The Sagami Bay region is situated in the eastern part of the South Fossa Magna on the Pacific coast of central Japan, where the north-south running Izu-Bonin arc on the eastern margin of the Philippine Sea (PHS) plate joins with the Honshu arc. In this region, the non-volcanic Izu outer arc (IOA) lithosphere is being subducted beneath the Kanto district north-northwestward, and the volcanic Izu inner arc (IIA) on the west of IOA is colliding with Honshu to the north of the Izu Peninsula because of its buoyancy. I infer that the subduction entrance of the IOA lithosphere is at the southern margin of the Sagami Basin in Sagami Bay, and estimate the depth of the upper surface of the IOA lithosphere beneath the Ashigara Plain on the north of Sagami Bay at more than 10km. On the other hand, the surface collision boundary between the IIA lithosphere and the Honshu block is located about 50km north from the subduction entrance of the IOA lithosphere. Therefore, there should exist a north-south striking fracture within the PHS plate which separates the subducting IOA lithosphere from the IIA lithosphere at rest on the earth's surface.

View full abstract

The pre-20th century history of earthquake swarms and volcanic activity in the Higashi Izu monogenetic volcano field, Japan, was examined mainly by a critical review of historical documents. Four descriptions of historical earthquakes (in A. D. 1868 or 70, 1816-17, 1737, and 1596) and two descriptions of possible volcanic activity (in 1854 and 1777) were critically read and their reliabilities were checked. The 1868 (or 1870) earthquake (and possibly the 1816-17 earthquake) corresponds to a historical analogue of the earthquake swarms off the east coast of Izu Peninsula, which occurred in 1930 and have been episodically occurring since 1978. The recurrence intervals of these four earthquake swarms (1816-17, 1868 or 70, 1930, and 1978-present) are about 50-60 years. While the duration of each swarm is 1-4 months for the three swarms before 1978, the 1978-present swarm has episodically continued for at least 21 years. Two possible volcanic events in the Higashi Izu monogenetic volcano field were rejected because of lack of other reliable evidence.

View full abstract

There are two different ideas about the tectonics of earthquakes along the Kozu-Matsuda fault, which is situated on the on-shore extension of the Sagami trough, central Japan. One is derived from geological evidence of the deformed Quaternary deposits in the onshore areas. The idea is that the Kozu-Matsuda fault upheaves the Oiso Hills and the Tanzawa Mountains and causes huge earthquakes with recurrence intervals thousands of years long. The fault moves without any relationship to the mega-thrust beneath the Sagami Trough from which 1923 Kanto earthquake occurred. The other is that the Kozu-Matsuda fault is a subsidiary fault of the plate boundary thrust along which huge earthquakes occur every two or three hundred years. Several seismological interpretations of the 1923 Kanto earthquake support this opinion. Based on this, the Kozu-Matsuda fault would never cause a huge earthquake without any link to the plate boundary fault. To test these competing hypotheses, the authors reconstructed the recent behavior of the Kozu-Matsuda fault based on the trench logs excavated at the foot of the fault scarp. Although no Holocene fault displacement was found from the trench walls, four or five seismic events since 3, 000 years ago were detected from the startigraphy and chronology of the landslide or slope failure deposits inferred to be of seismic origin. Only one event, which is distinct and occurred about 3, 000 years ago, is recognized to be caused by activity of the Kozu-Matsuda fault. As the other faint events include the event caused by the 1923 Kanto earthquake, these events are thought to be related the earthquakes from the Sagami trough. The trench survey seems to confirm the idea that the Kozu-Matsuda fault and the earthquake from the Sagami Trough have long recurrence intervals and that the last event of the Kozu-Matsuda fault occurred 3, 000 years ago.

View full abstract

It has been pointed that an M7 class earthquake might occur in the near future around northwestern Sagami Bay, southwest Kanto district, central Japan, and this hypothetical earthquake has been named the “Western Kanagawa Prefecture Earthquake (WKPE)”. In order to investigate the mechanism of WKPE, the Hot Springs Research Institute of Kanagawa Prefecture (HSRI) has carried out continuous monitoring of seismicity and crustal deformation since 1989, covering the anticipated rupture zone estimated from historical data. In this paper, the relationship between seismicity and geological structure in and around northwestern Sagami Bay is investigated in detail based on a hypocenter precisely determined by the HSRI during the period 1990-1998.
Hypocenters of earthquakes are calculated from 16 stations for the present seismicity analysis. Seismic stations are deployed 2-10km apart. The accuracy of the hypocenter determination in this study area was within 2-3km with application of the method of prediction analysis. The minimum magnitude of observable earthquakes was estimated to be M0 in the Hakone area and M1 in northwestern Sagami Bay.
Comparison between seismicity and geological structure can be summarized as follows:
(1) The higher seismicity areas are located in the prefectural border between Kanagawa and Yamanashi, and in the Hakone area. This higher seismicity can be attributed to the uplift of the Tanzawa Mountains and to volcanic activity, respectively, in the two areas.
(2) The cutoff depth of seismicity becomes deeper from the center toward the eastern flank of Hakone volcano. The depth change of the seismogenic layer should be attributed to geothermal effects of the Hakone volcano.
(3) The lower seismicity areas are located in the area (Ashigara Mountains and Oiso Hills) where the Philippines Sea plate is colliding with central Honshu, and in Quaternary volcanic areas (Yugawara and Taga).
(4) Seismicity shallower than 10km has not been observed around active faults, such as Kannawa and Kozu-Matsuda faults, during observation periods.
(5) Two remarkable seismic events occurred in 1990 (M5.1) and 1994 (M4.8), the first earthquakes in 57 and 74 years, respectively. Some researchers had postulated a period of seismic quiescence in the area, but this seems unlikely.

