Preliminary scientific assessment

Objective 1: identify and sample the fault that ruptured during the Tohoku earthquake

The primary scientific motivation for this expedition was to gather time-sensitive information on the stress state, chemical processes, and energetics of the Tohoku earthquake rupture at the toe of the prism. A prerequisite to achieving these goals was to locate, sample, and instrument the fault that slipped during the Tohoku earthquake. To this end, a major accomplishment of the expedition was the successful completion of three holes at Site C0019 (JFAST-3) that reached the depth of the regionally distinct seismic reflectors generally interpreted as the top of the igneous basement of the subducting Pacific plate (e.g., Tsuru et al., 2002; Kodaira et al., 2012). After the completion of the LWD/MWD hole to the target depth, we confidently assumed we had drilled across the zone of the Tohoku earthquake rupture as well as across the plate boundary interface. Moreover, contrary to expectations prior to drilling, borehole stability was not a significant issue at any depth in the hole. Thus, we concluded that hole conditions would not prevent installation of a temperature measurement observatory and that from temperature measurements we could verify the location of the earthquake rupture, provided we could identify, locate, and instrument the fault(s).

The original operations plan to use LWD/MWD data to locate the fault zone that was ruptured during the Tohoku earthquake required an extremely quick analysis that, in hindsight, was overly optimistic. As a result of operational delays, however, we had additional time to process and analyze the logging data more completely. Thus, we were able to confidently identify two significant fault targets for instrumentation and coring. The shallower fault target, the 720 fault, appeared as a meter-thick, highly conductive layer within a zone possibly tens of meters thick containing variably dipping sediment cut by conductive and resistive fractures. This feature seemed to match expectations for an earthquake rupture zone (i.e., a localized shear zone), possibly composed of gouge, within a broader zone of fractured and brecciated sediment (e.g., Sibson, 2003; Vannucchi and Tobin, 2000). However, the findings of no apparent change in lithology, bedding orientation, or borehole breakout patterns across the zone does not support the interpretation of a large displacement fault. The deeper fault target, the 820 fault, was identified on the basis of an apparent marked change in structure and lithology. Specifically, image log data showed the prism comprises a northeastward striking but variably dipping stack of sediments, as would be produced by folding and faulting during northwestward convergence and accretion. Steeply dipping beds near the base of the prism are truncated and in sharp contact with a very shallow dipping, concordant sequence of interbedded sediments that appeared to coincide with the regional seismic reflector associated with the top of the oceanic crust of the Pacific plate. The resistivity and gamma ray log response of the sediments below the contact indicate distinct lithostratigraphic units, unlike the log response of the sediments in the prism above. Overall, the abrupt structural and lithologic change across the contact is strongly suggestive of a plate boundary décollement between a deformed prism above and subducting oceanic crust below. However, the contact itself is unremarkable in image logs and displays little evidence of an associated fractured or brecciated zone with lowered resistivity as might be expected for a plate boundary fault zone recently ruptured by a great earthquake.

The two faults identified on the basis of LWD/MWD data were treated as equally important targets in planning the temperature measurement observatory and the coring hole. Accordingly, the location and spacing of the MTLs in the observatory was modified to acquire adequate measurement of temperature transients near either of the targets. The installation of the temperature measurement observatory was prioritized up to the point in time the underwater television (UWTV) cable failed and could not be repaired. Coring operations commenced once it was clear the observatory could not be installed, but the remaining time was very limited. Accordingly, we devised a plan for coring only across each of the primary targets and that included taking very short cores over the several meters spanning the fault targets.

