IODP Proceedings Volume contents Search | |||||||||||||
![]() | |||||||||||||
Expedition reports Research results Supplementary material Drilling maps Expedition bibliography | |||||||||||||
doi:10.2204/iodp.proc.343343T.203.2015 IntroductionThe slip behavior of major fault zones, especially at shallow depths, is controlled by the frictional behavior of the active shear zone. The 2011 Mw 9.0 Tohoku earthquake produced largely unexpected 30–80 m of shallow coseismic slip that breached the Japan Trench (Fujiwara et al., 2011; Ide et al., 2011; Ito et al., 2011). That the shallowest reaches of the Japan Trench in the Tohoku area facilitated such a large amount of coseismic slip propagation is inconsistent with current understanding of fault mechanics, and explaining this phenomenon therefore requires thorough characterization of megathrust fault zone frictional properties in this region (e.g., Lay and Kanamori, 2011). Drilling during Integrated Ocean Drilling Program (IODP) Expedition 343, the Japan Trench Fast Drilling Project (JFAST) penetrated the toe of the prism at the Japan Trench subduction zone ~5 km landward of the trench axis, within the area of high coseismic slip and directly updip from the Tohoku earthquake hypocenter (Fig. F1). Successful recovery of core samples of the fault zone that hosted the 2011 Tohoku earthquake, as well as the surrounding wall rocks, in the ~840 m deep borehole has provided the opportunity to investigate the frictional properties of the shallow plate boundary megathrust in the Tohoku. Sample descriptionsThe sample suite in this study consists of 21 samples recovered by rotary barrel coring between 648 and 837 meters below seafloor (mbsf) (Table T1). Seventeen of these samples are siliceous mudstones that make up the hanging wall prism and underthrust footwall (lithologic Units 2, 3, and 5) and are relatively structureless (see the “Site C0019” chapter [Expedition 343/343T Scientists, 2013b]). The remaining samples include one sample from a highly sheared scaly clay layer (Unit 4) interpreted to be the plate boundary fault zone at 821.5–822.5 mbsf (Chester et al., 2013), and three samples of a multicolored, stratified pelagic clay layer located ~10 m below the fault zone (Unit 6). Unlike the fault zone, the pelagic clay exhibits no shear fabric (see the “Site C0019” chapter [Expedition 343/343T Scientists, 2013b]). Seven of the samples were recovered in a sufficient condition to be tested intact: four hanging wall mudstones, one from the fault zone, and one footwall mudstone. The final intact sample was recovered in Core 343-C0019E-21R from the bottom of the borehole; however, this sample is a mudstone rather than the chert that defines Unit 7, and therefore is probably a Unit 3 sample that fell into the borehole. All 21 samples were tested as gouge powders; therefore, the seven intact samples can be directly compared with a powdered sample of equivalent composition. Rate- and state-dependent frictionMany studies over the years have demonstrated with laboratory shearing experiments that fault strength is dependent on the sliding velocity (see Marone, 1998, and references therein). In such experiments, the coefficient of sliding friction µ is calculated from the measured shear strength:
The velocity dependence of friction is then evaluated by instantaneously increasing (or sometimes decreasing) the sliding velocity from an initial value v0 to a new value v in a stepwise fashion. The frictional response to an imposed velocity step is described by an empirically derived constitutive friction law (Dieterich, 1979, 1981):
where a, b1, and b2 are unitless constants and Dc1 and Dc2 are critical slip distances, over which friction evolves over durations measured as θ1 and θ2, the state variables (Fig. F2). Because the frictional response depends on both the velocity v and the state variable θ, this formulation is known as rate- and state-dependent friction (RSF). If the friction level at the higher velocity v has reached steady-state, Equations 2 and 3 can be reduced to:
