BACKGROUND
The scientific importance of establishing long-term geophysical stations in deep oceans has been acknowledged by the earth sciences and the Ocean Drilling Program (ODP) communities and is expressed in various articles (COSOD II, 1987; Purdy and Dziewonski, 1988; BOREHOLE, 1995; Montagner and Lancelot, 1995; Ocean Drilling Program Long Range Plan, 1996). In essence, we want to understand active processes driving Earth's dynamics from a global to a regional scale, but 71% of Earth's surface is covered by oceans that can only be probed by using state-of-the-art digital sensors linked with land-based stations. Many sensors, whose locations will be carefully selected to maximize results, are needed around the world to attain the goals of the international geoscience programs. We have selected the western Pacific area for installation of ocean-bottom sensors because it is ideal for addressing problems related to plate subduction.
In the Japan Trench area, seven large (magnitude [M] > 7) interplate events occurred in the past 30 yr between 38° and 41°N. Recent large events are the 1968 Tokachi-Oki earthquake (at ~41°N with a moment magnitude [Mw] of 7.9) and the 28 December 1994 Far-off Sanriku earthquake (at ~40°N with Mw = 7.7). These events, however, are not sufficient to account for the subducting rate of about 10 cm/yr. Thus, the seismic coupling seems much smaller along the Japan Trench (35°-41°N) than compared with the Kurile Trench or Nankai Trough regions, which have a higher seismic energy release rate. Subduction at the Japan Trench may be proceeding mainly by stable sliding with unstable sliding events that are either relatively small (surface-wave magnitude [Ms] <8) and occur frequently, or with truly large events that occur infrequently.
There is a third important category whereby the subduction rate is accommodated by episodic aseismic events of time constants on the order of 10 min to several days (slow earthquakes). Such events, if they exist, are presently extremely difficult to detect. Kawasaki et al. (1995) reported that an ultra-slow earthquake, estimated to have a Mw of 7.3-7.7, accompanied the 1992 Off-Sanriku (located at 39.42°N, 143.33°E; Mw = 6.9) earthquake based on strain records observed ~120-170 km away from the source. A postseismic strain of 10-7 to 10-8 with a time constant of about a day was observed by quartz-tube extensometers (devices that measure relative strain). Historically, in the same area, the 1896 Sanriku tsunami earthquake (Mw ~8.5 but body-wave magnitude [Mb] ~7) killed about 22,000 people. Tsunami earthquakes rupture over a much longer time than normal earthquakes (Tanioka and Satake, 1996), supporting the notion that slow earthquakes may occur off the Sanriku coast.
More recently, the Japanese global positioning system (GPS) network has revealed a postseismic motion of northern Japan after the 1994 Far-off Sanriku earthquake (M = 7.2), which can be explained by a stress diffusion model that assumes slow slip on the earthquake fault (Heki et al., 1997). A different, but previously more prevailing interpretation, is that the postseismic deformation is caused by aseismic slip at a deeper depth extending down from the seismogenic zone. If such a slow slip really occurred in the vicinity of the normal seismogenic zone, then strainmeters in the proximity would have not only recorded signals much larger in magnitude, but also would have resolved how and where the slip initiated relative to a normal earthquake, and how it proceeded. Furthermore, one can test if aseismic and episodic slips occur irrespective of normal earthquakes.
The strain waveforms of slow earthquakes are of a ramp type. The amplitudes of strain steps decay inversely proportional to the distance cubed, much more rapidly than seismic waves. It is essential, therefore, to measure the strain signatures as near to these events as possible (within 20 km for an event equivalent to Mw of 7.0; e.g., Johnston et al., 1990) to estimate how the regional tectonic stress affects earthquake occurrences.
The primary objective is to establish long-term borehole observatories, and so once the instruments are installed, they must be serviced for data analyses, distribution, and archiving. There is an ongoing national program within Japan to achieve this (Ocean Hemisphere Network Project). Initially, power will be supplied to the observatories by a battery pack and data will be retrieved by a remotely operated vehicle (ROV). Eventually, the goal is to connect the observatories to a fiber- optic cable that will supply power and allow the data to be retrieved. A new fiber-optic cable owned by the University of Tokyo already exists and currently terminates near Site JT-1C. Once Site JT-1C proves operational, connections will be made to supply power, send commands, and retrieve data in real-time on land. A 50-km cable extension is planned to connect Site JT-2G as well. These stations will make invaluable additions to the existing geophysical network over the western Pacific. The data will eventually become accessible worldwide through the Internet.