A theoretical study has been made to derive the energy spectrum of random-pulse time-series. Since the pulse width corresponds to a rupture time of a random fault patch, the random time-series approximates short-period seismic radiation from the heterogeneous faultings. Assuming exponentially decaying pulses, a parameter S is introduced as the inverse of pulse relaxation times. Energy spectra of the random time series are expressed by a Lorentzian-like spectrum of 1/(ω2+S2) with a weighting function P(S), where ω is angular frequency. Cases that P(S) represents the probability density for Gaussian, uniform, and power-law distributions are studied. It is found that the power-law distributions are related to the probability density distribution specified by the Cantor set. Spectra converge respective constant values at the low frequency end, whereas in high frequencies they show the ω-2 decay. The energy spectrum for the Gaussian is described by the Lorentzian with a spectral corner frequency of the mean value of S. Linear and quadratic power-law distributions of S suggest the Lorentzian-like energy spectra. Fractal spectrum is obtained from the uniform distribution and distributions specified by the Cantor set, showing the 1/f and 1/f1+δ spectra, where f is the frequency and δ is a small fractional number between 0 and 1. The fractional power of spectral decay is related to the variance of the probability distribuition.
The characteristic structure of the Wadati-Benioff zone (WBZ) beneath Kyushu is presented from detailed examinations of the space distribution of intermediate-depth earthquakes. The data for this study are from the Japan Meteorological Agency Seismological Bulletins from January 1983 to April 1987. The WBZ beneath Kyushu, north of the Tokara Channel which lies above the significant transverse discontinuity of the zone, is divided into three segments by two transverse discontinuities which are located in central Kyushu and southern Kyushu (the central part of Kagoshima prefecture). In the northern segment of about 130km in strike length, the WBZ of 10 to 20km in thickness dips at an angle of about 60° at depths of around 80km, and about 65° in the lower portion extending to a depth of about 140km. The central segment of about 125km length changes strike from that of the northern segment, N15°E, to N27°E. The WBZ of 20 to 30km thick dips at an angle of about 60° at depths of 80 to 100km, and 60° to 70° at depths of about 100km or more. The deepest portion of the zone is around 170km. The southern segment of about 250km length strikes to N24°E. The zone of 50 to 60km in thickness dips at angles of 55° to 60° at depthes of about 80km and greater, and extends to about 200km in depth. The major structure defined by the WBZ seems to be mirrored in the structure of the oceanic region ranging from the zone to the trench axis. The transverse discontinuity separating the northern and central segments may be connected to the tectonic line separating source regions of the 1968 and 1984 large Hyuga-nada earthquakes from those of the other large earthquakes of 1931, 1941 and 1961. The discontinuity separating the central and southern segments connects probably with the tectonic line separating Hyuga-nada region and the southeast of Kyushu region. The active volcanoes in southern Kyushu (the central and southern segments), which is characterized by a fairly good volcanic lineament, are situated on the WBZ at depths of about 80 to 100km. This is a general characteristic for the active volcanoes in the island arcs. In the northern segment, however, the volcanoes seem to be hardly related to the WBZ. For instance, Asosan volcano lies to the west of the zone, though this volcano is petrologically referred to the Ryukyu type as those in southern Kyushu.
Since late 1986, we have been operating a three-component set of STS seismometers (STS-1) at Inuyama Observatory, Nagoya University, Japan. This is a preliminary report of the two years observation. The STS-1 seismometer equips with three component BRB (BRoadband Band) outputs and three component LP (Long Period) outputs. The lowpass filtered LP outputs are digitized and recorded continuously at a sampling interval of 10s in a dynamic range of about 120dB. These continuous data have been used so far to determine seismic moment or centroid moment tensor of large Japanese earthquakes. The BRB output is recorded on an event recordering system, where the BRB and LP outputs after an eight-pole analog anti-aliased filtering are digitized and recorded for 50 minutes at sampling intervals of 0.1 and 0.4s, respectively. The dynamic range of this system is limited at present by the A/D converter to about 90dB. The triggered BRB data have been used so far to determine the source time functions of large Japanese earthquakes, which suggest the occurrence of backswing of fault motion. To maintain the horizontal component instruments in a stable condition, a care must be taken for variation of atmospheric pressure. The results of the analyses indicate a unique importance of nearby observation of large earthquakes by broadband, wide dynamic range and high precision seismometers such as the STS instruments.
A magnetotelluric survey was carried out at the Hanaori fault area in northeastern Kinki district. Apparent resistivities at four frequencies of ELF band (7.8, 14.0 and 20.4Hz) and VLF band (17.4kHz) were measured along two survey lines perpendicular to the strike of the fault. Two-dimensional analysis using the finite-element method gives an underground resistivity model across the Hanaori fault. The model indicates that a low resistivity zone is 1.4km wide and extends to a depth of more than 800m. We interpret this zone as resulting from fault activities.
