Introduction

Assessing how major plate boundary faults fail, particularly as large earthquakes, requires analysis of how natural material from the fault zone and wall rock behave mechanically. Therefore, obtaining such samples has provided the impetus for large-scale scientific drilling projects (e.g., Ma et al. 2006; Tobin and Kinoshita 2006; Zoback et al. 2011). The Deep Fault Drilling Project (DFDP) was initiated in 2011 to investigate the Alpine Fault in New Zealand via coring, sampling, and wireline logging (Townend et al. 2009). The Alpine Fault is a highly appropriate target for study due to its status as a major plate boundary fault capable of large (M w  ~ 8) earthquakes, as well as the inference that it may be nearing the end of its earthquake cycle (Sutherland et al. 2007; De Pascale and Langridge 2012).

Laboratory measurements of frictional properties are an essential component of studies targeting fault slip behavior. Measurements of frictional strength and how this strength changes as a function of slip velocity (friction velocity dependence) provide valuable information constraining the location and likelihood of earthquake slip nucleation. Simulating natural conditions on major fault zones at depths where large earthquakes nucleate (i.e., several km depth) is not always straightforward. However, it is a critical component of experimental studies because the frictional properties of fault material can be highly dependent on experimental conditions such as stress, temperature, and the presence of pore fluids (e.g., Handin 1969; Blanpied et al. 1998; Collettini et al. 2009; Carpenter et al. 2012; den Hartog and Spiers 2013; Ikari et al. 2013, 2014; Boulton et al. 2014). We present results of laboratory measurements of frictional strength and velocity-dependent friction at pressures and temperatures representing in situ conditions at several km depth using intact and disaggregated samples from the two pilot boreholes of a multi-phase scientific drilling project on the Alpine Fault. The data we present here are part of a larger effort to comprehensively quantify the mechanical behavior of the Alpine Fault and support other recent work with similar goals (e.g., Sutherland et al. 2012; Townend et al. 2013; Boulton et al. 2014; Carpenter et al. 2014; Ikari et al. 2014), as well as providing a foundation for further drilling during Phase 2 of DFDP.

Geologic setting

The Alpine Fault, South Island, New Zealand

The Alpine Fault forms part of the boundary between the Pacific Plate and the Australian Plate along the West coast of New Zealand’s South Island (Fig. 1a). The fault is dextral-reverse and has accumulated ~400–500 km of cumulative offset at a displacement rate of ~23–27 mm/year over the past ~50,000 years, based on offset of geologic markers and radiocarbon dating (Cooper and Norris 1994; Norris and Cooper 2000; Sutherland et al. 2006). Major earthquakes (M w  > 7.0) are inferred to have occurred over the past 8,000 years with a recurrence of ~260–400 years (Bull 1996; Berryman et al. 2012). No large earthquake has occurred in the past 300 years, suggesting that a hazardous, large-magnitude earthquake on the Alpine Fault may be imminent (Sutherland et al. 2007, 2012; Townend et al. 2009, 2013). Geodetic measurements along the central segment of the fault indicate that the Alpine Fault is fully locked at depths of 5–8 km and partially locked up to ~18 km, and is loaded from below by the lower crust at a rate representing 50–70 % of the plate convergence rate (Beavan et al. 1999, 2007; Norris and Cooper 2000; Wallace et al. 2007). The lack of measurable historic creep at the surface indicates that tectonic strain is balanced by earthquake slip (Beavan et al. 1999).

Fig. 1
figure 1

a Map of the New Zealand area. Star indicates location of DFDP drilling. b Geologic cross-section showing DFDP-1A and DFDP-1B boreholes (modified from Sutherland et al. 2012)

