We analyzed the fractal natures of swarm earthquakes which occurred off the eastern coast of Izu Peninsula, central Japan, using the JMA Earthquake Catalogue data during the period of 1983-1997. The spatial fractal dimension D- and the b-value of the Gutenberg-Richter formula were used for the seismic fractal analysis.
The significant temporal variations of both parameters were detected for 8 typical swarm activities and classified in two groups. One is a positive correlation, and the other is a negative correlation between D- and b-values.
The former case seems to correspond with ordinary swarm activities, while in the latter case some direct volcanic activities such as eruptions are accompanied by earthquake swarms. Consequently a negative correlation in D- and b-values may play an important role as an indicator of volcanic eruptions.

View full abstract

The distribution of seismic intensity I=VII (very disastrous) for the 1995 Hyogoken-Nanbu earthquake was reported by the Japan Meteorological Agency (JMA). It was the first announcement since the seismic intensity in JMA scale had been revised in 1949 to include the highest class of I=VII. Originally the seismic intensity of I=VII was defined as “strong ground motion with collapse more than 30% of wooden houses”, which was based on destructive damage at the 1948 Fukui earthquake. During the last half century, aseismic design of wooden houses has progressed especially due to the popularization of the standard building code published in 1950. Recent cities like Kobe include various kinds of buildings in seismic performance from modern earthquake-resisting structures to old and vulnerable residences. Therefore the seismic intensity of I=VII should be recognized as “with collapse more than 30% of less aseismic wooden houses such as those built before 1950”. From these points, it must be confirmed whether the increase of the seismic performance of buildings was taken into account in the reported distribution of I=VII. First we review the term of “collapse of houses”. Then relationship between the collapse rate of houses and the overturning acceleration of tombstones is investigated and analyzed using damage data obtained from the 1995 Hyogoken-Nanbu earthquake. The analysis result and its comparison to the relationships for past earthquakes show that the average of the seismic performance has increased by 40-60% from 1948 to 1995, and that the collapse rate of 10% at the Hyogoken-Nanbu earthquake corresponds to that of 30% at the Fukui earthquake for the same overturning acceleration. Comparing the reported distribution of I=VII to the area with collapse more than 10%, historical continuity of the seismic intensity in JMA scale is discussed.

View full abstract

Waveform records obtained from the temporal aftershock observation in Nagata Ward, Kobe, were simulated taking into account the irregularity of the underground structure. The strong motion observation of the aftershocks was carried out by the linear array which consists of six stations from the Rokko Mountains to the coast of Osaka Bay, shortly after the 1995 Hyogo-ken Nanbu earthquake. The underground structure under these stations was determined from the results of geophysical surveys, the spectral ratio of observed horizontal components at the sediment stations to those at KWC on granite, and the difference between the travel times of the SP-converted waves and the direct S-waves at the sediment stations. The obtained 2-D structure model shows that the granite basement is steeply deepened along the observation line between the stations KWC and NGH forming a basin edge, and the thickness of sediments increases from 1.0km (NGH) to 1.2km (KMG). The velocity seismograms at the five stations on sediments were synthesized from the observed records at the KWC station on granite for the three aftershocks, using the 2-D structure in-plane and anti-plane models by the finite element method. The synthesized waveforms explain well not only the direct S-wave part, but also the conspicuous later phase part that were found in the observed records at the SYP and KMG stations near the coast of Osaka Bay. Next, we input the Ricker wavelet with a characteristic period of 0.8s from the basement to the same 2-D structure model. The synthetic waveforms were decomposed into two wave types to elucidate the origin of the later phases at the SYP and KMG stations. The one was the response of the flatly layered model at each point and the other was the diffracted wave from the basin edge. It was concluded from the results of these analyses that the later phases were due to constructive interference between the reflected S-waves from the basement and the diffracted waves from the basin edge. It was also found that the Rayleigh and Love waves of the higher modes were predominant in the diffracted waves at the KMG station.

