T22B-01 INVITED
Interrogating Cascadia in Nankai (and Vice Versa)
In global subduction-zone "thermometry", Cascadia and Nankai are both warm-slab end members and share a sharp contrast with cold-slab subduction zones. A direct consequence of the warmth of the young subducting slab is its shallow dehydration. The absence of intraslab earthquakes below 80-90 km depths in Cascadia and Nankai is understood to be related to the shallow dehydration of the slab. The slab releases most of its aqueous fluids before reaching the volcanic front where mantle temperature is high enough to cause melting of H2O-rich peridotite. The scarcity of fluids beneath the arc results in feeble volcanism (more so in Nankai), but the abundance of fluids in the forearc facilitates serpentinization of the mantle wedge. A high degree of mantle wedge serpentinization and the availability of free fluid may be responsible for the occurrence of episodic non-volcanic tremor around the wedge tip with accompanying slow slip of the subduction interface (the ETS events). The warmth of the slab also causes the megathrust seismogenic zone to be located shallower than in cold-slab subduction zones. In addition to these fundamental similarities associated with warm slabs, Cascadia and Nankai also have many "fortuitous" similarities due to regional tectonics and geology, such as the development of a large accretionary prism, the presence of a splay fault that potentially contributes to tsunami generation, the margin-parallel direction of compressive stress in the continental forearc, and the margin-normal direction of interseismic shortening. We have learned from the similarities between the two subduction zones, and we are beginning to learn from their differences. For example, great earthquakes at Cascadia tend to rupture very large distances along strike and have long recurrence intervals (~500 yrs; though with ~250 yr intervals along the southern margin), as suggested by offshore paleoseismic evidence, but those at Nankai tend to have shorter rupture lengths and recurrence intervals (~100 yrs). The reason may be that the plate interface at Cascadia is much smoother than at Nankai where the subducting plate carries many seamounts and suffers severe deformation during subduction. At Nankai, there are numerous very-low-frequency earthquakes in the shallow outer wedge offshore and in the deep ETS region. A preliminary search for these events at northern Cascadia using available seismic data indicates that similar events may also be present at Cascadia but by no means as abundant as at Nankai. The difference may be related to their different lapse times since the previous great earthquake and states of stress relaxation: A great earthquake occurred only a few decades ago at Nankai but over three centuries ago at Cascadia. One obvious difference surely caused by the different lapse times is the geodetically constrained interseismic elastic strain rate in the forearc: The rate is much higher at Nankai than at Cascadia. Some differences between the two subduction zones are yet to be investigated. For example, many earthquakes have been detected in the trench area at Nankai, but there is an apparent seismic quiescence in the same area at Cascadia. A plan is being developed to investigate the apparent quiescence at Cascadia using ocean bottom seismographs.
T22B-02
Changes in In Situ Stress Across the Nankai and Cascadia Convergent Margins From Borehole Breakout Measurements During Ocean Drilling
Borehole breakouts are commonly observed in borehole images shortly after drilling of continental margin sites. This study aims to compile and compare these results to determine what in situ shallow stress measurements can tell us about the larger scale tectonic regime. Recent Logging While Drilling resistivity images across the Kumano transect of the Nankai subduction zone, during Expedition 314, Stage 1 of the IODP NanTroSEIZE project, add to this dataset. Expedition 314 site data within the prism (C0001, C0004, C0006, including the megasplay fault system which may overlie the seismogenic updip limit) suggest maximum compressive stress (SHmax) is perpendicular to the margin (not parallel to the convergence vector) but is rotated through 90° at the forearc basin site (C0002). These results may point to changes in stress state of the shallow forearc from east to west: compression in the aseismic active prism (with evidence of strain partitioning of oblique convergence); and extension above the updip seismogenic zone suggesting focus of plate coupling at the plate boundary and not in the shallow forearc. Further south, ODP Leg 196 drilled the prism toe (808) with breakouts indicating SHmax parallel to the convergence vector, in contrast to Exp. 314 results. The stress state in the shallow prism at Site 808 may be affected by nearby seamount subduction or may represent differences in strain partitioning. On the Cascadia margin, two drilling legs have collected LWD borehole images (Leg 204 and Exp. 311). Leg 204 drilled 3 sites at hydrate ridge in the C Cascadia outer prism with breakout orientations variable between closely spaced sites. Prism fold axes are parallel to the margin so we might expect SHmax perpendicular to the margin as in Exp. 314. Deviations from this orientation may reflect local and surface effects (Goldberg and Janik, 2006). Exp. 311, N Cascadia, drilled 5 sites across the prism with breakouts in LWD images. Subduction is not oblique here, in contrast to the other sites discussed, and most sites indicate SHmax almost parallel to convergence and normal to major fold axes. In one case, the in situ stress orientation is also compatible with shallow normal faulting from seismic data. Site 1325, in a slope basin, deviates from this orientation and may reflect local processes. Borehole breakouts within the shallow forearc of convergent margins are often in agreement with other indications of regional tectonic stress and may be indicative of processes at depth. Deviations may represent local stresses due to gravitational processes.
