal faulting as well as for the series of submarine canyons west of Adak Canyon. Between 50km and 100km a narrow band of hypocenters dipping at 45° defines the Benioff zone. At 100km there is an apparent bend in the Benioff zone, and earthquakes extend to 275km at an apparent angle of 65°. Analysis of data from networks of ocean bottom seismographs deployed south of the island arc in 1978 and 1979 suggests that the shallow earthquakes in the main thrust zone have a wider depth range than determined from the land-based network alone and that the apparent bend in the Benioff zone is produced by lateral heterogeneity in the mantle caused by the subducting slab and does not exist. The distribution of hypocenters and focal mechanisms of deep earthquakes is consistent with a subducted slab which is warped at depth. Analysis of the seismccty preceding all the earthquakes with m_b≥4.5 suggests that detectable foreshocks precede less than 10% of these larger events. Investigations of the focal mechanism of small earthquakes preceding three larger earthquakes in the main thrust zone indicate a possible change of mechanism in the period 1-10 month prior to the larger events.

View full abstract

The ground motions caused by a shallow strike-slip fault with barriers have been calculated from a three-dimensional crack model. The faulting motions have been modeled as spontaneous three-dimensional crack propagation in an elastic half space. The ground displacements and particle velocities are obtained at each grid point of a finite difference method. The results indicate that there appear two different regions with high particle velocities. One is the end zone of rupture propagation and the other is just above the barrier. If we assume that earthquake damage depends upon the particle velocity on the ground surface, it may be suggested that the damage would be largest near the barrier and at the end zone of rupture propagation. If this is the case, the positions of barriers and end points of rupture propagation for large inland shallow-focus earthquakes would be located from the distribution of damage.

View full abstract

Correlations of earthquake occurrence among the regions in the west side of Suruga trough are investigated. A fine correlation has been found between seismic activity at the center of west coast of Suruga Bay (region A) and that at the boundary area of Shizuoka, Aichi and Nagano prefecture (region B), situated to the north-west direction of region A.
An earthquake in the region A has a tendency to occur preceding an earthquake in the region B. Propagation velocities of some disturbance connecting these two earthquakes enter almost in the range from 100km/year to 300km/year. It is considered that the correlation of earthquake occurrence between region A and region B is related to the tectonic line which has been proposed to exist in that direction.

View full abstract

Although several techniques have been proposed for a computer-based automatic detection of P and S onset times of small earthquakes, some seismologists have been felt hesitation in using such onset time data on a routine basis. This is mainly due to a lack of parameters judging precision of detected onset times. This paper aimed to introduce a new mathematical aspect for the confidence interval of onset time estimate and to apply the practical tests.
After the autoregressive model parameters have been estimated at two windows immediately preceding and behind P or S onset of an earthquake, the likelihood functions for the two models are defined by assuming normality of probability density function of residual time series. The log-likelihood function of the joint model, which is introduced by using of the above mentioned likelihood functions, gave a basis for onset time picking and statistical inference of a confidence interval time. P and S onset times and their 90% confidence interval were automatically estimated for more than one hundred seismograms of microearthquakes. The result made good comparisons with the manual pickings by several persons. When the signal-to-noise ratio is about 10dB, the 90% confidence interval times are estimated to be 0.2 and 0.8seconds for P and S onsets, respectively.

View full abstract

Theoretical study has been made to investigate the excitations of short-period P- and S-waves on the basis of a dynamic stochastic fault model. The long-wave description of the earthquake source is obtained through a linear superposition of double-couple point sources over a fault, and the seismic directivity is represented by the Doppler effect. Meanwhile, the short-wave description would be obtained by an energy additive superposition of the fracturing fault patches. The seismic directivity of short-waves would be totally under the controll of a patch multiplexing effect, which represents the degree of overlapping of random pulses due to the random fracturing of fault patches. It has been shown that the excitation of short-period S-waves is by far prominent than that of P-waves. This is confirmed through an analysis of energy density spectra of P- and S-waves from the dynamic stochastic source. This result would provide us with an answer to the question why the strong ground motion due to large earthquakes in short epicentral distances is always controlled by S-phases not by P-phases.

View full abstract