View full abstract

The Noto Peninsula coast has been suffered from the tsunamis originated along the eastern margin of the Japan Sea. According to the old documents, the 1833 Yamagata-Oki tsunami reached 5.3m at Wajima (tip of Noto Peninsula), and killed 47 persons. Inundation heights of the 1964 Niigata, 1983 Nihonkai-Chubu and 1993 SW. Hokkaido tsunamis were 2-4m at Wajima and its neighboring area. These tsunami heights are more than 2-3.5 times larger than the ones expected from the average tsunami magnitude. Based on the refraction diagrams and the shoaling-refraction factors around the peninsula, amplification factors estimated by the Green's formula are 3.0-4.0 at the north coast, 1.5 at the west coast and 1.0 at the east coast. The distribution patterns of the calculated factors nearly agree with those of the inundation heights for each tsunami. In case of the 1964 Niigata tsunami having the period of about 20min, the seishe of Nanao Bay (east side of peninsula) may be excited, because the maximum wave occurred 2.7 hour late from the initial wave. Tsunamis generated off Yamagata and SW. Niigata are strongly affected by the shoaling-refraction.

View full abstract

The regional gravity anomalies were calculated in Southwestern Japan presuming a three-dimensional density structure. The gravitational effect was calculated by the structural model represented by aggregate bodies of triangular vertical prisms which was generated by compiling the existing vast amount of data on the depths of lower surface of the crust and upper surface of the plate slab in this region. The densities of the Pacific and Philippine Sea plate slabs, upper and lower crusts and asthenosphere were presumed referring to past research. It was shown by the calculation that the gravitational effect of the crust has larger amplitude than that of the plate slabs in this region, while variation of the presumptive density of the plate slab is more effective than that of the crust. Comparison of the regional gravity correction by means of the above method with the ones by the spatial filtering method and the least square method shows that the former derives more accurate local gravity anomaly than the latter. The regional gravity anomaly can be calculated by the above method at an arbitrary point in Southwestern Japan, and then more accurate regional gravity correction can be performed for the analysis of local gravity structure.

View full abstract

New geophysical and geologic findings show, for the first time, subsurface branching and Holocene activity in the Kongo fault system. The system extends 18km along the southwest side of the valley that contains the ancient capital city of Nara. It continues southward to the Median Tectonic Line, towards which it swings westward. In the north it contains three faults of parallel strike, called Yamaguchi, Kongo, and Yamada. Seismic reflection surveys show that the Yamada and Kongo faults merge about 300m below the surface. They displace a prominent reflector, interpreted as the contact between granite and Pleistocene deposits, by 240-350m. The displacement of Pleistocene beds and terraces implies vertical slip rates in the range of 0.1-0.6mm/year. The last major surface rupture produced faults seen in outcrop and in a trench. It probably occurred in the first few centuries AD, as judged from a dozen radiocarbon ages. The most closely limiting of these ages are 1900±60 ¹⁴C yBP (the faulted strata) and 1780±50 ¹⁴C yBP (overlying, unfaulted strata), or between 40 BC and AD 380. A minimum vertical displacement of 1.2m, measured in the trench, implies an earthquake magnitude close to 7, comparable to the 1995 Kobe earthquake. The average recurrence interval is probably in the range of 2000-12, 000 years. This range is broad because the slip rate is poorly known, and results probability estimates for the next earthquake, on the basis of the characteristic earthquake model, highly variable.

View full abstract

Old seismograms of the Tokyo earthquake in the Meiji era (June 20, 1894) were analyzed to retrieve source parameters. This earthquake gave considerable damage on the Tokyo metropolitan area. The focal depth was estimated to be about 50km or about 80km from S-P times of several seismograms observed at Tokyo. The magnitude, which was calculated from data of maximum amplitudes, was 6.6. The focal mechanism and the seismic moment were inferred from waveform fitting with seismograms observed by a Ewing-type strong motion seismograph and a Gray-Milne-Ewing-type seismograph. The estimated moment is 1-3×10¹⁸Nm (M_w=6.0-6.3). The earthquake is considered to have a nearly vertical nodal plane with N-S strike, and to have occurred in the Pacific plate or the Philippine Sea plate. Stress drop of the event was larger than the average value in the region.