View full abstract

This article reviews recent studies on the submarine topography and geological structure of Sagami Bay. Sagami Bay straddles a plate boundary that poses seismic hazards to densely populated parts of Japan, including Tokyo just 70km to the northeast. Beneath the bay, the northeastern part of the subducting Philippine Sea plate descends under the continental plate that includes northeastern Japan. This subduction produced devastating earthquakes in 1703 and 1923, and such earthquakes are likely to occur again in the future. Bathymetric surveys show that Sagami Bay contains many knolls and canyons. The bay also contains a deep basin. Submarine volcanic knolls dot the southwestern part of the bay. Some seismic reflection and refraction surveys have revealed clearly the geological structure of Sagami Bay to a depth of 5 to 10km from the seabed: the basement reflector in the Sagami Basin inclines toward the northeast from the Izu Peninsula side. The trough-fill layer is more than 4km thick. However, these data do not attest the plate subduction distinctly or the existence of the West Sagami Bay Fracture.

View full abstract

Five subparallel active faults striking NWW-SEE truncate hilly Miura Peninsula, on the west side of Tokyo Bay. Right-lateral offset has predoiminated on these faults during the late Quaternary, associated with some vertical component. Seismic risk is a concern in Miura Peninsula, where the densely populated city of Yokosuka is built near or across these faults.
Eleven sites were drilled or excavated across three northern active faults, Kinugasa, Kitatake, and Takeyama faults, in order to obtain faulting history, including the time of faulting and the amount and nature of offset. This research has been carried out as official project of city or prefectural agcucies, supported by funding from the Science and Technology Agency, Japan, especiaally since the 1995 Kobe earthquake, as well as by personal research projects.
I summarize here the results of these projects. On Kitatake Fault, the latest activity is estimatetd to have occurred between 1, 200-1, 400yrs BP at seven sites. Repeat time is inferred to be 1, 500-2, 000 years, based on differential thickness of strata on the two sides of the fault. Three events during the last 5, 500 years were identified at Tsukui trench log along the Taleyama Fault. The latest event occurred 2, 000-2, 200yrs BP, certainly older than that of the Kitatake Fault., implying that these two faults activated independently. On the northernmost Kinugasa Fault Neogene rocks are intensively shattered and Holocene deposits are clearly deformed. This confirms that Kinugasa Fault is certainly an active fault which activated two times since ca. 13, 000 years BP.
Repeat time is estimated to be 1, 500-2, 000yrs for Kitatake Fault and to be average 2, 000 years along Takeyama Fault. But it may be shorter, because they probably converge in one major fault underground. In any case, these two faults are approaching the next period of activity.

View full abstract

Paleo-seismic events that occurred around the Sagami Trough during the past 10, 000 years were recorded by sandy beds in Holocene bay-floor mud beneath four alluvial plains on the Boso Peninsula and Miura Peninsula of central Japan. Facies analysis indicates that normal deposition of bioturbated mud was interrupted by more than ten thin sand and sandy gravel beds, each of which fines upward from an erosional base. Some sandy beds contain cross-lamination indicative of both landward and seaward currents. These coarse deposits in the bay-floor mud are called “event deposits”.
Quantitative and qualitative analyses of fossil assemblages from the view point of taphonomy revealed the source and depositional processes of the event deposits. The event deposits include transported and mixed molluscan assemblages such as muddy bay floor and rocky coasts. This suggests the existence of reworking events around the bay area. Invasions of open marine water into the embayments account for fully marine ostracodes being found mixed with inner-bay assemblages in two of the sandy beds.
Radiocarbon ages from 137 samples support correlation of seven of the sandy beds beneath four of the plains. They designate the repeated occurrences of reworking events near Sagami Bay. Five of the beds gave ages similar to those previously obtained for emergent processes of Holocene terraces in the region. Age agreement strongly suggests that these event deposits originated from earthquake-induced tsunamis.
Rapid lowering of sea-level was shown in two event horizons with sand sheets by analyzing ostracode and molluscan assemblages. These events were probably coseismic uplifts accompanied by tsunamis.
Detection of the coseismic events from bay-floor deposits supplies a powerful information to paleo-earthquake research and also contributes to seismo-tectonic studies as described in this paper.

