Shingo Yoshida

Electric potential changes associated with slip failure of granite: Preseismic and coseismic signals

Electric potential changes were measured for stick-slip events in granite samples with a three-block direct shear arrangement at 8 MPa normal stress. Two electrodes were mounted on the left- and right-hand blocks, and the electric potential difference between each electrode and the ground was measured with a high input impedance recording system of frequency range from DC to 100 Hz. As well as coseismic electric signals of about 1.5 V which appeared the moment of the dynamic slip event, preseismic signals were detected just before the slip event. The coseismic signal rises stepwise with opposite polarities at the two electrodes and exponentially decays with a time constant of e/s, where e is the permittivity and s is the conductivity of the rock sample. We conducted a simple test of rapid stress drop without slipping and observed almost the same electric signal as the coseismic signal. This suggests that the electric signal is generated by the piezoelectric effect. We proposed a generation model based on the piezoelectric effect and the resultant relaxation process and obtained a theoretical frequency response, which is in agreement with experimental data. The preseismic signal appears about 2–3 s before the dynamic event with an amplitude of about 50 mV. The local strains along two sliding surfaces were also measured to monitor the growth of the rupture nucleation zone. When the growth of the rupture nucleation zone occurred on the left sliding surface, a clear preseismic signal was detected at the electrode mounted on the left granite block. When the growth occurred on the right-hand surface, a signal was detected at the electrode on the right block. This shows that the preseismic electric signal is caused by stress change in the rupture nucleation zone. These preseismic and coseismic signals were also detected with an antenna, which was placed away from the sample surface.

Comparison of the coseismic rupture with the aftershock distribution in the Hyuga‐nada Earthquakes of 1996

On October 19, 1996, a large underthrusting earthquake (Ms = 6.7), the Hyuga-nada, Japan, earthquake occurred along the southern end of Nankai trough. About two months later, a second large earthquake (Ms = 6.7) occurred in the adjacent region. We study the source process of the two large earthquakes in the Hyuga-nada region and compare the coseismic rupture area with aftershock distribution. The main source parameters obtained for the first mainshock are: (strike, dip, rake)= (210°, 12°, 81°); the seismic moment Mo = 2.3×1019 Nm (Mw = 6.8); the rupture area S = 20 × 15 km², and the source duration T = 17 s. For the second main-shock, (strike, dip, rake) = (210°, 12°, 87°); the seismic moment Mo = 1.5×1019 Nm (Mw = 6.7); the rupture area S = 18 × 18 km², and the source duration T = 15 s. The coseismic rupture areas do not overlap the aftershock area, while the aftershock areas of the two main-shocks mutually overlap. This implies that the common aftershock region takes a role of barriers to dynamic rupture. It is also seen that the aftershock area expanded during the first one day.

Convection current generated prior to rupture in saturated rocks

Laboratory experiments were performed to study the generation mechanism of electric signals during deformation and rupture of rock, with the focus on the effects of pore water movement. We have developed a triaxial apparatus that is specially designed for this purpose, within the framework of the International Frontier Program on Earthquake Research of the Institute of Physical and Chemical Research (RIKEN). This apparatus can deform a rock specimen that is electrically isolated from the surroundings at a very slow strain rate. Using a cylindrical specimen of intact rock (Inada granite, Makabe granite, Muroto gabbro, and Fujieda sandstone) saturated with distilled water, we carried out triaxial deformation tests under a constant pore pressure at room temperature. During a deformation test we continuously measured the electric current generated in the specimen. In addition, local strains, pore pressure, and pore fluid movement from the intensifier were recorded at a sampling frequency up to 1 kHz. The volume change due to dilatancy was obtained from the average of the local strain measurements at four positions. We found that the convection current flowed before the main fracture, showing good correlation with the dilatancy rate and the water flow rate. The current density of the signal was about 1 mA/m2. This result demonstrated that the electric current is caused by an electrokinetic effect due to the water flow associated with accelerating evolution of dilatancy before the fracture.