View full abstract

Conventional triangulation and leveling data are analyzed to estimate crustal movements associated with the 1948 Fukui earthquake and its fault model. Horizontal displacement vectors at 84 triangulation points and vertical displacements of 82 leveling benchmarks are inverted to estimate slip distribution on the fault plane. Although two surface traces of faults were found after the earthquake, most of the seismic moment was released from a main fault on the west, and an eastern sub-fault played only a complementary role. The dip angle of the main fault is not well constrained. However, geodetic data are fairly consistent with an assumption of a vertical fault. Estimated fault mechanism is mostly left-lateral strike slip with the maximum slip of 6m. The seismic moment of the Fukui earthquake is estimated as 2.4×10¹⁹Nm (M_w=6.8), which is consistent with another estimation based on seismic data. The Fukui earthquake was comparable to the 1995 Kobe earthquake in its size, but the heterogeneity of slip distribution is different each other. The Fukui earthquake fault had a much longer preparatory period before the 1948 event, and fault strength might be completely recovered before the earthquake, which resulted in a rather homogeneous slip distribution. In the case of the 1995 Kobe earthquake, a short recurrence time along the Rokko-Awaji fault after the 1596 Keicho-Fushimi earthquake might result in a rather heterogeneous slip distribution.

View full abstract

Seismograms in 1940's are usually recorded on smoked paper. Owing to the recent development of photocopy and image processing techniques, we can reconstruct feasible waveform data by tracing, digitizing, and correcting the arc effect due to inclined lever. Using low gain strong motion data at several observatories of Japan Meteorological Agency and Kyoto University, we investigate the source process of the 1948 Fukui earthquake. In the waveform inversion we consider the uncertainty of instrumental constants (pendulum period, magnification and damping constant), timing, and chart speed. The rupture process is then modeled by moment-rate functions at grid points on a fault plane with a spacing of 10km. It is shown that a rupture initiated at a depth of about 10km, moved upward, and then propagated mainly to the south. The rupture front velocity was about 2.5km/s. The largest moment release occurred around 10km south of the epicenter. The main source parameters are: strike=170±50°, dip=70±5°, rake=-10±10°, seismic moment=2.1×10¹⁹Nm, M_w=6.8, fault area=30×10km², average dislocation=2.3m, stress drop=10MPa.

