The relationship between volcanic and seismic activities observed around Iwate volcano in 1998 are discussed. The first subject is the relation between the seismic activity and the ground deformation, which were simultaneously observed in the period from February to August, 1998. The seismic activity is a characterized by westward extension of the focal region in the western part of the volcano. The ground deformation is explained by moving sources of tensile faults and pressure sources located near the west end of the focal region at each twomonth period considered. The synchronism of these phenomena is explained by the idea that both of the activities are two different aspects of volcanic activity caused by the westward migration of magma in the western part of the volcano. The idea is confirmed by the examination using the temporal change of the Coulomb failure function (ΔCFF). The second subject is the relation between the volcanic activity mentioned above and a moderate earthquake (M6.1) of Sept. 3, 1998, that took place at 10 km SW from the summit of the volcano. The ΔCFF calculated from the source model for the volcanic ground deformation shows no evidence that the volcanic activity triggered the earthquake. On the other hand, temporal changes in seismicity in the volcanic area are explained as the effects of the stress changes caused by the earthquake. Nevertheless, some relations between the seismic activities and the volcanic activity are explained by changes in static stress, but there remain other phenomena that were observed at the same time and whose relationships are not yet understood.
A change of Coulomb Failure Stress (CFS) is commonly used to explain a mechanism of earthquake triggering due to both earthquake faulting and volcanic sources. First, we introduced crustal deformation and source model for five episodes of volcanic activities off Ito, central Japan and around the Mt. Iwate volcano, northern Japan based on dense geodetic measurements. And then, we calculated spatial distribution of a CFS change due to the estimated source models and compared the observed seismic activities. In the case of the 1998 earthquake swarm off Ito, the region where opening of dike and the largest earthquake (M5.7) increase CFS was concordant with the hypocenter distribution observed after the largest earthquake. However, the largest earthquake occurred in the area where CFS decreased. In the case of the Iwate volcano, a volcanic inflation source increased CFS by 0.33 MPa at the hypocenter of the M6.1 earthquake that occurred on September 3, 1998. The calculated CFS change is 7 % of the coseismic stress drop. The M6.1 earthquake occurred at the northern edge of the Nishine Fault found by geographical and geological studies. It suggests that the inflation promoted the rupture of a known Quaternary fault. Stress shadow caused by the M6.1 earthquake explains the observed quiescence of the seismic swarm activity west of E140.95° after the occurrence of the earthquake.
Great earthquakes, inland shallow earthquakes and volcanic eruptions occurred historically in and around the northern Honshu region, the northeastern part of the Japan island arc system. We studied statistically the time-space correlation among great offshore earthquakes with magnitudes (M) larger than 7.8 near the Japan trench, inland shallow earthquakes with M 6.2 and the volcanic eruptions, for the period from 1851 to 1996. We focus on the northern Honshu region in this study. During the period, there were four offshore great earthquakes near the Japan trench. Their magnitudes are M7.9 (1856), M8.2 (1896), M8.1 (1933), and M7.9 (1968). To estimate the temporal correlation among these events, we applied a statistical analysis that is one of the currently used methods in earthquake prediction research works. Our analysis suggests that alarm-like information could be issued for the inland volcanoes and the large inland earthquakes in the northern Honshu, at the occurrence time of the four great earthquakes near the Japan trench. We then found some simultaneous occurrence pattern between the large inland earthquakes and the volcanic eruptions, just prior to the four great earthquakes or just afterwards. As a result, the temporal correlation among the volcanic eruptions, the large inland shallow earthquakes and the four great earthquakes is not statistically negligible. This result suggests that the space-time relationship among the volcanic eruptions, the large inland shallow earthquakes and the great earthquakes near the trench is not only useful for researches on long-term earthquake prediction and volcanic eruption, but also important for understanding seismotectonics in the northern Honshu region.
The temporal relationship between the interplate and inland earthquakes of Kyushu Island region was investigated. We analyzed this relationship by stacking the temporal frequency of the inland earthquakes with reference to the occurrence time of each major interplate earthquake in the Hyuga-nada region, that occurred between 1900 and 2000. A good coincidence of occurrence between the inland and the interplate earthquakes is recognized. While the tendency of the occurrence after the interplate events has already been pointed out, we found that the inland earthquakes also tend to occur before the interplate events. Abrupt activation of the inland earthquakes within several months before major interplate earthquakes is recognized in addition to the tendency to occurr afterwards. The preceding inland earthquakes may indicate that they are triggered by precursory slow slips around the hypocenters of the Hyuga-nada earthquakes.
