Numerical simulations of the trans-oceanic tsunami propagation require a well designed computation program. In order to obtain good, reliable results, spatial and temporal grid sizes should be carefully determined. The area included in and the time for the computation are so wide and long that a huge computer memory is inevitable. There are two major sources of error which are closely related with grid size. One is the dispersion, physical or numerical, and other is the accumulation of round-off errors.
After a comparison of the maginitude terms in the fundamental equations for long waves, it is concluded that the convection term can be negligible and that the linear Boussinesq equations including the Coriolis force, expressed in the longitude-latitude coordinates, should be used.
Effects of the grid size on the computed results is examined as the one-dimensional problem for four different grids assuming that leap-frog scheme is applied to the linear long wave and linear Boussinesq equations. For comparisons, the linear surface wave equation is considered as the best equation, because it fully includes the effects of dispersion. The linear Boussinesq give better results with finer grid size, while the linear long wave equations give the best result with grid size 10km, with which the numerically introduced dispersion approximates well the physical dispersion.
Diagrams are provided to estimate errors in wave profile and wave length as a function of the grid size, travel distance and original wave length, for both the linear long wave and linear Boussinesq equations.
As an example of practical application, the Great Alaska Earthquake Tsunami of 1964 in the North Pacific Ocean is simulated with the linear long wave and linear Boussinesq equations.

View full abstract

Seismic waves from explosions at nine sites were observed at a number of temporary stations in the southwestern Kanto plain. Travel time analyses were carried out along the several surveying lines, so that the contradiction at cross points of the lines becomes as small as possible. Underground structures along the surveying lines were studied to reveal the three dimensional features of the region. The results are summarized as follows:
1) The structure consists of four layers which have P-wave velocities of 1.8, 2.8, 4.8 and 5.5km/s, respectively. Around the Yumenoshima explosion site, however, the thickness of the 4.8km/s layer seems to be very thin.
2) The depth to the 4.8km/s layer becomes largest around Yokohama as more than 4km. The thickness of this layer is more than 3km in the central part of the area considered, such as Okazu and Hiratsuka. However, it becomes shallower near the Kurokawa and Higashi-ohgishima explosion sites.
3) A step-like structure was found on the top of the 5.5km/s layer at points along the several surveying lines in the southeastern extension of the Tachikawa fault toward the Tokyo Bay area. On the southwestern side of the step-like structure, the depth to the 5.5km/s layer is more than several kilometers and this layer exists near the ground surface of the Kanto Mountains.

View full abstract

Along the Ryukyu Islands, 11 tsunamis have been recorded since 1664. Among them, there was the 1771 Yaeyama tsunami which hit the Ishigaki and Miyako Islands. About 12, 000 persons were drowned and 3, 200 houses washed away. In this paper, the mangnitudes of the principal tsunamis are discussed by the author's method (HATORI, 1986), using the diagram of attenuation of tsunami height with distance. Refraction diagrams of representative tsunamis are drawn to see travel times and the shoaling effects for each island.
The run-up heights at Ishigaki and Miyako Islands facing the tsunami source reached 20-30 meters, while those at the coasts of the East China Sea side decreased to 4-5 meters. And the run-up heights at Taketomi and Iriomote Islands etc. laying the wide coral reefs conspicuously decreased, too, suggesting that the short-period waves were predominant. The source area extends 100km along the mean depth 2, 500m which the location is expected with the wave preiod 10min. The wave rays projected from the half margin of source area concentrate between Miyako and Iriomote Islands. The wave rays of the hypothetical sources for the 1911 Amami-Oshima and 1938 Miyakojima tsunamis also concentrate within the segment of about 200km. Judging from the tsunami height-distance diagram, the magnitude of Imamura-Iida scale for the 1771 Yaeyama tsunami was m=4, but that of the Amami-Oshima tsunamis in 1901 and 1911 was determined to be m=1 and m=1.5, respectively, which are somewhat larger than the former values.
Tsunamis do not take place so frequently in the region of the Ryukyu Islands as that around the mainland of Japan, but those of the various types have been generated by the low-frequency or deep earthquakes. Most of historical small or moderate tsunamis may be missed, comparing with the recent seismic activity. In a map of the source distribution, there are manygaps of tsunami source area.