where the parameter a-b quantifies the rate-dependence of friction. This is the most important and widely used parameter because it controls the occurrence of slip instability that results in earthquake nucleation (e.g., Scholz, 1998). If a-b > 0, known as velocity-strengthening friction, slip is expected to be stable. On the other hand, a-b < 0 defines velocity-weakening behavior, which is a necessary condition for unstable slip. The RSF parameters, including a-b, have been used in numerical modeling studies to successfully simulate and describe a wide range of fault-slip phenomena, including earthquake nucleation and rupture (e.g., Okubo, 1989; Dieterich, 1992), afterslip (e.g., Perfettini and Avouac, 2007), and transient slow slip events (e.g., Liu and Rice, 2009). Experimental methodsLaboratory friction experiments were performed by shearing the samples in a single direct shear device (see Ikari and Kopf, 2011) (Fig. F3). In this apparatus, the sample volume is a vertically oriented cylinder 25 mm in diameter and with a height of ~12–20 mm within a cell consisting of two flat-lying steel plates. Intact samples were carved from whole-round cores to fit exactly into the sample cell. Powdered samples were prepared by drying the samples at room temperature, disaggregating by hand in a mortar and pestle, and sieving to a maximum grain size of 125 µm. The powders were then mixed with a small amount of 3.5% NaCl brine to form a paste, which was then pressed into the sample cell. Samples were then loaded to a target normal stress (σn) of ~5–7 MPa to match in situ effective vertical stresses estimated from shipboard measurements of bulk density (minus seawater density) and assuming hydrostatic fluid pressure conditions (see the “Expedition 343/343T summary” and “Site C0019” chapters [Expedition 343/343T Scientists, 2013a, 2013b]) (Table T1). Although the pore pressure cannot be directly controlled in this system, samples were allowed to consolidate overnight after application of normal load and allowed to freely drain through porous metal frits at the top and bottom of the sample. Shearing was only initiated after the sample height reached a steady value, indicating that any excess pore pressure had dissipated and the applied normal stress (σn) is equal to the effective normal stress (σn′). Shear is induced by relative displacement of the plates normal to the cylinder axis; for intact samples this means that the shear plane is not necessarily aligned with sample fabric, and this difference depends on the fabric dip. Deformation is in the apparatus is planar and therefore enforced to be localized. All tests were conducted under fluid-saturated conditions with 3.5% NaCl brine as the pore fluid. To ensure saturation during the tests, the sample cell is submerged in a pore fluid reservoir. The experimental procedure involves shearing at a constant rate of 10 µm/s to a displacement of ~5 mm, in order to reach an approximate steady-state shear strength in which low-displacement effects (e.g., from fracturing near the peak strength) are minimized. In order to measure the velocity (or rate) dependence of friction, the shearing velocity was then increased in a series of steps within the range 0.1–30 µm/s (Fig. F3). The upstep velocity is v = ~3v0, where v0 is the initial velocity of each step. Although the parameter a-b may be calculated by directly measuring the steady-state friction change from a velocity step (Equation 4), the individual RSF parameters a, b1, b2, Dc1, and Dc2 must be calculated with modeling techniques due to covariance. An expression for the system stiffness k (friction/displacement) is incorporated:
where (vlp – v) is the difference between true slip velocity v and the remotely recorded load point velocity vlp due to apparatus deformation. The system stiffness includes the combined effects of apparatus and sample stiffness; for most testing equipment this is dominated by the apparatus stiffness. Equations 2, 3, and 5 are solved with a fifth-order Runge-Kutta numerical integration, and best-fit RSF parameters are obtained by solving the inverse problem with an iterative least-squares method (Fig. F3) (Reinen and Weeks, 1993). Due to variations in the experimental data caused by signal noise or excursions in the measurement, standard deviations for the modeled values are also calculated. In some cases, the data are well fit using one state variable, where b2 = 0. Input parameters for the inversion include visual estimations of the RSF parameters, the initial friction level immediately preceding the velocity step µ0, friction slip dependence η = dµ/dx at v (where x is slip displacement), and the system stiffness k (friction/displacement). |