Full size image

Deep Fault Drilling Project and sample description

The Alpine Fault was chosen for scientific drilling because it is nearing the end of its earthquake cycle, high exhumation rates that have resulted in well-exposed outcrops of representative lithologies from depth, well-constrained Quaternary slip rates, the presence of an extensive geophysical monitoring network, and a fault dip of ~45° which allows penetration by vertical boreholes (Townend et al. 2009). Drilling on the central portion of the Alpine Fault commenced as part of the Deep Fault Drilling Project (DFDP) within the framework of the International Continental Scientific Drilling Project (ICDP). The initial phase of the project resulted in two pilot boreholes located 80 m apart: DFDP-1A drilled to 96-m depth, and DFDP-1B drilled to 151-m depth (Fig. 1b). A deeper borehole with a target depth of ~1,500 m is planned for the second phase of the project, DFDP-2. The hanging wall consists of five major lithologies: dark-gray highly fractured ultramylonite, laminated brown-green-black ultramylonite, foliated and unfoliated greenish-gray cataclasite, and gouge (Toy et al. 2014). Both boreholes successfully penetrated the fault zone proper, which is identified in recovered drill cores as a cm-scale package of fine-grained, clay-rich gouge interpreted to be a principal slip zone (PSZ) (Sutherland et al. 2012; Townend et al. 2013). The PSZ in borehole DFDP-1A is ~3 cm thick and located at ~90.6-m depth, under which Quaternary fluvial gravels were recovered as the footwall. In DFDP-1B, two PSZs were identified (Toy et al. 2014). The main (upper) PSZ is ~20 cm thick and located within green cataclasites at a depth of ~128 m, as measured in the recovered core. The second PSZ is located at ~144-m depth; it is yet to be determined which PSZ was most recently active. The footwall in the DFDP-1B borehole consists of granitoid cataclasites underlain by gneisses. From Borehole 1A, we tested two hanging wall ultramylonite samples, a poorly indurated gouge sample from 30 cm above the PSZ, and one sample of footwall gravel (Table 1). We emphasize that the gouge sample is located near but not within the zone interpreted as the PSZ, rather it is part of a larger zone of cataclasite-derived gouge. From Borehole 1B, we tested one ultramylonite sample and one sample of (proto-) cataclasite from the hanging wall and one granitoid cataclasite sample from below the first PSZ but above the second PSZ (Table 1).

Table 1 Sample and experiment details
Full size table

Experimental methods

Shearing experiments were conducted in a single-direct shear apparatus under controlled normal stresses up to 150 MPa, temperatures up to 225 °C, and with constant, monitored pore fluid pressures. The sample cell consists of a cylindrical volume (diameter 63 mm, height ~30–50 mm) within two hardened steel blocks. The top face of the sample (not the forcing block) is loaded vertically, and the lower block is displaced relative to the upper block using a near-frictionless roller mechanism at a controlled displacement rate, inducing shear perpendicular to the cylinder axis (Fig. 2). Although the total sample height is ~3 cm, the deformation is planar and thus simulates localized shear, although the actual shear surface of the sample may be slightly irregular. O-ring seals located between the two sample cell blocks, and between the upper cell and driving piston, confine deionised water pore fluids. Isolated ports in the lower block allow the pore pressure in at the sample bottom (backpressure) and also along the shear surface to be monitored separately (Fig. 2, inset). Pore fluid pressures were kept low (~500–600 kPa) by manually draining the sample with an external vent so that the applied normal stress σ n may be considered to be the effective normal stress \(\sigma_{n}^{{\prime }}\). The temperature is constantly monitored in two locations: within the oven external to the sample cell and via a thermistor in one of the ports in the lower cell in contact with the pore fluid.

Fig. 2
figure 2

Schematic illustration of the heated single-direct shear device

Full size image

We conducted two sets of experiments. In the first set, we used one sample of mylonite, obtained from 77.8 m depth in the DFDP-1A borehole. This sample is highly friable and was thus tested as disaggregated, mm-scale rock chips similar to borehole cuttings. For these experiments, only external (oven) temperature monitoring and, for tests with pore fluid, only backpressure fluid monitoring were available. This set of experiments was done using one sample in order to explore the effect of pressure and temperature, as well as the presence or absence of pore fluid on frictional behavior. In the second set of experiments, we used a selection of samples from both the DFDP-1A and DFDP-1B boreholes. These were either tested as disaggregated rock chips, or further ground to a maximum grain size of 250 µm. When possible, some samples were tested as intact rock cylinders aligned perpendicular to the core axis. These samples were either drilled or otherwise carefully sculpted from an original whole-round core diameter of 83 mm down to 63 mm and were saturated with deionised water under a vacuum overnight. For this set of experiments, all samples were sheared with the presence of pore fluid at 100 MPa effective normal stress and 160 °C as monitored by pore fluid temperature, in order to simulate realistic in situ conditions at several km depth. This set of experiments facilitates comparison between samples from the major lithologic units in the DFDP-1 boreholes at consistent pressure, temperature, and fluid saturation conditions.