View full abstract

The Applicability of the seismic tomography technique that is designed to determine a 2-D velocity structure model from seismic refraction profilings is numerically examined. The tomographic method used in this study was originally developed for seismic reflecetion data, and it is modified to treat refraction data by the author.
The numerical experiment consists of two steps. The first one is aimed for examining fundamental features of this method, such as the influence of the error contained in the observed traveltime data on the result, by using a simple velocity structure model as the standard one. The second is intended to investigate the practical efficiency of this method by applying to cases of reproducing a low velocity zone from the initial model with no lateral heterogeneity. In this step, the deployment of sources and receivers simulates the dense observation of real OBS-airgun refraction profilings, such as evenly located 4 sources and 179 receivers on the area of 96km width.
Main results obtained from experiments are described in the following. First, if the traveltime residual of the result is less than the error in the observed traveltime data, this result may not be the best one. Second, it is very difficult to determine the shallow velocity structure between sources uniquely without any effective restrictions. Third, the distance between two neighboring sources closely relates to the minimum distinguishable size of the velocity anomaly zone. It is better to increase the number of sources for obtaining more reliable result, if the total numer of observed traveltime data is nearly the same.

View full abstract

Intraplate seismicity in the Tohoku district is high in the regions along the volcanic front and in the belt-like zone parallel to the Oga-Oshika tectonic line proposed by MOGI (1985a). Almost all large earthquakes in the Tohoku district in the past 100 years occurred in the Oga-Oshika tectonic belt and a significant linear distribution of small earthquakes is occasionally observed in the zone. Further, seismic activity after the 1983 Japan Sea earthquake migrated to the south along the tectonic belt [YOSHIDA and HOSONO (1987)]. Another noticeable feature is that earthquakes occur in the lower crust in the zone. On the other hand seismicity in the regions along the volcanic front is characterized by shallow earthquakes. Space-time distribution of earthquakes with M≥4 in the Oga-Oshika tectonic belt in the past 25 years exhibits that the seismicity in the zone became active just before and after three large earthquakes which occurred in the zone or in its extended area. The existence of a tectonic belt in almost the same site has been proposed by geologists based on geological observations and tectonic considerations. Recently TADA (1986) presented an idea that the Honjo-Matsushima tectonic zone, which had been proposed by geologists [OIDE and ONUMA (1960)] divides Northeastern Japan Arc into the northern and the southern blocks, investigating crustal strain field in the past 90 years in the Tohoku district. Submarine topographies and geomagnetic anomalies in the Japan Sea also suggest an existence of boundary of crustal structure in the zone. Neogene sediments are widely distributed in the zone as well. These all features indicate that the Oga-Oshika tectonic belt is a deep-seated weak zone, reaching to the lower crust, which corresponds to the boundary of crustal blocks. Characteristics of seismic activity in this zone such as concurrent activation and migration also suggest that the zone makes a stress guide in mechanical interactions between blocks. Its origin may be related to the difference of coupling form between the oceanic and the land plates in the northern part and the southern part of the Tohoku district.

View full abstract

For the purpose of estimating a tsunami hazard statistically, a mean tsunami height H_n is defined as a geometrical one of log H_n=(1/n) ∑ⁿ_i=1 log H_i, in which H_i is the i-th value of n observed heights in 20-40km segment of the coast, and calculated in the most part of the coast of Japanese Islands. The mean is obtained for the tsunami, having the tsunami magnitude larger than 1, observed during a period from 1600 to 1986 in each segment. A cumulative tsunami energy is estimated by summing up H²_n for all the tsunamis concerned. The result shows that the cumulative energy takes the highest value on the coast of Iwate Prefecture, and the 2nd, 3rd, 4th and 5th highest values on those of Miyagi, Kochi, Mie and Chiba Prefectures, respectively.
Frequencies of H_n exceeding 0.5, 1, 2, 5, 7 and 10m are also obtained. It is noticeable that H_n exceeds the 5m level 4 times at maximum for all the segments in Iwate and Miyagi Prefectures. These frequencies are converted into the probability of tsunamis exceeding a certain prescrived height on the assumption of the Poisson process in the generation of earthquakes. The probability of a tsunami higher than 2m hitting the coast of Iwate Prefecture during a 10-year period amounts to 22%, while that of a tsunami higher than 5m 7.5%. These probabilities certainly representing the long-term tsunami activity averaged over the 387-year period in the past, they differ from the one estimated by taking short-term earthquake prediction into account.

