To study earthquake generation mechanisms, we carried out two seismic experiments in the Izu-Bonin Trench and the Japan Trench subduction zones using OBS and controlled sources in 1996 and 1998, respectively. In the Izu-Bonin subduction zone, we inferred the presence of serpentinized rocks along the top of the slab (Kamimura et al., 2002). Between 38-39° N along the Japan Trench forearc slope, we observed a good correlation between aseismicity and PP reflection phase from the top of the slab (the plate boundary) (Fujie et al., 2002). To confirm the results of the survey in the Izu-Bonin Trench subduction zone, we calculated synthetic waveforms from the velocity structure obtained by travel-time tomography and seismic attenuation (Q model). A comparison of observed and synthetic waveforms suggests the existence of a low Qp ( 10) layer beneath the Torishima serpentine seamount and extending from the seamount 80 km from the west of the seamount along the top of the slab. This modeling supports the presence of serpentinized peridotite at the top of the slab. We suggest that such a layer may allow an aseismic slip during subduction of the oceanic plate, primarily due to the low frictional coefficient of the crysotile (a phase of low temperature serpentine). In the Japan Trench subduction zone, we carried out a more extensive OBS-controlled source seismic experiment to confirm the results from 1996. We shot airguns along 7100 kmlong path that runs nearly parallel to the trench axis, and this paper presents the results from the western 5 lines. We observed very intense PP reflections from the top of the slab over the aseismic zone. By including a thin low velocity layer just above the slab, we obtained similar strong PP reflection phases in synthetic waveforms from the top of the slab. These observations and models suggest that a layer with a P velocity as low as 2-4 km/s and a thickness of 100-400 m is needed at the top of the slab to explain the observed PP reflection intensity. Such a layer can comprise aqueous fluid, clay, and/or serpentine-chlorite generated by hydration/dehydration of the subducting oceanic crust and the wedge mantle peridotite. The top of the slab at 18 a path of km exhibits stronger PP reflections than it does at 12 a depth of km depth, and it suggests upwelling of serpentinized and/or clay-rich material from the wedge mantle to the shallower plate boundary. In the Izu-Bonin Trench and the Japan Trench regions, we conclude the presence of low P wave velocity and low Q materials along the top of the slab. This result strongly suggests aseismic slip during the subduction of the Pacific Plate in the shallow part of the Izu-Bonin Trench subduction zone and in the aseismic zone between 38-39° N over the Japan Trench forearc slope.
Resent studies on relationship between sites for co-seismic slip and a-seismic slip are reviewed. Through analysis of seismic wave and Global Positioning System (GPS) data set, the sites for co-seismic slip and a-seismic slip have been obtained in Hyuga-nada and Sanriku region. Those studies show three important results : (1) the sites for co-seismic slip, episodic a-seismic slip, and continuous a-seismic slip do not overlap but complementarily share the plate boundary : (2) after-slip (post-seismic slip) may play an important role especially in the triggering large earthquakes : (3) the depth range of a-seismic slip may be controlled by not only a thermal effect but also some other factors such as lateral heterogeneity of fluidpressure and serpentinization.
Non-volcanic deep low-frequency tremors detected in southwest Japan are distributed in the forearc side along the strike of the descending Philippine Sea plate. The source depth of the tremor corresponds to the slab interface or the Moho discontinuity. The time sequence of the tremor activity is characterized by long durations from hours to weeks. The mobility and the successive occurrence of the tremor are thought to be related to the existence of fluid liberated from the slab by a dehydration process. The spatial distribution of the tremors is not homogeneous in a narrow belt but is spatially clustered. The major activity of the tremors with relatively long time durations is also clustered periodically, with a period of 2-3 months in the east and middle of Shikoku area and about 6 months in the west of Shikoku. On the other hand, tremors are sometimes triggered by local earthquakes or teleseismic waves. The periodicity of the tremor activity may represent a stable accumulation of fluid with a stable subduction process and the triggering phenomenon implies the unstable condition of the occurrence of tremors.
We have examined the effects of stress applied to a fault system on the dynamic rupture process of earthquakes. First, we investigated a feasibility of estimating the stress field from final slip distribution during an earthquake. The result of a numerical simulation suggests that principal stress directions can be estimated within an error of 5 degrees but stress ratio could not be well resolved. Then, a dynamic rupture process of the 2000 western Tottori earthquake (Mw6.6) was modeled assuming a tri-axial homogeneous stress field. The result was that the maximum principal stress direction of N105°E was the easiest for the rupture to propagate. This is mainly due to the initial shear stress distribution on the fault. This principal stress direction is consistent with that estimated by a stress tensor inversion of aftershock moment tensors of this earthquake. These results suggest a possibility to obtain a stress field applied to the fault by analyzing slip distributions of past earthquakes.