Shipboard observations of the samples collected from the second coring hole confirmed the inferences drawn from the LWD/MWD data set and clarified the nature of the targeted fault features. All cores taken from the prism (i.e., from depths shallower than the 820 fault) consist of clayey to silty mudstones comprising terrigenous silt and clay, vitric ash, and biogenic silica. Measurements of bedding orientation in cores from the prism show locally variable dip, as well as the existence of discrete changes in dip as might occur across faults. The entire suite of core samples taken from the prism is variably dissected by dark shear bands and open fractures that are tectonic in origin or other tectonic discontinuities that were opened by drilling. A prominent fractured and brecciated zone extends from 719 mbsf to deeper than 725 mbsf and bounds the prominent low-resistivity 720 fault. Several core samples from the fractured and brecciated zone contain minor faults, the largest and most prominent of which is a high-angle reverse fault that occurs at the same depth as the low-resistivity feature in the image logs. Although no geochemical anomaly was identified in the fractured and brecciated zone near the 720 fault, a local H2 and chlorinity anomaly was documented near 700 mbsf. The increase in H2 may have been generated by recent faulting, and the local reduction in chlorinity could reflect focused flow of fluids from a deeper source.

The data from core samples support the overall interpretation made on the basis of logging data that the 820 fault is the plate boundary décollement between the deformed sediments of the prism above and the basal sedimentary strata and igneous oceanic crust of the incoming Pacific plate below. Core samples clearly document an abrupt and marked change in bedding dip at the 820 fault; although it is not at all apparent in the logging data, a zone of highly sheared clay is present below the contact and was sampled by the coring. The zone is between 1 and 5 m in thickness and is characterized by a scaly fabric that, locally, is penetrative to the millimeter scale. Immediately above and below the sheared layer, visual and X-ray CT observations document that the deformation intensity of the sediment decreases rapidly with distance from the layer in a manner consistent with typical fault-related damage patterns. Lithologically, the multicolored clay layers and chert in the footwall of the 820 fault closely match the descriptions of the pelagic sediments recovered from the base of the drill hole on the Pacific plate east of the Japan Trench (Leg 56 Site 436; Shipboard Scientific Party, 1980b) and contrast markedly from the terrigenous sediments in the hanging wall of the décollement. The coincident lithologic and bedding discontinuity across the layer, the presence of deformed sediments bounding the layer, and the shear fabric of the clay in the layer are all compatible with the interpretation of a large displacement, plate boundary décollement.

The location of the décollement coincides with the uppermost reflector of the several distinct and parallel reflectors that demarcate the top of the subducting oceanic plate. As shown by the in-line seismic profile that passes through the drill site, these reflectors are continuous for ~1 km to the east-southeast and several kilometers to the west-northwest. The cross-line seismic profile that passes through the drill site indicates the reflectors are continuous several kilometers along strike to the north-northeast and south-southwest. Approximately 1 km in-line to the east-southeast, the basement is down-dropped along normal faults to form a prominent sediment-filled graben that spans the axis of the trench. A distinct reflector continues from the reflectors at the top of the horst below the drill site to the east and into the sediment fill of the graben (Fig. F2). The seismic character of this reflector is consistent with the general structure of the décollement documented at the drill site (i.e., a hanging wall characterized as chaotic and seismically transparent and a footwall consisting of subhorizontal and conformable seismic reflectors representing the bedded sediments and underlying igneous basement of the subducting plate). The reflector continues from the horst into the graben and cuts some sediment layers in the footwall, and thus defines the base of the displaced and thickened prism (also see Kodaira et al., 2012). The seismic data support the simple interpretation that the prism extends some 3 km east from the drill site to the axis of the trench. A first-order palispastic reconstruction of the prism, assuming constant area balancing, implies the displacement on the décollement at the drill site is on the order of 3 km. Although the deformation associated with the décollement is much more localized than that seen at other subduction décollements (e.g., Nankai, Barbados), the structure and fabric of the décollement at the drill site is compatible with displacements of this magnitude. Thus it is expected that the décollement at the drill site is likely continuous with the deeper portions of the plate boundary interface tens of kilometers downdip. A plate boundary décollement of this size and position is hypothesized as the locus of tectonic displacement of the subducting plate in the geologic past, as well as during the recent rupture that propagated to the trench during the Tohoku event.

