The Tokachi-Oki earthquake occurred on September 26, 2003. Precise aftershock distribution is important to understand the mechanism of this earthquake generation. To study the aftershock activity, we deployed fortyseven ocean bottom seismometers (OBSs) and two ocean bottom pressure meters (OBPs) at thirty-eight sites in the source region. We started the OBS observation four days after the mainshock for an observation period of approximately two months. In the middle of the observation period, nine OBSs near the epicenter of the mainshock were recovered to clarify the depth distribution of aftershocks near the mainshock. From the data overall OBS, seventy-four aftershocks were located with high spatial resolution. Most of the aftershocks were located in a depth range of 15-20 km and occurred within the subducting oceanic crust, the 5.5-km/s layer of the landward plate and the plate boundary. No aftershocks were found in the mantle of the subducting plate. The low seismic activity beneath the trench area where the water depth is greater than about 2000 m suggests a weak coupling between the two plates. The depth of the mainshock is inferred to be 15-20 km from the aftershock distribution.

View full abstract

The 2003 Tokachi-oki earthquake with Mw 7.9 is the largest interplate earthquake occurred ever since the high dense broadband seismometer network, the National Research Institute for Earth Science and Disaster Prevention (NIED) F-net, has been established over Japan. We determine the spatial distribution of moment tensor solutions and centroid depths of the mainshock and aftershocks. All aftershocks are divided to three groups: (1) the thrust fault type whose nodal plane is similar to the main shock; (2) the other thrust type with nodal plane different from the main shock; and (3) the normal fault type. The type (1) shows a depth distribution inclined to NW gently, coincident to the upper boundary of descending Pacific Plate. The active area of the type (1) does not overlap with the co-seismic slip area of the main shock at all. On the other hand, the type (2) shows no characteristic depth distribution with centroid depth scattered above and beneath the upper plate boundary. The type (3) are distributed, mainly, at about 40 km depth above the upper plate boundary. P axes of some aftershocks occurred above the plate boundary show the direction from ENE-WSW to ESE-WNW that suggests the effect of the Hidaka collision.

View full abstract

We performed waveform inversion of long period body wave data to investigate whether it is possible to determine the moment release distribution in time and space, using the 2003 Tokachi-Oki earthquake as an example. Our results show that it is difficult to determine the precise spatial distribution, and therefore we focused on determination of the temporal distribution. We allowed changes of the source mechanism during rupture in the inversion scheme. The temporal change of the scalar moment of the subevents was consistent with the moment rate function inferred from other studies. The source mechanisms of the subevents were primarily reverse faults, matching preliminary reports of this earthquake. No significant change of the source mechanism during the rupture were inferred. Our results suggest that the rupture process of this earthquake was rather simple.

View full abstract

The spatio-temporal slip distribution of the 2003 Tokachi-oki, Japan, earthquake was estimated from teleseismic body wave and strong ground motion data. To perform stable inversion, we applied smoothing constraints to the slip distribution with respect to time and space, and determined the optimal weights of constraints using an optimized Akaike's Bayesian Information Criterion (ABIC). We found that the rupture propagates mainly along the dip direction, and the length of the rupture area is shorter than its width. The mean rise time in the shallow asperity is significantly longer than that in the deep asperity, which might be attributed to variable frictional properties or lower strength of the plate interface at shallower depths. The average rupture velocity of deep asperity extends to the shear-wave velocity. The derived source parameters are as follows: seismic moment Mo = 1.7×1021 Nm (Mw 8.0); source duration = 50 sec. We also estimated the shear stress change due to the mainshock on and around the major fault zone. It appears that many aftershocks on the plate boundary took place in and adjacent to the zones of stress increase due to the rupture of the mainshock.

View full abstract

A great earthquake, named the 2003 Tokachi-oki earthquake, occurred in the southern Kuril subduction zone on 26th September 2003, 4:50 JST (41.7797°N, 144.0795°E, 42 km depth; Japan Meteorological Agency). Its ground motion was recoreded at 655 stations of the nationwide strong motion networks, K-NET and KiK-net. A maximum peak ground acceleration of 988 cm/s2 (gal) was observed at station HKD100 and amplitudes greater than 200 cm/s2 were observed over a wide area of eastern Hokkaido. We used a multi-line linear waveform inversion method to estimate the rupture process from the strong motion data of supplied by 15 stations. We assumed a fault plane model of 140 km × 160 km with strike and dip angles of our fault model are N246°E and 18°, respectively, placed on the estimated upper boundary of the subducting Pacific Plate. The estimated total slip distribution consisted of three major slip areas; (a) around the hypocenter, (b) the northwest part of the fault with the maximum slip of 5.9 m, and (c) the northeast edge of the fault plane. The major asperity (b) was composed of two large slip areas with different slip rate functions: the duration of moment release in the sourtheast part is longer than 15 sec, but in contrast most of the seismic moment of the northwest part was released in a short period of less than 10 sec. Our estimation of the total seismic moment was 2.9 × 1021 N·m which corresponded to Mw = 8.2.