T22B-03
Constraining in situ stress tensors in the upper Nankai accretionary wedge using borehole wall failures observed in the LWD boreholes, C0001, C0002 and C0004
We analyze borehole failures (breakouts and drilling-induced tensile fractures) observed through logging- while-drilling (LWD) technology in three vertical boreholes C0001, C0002 and C0004 drilled during IODP Expedition 314. The main goal of this analysis is to constrain and/or estimate the orientations and magnitudes of the in situ stress field in the upper Nankai accretionary wedge. The three boreholes are located in different sites (respectively, at the southern forearc edge, within the forearc basin deposits, and at a shallow part of the megasplay site), penetrating multiple lithographic and structural units divided by major and minor geological discontinuities such as unconformities, thrust sheets and splay faults. In all three boreholes, clear and nearly continuous traces of drilling-induced borehole wall failures were detected in the LWD resistivity images. In each borehole, systematically oriented borehole wall failures indicate the azimuths of far-field in situ horizontal principal stresses. The maximum horizontal principal stress changes drastically between the forearc basin and the prism sites. A preliminary attempt to constrain stress magnitudes based on widths of logged breakouts, occurrence of drilling-induced tensile fractures, and physical-property derived rock strengths indicates that the magnitudes of the present-day in situ stresses vary depending on different lithologic domains divided by major structural boundaries. The stress ratio of horizontal to vertical stress is higher in the accretionary prism than in the forearc basin sediments. This observation indicates that the former domain is under more lateral compression than the latter, which suggests that the stress state in the accretionary prism is characterized by a reverse or strike-slip fault stress regime. The stress state in the forearc basin is instead represented by a normal fault stress regime. This result is compatible with regional and macroscopic scale faults observed in the seismic sections. Although the results will be more accurately constrained when we have laboratory-determined rock strength, an essential parameter used for breakout analysis, our up-to-date result suggests that in situ stress is segmented due to the major structural boundaries.
T22B-04
Breakouts in Soft Sediment: Applications to IODP Expedition 314
IODP/ODP boreholes show enlargements in shallow sediments interpreted as breakouts. These features begin at depths less than 200m in sediments with porosities > 50% (e.g. Sites C0001, C0004). From a theoretical perspective borehole failure in sediments should be treated using critical state soil mechanics, rather than the Coulomb failure criterion, applicable to rocks. Borehole failures in ODP/IODP holes in the Nankai Trough have been defined by resistivity imaging recorded about 3 m above the bit. Given typical penetration rates of 25 to 30 m/hr the borehole failure is then imaged < 10 minutes after the passage of the bit. Cutting the borehole provides a nearly instantaneous reduction in lateral confining stress. This reduction in confining stress results in a strain path moving away from the consolidation line towards the critical state line and consequent failure. In fine-grained sediments subject to deformation and failure in < 10 minutes, the stress path is undrained and constrained to a plane of no volume change. This stress path approximates Coulomb failure. Laboratory experiments with ODP/IODP samples from Nankai Trough holes with breakouts show angles of internal friction of 20-28 deg. and discrete surfaces of shear failure deviating 30 to 35 deg. from sigma 1, consistent with Coulomb failure. Most borehole failures are imaged as low resistivity zones, due to spalling that increases the amount of conductive fluid between the RAB sensors and the formation. But a Gulf of Mexico IODP Site (U1322) shows high resistivity RAB tracks down the borehole. This image may represent an incipient state of ductile failure where sediment is bulging into the borehole in the Sh min direction, prior to spalling and forming a breakout. Laboratory experiments support such ductile failure in samples from very shallow depths, whereas the Coulomb-style failure occurs at higher degrees of consolidation. The absence of these high resistivity borehole failures at convergent margins (observed to date) may reflect superimposed tectonic consolidation.
