A weak plate boundary
Available online at www.sciencedirect.com R
Earth and Planetary Science Letters 214 (2003) 589^603 www.elsevier.com/locate/epsl
Compositional and £uid pressure controls on the state of stress on the Nankai subduction thrust: A weak plate boundary K.M. Brown a; , A. Kopf a , M.B. Underwood b , J.L. Weinberger a a
Scripps Institution of Oceanography, UCSD, 9500 Gilman Drive, La Jolla, CA 92093-0244, USA b Geological Sciences, University of Missouri, Columbia, MO 65211, USA Received 23 August 2002; received in revised form 1 May 2003; accepted 12 July 2003
Abstract We show that both fault mineralogy and regional excess fluid pressure contribute to low resolved shear stresses on the Nankai subduction plate boundary off southwest Japan. Ring and direct shear tests indicate that saturated clay minerals in the fault possess intrinsically low residual friction coefficients (Wr ) at stress levels between 1.0 and 40 MPa. The direct shear Wr values for purified smectite are V0.14 1 0.02, for illite V0.25 1 0.01, and for chlorite 0.26 1 0.02 (for point load velocities of 0.0001 mm/s). These clay minerals dominate the Nankai subduction de¤collement zone. Illite (plus quartz) is mechanically important in the altered incoming Muroto section and the predicted de¤collement Wr should lie between 0.2 and 0.32. This low residual strength, together with elevated fluid pressure, limits shear stresses to below V4 MPa within the frontal V50 km of the subduction system, consistent with the low wedge taper in this region. A higher wedge taper off the Ashizuri peninsula indicates basal shear stresses rise slightly along strike towards this region. Our analysis indicates lower fluid pressures must predominantly be responsible because only small second order along strike variations in Wr are predicted to occur as a result of variations in smectite and total clay content. These variations should be further reduced at depth under the wedge as smectite is diagenetically altered to illite. However, our data suggest the low Wr values of the clay-rich de¤collement still limit shear stresses to between V17 and 29 MPa within the frontal V50 km of the wedge, consistent with other estimates of plate boundary weakness. Indeed, we propose that it should be expected that subduction plate boundaries like Nankai will be weak because of the intrinsic presence of clay-rich faults and moderate fluid overpressures. Our data do not support the hypothesis that the smectite-to-illite reaction directly controls the onset of seismogenic behavior deep in the Nankai system because there is already a mechanical dominance of illite (rather than smectite) in the shallow de¤collement zone, and we find all the clay phases tend to velocity strengthen. However, temperature-activated clay diagenesis and dehydration may cause secondary changes in the fault properties and state of stress across the up-dip limit of the seismogenic zone. > 2003 Elsevier B.V. All rights reserved. Keywords: frictional strength; state of stress; £uid pressure; convergent margin; seismogenic zone; Ocean Drilling Program
1. Introduction * Corresponding author. E-mail address: [email protected] (K.M. Brown).
We present the initial results of an ongoing ex-
0012-821X / 03 / $ ^ see front matter > 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0012-821X(03)00388-1
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
590
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
perimental study that examines the role that clay minerals play in controlling the frictional properties of subduction plate boundary faults and how, together with regional £uid pressure variations, this controls the state of stress on the subduction thrust. We also address the role that clay diagenesis plays in the onset of seismogenic activity. We focus here on the Nankai Trough, southwest Japan, because it is representative of sediment-rich subduction zones (including Alaska and Cascadia) that are capable of generating very large earthquakes (M V8 and greater) [1^4]. Two signi¢cant, seemingly contradictory attributes of these subduction zone faults are that they appear to be almost fully locked over much of their alongstrike length during interseismic periods even though the faults appear to be extremely weak [5]. Accretionary wedge environments provide the opportunity to test more speci¢c ideas as to how three-dimensional changes in parameters such as fault zone mineralogy, temperature, and £uid pressure (i.e. e¡ective con¢ning stress) can a¡ect the stress state and seismogenic properties of plate boundaries. We chose the Nankai subduction zone as a study area, because it has a historical record of very large earthquakes that dates back to AD 684 [6^8]. The recurrence interval in the western Nankai region varies between 262 and 92 years, and the most recent ruptures were the 1944 Tonankai (M 8) and 1946 Nankaido (M 8.2) events o¡shore of the Kii and Muroto peninsulas [6,9]. In addition, the nationwide permanent Global Positioning System network shows that the coastal forearc region is moving to the westnorthwest at the full convergence rate of V4 cm/ yr, which indicates the plate boundary is essentially fully coupled or locked during the interseismic period [10,11]. However, the plate boundary fault also seems to be extremely weak if the maximum principal stress currently has a trench-parallel orientation and the margin-normal stress appears to be less than the vertical stress in the backstop region. For this to occur, numerical modeling suggests the maximum shear stress on the subduction thrust should not exceed V17 MPa even at the stronger down-dip end of the Nankai seismogenic zone [5]. Coring by the Deep Sea Drilling Project
(DSDP) and the Ocean Drilling Program (ODP) has already revealed signi¢cant along-strike changes in the lithostratigraphy, heat £ow, clay mineralogy, and diagenesis of the incoming Shikoku Basin section seaward of the Nankai Trough. Sediment composition at Nankai appears to be controlled by changing patterns of heat £ow and thermally activated clay diagenesis [12^14]. These shallow, along-strike mineralogical changes mimic those within the fault zone with increasing depth, and thereby allow us to test whether such changes signi¢cantly in£uence the down-dip frictional properties of this plate boundary. In this paper, we focus on experimentally determining the frictional strength of materials entering the subduction toe in order to independently evaluate physical reasons for the apparent weakness of the subduction fault at Nankai. We also address the issue of whether or not the smectiteto-illite reaction can lead to signi¢cant down-dip changes in fault strength. Smectite has certainly been recognized as a weak clay mineral, but it is not clear whether the transformation to illite causes substantial changes in friction coe⁄cient. Our goal is to address how sediment composition impacts fault properties at e¡ective normal stresses up to 40 MPa using both a ring shear and a newly developed direct shear apparatus. A combination of natural Nankai samples, individual mineral standards, and mixtures of smectite (Sm), illite (Il), chlorite (Ch), and quartz (Qtz) were used. Based on the experimental data we also address the potential role that regional pore pressure changes may play in determining the lateral changes in the overall fault strength at Nankai. In addition to arguments about the general state of stress, we are also able to o¡er insights on the debate over the potential diagenetic controls on the up-dip limit on seismogenic activity [2].
2. Geological context of the Nankai Trough At the Nankai Trough, the Philippine Sea plate (Shikoku Basin) is subducting to the northwest under southwest Japan (Eurasian plate) at a rate of 2^4 cm/yr [15,16] ; the direction of convergence is slightly oblique to the plate margin (Fig. 1).
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
Fig. 1. Location map of southwest Japan showing the distribution of ODP and DSDP sites. Also shown are P- and T-axis diagrams for 22 seismic events that suggest the principal stress is parallel to the trench [5,56].
The subducting oceanic lithosphere of the Shikoku Basin was generated by the rifting of the IzuBonin backarc that started in the Oligocene and culminated in Shikoku Basin sea£oor spreading that lasted until V13^15 Ma [17^19]. Seamount volcanism (Kinan seamounts) apparently persisted after the cessation of backarc spreading and contributed to the current elevated thermal gradients near the paleo-spreading axis. Subduction and the collision of the Izu-Bonin arc against Honshu apparently started 15^10 Ma at a slow rate with the development of a deeply penetrating subducting slab and volcanic activity occurring by V8 Ma in southern Kyushu and by 6 Ma in southwest Japan [20,21]. Subduction of the Shikoku Basin, accumulation of a thick trench-wedge turbidite facies, and sediment accretion subsequently led to the development of a broad accretionary prism. The de¤collement near the toe occupies a stratigraphic position within the lower Shikoku Basin facies (Fig. 2).
591
Oceanic basement relief and thermal state signi¢cantly in£uence sediment dispersal, composition, facies architecture, and diagenesis within the critical Miocene Shikoku Basin strata [14, 22,23]. The upper Shikoku Basin facies consists of hemipelagic mudstone and volcanic ash. O¡ the Ashizuri Peninsula (DSDP Site 297, ODP Site 1177, Figs. 1 and 2A), the lower Shikoku Basin facies consists of hemipelagic mudstone with abundant terrigenous sand turbidites and volcanic ash. Shipboard ODP results suggest that Sm is abundant, particularly in the deeper portions of this section. The Ashizuri de¤collement appears to occupy a position on seismic re£ection pro¢les just above the top of a series of turbidites at the toe of the wedge [22,24]. In contrast, along the Muroto transect farther northeast (Fig. 1), the thinner Miocene section accumulated above basement highs associated with the fossil backarc spreading center and the Kinan seamounts. The sandy turbidites are absent there (Fig. 2), so that the Muroto de¤collement passes through monotonous mudstones (Fig. 2B). The temperature gradient remains elevated (180‡C/km [25]) near the Kinan seamount chain, which leads to signi¢cant clay diagenesis near the prism toe [23,26].
3. Methods and experimental procedures 3.1. Sample selection, preparation, and XRD The following three types of samples were tested in this study: (i) natural samples from the Shikoku Basin section (Fig. 2); (ii) mineral standards composed of Sm, Il, Ch, and Qtz; and (iii) mineral mixtures of Qtz/Sm and Qtz/Il. All the frictional tests were conducted under fully seawater-saturated conditions to ensure the appropriate chemical environment for the clay phases in a marine environment. Natural samples were collected from mudstone horizons cored at DSDP Site 297 (Figs. 1 and 2A). Digital X-ray di¡raction (XRD) data for the natural Nankai samples were generated at the University of Missouri using a Scintag PAD V di¡ractometer. The generic methodologies for the XRD analyses for
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
Fig. 2. Lithostratigraphy and clay mineralogy of the Ashizuri (DSDP Site 297 and ODP Site 1173) and Muroto (ODP Site 1177) transects.
592
bulk powder and clay minerals are described in [27]. The Wards mineral standards (Sm, Wyoming bentonite), Il (Rochester Shale), and Ch (Madison Co) were size separated ( 6 2 Wm fraction retained) by centrifuging after rehydration with seawater ( s 72 h) to reduce contamination associated with the ¢ne silt and above size fraction. Contamination of the Il and to some extent the Sm standards is particularly signi¢cant with up to 30% Qtz in the Rochester Shale [28]. Tests were also conducted on arti¢cial mixtures of the puri¢ed end member Sm+Qtz and Il+Qtz mixtures (using dry weights), and on the unpuri¢ed Rochester Shale sample for comparison with the natural Nankai samples. In order to account for the widely varying mineral grain densities of the different phases, all wt% data are converted to the volumetric ratios of the di¡erent components in the sediment framework. Smectite is particularly problematic in this regard because it hydrates under saturated conditions and has a corresponding low mineral density of V2.1 g/cm3 in contact with seawater at moderate to elevated stress levels V hydrate [29,30]). (15 A 3.2. Frictional testing We conducted frictional resistance tests using both a ring shear apparatus at normal stresses of 6 2 MPa [31,32] and a direct shear system at normal stresses of 5^40 MPa [33,34]). A primary objective was the determination of the relationship between mineralogy and frictional resistance. A secondary objective was to obtain some preliminary data of the variation in Wr with normal stress level and slip velocity. Changes with normal stress are particularly germane to investigations of the controls on plate boundary stress state. The frictional resistance of a clay-rich sediment or fault gouge at elevated displacements is typically de¢ned as: Wr = d/(cn 3Pf ), where Wr is the residual frictional coe⁄cient, d is the residual shear strength, cn is the total normal stress on the gouge layer, and Pf is the £uid pressure. We focus on Wr because clay-dominated sediments have distinctly lower residual shear strengths (undergo strain weakening) at higher displace-
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
ments because the tabular clay minerals reorient and align within the shear zones [31,35]. The residual shear strength was determined at low stress levels (1^2 MPa) utilizing a standard Wykeham^Farrance Bromhead ring shear apparatus (WF 25850) [32]. Samples included a mineralogically diverse subset of natural Nankai samples and arti¢cial mineral mixtures. Previous studies indicate that the frictional behavior of a sediment changes in response to the transitions in properties of its principal mineral components, with residual shear strengths decreasing as total clay and Sm contents increase under saturated conditions [31,36]. To assure initial drained conditions, the 3^4 mm thick ring shear samples were consolidated for s 48 h until no further changes in thickness were observed. In order to maintain conditions as near to drained as possible, subsequent frictional testing was then conducted at relatively slow displacement rates of 0.0015 mm/s. The samples were also typically sheared for s 70 mm (V12 h) until no further changes in shear stress occurred in order to verify a drained residual value had been achieved. At higher stress levels, we conducted tests on mineral standards in a modi¢ed GEOCOMP direct shear apparatus (Fig. 3) utilizing a tabular sample geometry of 50 mmU100 mmUV15 mm (WULUH). The sample is contained within a shear box. A piston forces the sample approximately halfway (i.e. V8 mm) into the base plate of the apparatus and holds it under the required normal load. The outer shear box is forcefully held down on the Te£on gasket by an external jig arrangement in order to prevent sample loss during initial sample consolidation (a problem with the pure smectite samples). The seawater-saturated direct shear samples were consolidated for several days under uni-axial consolidation conditions to ¢nal stress levels of between 5 and 40 MPa. Monitoring the pore pressure allowed us to determine when the sample had achieved a near drained state prior to the initiation of shear testing (excess pore pressure developed during loading 6 1% of the total normal stress). We utilized pore pressure ports at the base of the sample (Fig. 3) that made a connec-
593
Fig. 3. (A) Schematic diagram of the direct shear sample geometry and pore pressure monitoring system used during the frictional testing. (B) Photograph showing the fabrics developed within a typical clay gouge (Rochester Shale puri¢ed for illite) after a direct shear test at 20 MPa normal stress.
tion with the shear zone via permeable quartz silt pillars formed by drilling three 3 mm wide cylindrical holes into the lower part of the tabular sample. In later test runs these were modi¢ed into narrow (2 mm) slit-¢lled oblong slits (length 1.5 cm) oriented with their long axes parallel to the shear direction to maintain prolonged hydrologic connection. The load on the outer shear box is subsequently fully released during the shear phase of the test so that it slips freely sideways on the Te£on gasket (Fig. 3). During the subsequent shear testing, the horizontal and vertical loads and displacements, as well as pore pressure at the ports into the fault zone, were continually recorded (Fig. 3). The tests were conducted under constant normal load condition, with the vertical load being reduced during displacement to account for the decreasing contact area between the upper and lower halves of the sample. Frictional resistance is dependent on slip rate at higher stress levels [37,38]. A discussion of the full rate and state variables for velocity-dependent frictional stability is given elsewhere [37^39]. To evaluate this dependence we undertook shear rate or velocity stepping tests as part of the direct shear tests. Shear rates varied over ¢ve orders of
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
594
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
magnitude from 0.00001 to 0.1 mm/s once the samples had achieved their residual state. We report here the frictional coe⁄cients at a low, but arbitrary, slip velocity of 0.0001 mm/s.
