1. Introduction

Geologists study the past as a key to the present and the future, and paleoseismologists try to project the facts and rules of the past to the future. In this sense, active fault studies in Japan have emphasized the aspects of historical science. As it is the case in history, if a researcher does not have perspective and motivation in the present and the future, the science tends to be retrospective. The author's prejudice is that the Japanese active fault studies did not have enough perspective and were shaken hard by the Kobe Earthquake. This abstract is an attempt to summarize what we have and we have not done on Japanese active faults and desirably to propose what to be done.

2. Kobe earthquake of January 17, 1995

Behind the Kobe city along the foot of Rokko Mountains, very distinct active faults of the Rokko fault system exist. There was not any surface deformation along the Rokko fault system during the Kobe earthquake. Meantime, post-earthquake geophysical exploration is revealing concealed active faults under the coastal plain of Kobe city subparallel to the Rokko fault system [Endo et al. 1995]. Now there are two possible candidates for seismogenic faults in the Kobe section. One is the deeper part of the Rokko fault system and the other is the concealed fault beneath Kobe city. Detailed geodetic survey is necessary to define the seismogenic fault of the Kobe area.

From the viewpoints of paleoseismology, Rokko fault system and the ATL [Arima- Takatsuki Tectonic Line; figure 1] had presumably generated the M 7.0 +/- 1/4 [Usami, 1987] Keicho earthquake of 1596 A.D. Archaeologically dated reverse faulting along the ATL [Sangawa, 1992] and far more severe and massive liquefaction in Kobe area compared with the 1995 Kobe earthquake strongly indicate the activity of Rokko fault system in 1596 [Tsukuda, 1987]. The Kobe Earthquake of January 17, 1995 might not be a characteristic event on the Rokko fault system but an episodic event on subsidiary faults. Otherwise, the deeper part and the shallower part of the Rokko fault sytems might generate different series of characteristic earthquakes. We need a lot more paleoseismological information to answer the questions.

As to the extraordinary strong ground shaking and consequent intensive damage to Kobe, many hypotheses have been proposed but none of them can explain the complex disaster well. We are in urgent need to understanding the phenomena as a whole. Here the author should like to mention a non-scientific but unequivocal fact: The communities of Kobe and surrounding area were very poorly prepared for earthquakes. The people there believed that the Kansai district [i.e. Kyoto-Osaka-Kobe areas] was secure from earthquake hazards. It's a tragic superstition in the age of advanced seismology and active fault studies, and only 400 years after a devastating Keicho earthquake.

3. Active faults in Japan

Figure 3 shows active faults of Central Japan and adjacent sea floor. Since Japanese islands are in a convergence zone between Pacific/Philippine Sea Plates and Eurasian/North American Plates, most active faults are under E-W compressional stress. Therefore, the active faults in Central Japan are either N-S trending reverse fault, NE-SW trending right-lateral strike-slip fault, or NW-SE trending left-lateral strike-slip fault. There are few normal faults in the volcanic area of central Kyushu. In figure 3, active faults are very densely distributed in the area around Kobe and to the east. The areas close to the Pacific including Tokyo have less active faults, but these areas are prone to frequent gigantic earthquakes from the Nankai and Sagami troughs.

Each fault seldom exceeds 50 km in length, but combination of several faults often ruptures at a time to generate M 7 or larger events, as in the Kobe earthquake. Research Group for Active Faults of Japan [1980, 1991] precisely described the geomorphic evidences of each fault, but the evaluation of the these faults as sources of hazardous earthquake has not done yet.

Japanese islands are very young and active mobile zone. Present tectonic regime and topographic relief were established within a few million years. 70% of the territory is occupied by actively upheaving steep mountains slopes and 120 million people live in the limited flat areas. Most of the flat areas such as coastal plain or intermontane basin are the products of tectonic movements. The active faults has made the land habitable and most Japanese are destined to live by active faults. In Kobe and Osaka area, 10 million people live in alluvial plain and lowland bounded by active faults [figure 1]. Above the seismogenic fault of the Kobe earthquake 2 million people live as densely as 10,000 to 15,000 persons per square kilometer. Our inevitable way of life would increase riskss from active faulting.

4. Historic and archaeological seismicity

Like as super-novas are pretty well recorded in Japanese historic documents, written records on paleo-earthquakes are abundant in Japan. Usami [1987], for example, compiled about 350 damaging earthquake before 20th century back to 5th century A.D. [figure 4]. At the same time, seismologists made much efforts to compile and publish historic documents on earthquakes and their damages. The Earthquake Research Institute of the University of Tokyo has published more than 20,000 pages of historic records in a number of volumes. These volumes give sound bases for the examining former recognition of historic earthquakes and open ways to text critique.

Historic records usually describe disaster of earthquake, but they seldom report surface faulting. For almost all large earthquake before 19th century, geological study is necessary to identify a seismogenic fault In case of the 1586 Tensho Earthquake [located 36 degree N, 137 degree E in figure 4] there are numerous description on damages in an extensive area of Central Japan, where there are so many active faults [figure 3]. Although a number of researchers surveyed several major active faults in the area for over 20 years, the seismogenic fault was identified only lately after excavation of more than 10 trenches [Atera fault and Miboro fault in figure 6; Toda et al. 1994; Sugiyama et al. 1991].

