Kin'ya Nishigami

Induced earthquakes accompanying the water injection experiment at the Nojima fault zone, Japan: Seismicity and its migration

The 1995 Hyogo-ken Nanbu (Kobe) earthquake of M7.2 occurred on January 17, 1995. After the earthquake, a scientific drilling program called the Nojima Fault Zone Probe was carried out at the Nojima fault which ruptured during the mainshock. Water was injected during two periods, February 9–13 and March 16–25, 1997. The pumping pressure at the surface was about 4 MPa. Pressurized water was injected into a 1800-m-deep borehole and supplied to the surrounding rock at depths between 1480 and 1670 m. The total amount of injected water was 258 m3. An increase in earthquake activity was observed 4 or 5 days after the beginning of each water injection. The seismicity increased in the region around 3 or 4 km from the injection point. This suggests that these earthquakes are likely to have been induced by the water injections. The induced earthquakes were located between 2 and 4 km in depth and had magnitudes ranging from −2 to +1. The hypocenters of the induced earthquakes migrated with speeds of ∼2–40 m/h. The speeds decreased with time, suggesting a relationship with the diffusion of water. Values of intrinsic permeability of 10−14–10−15 m2 were obtained from the time dependency of induced seismicity change. The coefficient of friction in the area where the induced earthquakes occurred was estimated to be less than 0.3. Twenty earthquake clusters were found, and cross-spectrum analysis was applied to them. We could distinguish between the induced and noninduced earthquakes from the analysis. The induced earthquakes forming each cluster migrated with speeds of 20–80 m/h, which means microscale water migration or permeation. The present water injection experiment revealed that the Nojima fault zone was highly permeable and could slip with small (∼10% or less) increases in pore fluid pressure or shear stress.

A new inversion method of Coda waveforms to determine spatial distribution of Coda scatterers in the crust and uppermost mantle

We have developed a new inversion method of coda waveforms from local earthquakes to estimate the inhomogeneous spatial distribution of scattering coefficient in the crust and the uppermost mantle. A preliminary analysis by this method was applied to the data observed at the Hokuriku district, central Japan. Inhomogeneous structures have been revealed in both horizontal and vertical directions as follows: (1) Strong scatterers are distributed in some localized parts of the surface layer and the upper crust; (2) Some of the strong scatterers seem to be related to major active faults; (3) Horizontal variation of scattering coefficient is smaller in the uppermost mantle than in the crust. We have found that the method presented here is effective enough to investigate deterministically the inhomogeneity in the shallow part of the crust.

Deep crustal heterogeneity along and around the San Andreas fault system in central California and its relation to the segmentation

The three-dimensional distribution of scatterers in the crust along and around the San Andreas fault system in central California is estimated using an inversion analysis of coda envelopes from local earthquakes. I analyzed 3801 wave traces from 157 events recorded at 140 stations of the Northern California Seismic Network. The resulting scatterer distribution shows a correlation with the San Gregorio, San Andreas, Hayward, and Calaveras faults. These faults seem to be almost vertical from the surface to ∼15 km depth. Some of the other scatterers are estimated to be at shallow depths, 0–5 km, below the Diablo Range, and these may be interpreted as being generated by topographic roughness. The depth distribution of scatterers shows relatively stronger scattering in the lower crust, at ∼15–25 km depth, especially between the San Andreas fault and the Hayward-Calaveras faults. This suggests a subhorizontal detachment structure connecting these two faults in the lower crust. Several clusters of scatterers are located along the San Andreas fault at intervals of ∼20–30 km from south of San Francisco to the intersection with the Calaveras fault. This part of the San Andreas fault appears to consist of partially locked segments, also ∼20–30 km long, which rupture during M6–7 events, and segment boundaries characterized by stronger scattering and stationary microseismicity. The segment boundaries delineated by the present analysis correspond with those estimated from the slip distribution of the great 1906 San Francisco earthquake, and from the fault geometry as reported by the Working Group on California Earthquake Probabilities [1990], although the segment boundaries along the San Andreas fault in and around the San Francisco Bay area are still uncertain.

Spatial distribution of coda scatterers in the crust around two active volcanoes and one active fault system in central Japan: Inversion analysis of coda envelope