Waveform inversion for rupture process using a non-flat seafloor model: Application to 1986 Andreanof Islands and 1985 Chile earthquakes

Abstract Yoshida, S., 1992. Waveform inversion for rupture process using a non-flat seafloor model: application to 1986 Andreanof Islands and 1985 Chile earthquakes. In: T. Mikumo, K. Aki, M. Ohnaka, L.J. Ruff and P.K.P. Spudich (Editors), Earthquake Source Physics and Earthquake Precursors. Tectonophysics , 211: 45–59. We inverted teleseismic P waveforms to infer rupture processes of subduction zone earthquakes using Greens functions computed for realistic seafloor topography because reverberations in the water layer near the source affect the teleseismic waveforms. We modeled the seafloor by a series of several plane surfaces with appropriate dip and performed three-dimensional ray tracing to compute Greens functions. From the GDSN seismograms, spatial and temporal slip distributions were estimated for the 1986 Andreanof Islands earthquake ( M s 7.7) and the 1985 Chile earthquake ( M s 7.8). To suppress unstable parts of the least-squares solution, we introduced some stabilizing constraints. The 1986 Andreanof Islands earthquake was analyzed using two kinds of source models: the first one has a prescribed rupture velocity and the second one allows any point to slip at any time. The solutions by the two models gave similar patterns of slip distribution. A relatively small subevent initially occurred and then a dominant subevent with a moment of about five times larger occurred with a center about 80 km west of the hypocenter. For the 1985 Chile earthquake, a small initial subevent was also found and a major rupture with a ten times larger moment extended approximately from 20 to 70 km south of the hypocenter. For both the earthquakes, it was found that most of the one-day aftershocks occurred in areas where a small moment was released during the main shock.

Waveform inversion using ABIC for the rupture process of the 1983 Hindu Kush earthquake

Abstract An inversion method was developed to infer earthquake rupture process from far-field velocity waveforms. We devise a source model that has variable slip and variable rupture velocity. The fault plane is divided into many subfaults. Slip amplitude is assumed to be dependent on the subfaults, and slip velocity at a point is assumed to be a triangle function of time. Rupture time is expressed by an interpolation equation using the rupture times at the corners of the subfaults. The slip distribution and the rupture-front motion are estimated by constrained least-squares inversion. We incorporate smoothing constraints on slip distribution and on rupture propagation into the observed data. The degree of the constraints is controlled by hyperparameters. The optimal values of the hyperparameters are determined by minimizing Akaikes Bayesian information criterion (ABIC). Under the assumption of ABIC, we can obtain a unique solution with reasonable resolution. We apply this inversion method to the Hindu Kush earthquake of 30 December 1983 with a depth of 212 km. We found that the rupture process of this earthquake was composed of two subevents. The first subevent occurs near the initiation point of the rupture. The centre of the second subevent region is approximately 20 km to the west of the first subevent. The released moment of the second subevent is about seven times larger than that of the first subevent.

A fault model of the 1995 Kobe earthquake derived from the GPS data on the Akashi Kaikyo Bridge and other datasets

Co-seismic horizontal displacements, which have been obtained from recently released GPS observations on the Akashi Kaikyo Bridge, are examined for their consistency with displacements observed in the vicinity of the bridge. An E18.7°S displacement of 25.0 cm should be removed from them. The adjusted data indicate an additional fault segment beneath Akashi Strait. We construct a new fault model by adding this segment to a model assumed previously. We then recover the slip distribution over the new model by inverting these data together with other geodetic observations. The displacements calculated from the recovered distribution fit the observations well, and the distribution of slip indicates that the additional segment is closely related to the southern main segment of the previous model. Joint inversion of the geodetic and waveform datasets suggests large slips with longer duration in the shallow parts of these segments.