4. Results 4.1. Clay contents of the natural Nankai samples It was determined from the XRD analyses [27] that the total clay contents of the DSDP Site 297 samples analyzed as part of this study vary but generally increase with depth below the equivalent position of the Ashizuri de¤collement in the nearby wedge (Fig. 2). Sm and Il are the dominant clay phases together with minor Ch; the natural samples are thus relatively simple three-component mixtures of Sm, Il ( 1 Ch), and granular components such as Qtz ( 1 feldspar). At the equivalent level of the Ashizuri de¤collement (as based on seismic re£ection stratigraphic correlation), Sm contents are 6 10^20% of the bulk sediment (Fig. 2). The high interbed variability between the various clay components in the deeper section below the de¤collement is thought to re£ect changes in the provenance of the ¢negrained components between eroding accreted materials from the Shimanto Belt (high Il 1 Ch), and volcaniclastic and variably altered ash components from the Izu-Bonin arc (generally elevated Sm). Beds within the volcaniclastic-dominated unit at the base of the section may locally contain substantial amounts of vitric ash; also, overall smectite content of the clay-sized fraction is locally high (70^90%). XRD data provide no evidence for smectite diagenesis at depth in this region, which is consistent with the moderate thermal gradient of V50^56‡C/km measured at nearby DSDP sites [40]. For the shear tests on the natural samples, a subset of ¢ne-grained DSDP Site 297 samples was selected with a range of total clay, Sm, and Il contents [27]. The volumetric ratios were then calculated based on their wt% and estimated grain densities (with a error of V 1 5% to perhaps 1 5^ 10% in the XRD-based estimates of wt% ; see [27]).
Fig. 4. Results of ring shear tests conducted on natural Nankai samples (black dots) and arti¢cial mineral mixtures of Sm with Qtz. Data for tests we conducted on Il and Qtz mixtures are also shown. The residual friction (Wr ) data are plotted against the volumetric ratio of (A) smectite (VSm ) and (B) total clay (VSmþIlþCh ) within the solid fraction. A2 and M2 denote estimated altered deep de¤collement property ranges (see Fig. 8), A1 and AS1 are unaltered Ashizuri shallow de¤collement ranges and M1 is a shallow Muroto value (Fig. 8). a = Vitric ash containing samples.
4.2. Ring shear studies A compilation of the ring shear results is shown in Fig. 4. In general, total clay content and the presence of a weak Sm clay component are factors in controlling the mechanical response of sediments. Thus, we show ring shear frictional results plotted against both the volumetric ratio
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
of the hydrous Sm (VSm ) and total clay minerals (VSmþIlþCh ) components of the sediment framework. In Fig. 4, we additionally show the results of ring shear tests on arti¢cial mixtures of Sm+Qtz and Il+Qtz. These data are plotted together with the results of previous studies that were conducted on Sm+Qtz mixtures at 6 1 MPa [31] and at 50 MPa [36], with samples saturated in deionized water. The ¢rst-order results of the ring shear studies are as follows. The pure Sm, Il, and Ch phases were found to have Wr values of 0.12, V0.20, and V0.22, respectively, at normal stress levels 6 2 MPa (Fig. 4). Arti¢cial mineral mixtures show progressive decreases in frictional strength with increasing contents of Sm, Il and total clay minerals (Fig. 4). We plot these results together with previously published results for chlorite (Wr = 0.23) [41]. Our data for Sm+Qtz mixtures closely match the curves of two previous studies across much of the compositional range [31,36], except that their shear strength pro¢les £atten out at clay contents below 20 vol% and above 80 vol%. The Il+Qtz samples show a slightly stronger frictional resistance at higher clay contents than the Sm+Qtz samples (Fig. 4B). The unpuri¢ed saturated Rochester Shale (V30% Qtz) has a Wr value of 0.215 and falls in line with our predictions from the Qtz+Il mineral mixture curves (Fig. 4). The majority of the natural Nankai mudstone samples tend to be mechanically weak re£ecting their high clay mineral contents (Fig. 4). Most data for the Nankai samples do not plot on the Sm+Qtz mixture curves (Fig. 4A). Instead they scatter widely to weaker Wr values below the arti¢cial mineral mixture curves. However, even though the scatter is large, there is still a rough relationship between increasing smectite contents and lower Wr values. In contrast, a very close ¢t exists between the natural samples and the mineral mixture curves when the Wr values are plotted against total clay mineral volume or VSmþIlþCh (Fig. 4B). Thus, the total clay mineral content appears to be the primary control on the frictional properties of the Nankai sediments with smectite exerting only a secondary e¡ect. Three Wr values do plot well above the predicted total clay levels based on their values (Fig. 4B). These sam-
595
Fig. 5. Example of direct shear results for saturated illite at stress levels of 10, 20, and 30 MPa. Rates of velocity stepping tests are shown. Typically, a velocity rate change is followed by a direct initial change in the frictional resistance (a-value), which then subsequently comes to a steady state value at the new slip velocity (b-value). The magnitude of the decay to a steady state friction is typically reported as: a3b = vWr /vlnVel, where vWr is the change in steady state frictional strength and Vel is the sliding velocity [38]. We consistently report positive values of a3b for a velocity increase, implying a velocity strengthening or stable sliding mechanism is operating.
ples contain signi¢cant amounts of amorphous vitric ash that cannot be accounted for quantitatively in the XRD analyses of mineral abundance. 4.3. Direct shear studies conducted on saturated pure mineral samples In order to determine if the low intrinsic frictional resistance of phases such as illite is likely to persist at greater depths and e¡ective stresses along the Nankai plate boundary thrust, direct shear tests were conducted on the mineral standards at e¡ective con¢ning stresses up to 40 MPa. Typical results of direct shear tests are shown for illite at 10, 20, and 30 MPa normal stress (Fig. 5). The samples develop a peak strength at low displacements and subsequent lower residual shear strength after the ¢rst V10^12 mm of displacement. The post-peak strength drop to residual values becomes more pronounced at elevated stress levels. Owing to the sample thickness of V15 mm, we were able to produce a fully contained natural-looking shear zone during the tests. Typically, the shear zone was dominated by upper
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
596
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
and lower boundary shear with well-developed Riedel shear sets (Fig. 3B). Pore pressure measurements indicate that initial yielding and failure of the clay samples is marked by initial pore pressure increases ( 6 1^2% of the total normal load) around the primary failure zone of the clay samples (Fig. 6). The pore pressure rapidly rises until just after the peak failure (Fig. 6, inset) and then slowly falls as the sample reaches its residual state and the pressures drain. Small pore pressure variations were sometimes observed during the course of the velocity stepping tests in the low permeability clay sediments (Fig. 6). We are still working on these latter data. In general they seem to be too small to account for the observed changes in frictional strength with velocity (Fig. 6 and caption). In contrast, the pure quartz samples show little or no pore pressure response during either initial failure or subsequent velocity stepping tests presumably because their highly elevated permeabilities allow them to deform in a fully drained state. The main results of the direct shear tests are
Fig. 6. Example of the direct shear residual friction and pore pressure data for the 20 MPa illite test. Location of the expanded region is shown in the box within the inset ¢gure. In order to account for the drop in frictional resistance across the 0.01 to 0.00001 mm/s transition the pore pressure would have to increase by 2450 kPa. We do not see pore pressure changes of this order of magnitude in our data.
Fig. 7. Summary ¢gure of the ring shear and direct shear Wr versus normal stress results for various saturated puri¢ed clay minerals and quartz. The frictional coe⁄cient of non-saturated smectite [34] is also shown together with our and other [36] relevant test data for water-saturated smectite.
shown in Fig. 7. For comparison, we have also plotted the results from direct shear tests conducted on non-saturated smectite at room humidity [34] (dashed line in Fig. 7) and the pure saturated Sm end member for a study at 50 MPa [36]. Typically, the Wr values of the pure saturated clays remain relatively low and constant at the 10, 20, and 30 MPa normal stress levels (Fig. 7). The Wr values for smectite are V0.14 1 0.02, for illite V0.25 1 0.01, and for chlorite 0.26 1 0.02 (for point load velocities of 0.0001 mm/s). A small increase in Wr is seen relative to values recorded during low stress ring shear tests. This shift may be a real stress-dependent e¡ect or a minor o¡set generated by di¡erences between the two experimental geometries. Overall, however, the direct shear data support the contention that the basic frictional resistance of saturated clays will remain relatively weak at elevated stress levels. In addition, each type of clay (Sm, Il and Ch) and quartz exhibit velocity-strengthening behavior at all stress levels up to 30^40 MPa (for a representative illustration, see Figs. 5 and 6, illite).
5. Discussion The inherent frictional weakness of clay minerals and clay-rich mudstones has been noted since the birth of geotechnical engineering, with a large
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
body of work focussing on mitigating hazards caused by weak clay horizons at shallow depths [35]. However, measurements under seawatersaturated conditions and at higher stress levels are relatively rare, and we note a wide scatter of values for Wr in the literature particularly for illite (0.11^0.68) [28,42^45] and chlorite (0.1^0.43 [45,46]). Smectite seems to retain a relatively weak strength in all water-saturated studies we have been able to identify (predominantly 0.07^ 0.16) [31,41,44^48]. Our tests utilizing two di¡erent test geometries con¢rm the conclusions of the studies at low stress levels; the most common saturated clays in typical ocean margin sediments retain low Wr values at e¡ective stress levels up to 30^40 MPa (Fig. 7). Although we note the small o¡set in Wr between the direct shear and ring shear tests in our own studies, we suggest that unaccounted-for mineral impurities, saturation state, grain size distribution, and perhaps even drainage conditions may also be important factors to explain the variations seen in the literature. For example, typically the purity of the clay standards is often not ascertained and for Il standards such as Rochester Shale Qtz contents as high as V30% have been measured [28]. Given the pure clays’ insensitivity to changes of normal stress, we propose that the types of VSmþIlþCh vs. Wr relationships shown in Fig. 4 are likely to persist at least to normal stress levels of up to 30^40 MPa. This view is supported by the close match between our low stress ( 6 2 MPa) ring shear data and the results of Logan and Rauenzahn [36] for similar Qtz+Sm mixtures at 50 MPa (Fig. 4). In our experiments, we also monitor the potential e¡ects of retarded drainage and pore pressure transients during drained failure of low permeability clays. Initial positive transient pore pressure e¡ects during undrained failure in low permeability samples are reasonable to expect given the collapse of the clay fabric and increase in total stress during the initial part of the test. However, our measured increases are too small to account for the low Wr of the saturated clays in the latter parts of the test (Fig. 6). They also do not appear to be large enough to account for the changes in Wr that occur during the velocity stepping tests (Fig. 6). Pore pressure e¡ects are unlikely to be
597
broadly responsible for the general weakness of clays in any event. In natural settings weak saturated clays have long been known to be responsible for creeping or reactivation of landslides that move for decades; these time periods are too long for excess pore pressures to remain trapped within their basal shear zones [35,41,42,48,49]. Instead, we suggest that the weakness of the clays is more likely a result of the absorption of water onto the charged clay surfaces and the full hydration of the smectite interlayer [47,50^55]. Changes in pore £uid composition have also been shown to have secondary e¡ects on the frictional coe⁄cient of clays through the interaction with this surfaceabsorbed H2 O [47]. The importance of the chemical and physical e¡ects of water is illustrated by the o¡set between the saturated and non-saturated (partially hydrated) Sm curves (Fig. 7). The mismatch is particularly large at low normal stresses, where the Wr of the unsaturated Sm is considerably higher than the saturated phase. Intervening surface-absorbed H2 O may be the weakening factor in clays as it results in a reduced e¡ective frictional contact area between the faces of the aligned clay minerals in the shear zones. Essentially, this provides a molecular lubrication layer on the outer mineral surfaces or along the interlayer regions in the case of the hydrated Sm phase. Changes in the nature and ordering of clay fabrics due to the presence of H2 O may also be an important factor. We will continue to investigate how the behavior of this absorbed H2 O and any pore pressure changes in£uence the velocity dependence of Wr at elevated stress levels in future studies and primarily report the basic frictional strength data here. 5.1. General implications for the state of stress on the Nankai subduction thrust in 3D The inferred stress state in the Nankai forearc based on the forearc earthquake patterns (Fig. 1) [5,56] suggests that on a margin scale the basal shear stress on the subduction thrust is low. Here we make an independent evaluation as to whether a combination of low Wr values for the de¤collement materials and moderately elevated £uid pressures lead to low resolved shear stresses.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
598
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
This analysis is complicated by the observation that the wedge taper increases from 4.1‡ o¡ Muroto to 7.9‡ o¡ Ashizuri (Fig. 8), suggesting a correlated increase in the basal shear stress on the wedge and lateral variability in either Wr and/or £uid pressure [57,58]. Thus, we also attempt to evaluate the relative changes in the contributions of these two parameters to along strike variations in the basal shear stresses on the Nankai subduction thrust as well as the overall regional stress ¢eld. We summarize our estimated ranges for the spatial variation in de¤collement Wr values schematically in Fig. 8, taking into account the lateral and stratigraphic variations in total clay and smectite content between the Ashizuri and Muroto regions (Fig. 2). We have chosen end member points in the system. In Fig. 8, points A1 and A2, respectively, correspond to the correlated stratigraphic intervals in the incoming section and regions deep under the wedge o¡ Ashizuri, whereas points M1 and M2 correspond to the incoming and deep Muroto de¤collement zone, respectively. For completeness we also show estimated values for Site 808 o¡ Muroto at the toe of the wedge. We give Wr estimates for both the younger stratigraphic level of the de¤collement at the toe of the wedge and the older stratigraphic level (near the sediment basement contact) down to which the de¤collement generally steps at deeper levels in the system [22]. At point A1 o¡ Ashizuri (Fig. 8), the nearby de¤collement at the toe of the wedge lies in the younger stratigraphic units at V430 m depth where total clay contents are V50^55 wt%, illite contents are V20^40 wt%, and smectite contents are V10^25 wt% of the bulk sediment (Sites 297, 1173, Fig. 2). Based on the relationships in Fig. 4B for such compositions we would expect a likely maximum Wr value of V0.34^0.32 (at point A1s using the Il/Qtz curves in Fig. 4). In contrast, near the oceanic basement/sediment interface, the mudstones generally contain V60^80 wt% (V64^92 vol%) total clay, and closer to an average of V30^40 wt% (37^57 vol%) bulk smectite depending on whether we chose Site 297 or 1177 as a representative section (Fig. 2). For these compositions our measured unaltered Wr values are low
Fig. 8. Schematic 3D representation showing the estimates of the de¤collement zone Wr values for the Ashizuri (A1 to A2) and Muroto areas (M1 to M2) and the wedge taper angles. Values in the incoming section are given for the stratigraphic horizon corresponding to the de¤collement level at the toe and the older stratigraphic section down to which the de¤collement steps deep under the wedge (A2 and M2).
