High Pore Fluid Pressure May Cause Silent Slip in the Nankai Trough

See allHide authors and affiliations

Science  28 May 2004:
Vol. 304, Issue 5675, pp. 1295-1298
DOI: 10.1126/science.1096535

This article has a correction. Please see:


Silent-slip events have been detected at several subduction zones, but the cause of these events is unknown. Using seismic imaging, we detected a cause of the Tokai silent slip, which occurred at a presumed fault zone of a great earthquake. The seismic image that we obtained shows a zone of high pore fluid pressure in the subducted oceanic crust located down-dip of a subducted ridge. We propose that these structures effectivelyextend a region of conditionallystable slips and consequently generate the silent slip.

The Nankai Trough, where the Philippine sea plate is subducting beneath the Eurasian plate, is considered unique because it has a well-investigated recurrence interval and displays rupture zone segmentation of magnitude 8–class earthquakes (1). Tsunami wave (2, 3) and seismic wave (4) modeling studies have shown that the slip of the 1944 magnitude 8 Tonankai event propagated across all of segment C (Fig. 1) but that no slip extended into segment D (the Tokai segment). This means that the Tokai segment has been seismically quiet for longer than the average recurrence interval of megathrust earthquakes in the Nankai Trough.

Fig. 1.

Topographic map of central Japan and the location of the onshore-offshore wide-angle seismic profile. Stars indicate land explosions. Four segments, A to D, are rupture zones compiled from historic earthquake data (1). Blue contours show the slip distribution (from 0 to 3 m) of the 1944 Tonankai earthquake (EQ), determined from strong ground motion data (4). The dotted blue line indicates the seaward limit of the back stop of the present day accretionary system proposed by Nakanishi et al. (27). Thick pink lines and yellow line indicate locations of the subducted south ridge, the north ridge, and the highly reflective/high Poisson's ratio zone, respectively. The square, shaded in blue, represents the back slip rates of the plate interface obtained by GPS data (11). The red arrows indicate estimated silent slips at the subduction interface (5). NAP, North American plate; PP, Pacific plate; EP, Eurasian plate; PSP, Philippine sea plate.

In addition to these coseismic slip phenomena, a dense Global Positioning System (GPS) network detected an interseismic silent-slip event in the Tokai segment (∼2 cm of surface horizontal displacement) (5). This slip was localized in a region of about 60 by 60 km in the coastal area of the Tokai segment. Similar silent-slip events have occurred in this region episodically over the past 30 years (6). Silent slips may have been reported in other subduction seismogenic zones (7), but no structural evidence for the cause of the slips has been documented.

We carried out an active-source seismic experiment along the 485-km-long onshore-offshore profile (Fig. 1) to investigate the coseismic and interseismic slip phenomena in the Tokai segment. The wide-angle seismic data from land explosions (maximum 500 kg) were recorded by 70 ocean-bottom seismometers (OBSs) and 328 land-based stations. In addition, signals from an airgun array (197 L) were recorded by all OBSs and 63 land-based stations deployed at the southern part of the onshore profile (about 100 km from the coastline) [supporting online material (SOM) text and figs. S1 to S3]. Multichannel seismic (MCS) reflection data were also acquired along a part of the profile. Here, we present a seismic velocity image from the combined onshore-offshore data set. The onshore reflectivity data and a part of offshore reflectivity data were also processed. The densely deployed OBSs (spacing of ∼3 km), together with onshore recording of offshore shots, provided a high-resolution seismic image from the offshore to the coastal region of the profile. The large land explosions also resolved an image of the subducted plate along the onshore part of the profile beneath Japan. In addition to these active-source seismic studies, we compiled passive-source (earthquake) seismic tomography results for central Japan (8). First-arrival refraction tomography (9) was applied with the use of the active-source wide-angle seismic data to produce a velocity image, and forward modeling of the observed prominent reflection phases was used to map the geometry of the major reflectors (SOM text and figs. S4 and S5).