We introduce a criterion for an active period of seismicity around the source region of a great interplate earthquake. It is based on stress change caused by the great earthquake. We test this criterion by applying it to the seismicity in southwest Japan before and after the 1944 Tonankai and 1946 Nankai earthquakes. We then apply it to more recent seismicity in southwest Japan and in the San Francisco Bay area. The results show that the seismicity in southwest Japan is now in transition from a quiet to active period and that the San Francisco Bay area is still in a quiet period. In other words, the probability of large earthquakes in southwest Japan is higher than it was in the preceding quiet period. Although the seismicity in the San Francisco Bay area was active in the 1980's, the probability of a large earthquake occurring is low.
We investigated statistical characteristics in a space-time distribution of large shallow inland earthquakes in Japan during the period from 1600 through 2000 using data from the following three catalogues : Chronological Scientific Tables (National Astronomical Observatory, 2001) for 1600-1884, Utsu (1979, 1982, 1985) for 1885-1925 and the catalogue of the Japan Meteorological Agency for 1926-2000. It is shown that an alternation of active and quiet periods is commonly observed for each district : inland seismicity tends to be active in the periods before and after great interplate earthquakes. Further, we show that another large earthquake is likely to occur after one during the period of 10-20 years within a distance of several tens of kilometers. We think this feature of the successiveness of large earthquakes is significant and we can utilize it for assessing seismic hazards.
We show that conspicuous seismic quiescence occurred in large areas along the coast of the Japan Sea before the 1891 Nobi, 1964 Niigata, 1983 Central Japan Sea, and 2000 Western Tottori Prefecture Earthquake. If we had noticed the quiescence that had appeared since the late 1980s along the northern coasts of Shimane, Tottori, and Hyogo Prefectures and had remembered the feature that a large earthquake in the coastal region of the Japan Sea is likely to be preceded by a widely extended seismic quiescence along the coast, we could have paid close attention to the focal region of the Western Tottori Prefecture Earthquake. We think the occurrence of seismic quiescence in a large area before a large earthquake implies that the preparatory process proceeds not only in the focal region, but in the circumferential areas as well.
Previous studies have reported many examples of possible mechanical coupling between volcanic unrests and large earthquakes, which occurred around the volcano. This paper reviews these studies and reorganizes the types of mechanical coupling into the following five cases (cases A-1, A-2, B-1, B-2, and C) and 10 mechanisms : In case A-1, in which a large earthquake triggers activation of a volcano, the following four mechanisms can explain their coupling : (A-1-1) an increase in compressional stress, which was produced by earthquake source faulting, squeezes magma up to the surface; (A-1-2) an increase in differential stress (or magma pressure), which was produced by earthquake source faulting, promotes dike intrusion; (A-1-3) increase in tensional stress, which was produced by earthquake source faulting, triggers gas bubbling in magma; and, (A-1-4) dynamic stress change, which was associated with seismic wave, triggers gas bubbling in magma. In case A-2, in which a large earthquake triggers deactivation of a volcano, the following three mechanisms can explain their coupling : (A-2-1) increase in compressional stress, which was produced by earthquake source faulting, chokes a vent or prevents gass bubbling in magma; (A-2-2) decrease in differential stress (or magma pressure), which was produced by earthquake source faulting, prevents dike intrusion; and, (A-2-3) increase in tensional stress, which was produced by earthquake source faulting, drains magma back toward a chamber. In case B-1, in which a volcanic unrest triggers a large earthquake, coupling can be explained by mechanism B-1-1 : change in stress, which was produced by dike intrusion (or pressure change in a magma chamber), promotes earthquake source faulting. In case B-2, in which a volcanic unrest prevents a large earthquake, coupling can be explained by mechanism B-2-1 : change in stress, which was produced by dike intrusion (or pressure change in a magma chamber), prevents earthquake source faulting. In case C, in which a change in plate motion causes a stress change and then triggers (or prevents) a large earthquake (or a volcanic unrest), coupling can be explained by mechanism C : stress change promotes (or prevents) earthquake source faulting or ascending / intrusion of magma.