The rupture process of the 1987 east off Chiba (Chiba-ken-Toho-oki) earthquake is examined using a waveform inversion technique. The synthetic seismograms are calculated by the empirical Green's function method. The observed seismograms are obtained using velocity type or acceleration type strong motion seismographs. The parameters we estimate in this analysis are the moment release, the rupture time, the rise time, and the stress drop at each subfault on the mainshock fault surface. The rupture history obtained here shows the heterogeneities of the faulting process which are well related to the aftershock distribution and the large scale tectonic structures. The results are summarized as follows: strong moment release occurred in areas of low aftershock activity; the rupture propagation paused and the rise time increased around the discontinuity at the depth of about 30km; the stress drop was high at the rupture initiation point as well as the deeper and the shallower parts of the northern fault surface where the mainshock terminated and the aftershock activity extended to the northern upward direction.
We developed an inverse Radon-transform method to reconstruct slip velocity intensity (SVI) on the fault plane by using near-field seismograms. The observed source time function is approximated to the integration of the product of the SVI and the isochrone velocity along the isochrone, the shape of which depends on the fault plane-observation point geometry, and the rupture and wave velocities. Our approach combines the back-projection method with a windowing procedure. Our approach has the advantage of not requiring a priori assumption of faulting area over the previous inversion studies of seismic source processes. Some simulation models used to test the applicability of our method have shown that it can recover well the SVI distribution. This method is applied to the 1980 Izu-Hanto-Toho-Oki earthquake. Our result shows a heterogeneous SVI distribution over the fault plane. Relatively large values of SVI were found in the north and center areas of the fault plane from the aftershock occurrence and the regions correspond to those region with less aftershock occurrence. The heterogeneous rupture effectively radiating high-frequency seismic wave is restricted to a smaller region than the inferred fault plane.
The 1974 Izu-Hanto-Oki earthquake is studied in detail using near-field strong motion seismograms. A waveform inversion method is applied to deduce the dislocation distribution and the characteristics of the rupture propagation during this earthquake. This earthquake involved right-lateral strike-slip motion on the almost vertical fault plane with a strike of N47°W. The rupture initiated at the central deepest part of the fault plane and propagated both sides smoothly, as a bilateral rupture propagation. The total source process time is about 11sec. A dislocation larger than 1m occurred in the region ranging from 3km to 10km in depth, and its horizontal span is about 20km around the hypocenter. Except for the southeastern end of the fault plane, dislocation smaller than 0.5m occurred in the shallower region of the fault plane. The average dislocation on the whole fault plane is 1.0m, and the total seismic moment is 7.6×1025dyne·cm. Few aftershocks took place in the area where dislocation larger than 2m occurred during the main event. Surface fractures, associated with this earthquake, appeared in the meizoseismal area. The dislocation distribution seismologically obtained in this study is consistent with the fault displacement along these surface fractures.
In order to understand mechanisms of H2 emanation increase in groundwater or soil gas near active faults before an earthquake, we investigated the variation of H2 emanation from a Westerly granite specimen under uniaxial compression, using a ceramic sensor with high selectivity and sensitivity to H2. Creep tests were carried out under 120, 140, 160, 180 and 193MPa stress levels, to clarify relations between H2 emanation, Rn emanation, stress, acoustic emission and strain. The experimental results are as follows: (1) H2 and Rn emanation began to increase in company with generation of acoustic emission, that is, microfracturing; (2) when generation of microfracturing stopped, H2 emanation stopped whereas Rn emanation was maintained at the same emanation level; (3) before ultimate fracture, H2 and Rn emanation continuously increased together with increasing volumetric strain; (4) drastic increase of H2 and Rn emanation appeared after the ultimate fracture, which was caused by the increase of fracture surface. The results suggest that simultaneous increase of H2 and Rn near active faults means continuous generation of microfracturing under the ground, and it may possibly act as a precursor of an earthquake.
Gutenberg-Richters' formula (GR), truncated GR formula (TGR) and modified GR formula (MGR) are known to represent the magnitude vs. frequency relation of earthquake occurrence. We discuss parameter estimations and goodness-of-fits of the three formulae by using actual data and maximum likelihood method. In the case of using data of round magnitudes, an upperbound magnitude (parameter C) and maximum likelihood of TGR are not directly estimated by using the maximum likelihood method based on probability density function. However, this is indirectly estimated by using the maximum likelihood method based on probability distribution function induced from the density function. With the modified method and the information criterion AIC, goodness-of-fit of TGR can be discussed in comparison with other formulae. A formula with 4 parameters is derived from an exponential function polynomial. The formula (4PGR) includes new parameter α as a convex index near the upperbound magnitude. The 4PGR corresponds to TGR and MGR in the cases of α→∞ and α→0, respectively, and either TGR or MGR can be regarded as a special case of 4PGR. Goodness-of-fits of GR, TGR, MGR and 4PGR for earthquake data at 4 regions are compared by using AIC. The Result at each region indicates much better fit of the formulae with the upperbound magnitude than original GR. In these cases such as 200 earthquakes in sample size, the best fit does not seem to be 4PGR but to be TGR or MGR.