Abstract Coda waves observed by local seismographic networks were analysed by using the inversion method presented by Nishigami (1991) to estimate spatial distribution of coda scatterers in the crust. First, deviations of the decay curve of observed coda envelope away from a reference curve were measured for lapse times greater than about 1.5–2.0 times the S-wave travel times. We assume this kind of deviation is mainly caused by the non-uniform distribution of scatterers in the crust. Next, the 3-D distribution of relative scattering strength in the crust was obtained by solving the observational equation, where single isotropic scattering and a spherical radiation from the source are assumed and the observational data is given by the deviation of coda envelope stated above. This method was applied to two volcanic regions and one active-fault region in the central part of Japan. In each analysis, the target region with dimension of 50–160 km in horizontal and 30–80 km in depth was divided into about 1500–2000 blocks with one side of 2–10 km, and 600–900 waveform traces from about 100–250 events observed at 7–38 stations were used. The lapse time of coda waves analysed was taken to be as short as possible to reduce the multiple scattering effect. It was, in the shortest case, about 3–13 s for the Ontake region. The result of inversion analyses showed several significant features. Strong scatterers were found at shallow part (0–20 km) of the crust around the Fukui earthquake fault. A low seismicity area between active faults showed a correlation with the weak scattering area. The detailed distribution of scatterers was also revealed around two active volcanoes, Mt. Ontake and Mt. Nikko-Shirane. The vertical distribution of strong scatterers just below the volcano from 0 to 7 km in depth and another horizontal one at a depth of 7 km were estimated in the Ontake region. An inclined distribution of scatterers was estimated at depths from 0 to 20 km just below Mt. Nikko-Shirane. The location of scatterers around these active volcanoes agreed well with that of the S-wave reflectors in the crust obtained by the NMO correction analysis. Frequency dependence of scatterer distribution was also examined for the Ontake region. The horizontal distribution of scatterers at 7 km depth showed a similar pattern for frequencies between 5 and 20 Hz, while other smaller-scale scatterers appeared to have frequency dependence.

Monitoring of fault healing after the 1999 Kocaeli, Turkey, earthquake

The North Anatolian fault zone that ruptured during the mainshock of theM 7.4 Kocaeli (Izmit) earthquake of 17 August 1999 has beenmonitored using S wave splitting, in order to test a hypothesisproposed by Tadokoro et al. (1999). This idea is based on the observationof the M 7.2 1995 Hyogo-ken Nanbu (Kobe) earthquake, Japan.After the Hyogo-ken Nanbu earthquake, a temporal change was detectedin the direction of faster shear wave polarization in 2–3 years after the mainshock (Tadokoro, 1999). Four seismic stations were installed within andnear the fault zone at Kizanlik where the fault offset was 1.5 m, about80 km to the east of the epicenter of the Kocaeli earthquake. Theobservation period was from August 30 to October 27, 1999. Preliminaryresult shows that the average directions of faster shear wave polarization attwo stations were roughly parallel to the fault strike. We expect that thedirection of faster shear wave polarization will change to the same directionas the regional tectonic stress reflecting fault healing process. We havealready carried out a repeated aftershock observation at the same site in2000 for monitoring the fault healing process.

Microseismic activity around the western extension of the 1967 Mudurnu Earthquake fault

A temporary microseismic network was installed with nine stations around the western extension of the 1967 Mudurnu Earthquake fault. The area studied is a part of one of the seismic gaps of the first kind along the North Anatolian Fault Zone. The fault zone branches into two fault zone in the area. It is found that the hypocentres are concentrated on the northern branch. It seems that the seismicity close to the western end of the 1967 Mudurnu Earthquake fault is dominated by normal faulting.

Chapter 11 Imaging Inhomogeneous Structures in the Earth by Coda Envelope Inversion and Seismic Array Observation

Abstract In this chapter, we introduce two kinds of deterministic analyses of coda waves, that is, inversion analyses of coda envelopes and seismic array observations, and we show several studies that effectively estimate the inhomogeneous structures in the crust and uppermost mantle. The first one analyzes wave data obtained by local or regional seismographic networks. Nishigami (1991) presented an inversion analysis of coda waves from local earthquakes, to estimate 3‐D distribution of relative scattering coefficients. The deviation of coda envelopes from average decay curves is measured as the observational data, assuming a single isotropic scattering model. This method was applied to central California and the inhomogeneous structure around the San Andreas fault system was revealed (Nishigami, 2000). Asano and Hasegawa (2004) revised this method to estimate the absolute scattering coefficients. Revenaugh (1995a) proposed another analysis method, called Kirchhoff coda migration, in which the forward‐scattered energy in teleseismic P coda observed by a regional seismographic network is stacked. The second approach is seismic array observation with station spacing shorter than the wavelength of seismic waves. We first summarize several analysis methods of seismic waves propagating through the array. For example, scattered waves with weak energy can be detected by beam‐forming techniques. Coda waves are also decomposed into wave trains with various ray directions using analyses such as multiple signal classification or semblance coefficients. The energy of scattered waves in the coda can be evaluated by processing the slant‐stacked waveforms under the assumption of a single‐scattering model. For example, Matsumoto et al . (1998) applied this method to the source area of the 1995 Kobe earthquake (M7.3), and revealed the existence of strong scatterers just beneath the hypocenter of the mainshock. These studies analyzing the seismic network or array observation data seem to be effective to estimate the Earths inhomogeneous structures.