View full abstract

This paper describes Holocene marine deposits around Sagami Bay and discusses Holocene crustal movement in relation to crustal deformation associated with great earthquakes in the Sagami Trough. Flights of Holocene marine terraces are distributed on the coasts along the Oiso Hills and Miura Peninsula, which are uplifted regions located on the northeastern side of the Sagami Trough. Marine terraces located in the southern part of Miura Peninsula indicate that the emergences occurred around 6, 000yrs BP, 4, 000yrs BP, and 3, 000yrs BP on the basis of the ages of sea caves and archaeological data.
Remarkable uplifted features in the Holocene are observed along the northeastern part of Sagami Bay; to the north of this uplifted zone a belt of subsidence in the Holocene is located on the line connecting Hadano to Kanazawa-Hakkei through Ofuna.
The emergence of the southwestern part of the Oiso Hills around 6, 500yrs BP is inferred from the fossil molluscan assemblage, the K-Ah (Kikai-Akahoya) volcanic ash, and related carbon-14 dates. The Kozu-Matsuda active fault which delineates the western margin of the Oiso Hills did not move at the time of the emergence around 6, 500yrs BP. The Oiso Hills to the east of the Kozu-Matsuda fault have uplifted and the Morito River Lowland to the west of the fault has subsided along with the slip of the Kozu-Matsuda fault since 6, 500yrs BP.
Marine sediments that emerged around 4, 500yrs BP are present over 5m above the present sea level in Odawara city, which is located in the southwestern margin of the Ashigara Lowland. The upper limit of Holocene marine deposits descends eastward to the Morito River Lowland.
No geological evidence of emergence around 6, 000yrs BP is found in Izu Peninsula to the southwestern side of the Sagami Trough. The post-glacial Jomon tranagression continued until 4, 000yrs BP in the northern part of Izu Peninsula and until 3, 000 to 2, 000yrs BP in the southern part of the peninsula. Izu Peninsula became a site of emergence after these periods in its various region.
Tsunami deposits have been discovered in marine sediments located in submerged drowned lowlands around Sagami Bay. The ages of emergence of marine terraces have been inferred from the ages of these tsunami deposits. The history of coseismic vertical movements at Sagami Bay may be clarified by further study of tsunami deposits.

View full abstract

Studies on the sediments induced by paleoseismic events, especially tsunami deposits, are reviewed and examined.
Various types of sediments with conspicuous characteristics are produced by paleo-seismic events. The following are typical examples of paleo-seismic events in the Holocene coastal sediments that have been studied: (1) Change in sedimentary environment caused by submergence or emergence; (2) Sedimentary structure change of intruded sandy sediments caused by seismic shaking, and re-concentration of heavy minerals in the sediments; (3) Re-sedimentation of marine and lacustrine deposits by gravity flow deposits such as turbidite and subaquatic landslide sediments; (4) deposition of tsunami deposits.
Synchronous and regional characteristics are the most useful ones for the identification of these events. Relatively coarser sediments such as sandy or gravelly layers with grading structure are characteristic of tsunami deposits. The deposits also become finer towards the inland, and a frame-like structure often can be recognized at the top contact with the under lain soft sediments. Microfossils and geochemical substances of marine origin sometimes increase at the horizons of tsunami deposits.

View full abstract

Coseismic uplift probably contributed to the emergence of Holocene marine terraces around Sagami Bay. The highest of these terraces emerged about 6, 000 years ago. Its height distribution reflects subduction of the Philippine Sea Plate, collision of part of this plate (Izu Peninsula), and crustal shortening in the continental plate.
Two kinds of coseismic uplift probably occur at eastern Sagami Bay, on the southern part of the Boso Peninsula. One type, represented by the Kanto earthquake of 1923 (in the Taisho era), was due to interplate slip along the Sagami Trough. The other type is represented by the Kanto earthquake of 1703 (in the Genroku era), the hypocentral area of which consists of the area along Sagami Trough and its southeastern extension. Various combinations of Taisho and Genroku earthquakes can account for the height of the highest Holocene marine terrace on the Boso Peninsula. The best such combination gives a mean recurrence interval of 500-600 years for Taisho-type earthquakes.
Since the accuracy and resolution of the height and formative age have not been so high, a consistent explanation of the formative history of Holocene marine terraces throughout Sagami Bay area remains to be given.

View full abstract

Evidence of paleo-earthquakes, including cracks, quicksand, and landslides, can be recognized at archaeological sites around the Sagami Bay region. Based on this evidence, we can reconstruct the occurrences of paleo-earthquakes and damage caused by ground motion and faulting. In the fault zone at the northern margin of Izu Peninsula, earthquakes are estimated to have occurred at the following times: (1) about 5, 000yrs BP; (2) 2, 800-3, 000yrs BP; (3) from about 3, 000yrs BP to AD 100; (4) from AD 687 to 750; and (5) after 1707 (probably in 1853). At the southeast margin of the Tanzawa Mountains and the Oiso Hills, traces of earthquake occurrence point to the following time: (1) the 4th century and from the second half of the 5th century to the first half of the 6th century; (2) from 687 to 750; (3) the second half of the 8th century; and (4) the 9th century. From the Sagami River to the Miura Peninsula, traces of earthquakes that occurred from AD 100 to 400 and in the 9th century are concentrated along the linear zone connecting the Ebina and Kamakura areas.

View full abstract