The seismic velocity and reflectivity images show several regions of crustal thickening down to a depth of 45 km, which we interpreted as subducted ridges (Fig. 2). In the offshore part of the profile, we recognized subducting ridges at 300 km (the north ridge) and 350 km (the south ridge) with maximum thicknesses of 20 and 12 km, respectively (SOM text and fig. S6). The ridge structures in the Tokai segment support the rupture propagation of the 1944 Tonankai earthquake (10). GPS data (11) suggest a strongly coupled region where the back-slip rate is about 35 mm/year in the Tokai segment. The north ridge is situated in the area for which the strong coupling is predicted (12).

Fig. 2.

(A) Seismic velocity image with reflection points (red dots) of prominent reflection phases shown. Black dotted line shows the interpretation of the crustal block. The lighter colored regions indicate where no seismic rays were sampled. Triangles and circles along the top frame indicate the locations of the land explosions and OBSs, respectively. The initial model of the tomography was constructed by referring to previous studies (10, 28). Travel time residuals calculated from the final model are shown in fig. S5. Also see SOM text and figures for details of the modeling procedure and the resolution. (B) Seismic reflectivity image. The offshore part of the image consists of a combined section of the poststack depth migration of the MCS data and the prestack depth migration of the OBS data; it is modified from Kodaira et al. (10). The framed region was reprocessed with the prestack depth migration shown in Fig. 4. Also see supporting SOM text for details of reflectivity imaging. Labeled arrow-heads show reflectors: A, top of the subducted crust; B, intracrustal reflection interpreted as the deeper extension of the Median Tectonic Line (29); C, Island arc Moho; D and D′, top of the middle crust of Izu island arc; E, bottom of the Neogene-Quaternary accretionary prism (27); F, top of the subducted crust; G, bottom of the subducted crust. Numbers in white boxes indicate seismic P-wave velocities (km/s). Unlabeled arrowheads indicate ends of the reflector A, B, and C. A question mark means that a location of the reflector C is obscure in this section. NMO, Normal move out; PSDM, Prestack depth migration.

The subducted crust beneath the Japanese island, according to the image, is a highly reflective interface from a depth of 25 to 45 km (Fig. 2). A slightly upward doming structure (2 km high), recognized at 190 to 230 km (Fig. 2), might be attributable to an even deeper ridge system (SOM text and fig. S7). A landward-dipping zone with a high Poisson's ratio (more than 0.34) is located exactly at the highly reflective subducted crust (Fig. 3). Moreover, the compilation of the seismic tomography results indicates that, in the Nankai seismogenic zone, the observed landward-dipping zone with a high Poisson's ratio only exists in the Tokai segment, which is where the subducted ridge system is inferred.

Fig. 3.

(A) Poisson's ratio structure obtained from passive-source (earthquake) seismic tomography (8) superimposed on the seismic velocity structure from Fig. 2A (only contours and reflectors). A significantly high–Poisson's ratio zone is observed at the subducted crust. (B) Expected variation of critical stiffness, kc = –(ab) (σnPf)/Lc, (solid curve) and the slip modes across central Japan. The dotted curve shows a proposed variation of kc for a typical subduction seismogenic zone in which neither ridge subduction nor high–pore pressure zone are observed (24).

Laboratory experiments measuring P- and S-wave velocities as a function of confining pressure and pore pressure for basalt samples have shown that Poisson's ratio increases to more than 0.35 as pore pressure increases at a constant confining pressure (13). We therefore propose that a zone of high pore fluid pressure is the cause of the highly reflective nature and the high Poisson's ratio in the subducted crust. Notably, the zone of high pore fluid pressure reaches the subduction interface (the top of the subducted oceanic crust) where the silent slip may occur. This is because the highly reflective nature at the subduction interface indicates a large acoustic impedance contrast.