View full abstract

A preliminary source model composed of asperities for the 2003 Tokachi-oki, Japan, earthquake (MJMA = 8.0) was estimated by the empirical Green's function method. The source parameters for three asperities located on the fault plane were determined from the comparisons of the synthesized broad-band ground motions with the observed ones. We found that the pulsive waveforms observed in north direction of the hypocenter were generated by the forward rupture directivity effect. Furthermore, the estimates of the stress parameter for asperities are higher than the averaged ones for past inland and subduction earthquakes.

View full abstract

The strong motion and geodetic data were individually inverted for the source process of the 2003 Tokachi-oki, Hokkaido, earthquake with a hypocenter 25 km deep and a fault plane above the subducting Pacific slab. Both the results show a simple slip distribution with a single major asperity, but the strong motion inversion may include a trade-off between slip location and rupture time and the geodetic inversion does not have sufficient resolution for far slips. We then carred out a joint inversion of the two datasets in order to overcome these weaknesses of the single dataset inversions. The resultant slip distribution still retains the simple pattern and has a seismic moment of 2.2 × 1021 N·m (Mw 8.2). The asperity, with a peak slip of 7.1 m, is located in the center of the fault plane 50 km away from the hypocenter in the down-dip direction. The slip rate functions on subfaults around the hypocenter and asperity indicate that the rupture propagated with a supershear speed on the upper part of the fault plane and slowed down to 100-90% of the S-wave velocity on the middle and lower parts. These simple slip patterns and near-supershear rupture may imply the maturity of the Hokkaido subduction zone around the source region.

View full abstract

Observation of aftershocks of the 2003 Tokachi-Oki earthquake was conducted in the southern part of the Tokachi basin in Hokkaido, Japan for estimation of local site effects. We installed accelerographs at 12 sites in Chokubetsu, Toyokoro, and Taiki areas, where large strong motion records were obtained during the main shock at stations of the K-NET and KiK-net. The stations of the aftershock observation are situated with different geological conditions and some of the sites were installed on Pleistocene layers as reference sites. The site amplifications are investigated using spectral ratio of S-waves from the aftershocks. The S-wave amplification factor is dominant at a period of about 1 second at the site near the KiK-net site in Toyokoro. This amplification fits well with calculated 1D amplification of S-wave in alluvial layers with a thickness of 50 meters. In addition to the site effects, we detected nonlinear amplification of the soft soils only during the main shock. The site effects at the strong motion site of the K-NET at Chokubetsu have a dominate peak at a period of 0.4 seconds. This amplification is due to soft soils having a thickness of about 13 meters. Contrary to the results at the two areas, site effects are not significantly different at the stations in the Taiki area, because of similarity on surface geological conditions.

View full abstract

We relocated hypocenters of the March 4, 1952 Tokachi-oki (Mj 8.2) earthquake and its aftershocks by reassessing data of the Central Meteorological Station network at the time and found that the distribution of relocated aftershocks is clearly bounded on its eastern extension by the Kushiro canyon, which extends south-east from the coast of Hokkaido near Kushiro to the Kuril trench. The reevaluated aftershock pattern is quite similar to the pattern of the September 26, 2003 Tokachi-oki earthquake (Mj 8.0). Detailed seismic intensity maps based on the report of human perceptions in 1952 and the seismic intensity meter network in 2003 also show resemblance. Similar aftershock pattern and seismic intensity distribution imply that both earthquakes are a pair of characteristic earthquakes of the same size sharing the same source area and the same focal process.

View full abstract

During the sequence of aftershock activity of the 2003 Tokachi-Oki earthquake, anomalous events predominant in low-frequency components occurred very close to a subducted seamount around the junction between the Kuril Trench and the Japan Trench. The low-frequency wave train in the coda part reflects the strong excitation of surface waves and the source depth is considered to be very shallow. The focal mechanism of the low-frequency event is a type of the reverse fault. Considering with the source depth and the mechanism, the low-frequency event might have occurred on the plate boundary as a dynamic process of the tectonic erosion by the subducted seamount.