(excluding the vitric ash horizons) and mostly lie between 0.13 and 0.24 overlapping with both the Sm/Qtz and Il/Qtz end member curves (A1 range, Fig. 4B). Thus, there appears to be no relationship between the lowest Wr regions (i.e. highest total clay and smectite contents) in the incoming section and the initial de¤collement level at the toe of the wedge o¡ Ashizuri (Fig. 2). At Point A2, deep beneath the wedge o¡ Muroto (Fig. 8), the onset of clay diagenesis should result in the altered Wr value progressively coming to lying on the Il+Qtz curve, with a maximum range between V0.25 to V0.32 if we assume the total clay content is reduced to between V60% and 80 vol% as the low density Sm phase is reacted to illite (A2 range, Figs. 4 and 8). At point M1 in the incoming section o¡ Muroto (Site 1173, Figs. 2 and 8), the proto-de¤collement horizon correlates with a stratigraphic level at V400 m depth, below a small peak in smectite (Fig. 2). The total clay mineral content at the proto-de¤collement level is V58^60 wt% (V63 vol%) and comprises roughly equal volumetric proportions of smectite (V25 wt% or 31 vol%) and illite plus chlorite (V32 vol%). Therefore, we expect the Wr values in the proto-de¤collement region to lie approximately mid-way between 0.2 and 0.3 based on the Sm/Qtz and Il/Qtz end member curves in Fig. 4B (point M1). The Il/Qtz relationship seems reasonable because the Wr of chlorite is very close to that of illite. At point M2 (Fig.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
8), all the smectite will have reacted to illite in this high heat£ow region, increasing the altered Wr value towards V0.32, (Il/Qtz end member and V60 vol% clay, Figs. 4 and 8). Indeed, beause temperature gradients approach 180‡C/km [13], smectite diagenesis is known to be well advanced at the toe of the wedge (i.e. at Site 808, Fig. 8) so that the altered Wr value should already be representative for the subduction thrust at shallow levels. Thus, in summary, the residual strengths of mudstones in the subduction thrust at Nankai are generally low at all depths beneath the wedge because of the elevated clay mineral contents and any small secondary along-strike changes in Wr values in the incoming section will be further reduced with smectite diagenesis at depth (Fig. 8). Indeed, the smectite content did not even apparently control the de¤collement level at shallow depths o¡ Ashizuri where it is most elevated in the older section. Consequently, we conclude the observed along-strike changes in wedge taper and associated basal shear stress (Fig. 8) must primarily relate to regional variations in pore pressure in the subduction thrust. The types of small dynamic pore pressure changes we observed in response to velocity changes and initial yielding of the samples (Fig. 6) are transient phenomena that would be superimposed on such regional variations in pore pressure. We start our evaluation of the role that pore pressure variations may play in controlling stress and wedge taper angle in the more intensively studied Muroto region [22]. A simulation of overpressure distribution in the Nankai wedge o¡ Muroto [59] is shown in Fig. 9A where V* = Pf 3Ph /Pl 3Ph (Ph = hydrostatic, Pf = £uid, and Pl = lithostatic pressures). Typically, the degree of overpressuring in such numerical simulations often rises to a peak (in this case V*W0.8 [58,59]) at a certain burial distance back from the deformation front and subsequently remains elevated or drops slightly at greater depths depending on the parameterizations used in the numerical simulations (which include clay dehydration sources). The resulting e¡ective normal stress distribution on the Muroto de¤collement zone is shown in Fig. 9B. Initially values remain extremely low
599
( 6 V5 MPa) in the frontal V30 km of the wedge and begin to rise above V11 MPa only at distances s 50 km from the toe of the wedge. The calculated maximum shear stresses based on our experimentally derived puri¢ed mineral Wr values are also shown in Fig. 9B. All of the pure clay end members result in low shear strength because their inherent weakness is combined with excess £uid pressure. Even the slightly higher estimated Wr values of the diagenetically altered deeper Nankai de¤collement (M2, Fig. 8) results in a shear stress of 6 V4 MPa at V50 km from the deformation front (Fig. 9B). To see how these types of £uid pressures, when taken together with our Wr values, may account for the observed taper angle o¡ Muroto, we plot in Fig. 10A two parameters derived from the critical taper equation (equation 5) in [57] assuming the cohesion of the wedge and basal thrust is neg-
Fig. 9. (A) Numerically estimated [59] £uid pressure distribution in the frontal 50 km of the wedge. (B) Estimated shear stress on the Nankai subduction thrust o¡ Muroto derived from the £uid pressure distribution shown in A and the frictional properties of the puri¢ed clay minerals (Sm and Il) and Qtz (shown in Fig. 7). The estimated frictional resistances for the diagenetically altered Muroto de¤collement lithologies are also shown (M2, Fig. 8).
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
600
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
ligible and the Muroto taper angle is V4.1‡. The two parameters are the wedge’s basal friction coe⁄cient (Wrb ) and a Wedge Strength Parameter, (Kp 31)(13Vw ) consisting of both the wedge’s internal Vw condition and its passive shear strength parameter (Kp , see Fig. 10 caption for further de¢nitions). This latter parameter is directly related to the frictional properties of the sand-dominated accreted trench sediments. Typically Kp should be between V1.2 and 1.7 for clays and V3 and 3.7 for sands (also see caption, Fig. 10). In Fig. 10A we plot the region de¢ned by the range of £uid pressures along much of the de¤collement zone (Vw = 0.6^0.8) derived from the modeled pore pressure distribution at Muroto (Fig. 9A) and the maximum predicted frictional coe⁄cient of the altered de¤collement sediments (V0.32). The region de¢nes a range of wedge strength parameter values between V0.7 and 1.6. If we assume the sand end member Kp range for the trench turbidite-dominated wedge, then we predict the Vw value should lie between V0.2 and V0.7, with the midpoint being a reasonable average. From Fig. 9A we can see this range for the wedge is consistent with the simulation results shown in Fig. 10A because most simulations show that Vw values progressively increase downwards from hydrostatic (0.0) at the wedge surface to V0.6^0.8 at the de¤collement zone. We have no in situ pore pressure measurements or numerical simulations for the Ashizuri region, but turning around the above analysis allows us to predict what the £uid pressures on the de¤collement may be. The critical taper in the Ashizuri region (7.9‡) is much larger than o¡ Muroto. If we assume the accreted sandy trench wedge has similar properties and wedge strength parameter values (V0.7 and 1.6) in both regions the unaltered Ashizuri shallow de¤collement Wrb range of 0.32^0.34 (A1s, Fig. 8) would lead us to predict an average Vb value for the de¤collement near the toe of the wedge of between V0.2 and 0.55 for a taper angle of 7.9‡ (Fig. 10B). For a deep de¤collement lying in an altered, but more clay-rich and weaker older section (Wrb range of 0.25^0.32, point A2 in Fig. 8), Vb could lie between V0.0 and V0.5 (i.e. nearer hydrostatic). The range of shear stress we predict for this
Fig. 10. The relationship between the basal friction strength (Wrb ) and the wedge strength parameter (WSP) for various basal values for: (A) the Muroto critical taper angle of 4.1‡C and (B) the Ashizuri critical taper of 7.9‡. The relationships are derived from the standard critical taper equation (equation 5 in [49]) that relates basal frictional coe⁄cient and £uid pressure condition of the subduction thrust (respectively, Wrb and Vw ) and internal wedge (Vw , Kp ) to the resulting critical wedge taper (angle between the surface and basal thrust of the accretionary wedge). The Kp parameter is the passive earth coe⁄cient of soil mechanics [31] and is given by Kp = (1+sinP)/(13sinP) where P is the internal angle of friction of the wedge materials. Outlined regions represent the best guess estimates for the Nankai wedge de¤collement in the A1, A2, and M2 regions on Fig. 8.
margin based on these Wrb and £uid pressure estimates appears to be generally supportive of the independently derived estimates of Wang and He [5] for anything greater than hydrostatic £uid pressure conditions on the de¤collement. In their study, from numerical simulations, the average shear stresses on the Nankai subduction thrust decrease up-dip from a maximum of 6 V17^18 MPa at V90 km from the toe of the wedge. The stress will be even lower if the down-dip width of the zone of fault mechanical coupling is wider. Similarly o¡ Muroto, even with the altered deeper de¤collement compositions, we predict the shear stress in the frontal 30 km remains below 2 MPa and progressively rises towards and above 4 MPa at s 50 km from the toe of the wedge (Fig. 9B). O¡ Ashizuri, basal shear stresses at a distance of 50 km landward of the toe may rise to between V17 and V29 MPa for respective Vb approximate upper and lower limits of 0.55 and 0.0
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
(from Fig. 10) and a Wrb range between V0.32 and V0.25 estimated for an altered lower stratigraphic sequence (A2, Fig. 8). The range of our predicted maximum basal shear stresses developed in the two regions encompasses the maximum Wang and He stress estimate particularly if there is at least some moderate overpressuring in the system. We do, for example, feel that the hydrostatic end member is extreme and that Vb should be s 0.0 o¡ Ashizuri. Thus, the presence of weak clay-rich fault zones and at least moderately elevated £uid pressures in the subduction thrust easily account for the apparent weakness of this plate boundary, and no doubt the same factors account for why other subduction thrusts may be similarly weak [5]. Clearly, however, our analysis does suggest along-strike changes in £uid pressure and basal shear stress are likely to be important in the Nankai system. Our quick analysis indicates Vb values in the Ashizuri de¤collement region may be 6 V70% of the values o¡ Muroto (Fig. 10). In subduction zones, regional pore pressures are controlled by such factors as the system’s bulk permeability, rate of tectonic thickening (impacted by sediment input and subduction rate), and mineral dehydration reactions [58]. At Nankai, the rate of subduction does not change signi¢cantly along strike and, if anything, dehydration sources of £uids would increase with smectite content from Muroto to Ashizuri, in the wrong direction to drive the inferred £uid pressure variations. This leaves variations in the system’s bulk permeability as being the most likely reason for the apparent change in £uid pressure. This possibility has also been pointed out in a recent numerical evaluation of the hydrologic controls on the morphology of a number of accretionary wedges including east and west Nankai [58]. Perhaps the turbidite sands in the lower Shikoku Basin section o¡ the Ashizuri region form relatively e⁄cient lateral drains, reducing regional £uid pressures under the wedge in this region. Lower £uid pressures in the older turbidites may also explain why the de¤collement does not form in this level at the toe of the Ashizuri wedge despite the section containing units with very low residual strengths (A1, Fig. 8).