We sheared the samples at a constant velocity of 10 µm/s for up to 12 mm in order to measure the steady-state shear strength τ (Fig. 3). Most samples achieved steady-state (i.e., residual) strength by 9-mm displacement; however, some samples exhibited long-term strain hardening. In these cases, residual strength was measured at 9-mm displacement to facilitate comparison between samples. Most samples did not exhibit a significant peak in strength, but rather a gradual increase to a residual value. We calculate the coefficient of sliding friction (µ) as:

$$\mu = \frac{\tau }{{\sigma_{n}^{{\prime }} }}$$
(1)

(Handin 1969) assuming that cohesion is negligible in our disaggregated samples or is lost after significant displacement in intact samples. After ~9- to 12-mm displacement, velocity-stepping tests were initiated in which the shear velocity was increased in discrete threefold (half order of magnitude) velocity steps in the range 0.1–30 µm/s. From these tests, we quantify the rate dependence of friction with the parameter a−b:

$$a - b = \frac{{\Delta \mu_{ss} }}{{\ln \left( {V/V_{0} } \right)}}$$
(2)

where ∆µ ss is the change in steady-state coefficient of friction upon an instantaneous change in sliding velocity from V 0 to V (e.g., Marone 1998; Fig. 3). Materials that exhibit positive values of a−b, or velocity-strengthening behavior, are expected to slide stably and therefore would be unable to host earthquake nucleation, and also resist earthquake propagation. Negative values of a−b, or velocity-weakening behavior, are required for frictional instability that results in earthquake nucleation, depending also on the elastic conditions in the fault zone (Scholz 1998, 2002). In the laboratory, a velocity-weakening material sheared in a sufficiently compliant testing apparatus may result in stick–slip-type instability, which is considered analogous to the earthquake cycle (e.g., Brace and Byerlee 1966).

Fig. 3
figure 3

a Example of experimental data, indicating steady-state friction measurement and velocity-stepping tests. Box indicates data shown in b as an example of a frictional response to a velocity step. Data are overlain by an inverse model

Full size image

Equation 2 is a reduced form of phenomenological constitutive law developed by Dieterich (1979, 1981) and is known as rate- and state-dependent friction:

$$\mu = \mu_{0} + a\ln \left( {\frac{V}{{V_{0} }}} \right) + b_{1} \ln \left( {\frac{{V_{0} \varTheta_{1} }}{{D_{c1} }}} \right) + b_{2} \ln \left( {\frac{{V_{0} \varTheta_{2} }}{{D_{c2} }}} \right)$$
(3)
$$\frac{{{\text{d}}\varTheta_{i} }}{{{\text{d}}t}} = 1 - \frac{{V\varTheta_{i} }}{{D_{ci} }},\quad i = 1,2$$
(4)

where a, b 1, and b 2 are empirically derived constants (unitless), Θ 1 and Θ 2 are the state variables which represent the time over which grain-scale asperity contacts are renewed (Dieterich and Kilgore 1994) or necessary to reach a critical granular porosity during shear (e.g., Marone and Kilgore 1993). The critical slip distances D c1 and D c2 are the displacements that correspond with the state variables and represent displacement necessary to re-establish a steady-state shearing condition (Fig. 3b).

We use an iterative least-squares inversion method to extract the rate-dependent friction parameter a−b as well as the individual parameters a, b 1, b 2, D c1, and D c2 (Reinen and Weeks 1993; Saffer and Marone 2003; Ikari et al. 2009). In many cases, the data are well described with only one state variable; in such cases, Eqs. 3 and 4 are simplified by setting b 2 = 0. However, in some cases, employing two state variables (Θ 1 and Θ 2) better describes friction data from velocity step tests compared to a single state variable model. When two state variables are employed, we define b = b 1 + b 2 (thus, a−b = a−Σb). The physical mechanisms which dictate the number of state variables, however, are not well known (e.g., Blanpied et al. 1998).