Traditionally, long-term earthquake probability has been estimated using regional seismicity or a characteristic earthquake hypothesis without considering any recent stress perturbations caused by sudden crustal deformation or nearby earthquakes. The conventional calculation method ignores such processes observed time dependent clustering of earthquakes, and the occurrence of aftershocks or anti-shocks (seismicity rate decreases). I have thus introduced a method to seek the time-dependent seismicity rate based on stress interaction, incorporating the rate-and state-dependent friction law. Regarding earthquake productivity response to the stress state, coseismic stress step controls amplification of seismicity rate increase. In addition, assuming that the constitutive parameter and normal stress are constant throughout time, the aftershock or anti-shock duration is inversely proportional to the regional tectonic stressing rate. Thus, influence of the stress step lasts longer where the loading rate is slow, and the long-term probability retains the stress-related change longer. In addition, change in the loading rate proportionally causes the change in the earthquake productivity, in other words earthquake probability, with some delay estimated by a new loading rate. The response time to the sudden loading rate change depends on a new loading rate as predicted by the rate-and state-friction theory. To make further validations about this theory and method, I have investigated two cases of stress triggering, 2000 seismic swarm activity in and around the Izu Islands, and the October 6, 2000, M7.3 Tottori-ken-seibu earthquake. In the Izu swarm activity, the factor of the stressing rate changes was focused. Observed hundreds-to thousands-fold increases of seismicity rate than usual during the active two-month period are almost equivalent to the increases in the tectonic stressing rate caused by a dike intrusion. In contrast, long aftershock duration and long influence of the stress perturbation associated with a low stressing rate were tested in the Tottori case. Even though we need to incorporate additional factors such as viscoelastic behavior and change in earthquake size distribution, this method is expected to contribute more precise long-term earthquake probabilistic forecasting together with welldetermined data from recent GPS and seismic networks.
In the Tamba Plateau, an earthquake swarm area in the Kinki district, Central Japan, seismicity was activated just after the Hyogo-ken Nanbu (Kobe) Earthquake (M7.3), which occurred in an adjacent area in 1995. We found that micro-earthquake activities in the Tamba Plateau corresponded to moon phase. Occurrences of micro-earthquakes increased after a new moon and a full moon during 1995 and 1996. Before 1995, such a correlation could not be found. The present study suggests a possibility that the stress change caused by the Hyogoken Nanbu Earthquake made seismicity in the Tamba Plateau sensitive to tidal forces.
We conducted a numerical simulation to clarify the effects of the earth tide on earthquake occurrence. In the simulation, fault planes, having different initial stresses, are loaded by constantly increasing tectonic stress and cyclic stress due to the earth tide. Earthquakes are assumed to occur when the shear stress reached a certain threshold level. The result of the simulation indicates that the tidal effect on earthquake occurrence is strongly controlled by the ratio of stress change rate between earth tide and tectonic stress accumulation; earthquakes concentrate near phase angle 0° (maximum tidal stress) when the ratio is large, and they are distributed with a peak around-90° (maximum acceleration of tidal stress) when the ratio is small. This phase selectivity is very similar to the observations of tidetriggered earthquakes reported so far, and rock failure experiments under cyclic loading, suggesting that our approach may provide an important clue for clarifying the physical mechanism of tidal triggerings of earthquakes. Schuster's test has been widely used for detecting tidal effect on earthquake occurrence. However, it is shown by the numerical simulation that the result of a test strongly depends on the size of a data set when earthquakes have phase selectivity by nature, and that Schuster's p-value is not appropriate to represent the strength of tidal effects. Alternatively, we propose to use α, which is the amplitude of a sine curve fitted to the frequency distribution of earthquakes against phase angle, to evaluate the strength of tidal effects. We also emphasize that the effect of ocean loading is an important component of the earth tide, and cannot be neglected in a study of tidal triggerings of earthquakes.
This paper briefly reviews the triggering characteristics of injection-induced seismicity. Water injection experiments were carried out in the Nojima fault, southwest Japan in 1997 and 2000 to detect the healing process of the fault zone after being ruptured by the MJMA 7.3 Hyogo-ken Nanbu (Kobe) earthquake in 1995. During the experiment in 2000, ultramicroearthquakes of M-1.2 to 1.0 were induced at about 2.5-4.5 km from the injection point and about 4-6 days after the beginning of injection. This space-time migration can be explained by a 2-D diffusion process of pore water pressure. Permeability estimated near the surface, at about 540-800 m depth, is extrapolated well to a depth of 2-4 km where induced events occurred. Other experiments at Matsushiro, central Japan and KTB, Germany also showed similar space-time relationships of induced seismicity. From observations in the Nojima experiment, we obtained characteristic states that suggest differences in the generating process between induced and normal (tectonic) earthquakes. Our findings are as follows : (1) high-frequency component is richer in the waveforms of tectonic events, and (2) the clustering of hypocenters is more dominant in induced events. Further analyses of these observations will lead to elucidating the generating process of induced earthquakes by increasing pore water pressure.