601
5.2. Implications for seismogenic processes As we expect the subduction thrust to be extremely weak at all depths, the implication is that the dynamic stress drop occurring during any earthquake must also be limited because it cannot exceed the peak frictional strength of the sediments just prior to failure. Thus, the small observed stress drops that occur during subduction earthquakes (typically between 1 and 10 MPa, e.g. [60]) seem entirely consistent with our low frictional strengths. It has also been hypothesized that the temperature range (60^150‡C) of the smectite-to-illite transition may be one critical factor in determining the onset of seismogenic activity if it coincides with a change from a velocitystrengthening to velocity-weakening response [61]. However, we did not observe velocity-weakening responses during tests on these clays, and particularly for the critical Il phase at stress levels below 40 MPa (Fig. 6). Additionally as we noted previously, the de¤collement beneath the shallow regions of the accretionary wedge does not lie in sediments where smectite is mechanically dominant. Thus, it seems unlikely that the Sm to Il reaction is directly (in terms of the frictional response) involved in the onset of unstable sliding and seismogenic activity. On the other hand, the combined e¡ects of water release during the Sm to Il reaction and any temperature-induced progressive chemical changes such as pressure solution/ cementation (+any clay diagenesis) could be important factors [12]. For example, a combination of a decrease in water release rate from clay dehydration and increasing Wr and temperature values could result in progressively elevating normal and shear stresses and chemical activity as the seismogenic zone is approached thereby, for example, enhancing pressure solution and a fault healing increasingly with depth.
Acknowledgements We thank Eli Silver and Kelin Wang for their timely and very helpful reviews that greatly assisted in the publication of this paper. This work was supported by NSF Grants OCE 0203799 and
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
602
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
9618166 (to K.M.B.), the NSF Margins program, the Humboldt Foundation (A.K.), and USSSP Grant 190-F001281 (to M.B.U.). This research utilized samples provided by DSDP/ODP, sponsored by the U.S. National Science Foundation and participating countries under management of Joint Oceanographic Institutions (JOI) Inc. [BOYLE]
[14]
[15]
[16]
References [17] [1] D.A. Oleskevich, R.D. Hyndman, K. Wang, The updip and downdip limits to great subduction earthquakes: Thermal and structural models of Cascadia, south Alaska, SW Japan, and Chile, J. Geophys. Res. 104 (B7) (1999) 14965^14991. [2] R.D. Hyndman, M. Yamano, D.A. Oleskevich, The seismogenic zone of subduction thrust faults, Island Arc 6 (1997) 244^260. [3] K. Satake, Depth distribution of coseismic slip along the Nankai Trough, Japan, from joint inversion of geodetic and tsunami data, J. Geophys. Res. 98 (1993) 4553^4565. [4] K. Satake, K. Shimazaki, Y. Tsujii, K. Ueda, Time and size of a giant earthquake in Cascadia inferred from Japanese tsunami records of January 1700, Nature 379 (1996) 246^249. [5] K. Wang, J. He, Mechanics of low-stress forearcs: Nankai and Cascadia, J. Geophys. Res. 104 (1999) 15191^ 15205. [6] M. Ando, Source mechanisms and tectonic signi¢cance of historical earthquakes along the Nankai Trough, Tectonophysics 27 (1975) 119^140. [7] M. Ando, A fault model of the 1946 Nankaido earthquake derived from tsunami data, Phys. Earth Planet. Inter. 28 (1991) 309^322. [8] H. Kumagai, Time sequance and the reoccurance models for large earthquakes along the Nankai Trough revisited, Geophys. Res. Lett. 23 (1996) 1139^1142. [9] H. Kanamori, Rupture processes of subduction-zone earthquakes, Annu. Rev. Earth Planet Sci. 14 (1986) 298^322. [10] S. Miyazaki, K. Heki, Crustal velocity ¢eld of Southwest Japan: Subduction and arc-arc collision, J. Geophys. Res. 106 (2001) 4305^4326. [11] S. Zhao, X. Wu, T. Hori, A. Smith, Y. Kaneda, S. Takemoto, Deformation and stress localization at the Nankai subduction zone, southwest Japan, Earth Planet. Sci. Lett. 206 (2003) 145^160. [12] C. Moore, D. Sa¡er, Updip limit of the seismogenic zone beneath the accretionary prism of southwest Japan: An e¡ect of diagenetic to low grade metamorphic processes and increasing e¡ecive stress, Geology 29 (2001) 183^186. [13] M.B. Underwood, K. Pickering, J. Gieskes, M. Kastner, R. Orr, Sediment geochemistry, clay mineralogy, and dia-
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
genesis: A synthesis of data from Leg 131, Nankai Trough, in: I.A. Hill, A. Taira, J.V. Firth (Eds.), Proc. ODP Sci. Results 131 (1993) 343^363. M.B. Underwood, D. Sa¡er, J. Steurer, O.L.S. Party, Along-strike variations in clay mineralogy and diagenesis: Implications for the up-dip limit of the Nankai seismogenic zone, EOS Trans. AGU 81 (2000) 1065. T. Seno, The instantaneous rotation vector of the Philippine Sea Plate relative to the Eurasian Plate, Tectonophysics 42 (1977) 209^226. D.E. Karig, C.L. Angevine, Geologic constraints on subduction rates in the Nankai Trough, Init. Rep. DSDP 87 (1986) 789^796. K. Okino, Y. Ohara, S. Kasuga, Y. Kato, The Philippine Sea: New survey results reveal the structure and history of the marginal basins, Geophys. Res. Lett. 26 (1999) 2287^2290. K. Okino, Y. Shimakawa, S. Nagaoka, Evolution of the Shikoku basin, J. Geomagn. Geoelectr. 46 (1994) 463^ 479. B. Taylor, Rifting and the volcanic-tectonic evolution of the Izu-Bonin-Mariana arc, in: B. Taylor, K. Fujioka (Eds.), Proc. ODP Sci. Results 126 (1992) 627^651. M. Takahashi, Large felsic magmatism of the Miocene outerzone of southwest Japan (in Japanese), Earth Monit. Spec. 23 (1999) 160^168. H. Kamata, K. Kodama, Tectonics of an arc-arc junction: an example from Kyushu Island at the junction of the southwest Japan Arc and Ryukyu Arc, Tectonophysics 233 (1994) 69^81. G.F. Moore, A. Taira, A. Klaus et al., New insights into deformation and £uid £ow processes in the Nankai Trough accretionary prism: Results of Ocean Drilling Program Leg 190, G-cubed 2 (2001) 10^25. M.B. Underwood, K. Pickering, Clay-mineral provenance, sediment dispersal patterns, and mudrock diagenesis in the Nankai accretionary prism, southwest Japan, Clays Clay Minerals 44 (1996) 339^356. J.K. Morgan, D.E. Karig, De¤collement processes at the Nankai accretionary margin, Southeast Japan, J. Geophys. Res. 100 (1995) 15221^15231. A. Taira, I. Hill, J. Firth, U. Berner, W. Bruckmann, T. Byrne, T. Chabernaud, A. Fisher, J.P. Foucher, T. Gamo, J. Gieskes, R. Hyndman, D. Karig, M. Kastner, Y. Kato, S. Lallemant, R. Lu, A. Maltman, G. Moore, K. Moran, G. Ola¡son, W. Owens, K. Pickering, F. Siena, E. Taylor, M. Underwood, C. Wilkinson, M. Yamano, Y. Zhang, Sediment deformation and hydrogeology of the Nankai Trough accretionary prism ^ Synthesis of shipboard results of ODP Leg 131, Earth Planet Sci. Lett. 109 (1992) 431^450. K.M. Brown, D.M. Sa¡er, B.A. Bekins, Smectite diagenesis, pore-water freshening, and £uid £ow at the toe of the Nankai wedge, Earth Planet. Sci. Lett. 194 (2001) 97^ 109. M. Underwood, N. Basu, J. Steurer, S. Udas, Data report: Normalization factors for semi-quantitative x-ray
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35] [36]
[37]
[38] [39] [40] [41] [42] [43]
[44]
[45]
di¡raction analysis, with application to DSDP site 297, Shikoku basin, in: G.F. Moore, A. Taira, A. Klaus (Eds.), Proc. ODP Sci. Results, Online publication HTML//, 190 (2003). C. Marone, D.M. Sa¡er, K. Frye, S. Mazzoni, Laboratory results indicating intrinsically stable frictional behavior of illite clay, EOS Trans. AGU 47 (Suppl.) (2001) 1248^ 1249. K.M. Brown, B. Ransom, Porosity corrections for smectite-rich sediments: Impact on studies of compaction, £uid generation, and tectonic history, Geology 24 (1996) 843^846. T.G. Fitts, K.M. Brown, Stress-induced smectite dehydration: Rami¢cations for patterns of freshening and £uid expulsion in the N. Barbados accretionary wedge, Earth Planet. Sci. Lett. 172 (1999) 170^197. J.F. Lupini, J.F. Skinner, A.E. Vaughan, The drained residual strength of cohesive soils, Ge¤otechnique 31 (1981) 181^213. A.W. Bishop, G.E. Green, V.K. Garga, A. Andersen, J.D. Brown, A new ring shear apparatus and its application to the measurement of residual strength, Ge¤otechnique 21 (1971) 273^328. C. Marone, C.B. Raleigh, C.H. Scholz, Frictional behavior and constituitive modelling of simulated fault gouge, J. Geophys. Res. 95 (1990) 7007^7025. D.M. Sa¡er, K.M. Frye, C. Marone, K. Mair, Laboratory results indicating complex and potentially unstable frictional behavior of smectite clay, Geophys. Res. Lett. 28 (2001) 2297^2300. T.W. Lambe, R.V. Whitman, Soil Mechanics, John Wiley and Sons, New York, 1979, 553 pp. J.M. Logan, K.A. Rauenzahn, Frictional dependence of gouge mixtures of quartz and montmorillonite on velocity, composition and fabric, Tectonophysics 144 (1987) 87^108. C. Marone, Laboratory-derived friction laws and their application to seismic faulting, Annu. Rev. Earth Planet. Sci. 26 (1998) 643^696. C.H. Scholz, Earthquakes and friction laws, Nature 391 (1998) 37^42. C.H. Scholz, The Mechanics of Earthquakes, Cambridge University Press, Cambridge, 1990, 439 pp. D.E. Karig, J.C. Ingle, L.S. Party, Initial Reports, 1975, 927 pp. A.W. Skempton, Long-term stability of clay slopes, Ge¤otechnique 14 (1964) 75^101. G. Mesri, A.F. Cepeda-Diaz, Residual shear strength of clays and shales, Ge¤otechnique 35 (1985) 269^274. G. Mesri, A.F. Cepeda-Diaz, Residual shear strength of clays and shales: Discussion, Ge¤otechnique 36 (1986) 269^281. C.A. Morrow, L.Q. Shi, J.D. Byerlee, Permeability of fault gouge under con¢ning pressure and shear stress, J. Geophys. Res. 89 (B5) (1984) 3193^3200. C.L. Chu, C.Y. Wang, W. Lin, Permeability and frictional properties of San Andreas fault gouges, Geophys. Res. Lett. 8 (1981) 565^568.
603
[46] C.A. Morrow, L.Q. Shi, J.D. Byerlee, Strain hardening and strength of clay-rich fault gouges, J. Geophys. Res. 87 (B8) (1982) 6771^6780. [47] R. Moore, The chemical and mineralogical controls upon the residual strength of pure and natural clays, Ge¤otechnique 41 (1991) 35^47. [48] A.W. Skempton, Residual strength of clays in landslides, folded strata, and the laboratory, Ge¤otechnique 35 (1985) 3^18. [49] J.B. Burland, On the compressibility and shear strength of natural clays, Ge¤otechnique 40 (1990) 329^378. [50] P.G. Slade, J.P. Quirk, K. Norrish, Crystalline swelling of smectite samples in concentrated NaCl solutions in relation to layer change, Clays Clay Minerals 39 (1991) 234^ 238. [51] T. Sato, T. Watanabe, R. Otsuka, E¡ects of layer charge location, and energy change on expansion properties of dioctahedral smecites, Clay Clay Minerals 40 (1992) 103^ 113. [52] W.L. Huang, W.A. Bassett, T. Wu, Dehydration and hydration of montmorillonite at elevated temperatures and pressures monitored using synchrotron radiation, Am. Miner. 79 (1994) 683^691. [53] B. Ransom, H.G. Helgeson, A chemical and thermodynamic model of dioctahedral 2 layer clay minerals in diagenetic processes: dehydration of diotahedral aluminous smectite as a function of temperature and depth in sedimentary basins, Am. J. Sci. 295 (1) (1995) 245^281. [54] P.M. Bu«chler, The e¡ect of exchangable cations in the permeability of a bentonite to be used in a stabilization pond liner, Water Sci. Technol. 22 (1990) 23^26. [55] H. Frenkel, J.O. Goertzen, J.D. Rhoades, E¡ects of clay type and content, exchangable sodium percentage, and electrolyte concentration on clay dispersion and soil hydraulic conductivity, Soil Sci. Soc. Am. J. 42 (1978) 32^ 39. [56] W.H. Konomi, T.K. Ito, K. Kondo, Focal mechanism solutions of earthquakes in the eastern Shikoku, its vicinity, Annu. Rep. Disaster Prev. Res. Inst. Kyoto Univ. 35 (1992) 333^346. [57] D.M. Davis, The compressive mechanics of accretionary wedges applied to the Leg 78A study area near Barbados, in: B. Biju-Duval, J.C. Moore et al. (Eds.), Init. Rep. DSDP 78A (1984) 569^579. [58] D.M. Sa¡er, B.A. Bekins, Hydrologic controls on the morphology and mechanics of accretionary wedges, Geology 30 (2002) 271^274. [59] D.M. Sa¡er, B.A. Bekins, Episodic £uid £ow in the Nankai accretionary complex: Timescale, geochemistry, £ow rates, and £uid budget, J. Geophys. Res. 103 (1998) 30351^30370. [60] S.L. Bilek, T. Lay, Comparison of depth dependent fault zone properties in the Japan trench and middle America Trench, Pure Appl. Geophys. 154 (1999) 433^456. [61] R.D. Hyndman, K. Wang, M. Yamano, Thermal constraints on the seismogenic portion of the southwestern Japan subduction thrust, J. Geophys. Res. 100 (1995) 15373^15392.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
Earth and Planetary Science Letters 214 (2003) 589^603 www.elsevier.com/locate/epsl
Compositional and £uid pressure controls on the state of stress on the Nankai subduction thrust: A weak plate boundary K.M. Brown a; , A. Kopf a , M.B. Underwood b , J.L. Weinberger a a
Scripps Institution of Oceanography, UCSD, 9500 Gilman Drive, La Jolla, CA 92093-0244, USA b Geological Sciences, University of Missouri, Columbia, MO 65211, USA Received 23 August 2002; received in revised form 1 May 2003; accepted 12 July 2003
Abstract We show that both fault mineralogy and regional excess fluid pressure contribute to low resolved shear stresses on the Nankai subduction plate boundary off southwest Japan. Ring and direct shear tests indicate that saturated clay minerals in the fault possess intrinsically low residual friction coefficients (Wr ) at stress levels between 1.0 and 40 MPa. The direct shear Wr values for purified smectite are V0.14 1 0.02, for illite V0.25 1 0.01, and for chlorite 0.26 1 0.02 (for point load velocities of 0.0001 mm/s). These clay minerals dominate the Nankai subduction de¤collement zone. Illite (plus quartz) is mechanically important in the altered incoming Muroto section and the predicted de¤collement Wr should lie between 0.2 and 0.32. This low residual strength, together with elevated fluid pressure, limits shear stresses to below V4 MPa within the frontal V50 km of the subduction system, consistent with the low wedge taper in this region. A higher wedge taper off the Ashizuri peninsula indicates basal shear stresses rise slightly along strike towards this region. Our analysis indicates lower fluid pressures must predominantly be responsible because only small second order along strike variations in Wr are predicted to occur as a result of variations in smectite and total clay content. These variations should be further reduced at depth under the wedge as smectite is diagenetically altered to illite. However, our data suggest the low Wr values of the clay-rich de¤collement still limit shear stresses to between V17 and 29 MPa within the frontal V50 km of the wedge, consistent with other estimates of plate boundary weakness. Indeed, we propose that it should be expected that subduction plate boundaries like Nankai will be weak because of the intrinsic presence of clay-rich faults and moderate fluid overpressures. Our data do not support the hypothesis that the smectite-to-illite reaction directly controls the onset of seismogenic behavior deep in the Nankai system because there is already a mechanical dominance of illite (rather than smectite) in the shallow de¤collement zone, and we find all the clay phases tend to velocity strengthen. However, temperature-activated clay diagenesis and dehydration may cause secondary changes in the fault properties and state of stress across the up-dip limit of the seismogenic zone. > 2003 Elsevier B.V. All rights reserved. Keywords: frictional strength; state of stress; £uid pressure; convergent margin; seismogenic zone; Ocean Drilling Program