View full abstract

The 2003 Tokachi-oki earthquake of M 8.0, which occurred off the south-east coast of Hokkaido, Japan, was preceded by noticeable deep seismicity in the subducting slab, including a deep-focus earthquake of M 7.1. The 1952 Tokachi-oki earthquake of M 8.2 and the 1993 Kushiro-oki earthquake of M 7.5, which occurred also off the south-east of Hokkaido along the Kurile Trench, were also preceded by high deep seismicity including large deepfocus earthquakes of the M 7 or over (focal depths 200-600 km). The relations between the deep seismicity and the ‘shallow’ large earthquake (focal depth ≤ 100 km) in these three cases are quite similar. This result strongly supports the author's view (Mogi, 1973, 1988) that large ‘shallow’ earthquakes along the subduction zones are sometimes preceded by the occurrence of large deep-focus earthquakes.

View full abstract

Tsunami height survey was conducted immediately after the 2003 Tokachi-oki earthquake. Results of the survey show that the largest tsunami height was 4 m to the east of Cape Erimo, around Bansei-onsen, and locally at Mabiro. The results also show that the tsunami height distribution of the 2003 Tokachi-oki earthquake is clearly different from that of the 1952 Tokachi-oki earthquake, suggesting the different source areas of the 1952 and 2003 Tokachioki earthquakes. Numerical simulation of tsunami is carried out using the slip distribution estimated by Yamanaka and Kikuchi (2003). The overall pattern of the observed tsunami height distribution along the coast is explained by the computed ones although the observed tsunami heights are slightly smaller. Large later phase observed at the tide gauge in Urakawa is the edge wave propagating from Cape Erimo along the west coast of the Hidaka area.

View full abstract

We estimate the tsunami source area of the 2003 Tokachi-oki earthquake (Mw 8.0) from observed tsunami travel times at 17 Japanese tide gauge stations. The estimated tsunami source area (- 1.4 × 104 km2) coincides with the western-half of the ocean-bottom deformation area (- 2.52 × 104 km2) of the 1952 Tokachi-oki earthquake (Mw 8.1), previously inferred from tsunami waveform inversion. This suggests that the 2003 event ruptured only the western-half of the 1952 rupture extent. Geographical distribution of the maximum tsunami heights in 2003 differs significantly from that of the 1952 tsunami, supporting this hypothesis. Analysis of first-peak tsunami travel times indicates that a major uplift of the ocean-bottom occurred approximately 30 km to the NNW of the mainshock epicenter, just above a major asperity inferred from seismic waveform inversion.

View full abstract

The slip distribution of the 2003 Tokachi-oki earthquake is estimated from the 11 tsunami waveforms recorded at 9 tide gauges in the southern Hokkaido and eastern Tohoku coasts and two ocean bottom tsunami-meters (pressure gauges) off Kamaishi, Tohoku. The largest slip of 4.3 m is estimated on the subfault located off Hiroo. A large slip of 2.1 m is also estimated on the subfault located near Kushiro. The total seismic moment of the 2003 Tokachi-oki earthquake is 1.0 × 1021 Nm. The slip distribution estimated from the tsunami waveform inversion is similar to the slip distribution deduced by Yamanaka and Kikuchi (2003) from the inversion of the teleseismic body waves. The rupture area of the 2003 Tokachi-oki earthquake is similar to the western part of the rupture area of the 1952 Tokachi-oki earthquake estimated by Hirata et al. (2003).

View full abstract

To investigate the postseismic crustal deformation associated with the Tokachi-oki earthquake (MJMA = 8.0) of 26 September 2003 in Japan Standard Time (JST), off southeastern Hokkaido, Japan, we newly established thirty GPS sites just after the mainshock in the eastern part of Hokkaido. Rapid data analysis for one month after the mainshock clearly indicated postseismic displacements only in the horizontal components. Observed maximum horizontal displacement was 6.6 cm from 28 September to 24 October, 2003. Absence of the vertical suggests that afterslip occurred in and around the coseismic fault rather than at downdip extension. Time series of coordinates are characterized by logarithmic decay functions with 4-11 days relaxation times. This suggests that postseismic deformation was due to afterslip on the fault following the large earthquake.