View full abstract

We simulate strong ground motions during the 1948 Fukui earthquake with the JMA magnitude 7.1 based on a heterogeneous source model and the hybrid simulation technique. So far there are no existing source models available for simulating strong ground motions from the 1948 Fukui earthquake. Most of the source models have been assumed to have uniform slip distribution on rectangular fault plane. Such models could generate ground motions only available longer than several seconds, underestimating shorter period motions (<1sec) of engineering interest. The objective of this paper is to construct a heterogeneous source model for simulating strong ground motions in a broad period band during the 1948 Fukui earthquake. We assume two source models to examine: Model 1 is a reverse fault model determined from the analysis of geodetic data by YOSHIOKA (1974) and Model 2 is a normal fault model from strong motion displacement data by KIKUCHI et al. (1999). Heterogeneous slip distribution on fault plane is estimated based on the self-similar scaling relationships of seismic moment versus asperity areas and slips by Somerville et al. (1999). Then we obtained the standardized source model consisting of two asperities to have the average characteristics of asperities for the seismic moment of the Fukui earthquake. Relative locations and rupture times of the asperities on the fault plane are determined following the source model by KIKUCHI et al. (1999). The maximum asperity corresponding to the second event in their model has an area of 12×12km² and slip of 1.7m and is located under the most heavily damaged area along the buried fault, known as the Fukui earthquake fault. The smaller asperity corresponding to the first event is located north of the maximum asperity. Rupture was initiated at the northern edge of the smaller asperity, propagated toward south, then broke to start the maximum asperity 7 seconds after the initial rupture. Large ground motions from both models, Model 1 and 2, are spread over the Fukui basin, although peak velocity distributions are rather different between the two models. Areas over 30% collapse ratio during the Fukui earthquake correspond to those with peak velocity over 60cm/s for Model 1 and over 80cm/s for Model 2. The level of the peak velocity in the areas with more than 30% collapse ratio are estimated to be over 80cm/s connected with both results by MOROI et al. (1998) and MIYAKOSHI and HAYASHI (1998). Pseudo velocity response spectra in the center of the Fukui basin for Model 2 have almost the same level of the observed ones at Takatori (TKT) and the simulated ones at Fukuike (FKI) within the damage belt during the 1995 Hyogo-ken Nanbu earthquake. We conclude that the damage distribution during the Fukui earthquake is well explained by strong ground motions simulated for Model 2 combined with the normal fault model by KIKUCHI et al.. (1999) and a standardized heterogeneous source model developed by SOMERVILLE et al. (1999).

View full abstract

The 1948 Fukui earthquake provided the detailed data for damage of buildings and collapse or overturn of simple bodies, such as chimneys, grave stones, etc. The overturning directions of simple bodies and the falling directions of wooden houses seem to have characteristic features of distribution with scatters. From the northern to the central part of the fault trace (strike=170°), the fault normal directions are predominant. On the other hand, from the central to the south but apart from the fault trace, the fault parallel directions are predominant. These distributions are simulated by 3D dynamic rupture model, and the details of the fault rupture processes are suggested.

View full abstract

Some geophysical explorations were conducted to investigate the underground structure around the Fukui Earthquake Fault, FEF, reported by TSUYA (1950). A seismic reflection survey was carried out at a dry river bed along the Kuzuryu River across the FEF. The gravity survey was also performed around the seismic reflection survey line. The seismic reflection profile shows that the Quaternary sedimentary cover is 200 to 300m in thickness and tilts gently westward. The profile also shows an existence of somewhat peculiar block with width of several hundred meters at the FEF. This block corresponds to the lower gravity anomaly zone obtained from the local gravity analysis. We concluded that the FEF with many ground failures induced by the earthquake has not been effectively displaced in vertical movement and the successive fault movement related in the FEF has thickened fault gouge layer along the FEF.

View full abstract

Subsurface geological structure of the Fukui basin where the 1948 Fukui Earthquake (M7.1) occurred has been studied by P and S wave seismic reflection survey together with a trench excavation survey at the eastern edge of the basin. In the central part of the basin, no distinctive fault has been detected by the P wave seismic reflection survey. Also, at the eastern edge of the basin, no evidence of faulting associated with the 1948 event has been detected by S wave seismic reflection and trench excavation surveys. However, liquefaction of gravel layer has been revealed in a trench at the eastern edge of the basin and its date is estimated sometime between A. D. 1200. and A. D. 1400. These results suggest that the faulting of the 1948 Fukui Earthquake (M7.1) is dominant with strike slip component, and hence it is concluded that no earthquake fault appeared on the ground surface other than the intensive surface cracks reported as the evidence of earthquake fault.