1. Introduction * Corresponding author. E-mail address: [email protected] (K.M. Brown).
We present the initial results of an ongoing ex-
0012-821X / 03 / $ ^ see front matter > 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0012-821X(03)00388-1
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
590
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
perimental study that examines the role that clay minerals play in controlling the frictional properties of subduction plate boundary faults and how, together with regional £uid pressure variations, this controls the state of stress on the subduction thrust. We also address the role that clay diagenesis plays in the onset of seismogenic activity. We focus here on the Nankai Trough, southwest Japan, because it is representative of sediment-rich subduction zones (including Alaska and Cascadia) that are capable of generating very large earthquakes (M V8 and greater) [1^4]. Two signi¢cant, seemingly contradictory attributes of these subduction zone faults are that they appear to be almost fully locked over much of their alongstrike length during interseismic periods even though the faults appear to be extremely weak [5]. Accretionary wedge environments provide the opportunity to test more speci¢c ideas as to how three-dimensional changes in parameters such as fault zone mineralogy, temperature, and £uid pressure (i.e. e¡ective con¢ning stress) can a¡ect the stress state and seismogenic properties of plate boundaries. We chose the Nankai subduction zone as a study area, because it has a historical record of very large earthquakes that dates back to AD 684 [6^8]. The recurrence interval in the western Nankai region varies between 262 and 92 years, and the most recent ruptures were the 1944 Tonankai (M 8) and 1946 Nankaido (M 8.2) events o¡shore of the Kii and Muroto peninsulas [6,9]. In addition, the nationwide permanent Global Positioning System network shows that the coastal forearc region is moving to the westnorthwest at the full convergence rate of V4 cm/ yr, which indicates the plate boundary is essentially fully coupled or locked during the interseismic period [10,11]. However, the plate boundary fault also seems to be extremely weak if the maximum principal stress currently has a trench-parallel orientation and the margin-normal stress appears to be less than the vertical stress in the backstop region. For this to occur, numerical modeling suggests the maximum shear stress on the subduction thrust should not exceed V17 MPa even at the stronger down-dip end of the Nankai seismogenic zone [5]. Coring by the Deep Sea Drilling Project
(DSDP) and the Ocean Drilling Program (ODP) has already revealed signi¢cant along-strike changes in the lithostratigraphy, heat £ow, clay mineralogy, and diagenesis of the incoming Shikoku Basin section seaward of the Nankai Trough. Sediment composition at Nankai appears to be controlled by changing patterns of heat £ow and thermally activated clay diagenesis [12^14]. These shallow, along-strike mineralogical changes mimic those within the fault zone with increasing depth, and thereby allow us to test whether such changes signi¢cantly in£uence the down-dip frictional properties of this plate boundary. In this paper, we focus on experimentally determining the frictional strength of materials entering the subduction toe in order to independently evaluate physical reasons for the apparent weakness of the subduction fault at Nankai. We also address the issue of whether or not the smectiteto-illite reaction can lead to signi¢cant down-dip changes in fault strength. Smectite has certainly been recognized as a weak clay mineral, but it is not clear whether the transformation to illite causes substantial changes in friction coe⁄cient. Our goal is to address how sediment composition impacts fault properties at e¡ective normal stresses up to 40 MPa using both a ring shear and a newly developed direct shear apparatus. A combination of natural Nankai samples, individual mineral standards, and mixtures of smectite (Sm), illite (Il), chlorite (Ch), and quartz (Qtz) were used. Based on the experimental data we also address the potential role that regional pore pressure changes may play in determining the lateral changes in the overall fault strength at Nankai. In addition to arguments about the general state of stress, we are also able to o¡er insights on the debate over the potential diagenetic controls on the up-dip limit on seismogenic activity [2].
2. Geological context of the Nankai Trough At the Nankai Trough, the Philippine Sea plate (Shikoku Basin) is subducting to the northwest under southwest Japan (Eurasian plate) at a rate of 2^4 cm/yr [15,16] ; the direction of convergence is slightly oblique to the plate margin (Fig. 1).
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
Fig. 1. Location map of southwest Japan showing the distribution of ODP and DSDP sites. Also shown are P- and T-axis diagrams for 22 seismic events that suggest the principal stress is parallel to the trench [5,56].
The subducting oceanic lithosphere of the Shikoku Basin was generated by the rifting of the IzuBonin backarc that started in the Oligocene and culminated in Shikoku Basin sea£oor spreading that lasted until V13^15 Ma [17^19]. Seamount volcanism (Kinan seamounts) apparently persisted after the cessation of backarc spreading and contributed to the current elevated thermal gradients near the paleo-spreading axis. Subduction and the collision of the Izu-Bonin arc against Honshu apparently started 15^10 Ma at a slow rate with the development of a deeply penetrating subducting slab and volcanic activity occurring by V8 Ma in southern Kyushu and by 6 Ma in southwest Japan [20,21]. Subduction of the Shikoku Basin, accumulation of a thick trench-wedge turbidite facies, and sediment accretion subsequently led to the development of a broad accretionary prism. The de¤collement near the toe occupies a stratigraphic position within the lower Shikoku Basin facies (Fig. 2).
591
Oceanic basement relief and thermal state signi¢cantly in£uence sediment dispersal, composition, facies architecture, and diagenesis within the critical Miocene Shikoku Basin strata [14, 22,23]. The upper Shikoku Basin facies consists of hemipelagic mudstone and volcanic ash. O¡ the Ashizuri Peninsula (DSDP Site 297, ODP Site 1177, Figs. 1 and 2A), the lower Shikoku Basin facies consists of hemipelagic mudstone with abundant terrigenous sand turbidites and volcanic ash. Shipboard ODP results suggest that Sm is abundant, particularly in the deeper portions of this section. The Ashizuri de¤collement appears to occupy a position on seismic re£ection pro¢les just above the top of a series of turbidites at the toe of the wedge [22,24]. In contrast, along the Muroto transect farther northeast (Fig. 1), the thinner Miocene section accumulated above basement highs associated with the fossil backarc spreading center and the Kinan seamounts. The sandy turbidites are absent there (Fig. 2), so that the Muroto de¤collement passes through monotonous mudstones (Fig. 2B). The temperature gradient remains elevated (180‡C/km [25]) near the Kinan seamount chain, which leads to signi¢cant clay diagenesis near the prism toe [23,26].
3. Methods and experimental procedures 3.1. Sample selection, preparation, and XRD The following three types of samples were tested in this study: (i) natural samples from the Shikoku Basin section (Fig. 2); (ii) mineral standards composed of Sm, Il, Ch, and Qtz; and (iii) mineral mixtures of Qtz/Sm and Qtz/Il. All the frictional tests were conducted under fully seawater-saturated conditions to ensure the appropriate chemical environment for the clay phases in a marine environment. Natural samples were collected from mudstone horizons cored at DSDP Site 297 (Figs. 1 and 2A). Digital X-ray di¡raction (XRD) data for the natural Nankai samples were generated at the University of Missouri using a Scintag PAD V di¡ractometer. The generic methodologies for the XRD analyses for
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
Fig. 2. Lithostratigraphy and clay mineralogy of the Ashizuri (DSDP Site 297 and ODP Site 1173) and Muroto (ODP Site 1177) transects.
592
bulk powder and clay minerals are described in [27]. The Wards mineral standards (Sm, Wyoming bentonite), Il (Rochester Shale), and Ch (Madison Co) were size separated ( 6 2 Wm fraction retained) by centrifuging after rehydration with seawater ( s 72 h) to reduce contamination associated with the ¢ne silt and above size fraction. Contamination of the Il and to some extent the Sm standards is particularly signi¢cant with up to 30% Qtz in the Rochester Shale [28]. Tests were also conducted on arti¢cial mixtures of the puri¢ed end member Sm+Qtz and Il+Qtz mixtures (using dry weights), and on the unpuri¢ed Rochester Shale sample for comparison with the natural Nankai samples. In order to account for the widely varying mineral grain densities of the different phases, all wt% data are converted to the volumetric ratios of the di¡erent components in the sediment framework. Smectite is particularly problematic in this regard because it hydrates under saturated conditions and has a corresponding low mineral density of V2.1 g/cm3 in contact with seawater at moderate to elevated stress levels V hydrate [29,30]). (15 A 3.2. Frictional testing We conducted frictional resistance tests using both a ring shear apparatus at normal stresses of 6 2 MPa [31,32] and a direct shear system at normal stresses of 5^40 MPa [33,34]). A primary objective was the determination of the relationship between mineralogy and frictional resistance. A secondary objective was to obtain some preliminary data of the variation in Wr with normal stress level and slip velocity. Changes with normal stress are particularly germane to investigations of the controls on plate boundary stress state. The frictional resistance of a clay-rich sediment or fault gouge at elevated displacements is typically de¢ned as: Wr = d/(cn 3Pf ), where Wr is the residual frictional coe⁄cient, d is the residual shear strength, cn is the total normal stress on the gouge layer, and Pf is the £uid pressure. We focus on Wr because clay-dominated sediments have distinctly lower residual shear strengths (undergo strain weakening) at higher displace-
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
ments because the tabular clay minerals reorient and align within the shear zones [31,35]. The residual shear strength was determined at low stress levels (1^2 MPa) utilizing a standard Wykeham^Farrance Bromhead ring shear apparatus (WF 25850) [32]. Samples included a mineralogically diverse subset of natural Nankai samples and arti¢cial mineral mixtures. Previous studies indicate that the frictional behavior of a sediment changes in response to the transitions in properties of its principal mineral components, with residual shear strengths decreasing as total clay and Sm contents increase under saturated conditions [31,36]. To assure initial drained conditions, the 3^4 mm thick ring shear samples were consolidated for s 48 h until no further changes in thickness were observed. In order to maintain conditions as near to drained as possible, subsequent frictional testing was then conducted at relatively slow displacement rates of 0.0015 mm/s. The samples were also typically sheared for s 70 mm (V12 h) until no further changes in shear stress occurred in order to verify a drained residual value had been achieved. At higher stress levels, we conducted tests on mineral standards in a modi¢ed GEOCOMP direct shear apparatus (Fig. 3) utilizing a tabular sample geometry of 50 mmU100 mmUV15 mm (WULUH). The sample is contained within a shear box. A piston forces the sample approximately halfway (i.e. V8 mm) into the base plate of the apparatus and holds it under the required normal load. The outer shear box is forcefully held down on the Te£on gasket by an external jig arrangement in order to prevent sample loss during initial sample consolidation (a problem with the pure smectite samples). The seawater-saturated direct shear samples were consolidated for several days under uni-axial consolidation conditions to ¢nal stress levels of between 5 and 40 MPa. Monitoring the pore pressure allowed us to determine when the sample had achieved a near drained state prior to the initiation of shear testing (excess pore pressure developed during loading 6 1% of the total normal stress). We utilized pore pressure ports at the base of the sample (Fig. 3) that made a connec-
593
Fig. 3. (A) Schematic diagram of the direct shear sample geometry and pore pressure monitoring system used during the frictional testing. (B) Photograph showing the fabrics developed within a typical clay gouge (Rochester Shale puri¢ed for illite) after a direct shear test at 20 MPa normal stress.
tion with the shear zone via permeable quartz silt pillars formed by drilling three 3 mm wide cylindrical holes into the lower part of the tabular sample. In later test runs these were modi¢ed into narrow (2 mm) slit-¢lled oblong slits (length 1.5 cm) oriented with their long axes parallel to the shear direction to maintain prolonged hydrologic connection. The load on the outer shear box is subsequently fully released during the shear phase of the test so that it slips freely sideways on the Te£on gasket (Fig. 3). During the subsequent shear testing, the horizontal and vertical loads and displacements, as well as pore pressure at the ports into the fault zone, were continually recorded (Fig. 3). The tests were conducted under constant normal load condition, with the vertical load being reduced during displacement to account for the decreasing contact area between the upper and lower halves of the sample. Frictional resistance is dependent on slip rate at higher stress levels [37,38]. A discussion of the full rate and state variables for velocity-dependent frictional stability is given elsewhere [37^39]. To evaluate this dependence we undertook shear rate or velocity stepping tests as part of the direct shear tests. Shear rates varied over ¢ve orders of
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
594
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
magnitude from 0.00001 to 0.1 mm/s once the samples had achieved their residual state. We report here the frictional coe⁄cients at a low, but arbitrary, slip velocity of 0.0001 mm/s.