A questionnaire survey was carried out to estimate the detailed seismic intensities on the Iwate Chubu-engan Earthquake of magnitude 6.6 of January 9, 1987. 5032 questionnare sheets were delivered to the persons living in and around Morioka City, to investigate the relation between the detailed seismic intensities at every 500m×500m mesh and the subsurface structure based on the boring data. Other 5122 questionnare sheets were delivered to the persons living in another area of Iwate Prefecture. The relation of intensity attenuation with distance was deduced and the relation between the intensity deviation and general geological stucture was investigated. Obtained results are as follows: 1) In Morioka City and its suburban districts, the detailed seismic intensity considerably depends on the subsurface structure at the site. For example, deviations from the average intensity are -0.1--1.0 in the district at where base rock of granite and shalstein is exposed or covered by sediment with sand and gravel. Conversely, at place of the soft ground constituted by volcanic mudflow and loam, they are +0.1-+0.7 2) From a view point of general geological structure in the whole Iwate Prefecture, seismic intensities are 0.1-0.6 greater than those expected from empirical formula at the valley of the Kitakami river and Toono basin where Quaternary fluvial plain deposit is widely distributed. On the contrary, seismic intensities are 0.1-0.4 less than the average at the west of the Kitakami river where Tertiary system and Quaternary volcanic rocks are distributed, and the whole Kitakami mountains where pre-Tertiary is distributed.
Swarm-like activity of microearthquakes, with a remakable regularity in recurrence interval of about 14.5 months, has taken place around the Kamafusa Dam reservoir in Miyagi Prefecture, the northeastern part of Japan, since the beginning of impoundment in February 1970. The temporary observation network was set forth in June 1980 and was in operation through December 1982 for an extensive investigation on the nature of seismic activity around the reservoir. Three groups of microearthquakes were observed north of the reservoir during the period of the temporary observation, though the seismicity in this area was not so active as that in the preceding period. The present study together with the previous work of SAIJO et al. (1985) on the details of seismicity in this area suggests a migration of swarm activity away from the reservoir since the onset of the swarm activity in 1970. In order to investigate the nature of swarm activity around the reservoir area in more detail, source parameters are estimated from the P-wave analysis for one of the three groups of microearthquakes, which includes a relatively large earthquake with the focal mechanism solution determined. The seismic moment is estimated from the low frequency level of the spectrum. The relationship between the initial half cycle pulse width and the corner frequency is used to infer the rupture velocity for the earthquakes of the group. The source radius is then precisely estimated from the initial pulse width and the rupture velocity obtained for the group. It is found that the logarithm of seismic moment is linearly related to the magnitude of earthquake as logM0=1.50M+15.63 for 0.5≤M≤4.0. Rupture velocities of the earthquakes are estimated as 0.81 times the S-wave velocity. Source dimensions are found to range from 62m to 710m, and stress drops from 0.1bars to 7.9bars.
This paper reviews rock friction studies and its applications to earthquake source studies. Current researches on rock friction have revealed that a rate of decrease in frictional resistance from a static frictional stress to a dynamic one is finite. The frictional behavior of many rocks has been described by a slip distance and/or slip rate-dependent constitutive law. Many researches have shown that surface roughness plays an important role in the friction law and have suggested that the fault roughness can be a source for size-scale dependence of physical parameters such as the critical slip-weakening displacement and the apparent fracture energy of shear fracture. Numerical models of earthquake faulting are presented using a constitutive friction law. Those can simulate not only a strong motion of an earthquake but also a premonitory slip of one. These results give a better understanding of the slip instability, leading to a deep understanding of an earthquake generation process.
This paper reviews recent developments in computational techniques for wavefields in laterally heterogeneous media, mainly the Gaussian beam method. We start with a solution by the asymptotic ray theory as a high-frequency limit, and point out two major defaults in this approach. The first problem is a two-point ray tracing or time-consuming search for the ray connecting source and receiver. This difficulty can be removed by the paraxial ray approximation, extrapolating wavefields outsides of rays. The second is due to singularity in some critical regions such as caustics and shadow zones. By combining a point-source wave with a plane wave, the Gaussian beam method can avoid such singularity. Also, superposition of many beams naturally smooths out heterogeneities of media, simulating the effect of finite wavelength. In laterally homogeneous media, this superposition corresponds to the WKBJ seismogram of Chapman as a limit. The real difficulty lies in ambiguous weighting factors for the two wavefields: a point-source wave and a plane wave. No studies have ever revealed any physical basis to find appropriate values of the weighting factors. In the near future, one must develop any method to globally estimate those values, rather than empirically, by matching boundary conditions of wavefields on free surface or on interface, which will eventually give more reliable results.