A numerical simulation of seismic cycles on a plate boundary is performed using a rate- and state-dependent friction law. The spatial distribution of frictional constitutive parameters is determined so that the simulated seismic cycles may mimic the occurrence of large earthquakes in the northern Sanriku area along the Japan trench. Two patches with velocityweakening frictional properties are assumed to represent two large asperities that were ruptured in the 1968 Tokachi-oki earthquake (Mw=8.2). By choosing appropriate friction parameters, I can simulate the recurrence of large earthquakes that break both the velocityweakening patches at a recurrence interval of 90 years. During each interseismic period of the simulated large earthquakes, one of the velocity-weakening patches is broken, resulting in the occurrence of a smaller earthquake. This may correspond to the occurrence of the 1994 Sanriku-oki earthquake of Mw=7.7. The coseismic slip amounts in the simulation are similar to those estimated for the 1968 Tokachi-oki and the 1994 Sanriku-oki earthquakes, and postseismic sliding is reproduced in the velocity-strengthening region. These simulation results for seismic cycles at the northern Sanriku subduction zone suggest that the frictional constitutive parameters on the plate boundary can be estimated to some extent. The model with realistic friction parameters is expected to be useful for forecasting the sliding process on the plate boundary.
It is now widely believed that fluids exert significant effects on earthquake faulting. Recently, considerable attention has been directed to the effects of fluid migration on the spatio-temporal complexity of earthquake occurrence. We review a series of theoretical studies by the author on the mechanical effects of fluid on the spatio-temporal variation of seismicity. In particular, we discuss the generation mechanisms of aftershocks and earthquake swarms in detail. It is shown that diversity in the occurrence of aftershocks and earthquake swarms can be explained in a unified way by considering fluid migration.
The seismogenic fault in the Okitsu Melange, Shimanto accretionary complex, Shikoku is characterized by repeated rapid slips with a pseudotachylyte formation and a concentration of vein minerals. The Okitsu Melange is composed of duplex structure of imbricated oceanic stratigraphic sequence, and the seismogenic fault is emplaced at the roof thrust of the duplex structure. Comparable fault along the seismogenic roof thrust of the duplex structure was found at the present Nankai trough, and vein mineral precipitation is expected from the fluid flow along the fault. Some of the clasts of cataclacite were supported by a vein mineral matrix of euhedral ankerite and quartz crystals. This occurrence of vein minerals may indicate rapid precipitation due to fluid flow of super saturation with vein minerals along the fault zone. The shear strengthening with time and the stick slips were recognized by the block-slider analogue experiments which consisted of the chilled slid with ice and hot water saturated with alum on the shear surface. The result of the experiment indicates that the vein mineral precipitation make stick slip the fault.
Occurrences of fault rocks were analyzed along an exhumed brittle-plastic fault zone in the Earth's crust, the Hatagawa Fault Zone (HFZ) of NE Japan. A conspicuous cataclasite zone with a maximum width of 100 m extends continuously for at least 40 km along the HFZ, corresponding to a rupture size of an inland earthquake as large as M7. The cataclasite zone was formed at temperatures above 220 t, and its activity had terminated by 98.1 ± 2.5 Ma. Mylonite zones with a sinistral sense of shear are heterogeneously distributed for the entire 45 km length along the HFZ. The temperatures calculated by two-feldspar thermometry from most of the mylonite zones are above 360°C. On the other hand, there is a mylonite zone with a length along the HFZ of approximately 6 km where the temperatures calculated are below 360°C. Microstructures of the fault rocks in this zone indicate that the deformation condition was at the brittle-plastic transition. The distribution of fault rocks along the HFZ suggests that only limited areas of the brittle-plastic transition were plastically deformed when the present exhumed level was in the brittle-plastic transition zone. Such heterogeneity of plastic deformation in the brittle-plastic transition zone can result in a significant stress concentration and the nucleation of large earthquakes. The cataclasite zone along the HFZ was possibly formed by the propagation of an earthquake nucleated in the brittle-plastic mylonite zone.