View full abstract

The seismic activity in the Chubu region of Houshu, Japan after the M8.0 Nobi earthquake of 1891 has been remarkably active, with 7 earthquakes of M≥6.5. A probabilistic test concludes that this series of earthquake occurrences for the relatively short-period is significant and not random. The Coulomb Failure Function (CFF) is used to evaluate effects of static stress loading on the neighboring faults for the earthquakes which occurred in the Chubu region of Honshu. Five of the seven events fall in positive area of change for the Coulomb Failure Function values (ΔCFF). This consistency suggests that the Nobi earthquake changed the stress field of the crust around the hypocenter, triggering many earthquakes. The ΔCFF values are calculated for ±10° range of fault strike and dip angles for models of each earthquake. The values are found to vary significantly depending upon fault parameters, but the sense of ΔCFF (positive or negative) is stable throughout the test of different fault orientations. Preceding the M7.1 1948 Fukui earthquake, an earthquake of M6.3 occurred in 1930, possibly on the northern extension of the Fukui eastern margin fault system. Along the Fukui earthquake fault, high microearthquake activity has been observed to diminish with time since 1976, when the local seismic network was established. Whether or not the Fukui eastern margin fault system was involved in the Fukui earthquake of 1948 is not certain, and future work is needed to resolve the issue.

View full abstract

Fatality risks due to building collapses are investigated for three typical inland earthquakes in Japan; the 1891 Nobi earthquake, the 1948 Fukui earthquake, and the 1995 Hyogoken-Nanbu earthquake. The damage data obtained from these earthquakes are analyzed to evaluate relationships between fatality rate in population and collapse rate of wooden houses. A lethality function normalized for estimated numbers of inhabitants in collapsed houses is also estimated for investigation on normalized fatality risk with respect to building collapses. By comparing the analysis results, we discuss the circumstances of human casualties in these earthquakes: 1) Although different distributions of collapse rate were observed between the Nobi earthquake and the Fukui earthquake, the human casualties in both earthquakes show similar tendency that higher mortality was properly induced in the area with large damage of buildings; 2) The extensive damage by the Hyogoken-Nanbu earthquake was concentrated in limited area as compared with the other earthquakes, however high fatality rates are estimated even in relatively low-damaged regions; 3) The fatality risks in seriously damaged regions are higher than those in moderately damaged regions for every earthquake, while the fatality risks evaluated for the Hyogoken-Nanbu earthquake are two times as large as those of past earthquakes. From these results, we point out that the vulnerability to human casualties by earthquakes is still existing despite the advancement of seismic performance of buildings.

View full abstract

It is not only the direct damage but also the secondary damage of fire which spreads entirely over the city, that city suffers from when an earthquake occurs. Not only did the 1995 Hanshin-Awaji earthquake destroyed the quake-resistant structure, but caused a large amount of fire at the same time, especially firing the wooden-made housing areas for 10 days. It made it clear how weak the present city is against an earthquake. One of the most important problems is the shortage of extinguishing water. The available water in place of the hydrants being out of service was in short supply at the areas and that was the reason why the fire spread out largely to combustible wooden houses. This means the fire extinguishing system was heavily dependent upon hydrants. In order to avoid such situation, it is important to provide some stand-by systems in addition to the ordinary system. We have to prepare the systems which minimize the damage if fires should break out by an earthquake. Such system is based on the concept of “fail-safe”. The water supply system in San Francisco is a typical safety system that consists of three backup systems, AWSS, Cisterns and PWSS. Against the fires after Loma Prieta Earthquake occurred in October 17, 1989, PWSS effectively functioned while the other systems were out of operation by the damages. On the other hand, in Japan, channel water is used as water supply system for fire prevention in Shirakawa-go and Gujohachiman. Channels form the basis of towns and support residents' daily lives while they are made the best use of as water supply system for extinguishing fire. This kind of nearly based water is effective when an unexpected disaster occurs. Most of the Japanese houses have been made of wood because wooden houses are suitable for Japanese climate. That is, many regions are at risk of big fire when big earthquake occurs. However, our country abounds in natural water, which is conveyed to our living spaces through channels. In order to protect wooden houses from fires, to secure our lives and to preserve our culture, it is recommended that channel network be used as water supply system for fire extinguishing. This is the system rooted in culture and climate.

View full abstract