After analysis of core samples from both the 720 and 820 fault zones, the design of the MTL autonomous observatory was modified to primarily target the 820 fault using a dense array of sensors; however, sensor placement to characterize a possible temperature anomaly at the 720 fault has been retained. In addition, some sensors are programed to acquire data at higher frequencies at the time of scheduled retrieval in order to directly measure a temperature profile of the entire hole as the string is pulled out. Thus, the observatory is expected to provide valuable data when the sensor string is later retrieved, which is scheduled for October 2012 or February 2013. The observatory data will document the spatial and transient character of the temperature distribution in the lower reaches of the borehole. These results will not only allow confirmation that the rupture occurred at the décollement but will also lead to a robust estimate of the frictional heat generated during the dynamic slip of the Tohoku earthquake. The determination of total heat can constrain models used to estimate the dynamic shear strength of the fault and infer the magnitude of the coefficient of friction.

Objective 2: what was the stress state on the fault that controls rupture during the earthquake and was the stress completely released?

The most significant goal of this rapid response drilling project is to determine the dynamic frictional stress. Data from the measurement of the time decaying temperature anomaly associated with the earthquake slip will be used to estimate the frictional heat produced at the time of the earthquake, which can be used to infer the level of dynamic friction (e.g., Brodsky et al., 2009; Fulton et al., 2010). Measurements of current stress (i.e., postearthquake slip) also can be used to explore different models to explain how dynamic slip occurred and the degree to which stress was released. The in situ stress will also be determined through the use of LWD image logs to characterize borehole deformation (e.g., Zoback et al., 2003) and measurements of anelastic strain recovery (ASR) of core samples (e.g., Byrne et al., 2009). Several samples were collected for ASR, and these time-sensitive measurements were begun during this expedition. Measurements and analysis will be completed within several months of the end of the expedition and, if successful, could provide true 3-D determination of in situ stress in the prism.

The borehole breakouts evident in image logs from the LWD hole indicate several different in situ stress domains within the prism. At depths less than ~200 mbsf, the maximum horizontal compressive stress (SHmax) systematically varies from approximately parallel to the convergence direction to perpendicular (at 140 mbsf) and back toward parallelism again with greater depth. At intermediate depths (200–537 mbsf), SHmax orientation is variable and nonsystematic. At deeper levels in the prism (537–820 mbsf), SHmax displays a single preferred orientation ~20° clockwise from the convergence direction. Faults and bedding are variable in dip magnitude, but faults and bedding at all depths in the prism show a preferred northeast strike direction reflecting horizontal contraction and local extension approximately parallel to the plate convergence direction. That SHmax varies with depth in the prism likely indicates a postearthquake stress state in the prism in which SHmax is similar in magnitude to Shmin except possibly in the very basal part of the prism. The variation at shallow depths can be understood as a result of reduction in the magnitude of the horizontal stress in the direction of slip during the earthquake. If the magnitude of maximum horizontal stress is reduced to values similar to the minimum horizontal stress, then borehole breakout directions will reflect local perturbations and appear highly variable or they may change systematically with depth because of variation in mechanical properties or the presence of other tectonic discontinuities. Thus, these preliminary data on SHmax directions are consistent with a reduction of shear stress on the megathrust during the Tohoku earthquake, but further analysis is necessary to infer the magnitude of stress changes.

Objective 3: what are the characteristics of large earthquakes in the fault zone, and how can we distinguish present and past events in fault zone cores?

Success in recovering ~1 m of core from the high–shear strain zone and neighboring sediments from the plate boundary décollement provides material for shore-based mechanical and physical properties testing, as well as for geochemical, mineralogical, and microstructural analyses. Physical properties testing, particularly for thermal and fluid-flow properties, are important for constraining models of thermal-mechanical-hydrological processes important to the seismic cycle, modeling thermal transients in order to infer the dynamic strength of the Tohoku earthquake fault, and determining stress state of the prism over the seismic cycle. Several shore-based studies of frictional properties of intrafault materials as a function of slip rate, pressure, temperature, and fluids are planned; this information will feed into mechanical analyses of seismic faulting. In addition, three special interest structural whole-round samples taken from the sheared clay of the décollement, as well as two other structural whole-rounds capturing secondary faults, will provide material for coordinated nondestructive and destructive investigations of structure, chemistry, and mineralogy of the faulted sediments. These studies and those of geochemistry and other physical properties will be used to search for signatures of seismic slip and investigate processes of dynamic slip during large earthquakes.