There is now clear evidence that large earthquakes rupture fault planes dipping 23°-35° in the two rift areas (Woodlark and Corinth) where high opening rates (10-37 mm/yr) are accompanied by high seismicity rates (Abers, 2001). Waveform inversion studies place an Mw = 6.2 earthquake with ~32°N dip on the Moresby normal fault at 5.3 km depth (Abers et al., 1997). Water multiples in the seismograms require 3.0 ± 0.2 km of water overlying the source, a condition only fulfilled near the Moresby rift axis (Figs. F2, F8). The teleseismic location of this and nearby earthquakes better fits the bathymetry and the location of microseismicity determined by a local array of ocean bottom seismometers/hydrophones (OBS/H) if the teleseismic locations (biased by the presence of velocity anomalies associated with subducted slabs to the north) are shifted south by 15 km (B. Taylor and B.C. Zelt, unpubl. data)—as has been done in Figure F2.
In spite of the failure to reach the Moresby detachment at depth (the main objective at Site 1108), critical information was provided by studies of samples from the fault zone where it cropped out at Site 1117. Onboard analyses revealed a transition downward from ductile to brittle conditions beneath the detachment (Shipboard Scientific Party, 1999), but information concerning the conditions at which the associated structures formed is conflicting. Roller et al. (2001) showed that synmylonitic structures and vein fill mineralogy (mostly calcite) indicate exhumation from significant depth. In particular, the quartz-rich porphyroclasts in the mylonite commonly display signs of dynamic recrystallization that attest to ductile creep under (at least) lowermost greenschist facies conditions. The deformation fabrics also show evidence for multiple opening and healing of veins during retrograde evolution and hydrothermal mineralization. Within the veins, calcite twins were used by Roller et al. as empirical paleopiezometers to derive differential paleostresses on the order of 20-40 MPa at temperatures lower than 200°C, corresponding to a maximum depth of 2 km assuming the rift basin thermal gradient of ~100°C/km (Shipboard Scientific Party, 1999). These temperatures and depths are consistent with the observation that the 66-Ma gabbro beneath the fault at Site 1117 was not thermally reset by subsequent rifting events and so must have remained at shallow and cool (<250°C) levels in the crust (Monteleone et al., this volume). However, they are not consistent with the estimated 2.5- to 3-km depth and 250°-300°C temperatures that mark the beginning of plastic deformation evidenced in the mylonites. Furthermore, the differential pressures are not consistent with the fault geometry, which indicates that there was no more than a few hundreds of meters of overburden above Site 1117 prior to faulting. There is consequently a conflict between the estimated depths and temperatures at which the mylonites and calcite twins formed and all other indicators of these parameters at Site 1117. A possible resolution is to invoke a dynamic shear heating process to explain the mylonites and calcite twins.
As briefly stated above (see also Shipboard Scientific Party, 1999), the fault gouge and the mylonites at Site 1117 show evidence of strong hydrothermal mineralization, suggestive of fluid migration within the detachment fault zone. This has been further substantiated by hydraulic conductivity testing on minicores from Sites 1108, 1114, and 1117 (Kopf, 2001). These measurements show that the fault gouge at Site 1117 has a larger permeability (on the order of 10-14 m2), especially in the direction parallel to the tectonic fabrics, than both the rift basin sediments at Site 1108 and the footwall block basement at Site 1114 (10-15 to 10-17 m2). Vertical permeability measurements on samples from Sites 1109 (Stover et al., this volume) and 1108, 1115, and 1118 (Kemerer and Screaton, this volume) led to similar values that range from 10-16 to 10-18 m2. The increase in permeability in the direction parallel to the fabric obviously derives from shear compaction. Together with the decrease in permeability above the fault zone (due both to the shear fabrics and the cementation), it results in enhanced fluid migration within the fault zone, as well as fluid overpressure that may trigger the fault activity. Note that anisotropic permeability also occurs in fine-grained sediments outside of fault zones at Site 1108, due to aligned microfractures (Bolton et al., 2000).
The detailed study by Floyd et al. (2001) of an MCS profile that images the fault zone at depth (and on which Site 1108 was drilled) provides a completely independent approach to the question of fluid flow within the fault zone. Their inversion of seismic reflection data indicates the occurrence within the fault zone of a 33-m-thick layer with a low P-wave velocity (4.3 km/s) at depths between 4 and 5 km. Isolated sections of the fault even show velocities as low as 1.7 km/s that strongly suggest high porosities maintained by high fluid pressures associated with hydrothermal fluid flow within the fault zone. Although the measurement technique as well as the conditions of overburden are quite different, these velocities are even lower than those obtained from core measurements on the fault gouge at Site 1117 (2 km/s) (Shipboard Scientific Party, 1999) and require high porosities and near-lithostatic fluid pressures to maintain them. High porosities are confirmed by the concurrent decrease in Poisson's ratio and P-wave velocities in sections of the fault zone (inferred from the amplitude variation with offset of the common midpoint gathers), which is interpreted as a greater decrease in P-wave than S-wave velocity resulting from slower P-wave paths through pore fluids (Floyd et al., 2001).
In addition to the evidence for a permeable and porous Moresby fault zone at greater than hydrostatic pressures, the samples of the gouge from the fault at Site 1117 provide evidence for weak frictional properties. The fault gouge minerology is talc-chlorite-serpentinite-calcite (Shipboard Scientific Party, 1999)—an association commonly produced by alteration of mafic and ultramafic rocks (Deer et al., 1992), such as the gabbros at Site 1117. Talc, more than serpentinite, has the important mechanical property of maintaining low coefficients of friction to greenschist pressures and temperatures (C. Scholz, pers. comm. 2000). Our best estimates of the Moresby fault dip (27°-32°N) (Abers et al., 1997; Taylor et al., 1999) are consistent with the frictional properties expected for mature faults with well-developed gouge zones (Abers, 2001; Byerlee and Savage, 1992). As Abers (2001) demonstrates, very unusual fault mechanics may not be needed. Specifically, a combination of a cohesionless fault with only slightly reduced coefficients of friction (0.5) vs. surrounds with cohesion and typical friction (0.6) permits failure in the upper crust on faults of 20°-35° dip without hydrofracture even at hydrostatic fluid pressures.
Thus, the propensity for failure at the shallow dips (~30°) observed on the Moresby normal fault is overdetermined (i.e., the result of several congruent factors, each sufficient to produce the result). There is evidence for both:
These fault properties and failure modes are a logical consequence of a continent with an ophiolitic upper crust being rapidly extended adjacent to a seafloor spreading center. Such shallow-angle normal faults may be a common feature of strain localization during the transition from rifting to spreading (Abers et al., 1997; Taylor et al., 1999; Floyd et al., 2001; Pérez-Gussinyé and Reston, 2001). In this case, however, given the shallow dips of the opposing faults to either side of Site 1114 (Figs. F1, F8), at least one if not both of these faults initiated at shallow dips (i.e., did not rotate from high angles).