4. Results 4.1. Clay contents of the natural Nankai samples It was determined from the XRD analyses [27] that the total clay contents of the DSDP Site 297 samples analyzed as part of this study vary but generally increase with depth below the equivalent position of the Ashizuri de¤collement in the nearby wedge (Fig. 2). Sm and Il are the dominant clay phases together with minor Ch; the natural samples are thus relatively simple three-component mixtures of Sm, Il ( 1 Ch), and granular components such as Qtz ( 1 feldspar). At the equivalent level of the Ashizuri de¤collement (as based on seismic re£ection stratigraphic correlation), Sm contents are 6 10^20% of the bulk sediment (Fig. 2). The high interbed variability between the various clay components in the deeper section below the de¤collement is thought to re£ect changes in the provenance of the ¢negrained components between eroding accreted materials from the Shimanto Belt (high Il 1 Ch), and volcaniclastic and variably altered ash components from the Izu-Bonin arc (generally elevated Sm). Beds within the volcaniclastic-dominated unit at the base of the section may locally contain substantial amounts of vitric ash; also, overall smectite content of the clay-sized fraction is locally high (70^90%). XRD data provide no evidence for smectite diagenesis at depth in this region, which is consistent with the moderate thermal gradient of V50^56‡C/km measured at nearby DSDP sites [40]. For the shear tests on the natural samples, a subset of ¢ne-grained DSDP Site 297 samples was selected with a range of total clay, Sm, and Il contents [27]. The volumetric ratios were then calculated based on their wt% and estimated grain densities (with a error of V 1 5% to perhaps 1 5^ 10% in the XRD-based estimates of wt% ; see [27]).
Fig. 4. Results of ring shear tests conducted on natural Nankai samples (black dots) and arti¢cial mineral mixtures of Sm with Qtz. Data for tests we conducted on Il and Qtz mixtures are also shown. The residual friction (Wr ) data are plotted against the volumetric ratio of (A) smectite (VSm ) and (B) total clay (VSmþIlþCh ) within the solid fraction. A2 and M2 denote estimated altered deep de¤collement property ranges (see Fig. 8), A1 and AS1 are unaltered Ashizuri shallow de¤collement ranges and M1 is a shallow Muroto value (Fig. 8). a = Vitric ash containing samples.
4.2. Ring shear studies A compilation of the ring shear results is shown in Fig. 4. In general, total clay content and the presence of a weak Sm clay component are factors in controlling the mechanical response of sediments. Thus, we show ring shear frictional results plotted against both the volumetric ratio
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
of the hydrous Sm (VSm ) and total clay minerals (VSmþIlþCh ) components of the sediment framework. In Fig. 4, we additionally show the results of ring shear tests on arti¢cial mixtures of Sm+Qtz and Il+Qtz. These data are plotted together with the results of previous studies that were conducted on Sm+Qtz mixtures at 6 1 MPa [31] and at 50 MPa [36], with samples saturated in deionized water. The ¢rst-order results of the ring shear studies are as follows. The pure Sm, Il, and Ch phases were found to have Wr values of 0.12, V0.20, and V0.22, respectively, at normal stress levels 6 2 MPa (Fig. 4). Arti¢cial mineral mixtures show progressive decreases in frictional strength with increasing contents of Sm, Il and total clay minerals (Fig. 4). We plot these results together with previously published results for chlorite (Wr = 0.23) [41]. Our data for Sm+Qtz mixtures closely match the curves of two previous studies across much of the compositional range [31,36], except that their shear strength pro¢les £atten out at clay contents below 20 vol% and above 80 vol%. The Il+Qtz samples show a slightly stronger frictional resistance at higher clay contents than the Sm+Qtz samples (Fig. 4B). The unpuri¢ed saturated Rochester Shale (V30% Qtz) has a Wr value of 0.215 and falls in line with our predictions from the Qtz+Il mineral mixture curves (Fig. 4). The majority of the natural Nankai mudstone samples tend to be mechanically weak re£ecting their high clay mineral contents (Fig. 4). Most data for the Nankai samples do not plot on the Sm+Qtz mixture curves (Fig. 4A). Instead they scatter widely to weaker Wr values below the arti¢cial mineral mixture curves. However, even though the scatter is large, there is still a rough relationship between increasing smectite contents and lower Wr values. In contrast, a very close ¢t exists between the natural samples and the mineral mixture curves when the Wr values are plotted against total clay mineral volume or VSmþIlþCh (Fig. 4B). Thus, the total clay mineral content appears to be the primary control on the frictional properties of the Nankai sediments with smectite exerting only a secondary e¡ect. Three Wr values do plot well above the predicted total clay levels based on their values (Fig. 4B). These sam-
595
Fig. 5. Example of direct shear results for saturated illite at stress levels of 10, 20, and 30 MPa. Rates of velocity stepping tests are shown. Typically, a velocity rate change is followed by a direct initial change in the frictional resistance (a-value), which then subsequently comes to a steady state value at the new slip velocity (b-value). The magnitude of the decay to a steady state friction is typically reported as: a3b = vWr /vlnVel, where vWr is the change in steady state frictional strength and Vel is the sliding velocity [38]. We consistently report positive values of a3b for a velocity increase, implying a velocity strengthening or stable sliding mechanism is operating.
ples contain signi¢cant amounts of amorphous vitric ash that cannot be accounted for quantitatively in the XRD analyses of mineral abundance. 4.3. Direct shear studies conducted on saturated pure mineral samples In order to determine if the low intrinsic frictional resistance of phases such as illite is likely to persist at greater depths and e¡ective stresses along the Nankai plate boundary thrust, direct shear tests were conducted on the mineral standards at e¡ective con¢ning stresses up to 40 MPa. Typical results of direct shear tests are shown for illite at 10, 20, and 30 MPa normal stress (Fig. 5). The samples develop a peak strength at low displacements and subsequent lower residual shear strength after the ¢rst V10^12 mm of displacement. The post-peak strength drop to residual values becomes more pronounced at elevated stress levels. Owing to the sample thickness of V15 mm, we were able to produce a fully contained natural-looking shear zone during the tests. Typically, the shear zone was dominated by upper
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
596
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
and lower boundary shear with well-developed Riedel shear sets (Fig. 3B). Pore pressure measurements indicate that initial yielding and failure of the clay samples is marked by initial pore pressure increases ( 6 1^2% of the total normal load) around the primary failure zone of the clay samples (Fig. 6). The pore pressure rapidly rises until just after the peak failure (Fig. 6, inset) and then slowly falls as the sample reaches its residual state and the pressures drain. Small pore pressure variations were sometimes observed during the course of the velocity stepping tests in the low permeability clay sediments (Fig. 6). We are still working on these latter data. In general they seem to be too small to account for the observed changes in frictional strength with velocity (Fig. 6 and caption). In contrast, the pure quartz samples show little or no pore pressure response during either initial failure or subsequent velocity stepping tests presumably because their highly elevated permeabilities allow them to deform in a fully drained state. The main results of the direct shear tests are
Fig. 6. Example of the direct shear residual friction and pore pressure data for the 20 MPa illite test. Location of the expanded region is shown in the box within the inset ¢gure. In order to account for the drop in frictional resistance across the 0.01 to 0.00001 mm/s transition the pore pressure would have to increase by 2450 kPa. We do not see pore pressure changes of this order of magnitude in our data.
Fig. 7. Summary ¢gure of the ring shear and direct shear Wr versus normal stress results for various saturated puri¢ed clay minerals and quartz. The frictional coe⁄cient of non-saturated smectite [34] is also shown together with our and other [36] relevant test data for water-saturated smectite.
shown in Fig. 7. For comparison, we have also plotted the results from direct shear tests conducted on non-saturated smectite at room humidity [34] (dashed line in Fig. 7) and the pure saturated Sm end member for a study at 50 MPa [36]. Typically, the Wr values of the pure saturated clays remain relatively low and constant at the 10, 20, and 30 MPa normal stress levels (Fig. 7). The Wr values for smectite are V0.14 1 0.02, for illite V0.25 1 0.01, and for chlorite 0.26 1 0.02 (for point load velocities of 0.0001 mm/s). A small increase in Wr is seen relative to values recorded during low stress ring shear tests. This shift may be a real stress-dependent e¡ect or a minor o¡set generated by di¡erences between the two experimental geometries. Overall, however, the direct shear data support the contention that the basic frictional resistance of saturated clays will remain relatively weak at elevated stress levels. In addition, each type of clay (Sm, Il and Ch) and quartz exhibit velocity-strengthening behavior at all stress levels up to 30^40 MPa (for a representative illustration, see Figs. 5 and 6, illite).
5. Discussion The inherent frictional weakness of clay minerals and clay-rich mudstones has been noted since the birth of geotechnical engineering, with a large
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
body of work focussing on mitigating hazards caused by weak clay horizons at shallow depths [35]. However, measurements under seawatersaturated conditions and at higher stress levels are relatively rare, and we note a wide scatter of values for Wr in the literature particularly for illite (0.11^0.68) [28,42^45] and chlorite (0.1^0.43 [45,46]). Smectite seems to retain a relatively weak strength in all water-saturated studies we have been able to identify (predominantly 0.07^ 0.16) [31,41,44^48]. Our tests utilizing two di¡erent test geometries con¢rm the conclusions of the studies at low stress levels; the most common saturated clays in typical ocean margin sediments retain low Wr values at e¡ective stress levels up to 30^40 MPa (Fig. 7). Although we note the small o¡set in Wr between the direct shear and ring shear tests in our own studies, we suggest that unaccounted-for mineral impurities, saturation state, grain size distribution, and perhaps even drainage conditions may also be important factors to explain the variations seen in the literature. For example, typically the purity of the clay standards is often not ascertained and for Il standards such as Rochester Shale Qtz contents as high as V30% have been measured [28]. Given the pure clays’ insensitivity to changes of normal stress, we propose that the types of VSmþIlþCh vs. Wr relationships shown in Fig. 4 are likely to persist at least to normal stress levels of up to 30^40 MPa. This view is supported by the close match between our low stress ( 6 2 MPa) ring shear data and the results of Logan and Rauenzahn [36] for similar Qtz+Sm mixtures at 50 MPa (Fig. 4). In our experiments, we also monitor the potential e¡ects of retarded drainage and pore pressure transients during drained failure of low permeability clays. Initial positive transient pore pressure e¡ects during undrained failure in low permeability samples are reasonable to expect given the collapse of the clay fabric and increase in total stress during the initial part of the test. However, our measured increases are too small to account for the low Wr of the saturated clays in the latter parts of the test (Fig. 6). They also do not appear to be large enough to account for the changes in Wr that occur during the velocity stepping tests (Fig. 6). Pore pressure e¡ects are unlikely to be
597
broadly responsible for the general weakness of clays in any event. In natural settings weak saturated clays have long been known to be responsible for creeping or reactivation of landslides that move for decades; these time periods are too long for excess pore pressures to remain trapped within their basal shear zones [35,41,42,48,49]. Instead, we suggest that the weakness of the clays is more likely a result of the absorption of water onto the charged clay surfaces and the full hydration of the smectite interlayer [47,50^55]. Changes in pore £uid composition have also been shown to have secondary e¡ects on the frictional coe⁄cient of clays through the interaction with this surfaceabsorbed H2 O [47]. The importance of the chemical and physical e¡ects of water is illustrated by the o¡set between the saturated and non-saturated (partially hydrated) Sm curves (Fig. 7). The mismatch is particularly large at low normal stresses, where the Wr of the unsaturated Sm is considerably higher than the saturated phase. Intervening surface-absorbed H2 O may be the weakening factor in clays as it results in a reduced e¡ective frictional contact area between the faces of the aligned clay minerals in the shear zones. Essentially, this provides a molecular lubrication layer on the outer mineral surfaces or along the interlayer regions in the case of the hydrated Sm phase. Changes in the nature and ordering of clay fabrics due to the presence of H2 O may also be an important factor. We will continue to investigate how the behavior of this absorbed H2 O and any pore pressure changes in£uence the velocity dependence of Wr at elevated stress levels in future studies and primarily report the basic frictional strength data here. 5.1. General implications for the state of stress on the Nankai subduction thrust in 3D The inferred stress state in the Nankai forearc based on the forearc earthquake patterns (Fig. 1) [5,56] suggests that on a margin scale the basal shear stress on the subduction thrust is low. Here we make an independent evaluation as to whether a combination of low Wr values for the de¤collement materials and moderately elevated £uid pressures lead to low resolved shear stresses.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
598
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
This analysis is complicated by the observation that the wedge taper increases from 4.1‡ o¡ Muroto to 7.9‡ o¡ Ashizuri (Fig. 8), suggesting a correlated increase in the basal shear stress on the wedge and lateral variability in either Wr and/or £uid pressure [57,58]. Thus, we also attempt to evaluate the relative changes in the contributions of these two parameters to along strike variations in the basal shear stresses on the Nankai subduction thrust as well as the overall regional stress ¢eld. We summarize our estimated ranges for the spatial variation in de¤collement Wr values schematically in Fig. 8, taking into account the lateral and stratigraphic variations in total clay and smectite content between the Ashizuri and Muroto regions (Fig. 2). We have chosen end member points in the system. In Fig. 8, points A1 and A2, respectively, correspond to the correlated stratigraphic intervals in the incoming section and regions deep under the wedge o¡ Ashizuri, whereas points M1 and M2 correspond to the incoming and deep Muroto de¤collement zone, respectively. For completeness we also show estimated values for Site 808 o¡ Muroto at the toe of the wedge. We give Wr estimates for both the younger stratigraphic level of the de¤collement at the toe of the wedge and the older stratigraphic level (near the sediment basement contact) down to which the de¤collement generally steps at deeper levels in the system [22]. At point A1 o¡ Ashizuri (Fig. 8), the nearby de¤collement at the toe of the wedge lies in the younger stratigraphic units at V430 m depth where total clay contents are V50^55 wt%, illite contents are V20^40 wt%, and smectite contents are V10^25 wt% of the bulk sediment (Sites 297, 1173, Fig. 2). Based on the relationships in Fig. 4B for such compositions we would expect a likely maximum Wr value of V0.34^0.32 (at point A1s using the Il/Qtz curves in Fig. 4). In contrast, near the oceanic basement/sediment interface, the mudstones generally contain V60^80 wt% (V64^92 vol%) total clay, and closer to an average of V30^40 wt% (37^57 vol%) bulk smectite depending on whether we chose Site 297 or 1177 as a representative section (Fig. 2). For these compositions our measured unaltered Wr values are low
Fig. 8. Schematic 3D representation showing the estimates of the de¤collement zone Wr values for the Ashizuri (A1 to A2) and Muroto areas (M1 to M2) and the wedge taper angles. Values in the incoming section are given for the stratigraphic horizon corresponding to the de¤collement level at the toe and the older stratigraphic section down to which the de¤collement steps deep under the wedge (A2 and M2).