5. Geological paleoseismology

Common recurrence interval of earthquakes from intraplate active faults is longer than 1,000 years. This is much longer than the recurrence intervals of interplate earthquakes along the Nankai Trough [figure 5]. The historic record therefore covers at most one earthquake event from each active fault. To know the recurrence interval, geological data are indispensable for most active faults on land in Japan. More than 50 trenches have been excavated in Japan [figure 6] and long recurrence intervals were confirmed. However, as it is clear in figure 6, most of the excavated fault have historic records of the last event within a thousand years. This means that most excavation have succeeded to identify recent historic records with geologic records, and elapsed time since the last event was usually considerably shorter than recurrence interval. In these trenches, validity of trenching method was confirmed as well as the quiescence of the fault for coming several hundred years. These are happy cases for paleoseismology and precision in dating was not very critical. On the other hand, to announce possible surface faulting in the near future is more difficult and delicate matter.

The Gofukuji fault along ISTL [Okumura, 1994] and the Okamura fault along MTL [Tsutsumi et al. 1992; Okada, 1992] are among few examples that may evoke attention to coming events. The longer recurrence intervals bring certain difficulties to forecast future earthquakes. Because the longer the recurrence interval is, the larger is the magnitude of chronological uncertainty owing to fluctuation of actual recurrence time as well as to errors in dating. The accuracy of dating in Japanese trenches is usually 300 years or worse and probabilistic forecast does not make much sense when deviations are taken into account. For the common communities this inaccurate forecast is not very useful. Geological studies of active faults should refine the accuracy in chronology to less than 100 years. It is quite possible as in Califotnia but the new techiniques are not well accepted by Japanese paleoseismology yet.

References

Research Group for Active Faults of Japan, Active faults in Japan: sheet maps and inventories, University of Tokyo Press, Tokyo, 359p., 1980 [J/E].

Research Group for Active Faults of Japan, Active faults in Japan: sheet maps and inventories [revised edition], University of Tokyo Press, Tokyo, 437p., 1991 [J/E].

Awata, Y., K. Mizuno, Y. Sugiyama, K. Shimoakawa, R. Imura, and K. Kimura, Surface faults associated with the Hyogo-ken Nanbu Earthquake of 1995, Chishitsu-news, 486, 16-20, 1995 [J].

Endo, K., S. Watanabe, M. Makino, Y. Murata, K. Watanabe, and A. Urabe, Subsurface geology and disaster caused by the 1995 South Hyogo Earthquake, Proc. symposium on the Great Hanshin-Awaji earthquake and its geo-environments, 219-224, 1995 [J/E].

Kaizuka, S. and T. Imaizumi, Horizontal strain rates of the Japanese islands estimated from Quaternary fault data, Geogr. Rep. Tokyo Metropol. Univ., 19, 43-65, 1984 [E].

Kasahara, K., Earthquakes and tectonics, in K. Kasahara and A. Sugimura eds., Iwanami series on earth sciences, vol. 10 [296p. Tokyo], 33-88, 1978 [J].

Kikuchi, M., Source process of the Kobe earthquake of January 17, 1995, Chishitsu-news, 486, 12-15, 1995 [J].

Mizuno, K., H. Hattori, A. Sangawa, and Y. Takahashi, Geology of the Akashi district with geological sheet map at 1:50,000, Geological Survey of Japan, 90p., 1995 [J/E].

Okada, A., Proposal of the segmentation on the Median Tectonic Line active fault system, Mem. Geol. Soc. Japan, 40, 15-30, 1992 [J/E].

Okumura, K., K. Shimokawa, H. Yamazaki, and E. Tsukuda, Recent surface faulting events along the middle section of the Itoigawa-Shizuoka tectonic line --Trenching survey of the Gofukuji fault near Matsumoto, Central Japan--, Zisin Jour, Seismol. Soc. Japan, Ser 2, 46, 425-438, 1994 [J/E].

Sangawa, A., Seismo-archaeology, Chuko-library, 1096, Chuo-Koronsha, Tokyo, 251p., 1992 [J].

Sangawa, A., Research on paleoearthquakes u sing traces discovered in archaeological sites, The Quaternary Research [Tokyo], 32, 249-256, 1993 [J/E].

Sugiyama, Y., Y. Awata, and E. Tsukuda, Holocene activity of the Miboro fault system, central Japan, and its implications for the Tensho Earthquake of 1586, Zisin Jour, Seismol. Soc. Japan, Ser 2, 44, 283-295, 1991 [J/E].

Tada, T., Crustal deformation associated with the 1995 Kobe earthquake and a geodetic source fault model, Proc. symposium on the Great Hanshin-Awaji earthquake and its geo-environments, 7-10, 1995 [J/E].

Toda, S., D. Inoue, N. Takase, A. Kubouchi, and N. Takase, The latest activity of the Atera fault: a possibility of 1586 Tensho Earthquake, Zisin Jour, Seismol. Soc. Japan, Ser 2, 47, 73-79, 1994 [J/E].

Tsukuda, E., Migration of historical earthquakes, central Japan, Direction in Paleoseismology, U.S. Geological Survey Open File Report, 87-673, 271-284 [E].

Tsutsumi, H., A. Okada, T. Nakata, and M. Ando, Near surface structure and Holocene movements on the Okamura fault, an active segment of the Median Tectonic Line in central Shikoku - A case study of the 1988 spring trench survey -, Mem. Geol. Soc. Japan, 40, 1113-127, 1992 [J/E].

Usami, T., Catalog of damaging earthquakes in Japan, revised edition, Univ. of Tokyo Press, Tokyo, 434p., 1987

Utsu, T., The worst 20 large earthquake and tsunamis in Japan, T. Utsu, E. Shima, T. Yoshii, and K. Yamashina eds, Encycropedia of earthquake, Asakura-shoten, Tokyo, 568p., p. 456, 1987 [J].

[J] in Japanese only, [J/E] Japanese with English abstract, [E]: in English

zz

z