The Nojima fault appeared on the surface in the northern part of Awaji Island, central Japan as a result of the Hyogo-ken Nanbu earthquake (1995, M = 7.2). Active fault drilling was performed by the Disaster Prevention Research Institute (DPRI), Kyoto University, and core samples were retrieved from 1410 to 1710 m, which were composed of intact and fractured granodiorites. We obtained calcite samples and gas samples from the vein in marginal fracture and non-fracture zones. We analyzed the carbon and oxygen isotope ratios of calcite and carbon dioxide to investigate the characteristic isotope ratios of fluids in the active fault zone, to estimate the origins of fluids, and to determine the sealing process of fractures. The analyzed values of carbon and oxygen isotope ratios of calcite were -10.3 to -7.2‰, 18 to 23/00, respectively, and carbon isotope ratios of CO2 were -21 to -17‰. If carbon isotope ratios of calcite were at equilibrium with those of CO2, the precipitation temperature of calcite is calculated to be 30 to 50°C. This temperature is consistent with the present temperature of the depth where drilling cores were retrieved. Oxygen isotope ratios of H2O that, precipitated calcite were calculated to be -1.8 to -5.5‰. These values indicate calcite were precipitated from mixed fluids of sea water and meteoric water. Therefore, the marginal fracture zone of the Nojima fault was sealed with calcite, which was generated from mixing of sea water and meteoric water in situ.
To quantify the transport properties of faults, a series of laboratory experiments is performed on permeability change during deformation of simulated shale smear. The fluid sealing potential of faults is important for better understanding of the hydrocarbon migrationaccumulation process and abnormal fluid pressure build-up below faults in a sedimentary basin. One of the possible mechanisms of the fault sealing process is role of shale smear characterized by migration of mudstone into the fault plane, forming a barrier to the fluid flow. To evaluate the sealing potential of a fault, it is necessary to develop tools and methodology to monitor the permeability of faults during deformation, and to compare experimental results with field observations. First, the author investigated the smear continuity developed in deltaic sediments on Airport Road outcrop, Miri, Sarawak, Malaysia. At this site, the critical shale smear factor (SSFcrt : showing maximum continuous smear at a shale-thickness/throw data) was approximately 8, which is larger than previously reported data at other sites. Then the author conducted experiments using sandstone-shale interbedded specimens obtained from the Miri formation. The natural specimen was inserted into a part of an artificial fault of pre-cut Berea sandstone, and the permeability of the sample was measured. Similar to results of the previous experiments on siltstone by Takahashi (2003), permeability show the following three stages : (1) Regime 1, rapid reduction due to compaction prior to faulting, (2) Regime 2, constant and relatively lower permeability, and (3) Regime 3, permeability recovery caused by erosion of smear. Except for an experiment with a sandy specimen, moreover, all experiments show permeability recovery of 0.5 orders of magnitude after Regime 1. This characteristic permeability change from Regime 1 to Regime 2 for mud-rich specimens appears to indicate dilation induced by fault propagation in the over-consolidated specimen at starting fault sliding. Fault structures observed from the outcrop and from experimental products are similar. The laboratory experiments can thus accurately reproduce the mechanism of fault sealing by shale smear, suggesting that experimental studies are essential to understand fault sealing by shale smear.
Analyzing the development of anomalous high fluid pressure is important for basin analysis and fluid-oil migration analysis at depth. We measured the permeability and the porosity of sedimentary rocks in the Western Foothills of Taiwan where the 1999 Chi-Chi earthquake occurred. Permeabilities of all rocks decreased with increasing Pe (effective pressure), and all of them showed hysteresis. Permeability was generally higher in sandstone (10-15 10-18 m2) than in siltstone (10-1710-20 m2). Sandstones were divided into two groups. One group showed little sensitivity to Pe and the other group decreased moderately with increasing Pe. Porosity also decreased with increasing Pe and showed hysteresis. Porosities decrease to around 4-8% during compaction to 200 MPa. Specific storage decreased with Pe and the values changed from 10-8 Pa-1 to 10-10 Pa-1 at the 200 MPa of Pe. The fluid pressure distribution of the Western Foothills was calculated by the non-linear one-dimensional compaction flow model in sedimentary basins (Bethke and Corbet, 1988) using the experimental results of permeability and specific storage as a function of Pe. The result indicated that abnormal fluid pressure was generated from 4 km and increased with depth, showing a similar trend to published in-situ data. However, the results calculated from the simplified compaction flow model that permeability and specific storage can be assumed to be constant at these depths showed a strong difference with in-situ data. This indicates that we can not neglect pressure sensitivities to hydraulic properties when we construct a fluid flow model of a basin at depth.