Waveform inversion for rupture processes of two deep earthquakes in the Izu-Bonin region

Abstract Using a waveform inversion method, we studied the rupture processes of two deep earthquakes in the Izu-Bonin region. The earthquake source was modelled by space-time-dependent slip on a fault plane. The fault plane was divided into many subfaults, and the source was expressed in terms of the final dislocations on the subfaults, the rupture starting times at the subfault corners and the rise time of source function. From P-wave seismograms recorded by the Global Digital Seismograph Network, we first obtained the displacement waveforms by deconvolution techniques. Next, we inverted these waveform data to determine the model parameters. For the earthquake near Torishima on March 6, 1984, non-homogeneous stress distribution and variation in rupture velocity were revealed. The rupture initiated at a corner of a 54 × 36 km 2 fault, and high-stress drop occurred in the middle region of the fault, about 30 km away from the initiation. The earthquake south off Honshu of April 24, 1984, did not have as complex a rupture process. The rupture propagated unilaterally with almost constant velocity.

Single and double asperity failures in a large‐scale biaxial experiment

Stick slip experiments were performed in a direct shear apparatus using large Inada granite blocks. The dimensions of the pre-existing fault surface were 100 cm (length) × 10cm (width). By applying heterogeneous normal stress using three actuators by which forces could be independently controlled, two asperities with different strengths were formed on a fault plane. Under a certain normal stress distribution, single asperity failure and double asperity failure occurred alternately. The single asperity failure refers to a stick slip event in which a weaker asperity ruptures alone, without triggering the rupture of a stronger asperity. The double asperity failure is an event in which a rupture of a weak asperity triggers the rupture of a strong asperity, resulting in a failure of the whole fault. For triggering of a rupture, it is necessary that stress at the strong asperity is accumulated to a certain level just prior to the failure of the weak asperity.

Origin of transient self‐potential signals associated with very long period seismic pulses observed during the 2000 activity of Miyakejima volcano

Origin of the previously reported transient geoelectrical (self-potential, SP) signals in the Miyakejima 2000 activity, that repeatedly occurred concurrently with very long period (VLP) seismic pulses, was investigated. SP waveforms stacked across repeated VLP events showed a step-like rise followed by a gradual decay at all stations spread over the island of 8 km diameter. Within a realistic range of hydrological diffusivity, the short time constants of the SP signals cannot be explained by the electrokinetic effect caused by fluid flow within a limited volume, proposed earlier as a fluid injection hypothesis. On the other hand, poroelasticity predicts an island-wide distributed flow field to occur almost instantaneously upon VLP events due to the step of strain field imposed by the mechanical event. We propose that the observed SP signals resulted from the streaming current by this island-wide flow field. Our quantitative model, assuming a vertical tensile crack as a mechanical source, which has been suggested by preceding seismic studies, can explain the time constants and the amplitude of the SP signals (both spatial pattern and absolute amplitude), within a reasonable range of rock properties and the scalar moment of the mechanical source (VLP event). Location and attitude of the mechanical source were well constrained by grid search and are consistent with those estimated earlier from other types of data.

Interpretation of various slip modes on a plate boundary based on laboratory and numerical experiments

This paper discusses various slip modes on a plate boundary on the basis of a two-degree-of-freedom block-spring model and large-scale biaxial experiments, including a new experimental result on afterslip. We conducted slip experiments using large granite blocks with a pre-existing fault surface of 100 cm in length. Velocity-strengthening friction was given over a half of the fault length by inserting a thin Teflon sheet, while the other half retained velocity-weakening friction of the bare rock surface. Under a loading at a constant velocity, dynamic stick-slip repeated on the velocity-weakening region, causing afterslip on the velocity-strengthening region. The velocity-strengthening region experienced small coseismic slip as well, with the magnitude decreasing with the distance from the velocity-weakening region. The behaviors observed in the laboratory experiments were quantitatively simulated by a two-degree-of-freedom block-spring model, in which two blocks (Block 1 and Block 2) are connected by a liner spring and driven by a slowly moving driver. The friction on each block was assumed to obey rate and state dependent friction law. When a - b was assumed to be negative for Block 1, and positive for Block 2, afterslip occurred at Block 2. This model can also reproduce wide spectrum of slip modes by adjusting frictional parameters.