Continuous seismic observations at Iwo-jima, an active volcanic island belonging to the Izu-Ogasawara island arc, have detected remote triggering of microearthquakes in and around the island. The remote triggering at Iwo-jima is a phenomenon of an abrupt increase of microearthquake activity at the time of a passage of seismic waves from a distant large earthquake. We examined seismograms of a total of 21 earthquakes with magnitude larger than 7 and within an epicenter distance of 3000 km from Iwo-jima. Remote triggering phenomena were found at four events during the period from 1980 to 1993 : the 1983 west off Tohoku earthquake, the 1984 southeast off Kyushu earthquake, the 1993 southeast off Hokkaido earthquake, and the 1993 Mariana Island earthquake. The largest epicenter distance among them was 2009 km. The initial times of triggering coincide with the theoretical arrival times of surface waves and successive occurrences of earthquakes continued for 6 to 15 min, suggesting that dynamic stress or strain caused the remote triggering phenomena at Iwo-jima. As a well-developed hydrothermal system is suggested in shallow depths beneath Iwo-jima, volcanic fluids presumably play an important role in remote triggering.
Numerical simulations have revealed that a rupture jumps from one fault to another near the Earth's surface. We show that a rupture jump occurs at a deep portion under depthdependent stress. We investigated rupture behaviour at a gap region between two faults, by calculating spontaneous rupture processes on two parallel strike-slip faults in a 3-D elastic medium. Depth-dependent stresses introduced from the results of hydraulic fracturing experiments at the KTB site are assumed. The differential stress is assumed to be proportional to depth. The rupture processes under depth-dependent and uniform stress conditions are compared : similar rupture processes on the first fault, but different ones on the second fault are obtained. Under the depth-dependent stress condition, an upward rupture on the first fault can trigger a rupture at a deep portion of the second fault, because strength is reduced at shallower depths. Near the Earth's surface, a rupture on the second fault can be triggered by only a P-wave radiated from the first fault. Strength and stress drop heterogeneities on faults cause a rupture jump at a deep portion and the complexity of the rupture process. Our results suggest that the stress increase with depth is an important factor.
We attempt to simulate the activity of earthquakes of moderate to large sizes in and around the Japanese islands using a block and fault model, in which slip deficit rates were derived from triangulation and trilateration data covering one-hundred years. The original block and fault model has 104 faults. However, the original fault size is too large to simulate moderate earthquakes. Therefore, we divide each modeled fault into 5 × 5 elements to generate earthquakes as small as M5. In total there are 2600 elements. We assume stress accumulates according to the estimated slip deficit rates. Interaction between fault elements is represented by changes in Coulomb Failure Function (hereafter Δ CFF) induced by the movement of other faults that are determined by the geometrical relationship and the direction of slip deficit rates. When stress reaches the threshold level, accumulated CFF is released by a forward slip and redistributed to surrounding faults according to the CFF changes calculated above. If the redistribution of CFF induces the next rupture, the same process is repeated until there are no more rupturing elements. We assume rupture thresholds to be 2.5 MPa for interplate boundaries and 10 MPa for inland faults, respectively. We simulate seismicity for 10000 years with a time step of 1 year. In this simulation large events that rupture almost all elements of a fault rarely occur. This suggests that strain rates derived from geological data or historical earthquake catalog might be underestimated. Simulated seismicity does not satisfactorily fit the GutenbergRichter's law, because moderate events occur more frequently than small or large events. This suggests that we have to incorporate heterogeneity in the rupture threshold or the size of elements on a fault plane. The correlation between interplate earthquakes along the Nankai trough and inland events in southwest Japan is not clear, but there seems to be a complementary relationship in activities between both regions. Migration of large events along the Nankai trough is occasionally seen in this simulation, but its direction is different from time to time.
I investigated the generation mechanism of long range correlations of large earthquake occurrences that have been reported by many seismologists. I present a simple, one dimensional, stick-slip model based on Hashimoto's earthquake model (1998) which shows a phenomenon of unusual long-range correlations. This model consists of many blocks in contact with rough surface of a plate and assumes non-local or higher order couplings with other blocks, unlike conventional models. Simulations show that small-scale or non-local periodic patterns appear from time, to time, and these patterns interfere with one another, which results in coincidental occurrences of earthquakes in a quite wide region. The results show that long-range correlations are produced from this model, which may account for actual observations.