(excluding the vitric ash horizons) and mostly lie between 0.13 and 0.24 overlapping with both the Sm/Qtz and Il/Qtz end member curves (A1 range, Fig. 4B). Thus, there appears to be no relationship between the lowest Wr regions (i.e. highest total clay and smectite contents) in the incoming section and the initial de¤collement level at the toe of the wedge o¡ Ashizuri (Fig. 2). At Point A2, deep beneath the wedge o¡ Muroto (Fig. 8), the onset of clay diagenesis should result in the altered Wr value progressively coming to lying on the Il+Qtz curve, with a maximum range between V0.25 to V0.32 if we assume the total clay content is reduced to between V60% and 80 vol% as the low density Sm phase is reacted to illite (A2 range, Figs. 4 and 8). At point M1 in the incoming section o¡ Muroto (Site 1173, Figs. 2 and 8), the proto-de¤collement horizon correlates with a stratigraphic level at V400 m depth, below a small peak in smectite (Fig. 2). The total clay mineral content at the proto-de¤collement level is V58^60 wt% (V63 vol%) and comprises roughly equal volumetric proportions of smectite (V25 wt% or 31 vol%) and illite plus chlorite (V32 vol%). Therefore, we expect the Wr values in the proto-de¤collement region to lie approximately mid-way between 0.2 and 0.3 based on the Sm/Qtz and Il/Qtz end member curves in Fig. 4B (point M1). The Il/Qtz relationship seems reasonable because the Wr of chlorite is very close to that of illite. At point M2 (Fig.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
8), all the smectite will have reacted to illite in this high heat£ow region, increasing the altered Wr value towards V0.32, (Il/Qtz end member and V60 vol% clay, Figs. 4 and 8). Indeed, beause temperature gradients approach 180‡C/km [13], smectite diagenesis is known to be well advanced at the toe of the wedge (i.e. at Site 808, Fig. 8) so that the altered Wr value should already be representative for the subduction thrust at shallow levels. Thus, in summary, the residual strengths of mudstones in the subduction thrust at Nankai are generally low at all depths beneath the wedge because of the elevated clay mineral contents and any small secondary along-strike changes in Wr values in the incoming section will be further reduced with smectite diagenesis at depth (Fig. 8). Indeed, the smectite content did not even apparently control the de¤collement level at shallow depths o¡ Ashizuri where it is most elevated in the older section. Consequently, we conclude the observed along-strike changes in wedge taper and associated basal shear stress (Fig. 8) must primarily relate to regional variations in pore pressure in the subduction thrust. The types of small dynamic pore pressure changes we observed in response to velocity changes and initial yielding of the samples (Fig. 6) are transient phenomena that would be superimposed on such regional variations in pore pressure. We start our evaluation of the role that pore pressure variations may play in controlling stress and wedge taper angle in the more intensively studied Muroto region [22]. A simulation of overpressure distribution in the Nankai wedge o¡ Muroto [59] is shown in Fig. 9A where V* = Pf 3Ph /Pl 3Ph (Ph = hydrostatic, Pf = £uid, and Pl = lithostatic pressures). Typically, the degree of overpressuring in such numerical simulations often rises to a peak (in this case V*W0.8 [58,59]) at a certain burial distance back from the deformation front and subsequently remains elevated or drops slightly at greater depths depending on the parameterizations used in the numerical simulations (which include clay dehydration sources). The resulting e¡ective normal stress distribution on the Muroto de¤collement zone is shown in Fig. 9B. Initially values remain extremely low
599
( 6 V5 MPa) in the frontal V30 km of the wedge and begin to rise above V11 MPa only at distances s 50 km from the toe of the wedge. The calculated maximum shear stresses based on our experimentally derived puri¢ed mineral Wr values are also shown in Fig. 9B. All of the pure clay end members result in low shear strength because their inherent weakness is combined with excess £uid pressure. Even the slightly higher estimated Wr values of the diagenetically altered deeper Nankai de¤collement (M2, Fig. 8) results in a shear stress of 6 V4 MPa at V50 km from the deformation front (Fig. 9B). To see how these types of £uid pressures, when taken together with our Wr values, may account for the observed taper angle o¡ Muroto, we plot in Fig. 10A two parameters derived from the critical taper equation (equation 5) in [57] assuming the cohesion of the wedge and basal thrust is neg-
Fig. 9. (A) Numerically estimated [59] £uid pressure distribution in the frontal 50 km of the wedge. (B) Estimated shear stress on the Nankai subduction thrust o¡ Muroto derived from the £uid pressure distribution shown in A and the frictional properties of the puri¢ed clay minerals (Sm and Il) and Qtz (shown in Fig. 7). The estimated frictional resistances for the diagenetically altered Muroto de¤collement lithologies are also shown (M2, Fig. 8).
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
600
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
ligible and the Muroto taper angle is V4.1‡. The two parameters are the wedge’s basal friction coe⁄cient (Wrb ) and a Wedge Strength Parameter, (Kp 31)(13Vw ) consisting of both the wedge’s internal Vw condition and its passive shear strength parameter (Kp , see Fig. 10 caption for further de¢nitions). This latter parameter is directly related to the frictional properties of the sand-dominated accreted trench sediments. Typically Kp should be between V1.2 and 1.7 for clays and V3 and 3.7 for sands (also see caption, Fig. 10). In Fig. 10A we plot the region de¢ned by the range of £uid pressures along much of the de¤collement zone (Vw = 0.6^0.8) derived from the modeled pore pressure distribution at Muroto (Fig. 9A) and the maximum predicted frictional coe⁄cient of the altered de¤collement sediments (V0.32). The region de¢nes a range of wedge strength parameter values between V0.7 and 1.6. If we assume the sand end member Kp range for the trench turbidite-dominated wedge, then we predict the Vw value should lie between V0.2 and V0.7, with the midpoint being a reasonable average. From Fig. 9A we can see this range for the wedge is consistent with the simulation results shown in Fig. 10A because most simulations show that Vw values progressively increase downwards from hydrostatic (0.0) at the wedge surface to V0.6^0.8 at the de¤collement zone. We have no in situ pore pressure measurements or numerical simulations for the Ashizuri region, but turning around the above analysis allows us to predict what the £uid pressures on the de¤collement may be. The critical taper in the Ashizuri region (7.9‡) is much larger than o¡ Muroto. If we assume the accreted sandy trench wedge has similar properties and wedge strength parameter values (V0.7 and 1.6) in both regions the unaltered Ashizuri shallow de¤collement Wrb range of 0.32^0.34 (A1s, Fig. 8) would lead us to predict an average Vb value for the de¤collement near the toe of the wedge of between V0.2 and 0.55 for a taper angle of 7.9‡ (Fig. 10B). For a deep de¤collement lying in an altered, but more clay-rich and weaker older section (Wrb range of 0.25^0.32, point A2 in Fig. 8), Vb could lie between V0.0 and V0.5 (i.e. nearer hydrostatic). The range of shear stress we predict for this
Fig. 10. The relationship between the basal friction strength (Wrb ) and the wedge strength parameter (WSP) for various basal values for: (A) the Muroto critical taper angle of 4.1‡C and (B) the Ashizuri critical taper of 7.9‡. The relationships are derived from the standard critical taper equation (equation 5 in [49]) that relates basal frictional coe⁄cient and £uid pressure condition of the subduction thrust (respectively, Wrb and Vw ) and internal wedge (Vw , Kp ) to the resulting critical wedge taper (angle between the surface and basal thrust of the accretionary wedge). The Kp parameter is the passive earth coe⁄cient of soil mechanics [31] and is given by Kp = (1+sinP)/(13sinP) where P is the internal angle of friction of the wedge materials. Outlined regions represent the best guess estimates for the Nankai wedge de¤collement in the A1, A2, and M2 regions on Fig. 8.
margin based on these Wrb and £uid pressure estimates appears to be generally supportive of the independently derived estimates of Wang and He [5] for anything greater than hydrostatic £uid pressure conditions on the de¤collement. In their study, from numerical simulations, the average shear stresses on the Nankai subduction thrust decrease up-dip from a maximum of 6 V17^18 MPa at V90 km from the toe of the wedge. The stress will be even lower if the down-dip width of the zone of fault mechanical coupling is wider. Similarly o¡ Muroto, even with the altered deeper de¤collement compositions, we predict the shear stress in the frontal 30 km remains below 2 MPa and progressively rises towards and above 4 MPa at s 50 km from the toe of the wedge (Fig. 9B). O¡ Ashizuri, basal shear stresses at a distance of 50 km landward of the toe may rise to between V17 and V29 MPa for respective Vb approximate upper and lower limits of 0.55 and 0.0
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
(from Fig. 10) and a Wrb range between V0.32 and V0.25 estimated for an altered lower stratigraphic sequence (A2, Fig. 8). The range of our predicted maximum basal shear stresses developed in the two regions encompasses the maximum Wang and He stress estimate particularly if there is at least some moderate overpressuring in the system. We do, for example, feel that the hydrostatic end member is extreme and that Vb should be s 0.0 o¡ Ashizuri. Thus, the presence of weak clay-rich fault zones and at least moderately elevated £uid pressures in the subduction thrust easily account for the apparent weakness of this plate boundary, and no doubt the same factors account for why other subduction thrusts may be similarly weak [5]. Clearly, however, our analysis does suggest along-strike changes in £uid pressure and basal shear stress are likely to be important in the Nankai system. Our quick analysis indicates Vb values in the Ashizuri de¤collement region may be 6 V70% of the values o¡ Muroto (Fig. 10). In subduction zones, regional pore pressures are controlled by such factors as the system’s bulk permeability, rate of tectonic thickening (impacted by sediment input and subduction rate), and mineral dehydration reactions [58]. At Nankai, the rate of subduction does not change signi¢cantly along strike and, if anything, dehydration sources of £uids would increase with smectite content from Muroto to Ashizuri, in the wrong direction to drive the inferred £uid pressure variations. This leaves variations in the system’s bulk permeability as being the most likely reason for the apparent change in £uid pressure. This possibility has also been pointed out in a recent numerical evaluation of the hydrologic controls on the morphology of a number of accretionary wedges including east and west Nankai [58]. Perhaps the turbidite sands in the lower Shikoku Basin section o¡ the Ashizuri region form relatively e⁄cient lateral drains, reducing regional £uid pressures under the wedge in this region. Lower £uid pressures in the older turbidites may also explain why the de¤collement does not form in this level at the toe of the Ashizuri wedge despite the section containing units with very low residual strengths (A1, Fig. 8).
601
5.2. Implications for seismogenic processes As we expect the subduction thrust to be extremely weak at all depths, the implication is that the dynamic stress drop occurring during any earthquake must also be limited because it cannot exceed the peak frictional strength of the sediments just prior to failure. Thus, the small observed stress drops that occur during subduction earthquakes (typically between 1 and 10 MPa, e.g. [60]) seem entirely consistent with our low frictional strengths. It has also been hypothesized that the temperature range (60^150‡C) of the smectite-to-illite transition may be one critical factor in determining the onset of seismogenic activity if it coincides with a change from a velocitystrengthening to velocity-weakening response [61]. However, we did not observe velocity-weakening responses during tests on these clays, and particularly for the critical Il phase at stress levels below 40 MPa (Fig. 6). Additionally as we noted previously, the de¤collement beneath the shallow regions of the accretionary wedge does not lie in sediments where smectite is mechanically dominant. Thus, it seems unlikely that the Sm to Il reaction is directly (in terms of the frictional response) involved in the onset of unstable sliding and seismogenic activity. On the other hand, the combined e¡ects of water release during the Sm to Il reaction and any temperature-induced progressive chemical changes such as pressure solution/ cementation (+any clay diagenesis) could be important factors [12]. For example, a combination of a decrease in water release rate from clay dehydration and increasing Wr and temperature values could result in progressively elevating normal and shear stresses and chemical activity as the seismogenic zone is approached thereby, for example, enhancing pressure solution and a fault healing increasingly with depth.