T22B-05
In Situ Stress Determinations from Anelastic Strain Recovery (ASR): Preliminary Results and Comparisons to Borehole Breakout and Core-scale Fault Data from IODP Expeditions 314, 315 and 316 to the Nankai Trough
IODP Expeditions 314, 315 and 316 drilled several sites across the Nankai accretionary prism, penetrating and sampling both accreted sediments and the prism's sedimentary cover, including sediments of the Kumano Basin. As part of the coring program during Exp 315 and 316, we collected 4 whole-round sediment cores for ASR analysis; all samples were prepared for ASR data recovery and analysis following the guidelines in Lin et al. (2007). Anelastic normal strains, measured every ten minutes in at least nine directions, including six independent directions, were used to calculate the anelastic strain tensors. Three of the four samples showed coherent strain recovery over periods of 1 to 4 weeks whereas the fourth showed irregular strain recovery and was not analyzed further. Two of the samples with coherent strain recovery were from Unit III of the Kumano Basin (C0002B-45R-4 884.33-884.39 and C0002B-48R-2 911.765-911.905) whereas the third sample (C0006F-9R-1 467.56-467.67) was from the toe region of the prism, less than 1 km from the deformation front. All of the samples are composed of silty clays or hemipelagic muds with relatively high porosities (~40%). Calculations of the reduced stress tensors (orientations and ratios) show that both samples from the Kumano Basin record steep to nearly vertical maximum principal stress axes (sigma 1). A steeply plunging sigma 1 in Kumano Basin is consistent with the inversion of core-scale fault data from this site as well as seismic reflection data that show extensional structures throughout this region of the basin. In addition, one of the samples from the Kumano Basin has been reoriented with paleomagnetic data and shows that the minimum principal stress trends N40W, nearly identical to the trends of the minimum stresses interpreted from borehole breakout data (N38W) and the fault data (~~NW). These results confirm the validity of using the ASR technique in determining in situ stresses in drilled cores - even in partially lithified muds with relatively high porosities. Calculation of the reduced stress tensor for the sample from Site C0006 (near the prism toe) also yields a steeply plunging (~60°) maximum principle stress. Although core-based observations have documented abundant extensional structures consistent with the ASR results, seismic reflection data show numerous low-angle thrust faults. This part of the wedge therefore may be gravitationally relaxing between great earthquakes, for example, or there may be stress discontinuities within the wedge, for example across individual thrust sheets. Kinematic analysis of reoriented core-scale faults will help resolve these conflicting observations.
T22B-06 INVITED
Shallow very-low-frequency earthquakes around Japan: Recent studies and observation
Very-low-frequency (VLF) earthquakes have been observed in three regions around Japan. (1) Deep VLF earthquakes have occurred in the down-dip part of the Nankai subduction zone [Ito et al., 2007]. (2) Shallow VLF earthquakes have occurred within the accretionary prism in the up-dip portion of the Nankai subduction zone [Obara and Ito, 2005; Ito and Obara, 2006]. The stress drops of these shallow VLF events were very low, in the range 0.1--10kPa; this corresponds to 0.1--1% of the range for ordinary earthquakes [Ito and Obara, 2006]. Ito and Obara [2006] suggested that the largest shallow VLF earthquake (MW 4.0) occurred on a circular fault of radius ~5--10 km. They proposed that the shallow VLF events were related to numerous reverse fault systems located in areas of high fluid pressure within the accretionary prism. (3) Shallow VLF earthquakes have occurred in the region off Tokachi, northern Japan, along the Japan Trench [Asano et al., 2008], where the Pacific Plate subducts beneath the Japanese land area. The occurrence of these shallow VLF earthquakes suggests that VLF events can occur on the plate boundary at depths shallower than that of the main seismogenic zone [Asano et al., EPS, 2008]. The megasplay faults in the Nankai subduction zone are observed to generate a reverse-polarity reflection on seismic reflection profiles [Park et al.,2002]; this may indicate the existence of an elevated fluid process in the fault zones [Shipley et al., 1994]. Hydrodynamics phenomena responsible for the seismic signals detected by ocean bottom seismometers were first reported by Brown et al. (2005) using osmotically-driven fluid flow meters (CAT meters); these meters were used to detect temporal changes in the rate of cold seepage of a shallow subduction system in the regions of the Costa Rica subduction zone. The Pacific plate is subducting beneath Tohoku, northeastern Japan, at the Japan Trench. An aseismic slip has been observed to occur as a post- seismic slip following the occurrence of large earthquakes. Tsunami earthquakes, a type of slow earthquakes, have also occurred in the vicinity of the Japan Trench; thus far, there have been no observations of non-volcanic tremors, VLF earthquakes, and short-term slow slips in NE Japan, and the lack of observations can be attributed to the difficulty in detecting slow earthquakes near the trench. Recently, we have deployed ocean bottom seismographs and have set up geodetic and hydraulic stations in northeastern Japan to detect and observe shallow slow earthquakes and the corresponding transient hydrotectonic processes. The CAT meters were installed on cold seeps that were discovered by diving surveys of the manned submersible SHINKAI 6500. We found Calyptogena, which suggested the existence of a cold seep, during the diving survey. The seeps are distributed near a splay fault that was detected from a seismic reflection image [Tsuji et al., 2008], suggesting that the fluid in the cold seeps migrates from the splay fault to the seafloor. The Calyptogena colonies are distributed along the strike of the landward slope. We have also developed a simplified ocean-bottom benchmark (SOBB) that comprises three types of sensors; short-period seismometers, broadband seismometers, and pressure gauges.