A possible mechanism for generating a zone of high pore fluid pressure in subducting oceanic crust is a dehydration of hydrous minerals in the oceanic crust. According to a phase diagram of hydrous mid-ocean–ridge basalt (14), the dehydration reactions of low-grade hydrous phases occur as the temperature and pressure increase. The depth-temperature curve for the central Nankai crossing segment C (Fig. 1) (15), plotted on the stability diagram for these phases (14), shows that the highly reflective oceanic crust with a high Poisson's ratio is situated outside the stability area of the low-grade hydrous phases, and several dehydration reactions continuously could take place. Although the depth-temperature curves in the Tokai segment have not been published, a thermal structure at the Tokai segment may be as warm as the thermal structure at segment C. To determine this, we estimated the temperature for a steady-state subduction of a young oceanic lithosphere (16) and used the age (17) and the convergence rate (18) at the Tokai segment and segment C. A heat-flow observation (19), which shows a slight decrease of the heat flow from segment C to the Tokai segment, also suggests that the Tokai segment is not thermally affected by the present volcanism at the Izu-Bonin arc.

The precise seismic reflectivity image along part of the profile (Fig. 4) shows several fractures cutting into the subducted ridge, which are interpreted to have formed during the subduction and collision of the ridge system (20, 21). These fractures may function as conduits that transport water into the subducted crust, and consequently, concentrations of hydrous minerals may be expected within the subducted ridges.

Fig. 4.

(A) Prestack-depth migrated section of part of the MCS profile, and (B) interpretation. Location of this plot is shown in Fig. 2. The horizontal axis is measured from the northern end of the profile. Several fractures cutting through the subducted ridge (south ridge) are imaged. Similar fractures cutting into the crust were imaged in the incoming and subducted crust in the Tokai segment (20, 21). These fractures are considered to form conduits that transport water into the crust. See SOM text for details of reflectivity imaging.

The subducted ridge structures can explain the interseismic silent slips observed in the coastal region of the Tokai segment (5). According to laboratory experiments on rock friction (2224), stable and unstable (i.e., seismogenic) slips at subduction interfaces depend on the critical stiffness of the fault, kc = –(ab)(σnPf)/Lc, where σn is normal stress, Pf is pore fluid pressure, (ab) is the velocity dependence of steady-state friction, and Lc is the critical slip distance. When (ab) is positive, stable slip occurs without an earthquake (kc < 0). When (ab) is negative, slips are either unstable (k < kc, where k represents system stiffness considering the entire subduction system as a simple spring-slider model) or conditionally stable (0 < kc < k), which includes silent slip (24).

A synoptic model for depth variations of kc for a typical subduction zone interface has been proposed (24) (Fig. 3). The width of the conditionally stable region (0 < kc < k) is quite narrow in this model. In contrast, the expected variation of kc across central Japan, on the basis of our imaged structure (bold line in Fig. 3), suggests a wider region of conditionally stable slip. This is because Pf may approach σn (i.e., σnPf approaches zero) in a high–pore pressure zone, and the critical stiffness of the plate interface would decrease as a result, maintaining a conditionally stable region over the high–pore pressure zone. The slip mode of the deeper part of the high–pore pressure zone is stable because of crystal plasticity resulting from high temperatures. A recent geodetic study (25) shows an abrupt decrease in coupling at a depth of 20 km and full decoupling at a depth of 36 km. This result quantitatively supports our estimated variation of kc.

Yoshida and Kato (26) simulated silent slip by considering a block-spring model of two degrees of freedom in which one block (block 1) was assumed to be unstable (k < kc) and the other block (block 2) was conditionally stable (0 < kc < k). This is the situation observed in central Japan. It has been demonstrated (26) that block 2 slips as an episodic slow slip from the time when shear stress reaches the steady-state frictional strength after a locked period and is followed by a dynamic slip that is triggered by block 1. We thus believe, from the expected depth variation of kc (Fig. 3), that the slip behavior of the Tokai silent slip is consistent with that of the simulated block 2.

We conclude from our structural results that the Tokai silent slip was caused by the zone of high pore fluid pressure in the deeper portion of the locked zone. This zone of high pore fluid pressure is the result of dehydration of the subducted ridge, and it substantially extends the conditionally stable region. Episodic silent slip would make the recurrence interval in the Tokai segment more complex, because episodic silent slip in the deeper part would cause an increase in shear stress in the strongly coupled region (26). We also expect that silent slips may precede a catastrophic slip in the strongly coupled region where the subducted ridge exists.

Supporting Online Material

SOM Text

Figs. S1 to S7

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article