View full abstract

Ocean bottom bases (OBB), newly developed for long-term geodetic and geophysical observations, were installed on the seafloor of Sagami Bay, Central Japan. The OBB, made of 1m-square nonmagnetic concrete block, was deployed using a wire at a desired position owing to a precise underwater positioning system using acoustic transponders and a 30KHz bottom pinger. The OBB deployed was used to evaluate the accuracy of the underwater positioning system. The results of the experiment show that the acoustic positioning is accurate to 10^-3 of the slant range. It was confirmed later on a submersible “Shinkai 2000” that the OBB was successfully installed and was very stable on the flat mud bottom.

View full abstract

The seismological data of JMA (Japan Meteorological Agency) make epochs at 1961 and 1976. The detectability of seismic events improved from each epoch, and especially it rose remarkably from the latter epoch. Moreover, the result of reexamination of the seismological data for the period 1926-1960 by JMA, shows that the yearly frequency of seismic events in the period rose almost equal to the yearly frequency found during the period 1961-1975. The yearly frequency of events of M≥3.5 after 1976 is almost equal to that of events for the period 1961-1975, too. Therefore, if the events of M<3.5 after 1976 are eliminated from the data, it is possible to consider as the seismological data are almost homogeneous throughout the period 1926-1982. This means that it is possible to analyze seismicity quantitatively. The new seismic data are available for the statistical test of time-spatial change of seismicity around the dam areas effected by impoundment.
The present paper is a revision of the previous paper [TERASHIMA (1983)] surveyed qualitatively the seismicity change between pre- and post- impoundment for the 43 dam areas, and the revision was done by applying a hypothesis test for the new seismic data of the same dam areas surveyed in the previous paper.
The hypothesis test is done for the significant difference between the yearly frequency of seismic events from 1926 to the start of impoundment for each dam and the seismicity index that is defined here by the yearly mean frequency of seismic events for 5 years after the start of impoundment for each dam.
The significant difference is detected in 20 dam areaes. In the following 6 dam areas of them, that is, Kurobe, Nagawado, Kusaki, Hitotsuse, Kuzuryu and Midono, it is estimated that the significant differences are caused by the seismicity induced by the impoundment, taking a time series of seismicity around the each dam into consideration.
It is concluded that the seismicity of Kusaki dam area became quiescent after the impoundment and it of the other 5 dam areas did active after the impoundment.
In the remaining 23 dam areas, no significant difference is recognized by the statistical test. They show also no change between the actual seismicities before and after impoundment.
There are some differences in the results between the present paper and the previous one. The causes come from:
(i) use of the new seismic data for the period 1926-1982 in the present paper,
(ii) use of the seismicity in epicentral distance Δ≤30km centering around each dam in the present paper, but Δ≤50km in the previous one,
(iii) application of the statistical test for the standards of objective and quantitative judgment in the present paper, but only qualitative in the previous one, and
(iv) use of the started date of impoundment in the present paper instead of the date of completion of dams in the previous paper, for the date to divide the seismicity into two groups, before and after the impoundment (There are usually the difference of 1 to 2 years between the started date of impoundment and the date of completion of dams).

View full abstract

In Shikoku, earthquakes in the crust show a vertically steplike distribution across the Median Tectonic Line (MTL). Namely, the upper and lower boundaries of the distribution in the southern side of MTL are shallower than those in the northern side of the line. Since crustal microearthquakes mainly occur in the granitic layer (V_p=6.0km/s), the steplike distribution suggests a vertical dislocation in the granitic layer which was caused by uplifting of the southern part (Sanbagawa Metamorphic Belt) against MTL. On the other hand, most earthquakes in the mantle occur in the southern side of MTL in the central and eastern parts of Shikoku and occur in both the sides of MTL in the western part of Shikoku.

View full abstract

This paper gives a review of studies on source complexity such as multiple event nature of great earthquakes. It includes a brief note to historical background as well as inversions of complex body waves which have recently been developed to extract informations about fault asperities or barriers. It is emphasized that the multiple event nature is colsely related to some rate-determining process, such as stable frictional-sliding, in the areas surrounding the asperities.

View full abstract

The cutoff frequency f_max is a fundamental parameter for the prediction of strong ground motion. This paper, first, gives a review of two different interpretations about f_max, i. e., site-controlled f_max and source-controlled f_max, used in the standard prediction methods. Hanks and McGuire's approach assumes f_max as mainly site effects and the specific barrier model proposed by Papageorgiou and Aki as mainly source effects. Next, the recent status of observation related to f_max is reviewed. From a seismometrical point of view, we will have to examine whether f_max exists as source effects or not.

View full abstract