T22B-07
Frictional and Hydrologic Properties of Fault Gouge From a Mega-splay Fault at the Nankai Subduction Zone, Sampled by NanTroSEIZE IODP Drilling
Characterizing the frictional and hydrologic properties of natural fault gouge is crucial to understanding the generation and nature of earthquakes and the strength of crustal faults. Previous work on natural gouge analogues has shown that gouge mineralogy, specifically the presence of clays, may control shear strength, frictional stability, and permeability. However, there is a need for quantification of these properties in natural fault gouges under in-situ conditions. We report on laboratory experiments examining the frictional and hydrologic properties of natural fault gouge and wall rock from a major out-of-sequence thrust splay fault within the Nankai accretionary complex, collected during IODP Expedition 316 from Hole C0004 at a depth interval of 119-439 mbsf (meters below sea floor). Samples were dried at low temperature (to prevent clay transformation during drying) and disaggregated to a grain size of < 106 μm. Shearing and permeability experiments were conducted in a servo-controlled apparatus within a pressure vessel using the double-direct shear configuration under saturated, controlled pore pressure conditions and using brine of seawater composition as a pore fluid. Effective normal stress was either 3 MPa, approximating the in-situ conditions, or 25 MPa representing an extrapolation to greater depth. Pre- and post-shear, fault-normal permeabilities were measured by controlling the pore pressure gradient across the gouge layer and measuring the resulting flow rate. During shearing, pore pressure was controlled at one sample boundary, and the other boundary was closed to fluid flow; pore pressure was monitored at the closed boundary to evaluate pore pressure control and communication within the shearing layer. Layers were initially 5-6 mm thick with nominal contact dimensions of 6.1 cm by 5.4 cm. Experiments were conducted under constant shear velocity boundary conditions of 1-30 μm/s. Preliminary results at effective normal stress of 3 MPa show that the gouges exhibit coefficients of friction μ ranging from 0.43-0.55 across the depth range of the fault zone. Strain weakening reduces these values to 0.37-0.41 at shear strains of ~20. Velocity-stepping tests show that the gouges are strongly velocity-strengthening, with values of (a-b) > 0.0025. Slide-hold-slide tests indicate low healing and compaction rates. Permeability of the sheared samples ranges from 1.6 - 4.5x10-18 m2.
T22B-08
Frictional Properties of Fault Gouges Recovered from NanTroSEIZE Expedition 316 Drilling
Along subduction interfaces, the updip limit of the seismogenic zone is an important issue, notably regarding tsunami generation. One of the factors controlling this transition is the frictional properties of rock assemblages formed along active faults crossing accretionary prisms. We have thus begun a broad experimental survey of the frictional properties of fault materials retrieved from splay faults cored during NanTroSEIZE Expedition 316 drilling. Laboratory experiments are being conducted in a high-pressure, rotary-shear gas apparatus, employing estimated or measured in situ normal stresses, temperatures, pore pressures, and pore fluid chemistries. Importantly, the rotary-shear geometry allows for tests to large shear displacements of ~~1 m or more, allowing us to explore the tendency for shear localization and associated velocity-weakening frictional behavior necessary for unstable slip, over shear strains of at least 1000. Pore fluids (simulated seawater) are prepared by duplicating the major element chemistry (Na, Ca, Cl, Mg and K) of pore fluids recovered via shipboard sampling, achieved by mixing appropriate reagent-grade salts with deionized water. Unconsolidated samples are taken as-is from the sealed sample bag and placed between permeable sandstone forcing blocks. Samples are then saturated with simulated seawater, and the pore fluid pressure is subsequently maintained at a constant value throughout each experiment. A test on a 1-mm thick, clay-rich microbreccia recovered from the megasplay fault zone at a depth of 275 mbsf at Site C0004, Section 29R-2, for a normal stress of 31.6 MPa, pore pressure of 25.8 MPa, sliding velocity of 0.1 to 10 µm/s, and total slip of ~1.1 m, yields an initial peak value of the friction coefficient of ~0.6. With increasing strain, however, the friction coefficient decreases dramatically, obtaining a steady-state value of 0.21 at ~90 mm of slip followed by a monotonic increase to 0.26 at a slip of 1 m. This dramatic decrease in friction is accompanied by a change from 'velocity-strengthening' friction behavior for slip <200 mm to 'velocity-weakening' behavior for slips of 200 mm to 1.1 m. Microstructural analyses conducted after the experiment indicate that slip is localized on a discrete surface, suggesting that slip localization is associated with velocity-weakening friction. Velocity-weakening friction, a necessary condition for unstable slip, may allow earthquake ruptures that nucleate at depth to propagate unstably along splay faults, perhaps allowing for the generation of tsunamis. The experimental results to date highlight the likely critical role of slip localization and the associated velocity weakening frictional behavior in determining the stability of splay faults of the Nankai subduction zone.