View full abstract

The Mw 8.1 earthquake occurred on September 25, 2003, off the southeast coast of Hokkaido, Japan. Since 2000 we have conducted high-rate sampling GPS measurements and precise gravity surveys in Erimo Peninsula, the closest site to the source region of the 2003 event. Strong ground motion recorded by GPS at the point of Erimo Peninsula, located just above the second asperity of the earthquake, shows two major pulses as large as about 56 cm on the EW component. Displacements obtained from the integration of accelerograms very close to our GPS site are consistent with each other, showing the absolute displacement field generated by the magnitude 8-class earthquake. Synthetic seismogram from a similar fault model by Yamanaka and Kikuchi (2003) would predict the amplitude of the second pulse to be about one half of that observed. Synthetic NS component from the GSI fault model (2003) is not consistent with our observations both on amplitude and polarity. The amplitude of ground motions detected by our GPS observation is more than one order larger than the noise level of the GPS survey, so this discrepancy is not due to insufficient GPS observation. We rather think that this suggests that our observations closest to the earthquake would give an insight into the detail of the source processes of the earthquake, which cannot be resolved from observations away from the source region. Static deformation at the point of Erimo Peninsula is consistent with the GSI fault model but not with the Yamanaka and Kikuchi model. The static analysis of our GPS measurement evidently describes the continuous post-seismic deformation as well as the co-seismic displacement in the source region until November.

View full abstract

We analyzed 1-Hz GPS data observed at 14 stations of the GPS Earth Observation Network (GEONET) of the Geographical Survey Institute, Japan, associated with the 2003 MJMA8.1 Tokachi-oki earthquake, which occurred at the Kurile Trench. The GPS stations are located 70-240 km away from the epicenter. GPS data clearly captured rapid co-seismic ground displacements. At a GPS station 70 km away from the epicenter, coseismic displacements started 15 seconds after the origin time, and after 40 seconds at the stations 240 km away. Observed displacement amplitude exceeded 20 cm at GPS sites 240 km away from the epicenter. Displacement amplitudes attenuate with time and distance from the epicenter, oscillating with periods of 40-60 seconds. We compared the 1-Hz GPS data and displacement seismogram integrated from strong ground motion data, which showed fairly good agreements. In spite of careful screening of 1-Hz GPS data during 30 minutes preceding the main shock, no significant preseismic deformation over 1 cm in the horizontal components was recorded. 30 second sampling GPS data at 14 sites during 20 hours preceding the main shock did not show any significant pre-seismic deformation, either. These results indicate that pre-seismic strain change, if any, was smaller than 0.5-1.0×10.7 before the 2003 Tokachi-oki earthquake.

View full abstract

Groundwater level and flow rate at 44 wells are continuously observed by the Geological Survey of Japan and the Shizuoka and Gifu Prefectural Governments for monitoring seismic and volcanic activities. The 2003 Tokachi-oki earthquake (M8.0) occurred off the south coast of Hokkaido Island, Japan on September 26, 2003. The epicentral distance to the nearest observation well is about 250 km and that to the farthest is about 1200 km. At the 22 wells, we detected changes in groundwater level or flow rate in relation to the earthquake. Most of the changes are coseismic step-like changes and/or short-period oscillations. In the nearest two observation wells, long-period oscillations with the periods of 39 and 53 minutes were also observed for several days after the earthquake, which is likely due to tsunami. In comparison between distributions of changes in groundwater level and theoretical coseismic strain by the fault model, it is clear that step-like increases were found in the contraction area of the coseismic strain. The relationship between amounts of the observed step-like groundwater-level changes and theoretical ones, calculated by the fault model using strain sensitivities of groundwater level indicates that the groundwater levels in the several wells responded to the coseismic strain.

View full abstract

Changes in form of hypocenter distribution preceding the 2003 Tokachi-oki Earthquake (MJ = 8.0) were investigated by using the analysis of temporal variation in spatial fractal dimension D. In this study, it was found that the D value began to decrease in 1998, and had been very small for about one year before the main shock occurrence. Such a D decrease before the main shock occurrence is a characteristic of some recent large earthquakes. Therefore, the D decrease may be an earthquake precursor. The D value decrease is yielded by both of seismic activation and quiescence which have often been reported as an earthquake precursor, due to a property of the calculation method. Therefore, the D change can be detected, even if the seismic activation and quiescence occur simultaneously in which case the number of earthquakes does not change significantly. On account of this property, using the D value is advantageous to detect the precursory change of seismic activity before a large earthquake.

View full abstract