Acknowledgements We thank Eli Silver and Kelin Wang for their timely and very helpful reviews that greatly assisted in the publication of this paper. This work was supported by NSF Grants OCE 0203799 and
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
602
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
9618166 (to K.M.B.), the NSF Margins program, the Humboldt Foundation (A.K.), and USSSP Grant 190-F001281 (to M.B.U.). This research utilized samples provided by DSDP/ODP, sponsored by the U.S. National Science Foundation and participating countries under management of Joint Oceanographic Institutions (JOI) Inc. [BOYLE]
[14]
[15]
[16]
References [17] [1] D.A. Oleskevich, R.D. Hyndman, K. Wang, The updip and downdip limits to great subduction earthquakes: Thermal and structural models of Cascadia, south Alaska, SW Japan, and Chile, J. Geophys. Res. 104 (B7) (1999) 14965^14991. [2] R.D. Hyndman, M. Yamano, D.A. Oleskevich, The seismogenic zone of subduction thrust faults, Island Arc 6 (1997) 244^260. [3] K. Satake, Depth distribution of coseismic slip along the Nankai Trough, Japan, from joint inversion of geodetic and tsunami data, J. Geophys. Res. 98 (1993) 4553^4565. [4] K. Satake, K. Shimazaki, Y. Tsujii, K. Ueda, Time and size of a giant earthquake in Cascadia inferred from Japanese tsunami records of January 1700, Nature 379 (1996) 246^249. [5] K. Wang, J. He, Mechanics of low-stress forearcs: Nankai and Cascadia, J. Geophys. Res. 104 (1999) 15191^ 15205. [6] M. Ando, Source mechanisms and tectonic signi¢cance of historical earthquakes along the Nankai Trough, Tectonophysics 27 (1975) 119^140. [7] M. Ando, A fault model of the 1946 Nankaido earthquake derived from tsunami data, Phys. Earth Planet. Inter. 28 (1991) 309^322. [8] H. Kumagai, Time sequance and the reoccurance models for large earthquakes along the Nankai Trough revisited, Geophys. Res. Lett. 23 (1996) 1139^1142. [9] H. Kanamori, Rupture processes of subduction-zone earthquakes, Annu. Rev. Earth Planet Sci. 14 (1986) 298^322. [10] S. Miyazaki, K. Heki, Crustal velocity ¢eld of Southwest Japan: Subduction and arc-arc collision, J. Geophys. Res. 106 (2001) 4305^4326. [11] S. Zhao, X. Wu, T. Hori, A. Smith, Y. Kaneda, S. Takemoto, Deformation and stress localization at the Nankai subduction zone, southwest Japan, Earth Planet. Sci. Lett. 206 (2003) 145^160. [12] C. Moore, D. Sa¡er, Updip limit of the seismogenic zone beneath the accretionary prism of southwest Japan: An e¡ect of diagenetic to low grade metamorphic processes and increasing e¡ecive stress, Geology 29 (2001) 183^186. [13] M.B. Underwood, K. Pickering, J. Gieskes, M. Kastner, R. Orr, Sediment geochemistry, clay mineralogy, and dia-
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
genesis: A synthesis of data from Leg 131, Nankai Trough, in: I.A. Hill, A. Taira, J.V. Firth (Eds.), Proc. ODP Sci. Results 131 (1993) 343^363. M.B. Underwood, D. Sa¡er, J. Steurer, O.L.S. Party, Along-strike variations in clay mineralogy and diagenesis: Implications for the up-dip limit of the Nankai seismogenic zone, EOS Trans. AGU 81 (2000) 1065. T. Seno, The instantaneous rotation vector of the Philippine Sea Plate relative to the Eurasian Plate, Tectonophysics 42 (1977) 209^226. D.E. Karig, C.L. Angevine, Geologic constraints on subduction rates in the Nankai Trough, Init. Rep. DSDP 87 (1986) 789^796. K. Okino, Y. Ohara, S. Kasuga, Y. Kato, The Philippine Sea: New survey results reveal the structure and history of the marginal basins, Geophys. Res. Lett. 26 (1999) 2287^2290. K. Okino, Y. Shimakawa, S. Nagaoka, Evolution of the Shikoku basin, J. Geomagn. Geoelectr. 46 (1994) 463^ 479. B. Taylor, Rifting and the volcanic-tectonic evolution of the Izu-Bonin-Mariana arc, in: B. Taylor, K. Fujioka (Eds.), Proc. ODP Sci. Results 126 (1992) 627^651. M. Takahashi, Large felsic magmatism of the Miocene outerzone of southwest Japan (in Japanese), Earth Monit. Spec. 23 (1999) 160^168. H. Kamata, K. Kodama, Tectonics of an arc-arc junction: an example from Kyushu Island at the junction of the southwest Japan Arc and Ryukyu Arc, Tectonophysics 233 (1994) 69^81. G.F. Moore, A. Taira, A. Klaus et al., New insights into deformation and £uid £ow processes in the Nankai Trough accretionary prism: Results of Ocean Drilling Program Leg 190, G-cubed 2 (2001) 10^25. M.B. Underwood, K. Pickering, Clay-mineral provenance, sediment dispersal patterns, and mudrock diagenesis in the Nankai accretionary prism, southwest Japan, Clays Clay Minerals 44 (1996) 339^356. J.K. Morgan, D.E. Karig, De¤collement processes at the Nankai accretionary margin, Southeast Japan, J. Geophys. Res. 100 (1995) 15221^15231. A. Taira, I. Hill, J. Firth, U. Berner, W. Bruckmann, T. Byrne, T. Chabernaud, A. Fisher, J.P. Foucher, T. Gamo, J. Gieskes, R. Hyndman, D. Karig, M. Kastner, Y. Kato, S. Lallemant, R. Lu, A. Maltman, G. Moore, K. Moran, G. Ola¡son, W. Owens, K. Pickering, F. Siena, E. Taylor, M. Underwood, C. Wilkinson, M. Yamano, Y. Zhang, Sediment deformation and hydrogeology of the Nankai Trough accretionary prism ^ Synthesis of shipboard results of ODP Leg 131, Earth Planet Sci. Lett. 109 (1992) 431^450. K.M. Brown, D.M. Sa¡er, B.A. Bekins, Smectite diagenesis, pore-water freshening, and £uid £ow at the toe of the Nankai wedge, Earth Planet. Sci. Lett. 194 (2001) 97^ 109. M. Underwood, N. Basu, J. Steurer, S. Udas, Data report: Normalization factors for semi-quantitative x-ray
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
K.M. Brown et al. / Earth and Planetary Science Letters 214 (2003) 589^603
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35] [36]
[37]
[38] [39] [40] [41] [42] [43]
[44]
[45]
di¡raction analysis, with application to DSDP site 297, Shikoku basin, in: G.F. Moore, A. Taira, A. Klaus (Eds.), Proc. ODP Sci. Results, Online publication HTML//, 190 (2003). C. Marone, D.M. Sa¡er, K. Frye, S. Mazzoni, Laboratory results indicating intrinsically stable frictional behavior of illite clay, EOS Trans. AGU 47 (Suppl.) (2001) 1248^ 1249. K.M. Brown, B. Ransom, Porosity corrections for smectite-rich sediments: Impact on studies of compaction, £uid generation, and tectonic history, Geology 24 (1996) 843^846. T.G. Fitts, K.M. Brown, Stress-induced smectite dehydration: Rami¢cations for patterns of freshening and £uid expulsion in the N. Barbados accretionary wedge, Earth Planet. Sci. Lett. 172 (1999) 170^197. J.F. Lupini, J.F. Skinner, A.E. Vaughan, The drained residual strength of cohesive soils, Ge¤otechnique 31 (1981) 181^213. A.W. Bishop, G.E. Green, V.K. Garga, A. Andersen, J.D. Brown, A new ring shear apparatus and its application to the measurement of residual strength, Ge¤otechnique 21 (1971) 273^328. C. Marone, C.B. Raleigh, C.H. Scholz, Frictional behavior and constituitive modelling of simulated fault gouge, J. Geophys. Res. 95 (1990) 7007^7025. D.M. Sa¡er, K.M. Frye, C. Marone, K. Mair, Laboratory results indicating complex and potentially unstable frictional behavior of smectite clay, Geophys. Res. Lett. 28 (2001) 2297^2300. T.W. Lambe, R.V. Whitman, Soil Mechanics, John Wiley and Sons, New York, 1979, 553 pp. J.M. Logan, K.A. Rauenzahn, Frictional dependence of gouge mixtures of quartz and montmorillonite on velocity, composition and fabric, Tectonophysics 144 (1987) 87^108. C. Marone, Laboratory-derived friction laws and their application to seismic faulting, Annu. Rev. Earth Planet. Sci. 26 (1998) 643^696. C.H. Scholz, Earthquakes and friction laws, Nature 391 (1998) 37^42. C.H. Scholz, The Mechanics of Earthquakes, Cambridge University Press, Cambridge, 1990, 439 pp. D.E. Karig, J.C. Ingle, L.S. Party, Initial Reports, 1975, 927 pp. A.W. Skempton, Long-term stability of clay slopes, Ge¤otechnique 14 (1964) 75^101. G. Mesri, A.F. Cepeda-Diaz, Residual shear strength of clays and shales, Ge¤otechnique 35 (1985) 269^274. G. Mesri, A.F. Cepeda-Diaz, Residual shear strength of clays and shales: Discussion, Ge¤otechnique 36 (1986) 269^281. C.A. Morrow, L.Q. Shi, J.D. Byerlee, Permeability of fault gouge under con¢ning pressure and shear stress, J. Geophys. Res. 89 (B5) (1984) 3193^3200. C.L. Chu, C.Y. Wang, W. Lin, Permeability and frictional properties of San Andreas fault gouges, Geophys. Res. Lett. 8 (1981) 565^568.
603
[46] C.A. Morrow, L.Q. Shi, J.D. Byerlee, Strain hardening and strength of clay-rich fault gouges, J. Geophys. Res. 87 (B8) (1982) 6771^6780. [47] R. Moore, The chemical and mineralogical controls upon the residual strength of pure and natural clays, Ge¤otechnique 41 (1991) 35^47. [48] A.W. Skempton, Residual strength of clays in landslides, folded strata, and the laboratory, Ge¤otechnique 35 (1985) 3^18. [49] J.B. Burland, On the compressibility and shear strength of natural clays, Ge¤otechnique 40 (1990) 329^378. [50] P.G. Slade, J.P. Quirk, K. Norrish, Crystalline swelling of smectite samples in concentrated NaCl solutions in relation to layer change, Clays Clay Minerals 39 (1991) 234^ 238. [51] T. Sato, T. Watanabe, R. Otsuka, E¡ects of layer charge location, and energy change on expansion properties of dioctahedral smecites, Clay Clay Minerals 40 (1992) 103^ 113. [52] W.L. Huang, W.A. Bassett, T. Wu, Dehydration and hydration of montmorillonite at elevated temperatures and pressures monitored using synchrotron radiation, Am. Miner. 79 (1994) 683^691. [53] B. Ransom, H.G. Helgeson, A chemical and thermodynamic model of dioctahedral 2 layer clay minerals in diagenetic processes: dehydration of diotahedral aluminous smectite as a function of temperature and depth in sedimentary basins, Am. J. Sci. 295 (1) (1995) 245^281. [54] P.M. Bu«chler, The e¡ect of exchangable cations in the permeability of a bentonite to be used in a stabilization pond liner, Water Sci. Technol. 22 (1990) 23^26. [55] H. Frenkel, J.O. Goertzen, J.D. Rhoades, E¡ects of clay type and content, exchangable sodium percentage, and electrolyte concentration on clay dispersion and soil hydraulic conductivity, Soil Sci. Soc. Am. J. 42 (1978) 32^ 39. [56] W.H. Konomi, T.K. Ito, K. Kondo, Focal mechanism solutions of earthquakes in the eastern Shikoku, its vicinity, Annu. Rep. Disaster Prev. Res. Inst. Kyoto Univ. 35 (1992) 333^346. [57] D.M. Davis, The compressive mechanics of accretionary wedges applied to the Leg 78A study area near Barbados, in: B. Biju-Duval, J.C. Moore et al. (Eds.), Init. Rep. DSDP 78A (1984) 569^579. [58] D.M. Sa¡er, B.A. Bekins, Hydrologic controls on the morphology and mechanics of accretionary wedges, Geology 30 (2002) 271^274. [59] D.M. Sa¡er, B.A. Bekins, Episodic £uid £ow in the Nankai accretionary complex: Timescale, geochemistry, £ow rates, and £uid budget, J. Geophys. Res. 103 (1998) 30351^30370. [60] S.L. Bilek, T. Lay, Comparison of depth dependent fault zone properties in the Japan trench and middle America Trench, Pure Appl. Geophys. 154 (1999) 433^456. [61] R.D. Hyndman, K. Wang, M. Yamano, Thermal constraints on the seismogenic portion of the southwestern Japan subduction thrust, J. Geophys. Res. 100 (1995) 15373^15392.
EPSL 6769 5-9-03 Cyaan Magenta Geel Zwart