To clarify the deformation and slip mechanisms in seismogenic regions, we designed and developed a new gas-medium, high-pressure and high-temperature apparatus. The apparatus can attain confining pressures up to 200 MPa by argon gas, temperatures up to 800°C, and pore pressures up to 200 MPa by argon gas or water. Confining and pore pressures can be applied and servo-controlled. We report preliminary test results. The stress-strain curve of granite sample at a confining pressure of 200 MPa and a temperature of 800°C shows a ductile behavior. One of the friction parameters, the a-b value defined as the velocity dependence of steady-state friction, of quartz gouge was measured under hydrothermal conditions.
A quantitative interpretation was made on the observational parameters in an equation of time-dependent healing of frictional strength of faults. The interpretation is based on a physical model that explains the time-dependent growth curve of frictional strength as an increase of the real contact area of the fault surface due to asperity creep driven by the stress normal to the contacts. The creep is assumed to be rate-limited by some stress-assisted thermally activated process. The model thus predicts temperature and normal contact-stress dependences of the observational parameters of the time-dependent healing of frictional strength. The predictions are more or less consistent with available experimental data, which include different healing mechanisms associated with different types of elementary contact deformation mechanism, such as stress corrosion and pressure solution.
To investigate the critical phenomena between aqueous fluid and hydrous silicate magma in the Earth's interior, a new method for making direct observations of immiscible fluids was developed using X-ray radiography technique together with the Kawai-type double stage multianvil system driven by a DIA-type cubic press (SPEED-1500) installed at BL04B1, SPring-8, Japan. Basalt-H2O and Sr-plagioclase-H2O systems were investigated. A new sample container comprising of a metal (Pt or AuPd) tube with a pair of lids made of single crystal diamonds was used under pressures between 2 and 5 GPa and temperatures up to 1500°C. The sample inside the container can be directly observed through the diamond lids. At around 800-900°C and pressures up to about 3 GPa in both basalt-H2O and Srplagioclase-H2O systems, we observed light-gray spherical bubbles moving upward in the darkgray matrix. The light-gray spheres that absorb less X-rays represent an aqueous fluid, whereas the dark-gray matrix represents a silicate melt. At least up to 1200°C, immiscible two phases (i.e., both aqueous fluid and silicate melt) were observed up to about 3 GPa in the basalt-H2O system. At pressures above about 4 GPa, no bubbles appeared when heating to about 1200°C, suggesting the possibility that the second critical end-point in basalt-H2O system could be around 3 to 4 GPa. Our new technique can be applied to the direct observations of various coexisting 2-fluids under deep mantle conditions that could not be achieved by previous experimental methods.
To understand not only the mechanisms of earthquakes, but also the origins of the diverse behavior of faults and plate boundaries, one must integrate (1) field studies on faults to understand deep intrafault processes, (2) laboratory work to reproduce those processes and determine mechanical and transport properties of fault zones, (3) theoretical and numerical studies analyzing fault motion, including earthquake generation processes, based on the constitutive properties determined by laboratory studies, and (4) seismological and geodetic studies revealing dynamic fault motion during earthquakes and diverse aseismic fault behavior. Ideally, such integrated studies should be carried out for a selected fault that produced an earthquake with good seismic/geodetic records so the predictions from (1) to (3) can be fully tested against (4), rather than selecting favorite data from the literature. The 1999 Taiwan Chi-chi earthquake is an ideal example for such integrated studies because the fault motion during the earthquake is clearly analyzed based on very good near-field strong motion data, and because the Chelungpu fault zone, which caused the earthquake, is exposed on land and has been drilled at several places. Also, the IODP drilling project into seismogenic zones in the Nankai Trough will provide rare opportunities for such integrated fault and earthquake studies in the near future. This paper focuses on high-velocity frictional properties of faults for which frictional heating plays a crucial role, with special reference to dynamic fault motion during large earthquakes. Recent progress in high-velocity friction studies, particularly those in frictional melting, thermal pressurization, and high-velocity gouge behavior, are rapidly filling the gap between field/laboratory studies on faults and seismological/geodetic studies on earthquakes. Representative results from our recent studies are presented, revealing that the field/laboratory data can predict slip weakening distance, Dc, of the same order as determined by seismic data. We also show highlight data on frictional melting and argue that the effect of frictional melting on the dyanimic fault property can be predicted by solving a Stefan problem with moving boundaries. Seismic fault motion may be predicted not so long in the future based on the measured properties of a fault that caused an earthquake. Transition from ordinary friction to high-velocity friction that has been poorly explored to the present, should control the initial phase of earthquake generation and perhaps is critical to an understanding of the physical bases of earthquake prediction. This is perhaps the most important area for systematic studies in the near future.