Results

DFDP-1A mylonite (77.8 m depth)

Results of friction experiments for the mylonite sample show that under nominally dry conditions and a constant normal stress of 50 MPa, friction increases with temperature from µ = 0.57 at 75 °C to µ = 0.80 at 225 °C (Fig. 4). The presence of pore water tends to weaken samples, as evidenced in two cases: (1) at 50 MPa and 75 °C, where µ = 0.57 dry and µ = 0.46 wet, and (2) at 100 MPa, where µ = 0.55, dry at 200 °C and µ = 0.43 wet, at 160 °C. In contrast with the dry samples, increasing temperature in wet samples tends to decrease their frictional strength (Fig. 4). Under fluid-saturated conditions and at room temperature (22 °C), we find that increasing normal stress in the range 50–150 MPa has little effect on friction (µ = 0.60–0.62).

Fig. 4
figure 4

Coefficient of friction as a function of temperature for an ultramylonite sample from DFDPD-1A comparing wet and dry samples at a range of effective normal stresses

Full size image

For the mylonite samples, we observe mostly velocity-strengthening behavior, but also several instances of velocity-weakening friction (Fig. 5; Table 2). For dry mylonite samples at 50 MPa, a−b values are positive at 75 °C (a−b = ~0.001–0.006), but at elevated temperature (150 and 225 °C) velocity weakening (a−b = <−0.003) and instances of stick–slip are observed. Fluid-saturated samples at room temperature show velocity strengthening (a−b = ~0.001–0.005) at 50 and 100 MPa, but velocity neutral to weakening behavior at 150 MPa effective normal stress (a−b = 0 to ~−0.001). Values of a−b are mostly positive for wet and dry samples at the same effective stress and similar temperatures (e.g., wet and dry samples at 50 MPa and 75 °C, and wet and dry samples at 100 MPa and 160–200 °C), but a few instances of velocity weakening are observed for wet samples. No clear dependence of a−b on sliding velocity could be distinguished outside of the data scatter.

Fig. 5
figure 5

Velocity-dependent friction parameter a−b as a function of temperature for an ultramylonite sample from DFDPD-1A comparing wet and dry samples at a range of effective normal stresses

Full size image
Table 2 Velocity-dependent friction parameters for DFDP-1A ultramylonite, 77.8 m
Full size table

DFDP-1A and DFDP-1B core samples at elevated P–T

At 100 MPa, 160 °C and in the presence of pore fluid, samples from Borehole 1A show a large range in strength, and a simple relationship with lithology or sample state (disaggregated vs. intact) is not apparent. The two tested ultramylonite samples exhibit residual µ of 0.43 and 0.73 (Fig. 6). The coefficient of friction of the shallower cataclasite sample is µ = 0.51, whereas the gouge sample near the PSZ exhibited µ = 0.67. The footwall gravel sample was the weakest tested from either borehole, with µ = 0.37. For the DFDP-1B borehole, friction coefficients were more consistent, ranging from µ = 0.44 to 0.49 for the three different lithologies (ultramylonite, proto-cataclasite, granitoid cataclasite). The ultramylonite sample from Borehole 1B was the only sample that exhibited a slight peak, at µ = 0.45 (residual µ = 0.44).

Fig. 6
figure 6

Coefficient of friction with recovery depth for samples of major lithologic units from DFDP Boreholes 1A and 1B. All experiments in this figure were conducted water saturated at 100 MPa and 160 °C

Full size image

For both boreholes, a−b values range between ~−0.001 and 0.002 (Fig. 7; Table 3). All samples exhibited both velocity-weakening and velocity-strengthening behaviors, except the shallower cataclasite sample from Borehole 1A, which was exclusively velocity strengthening. Stick–slip instability was observed for two samples: the footwall gravel in Borehole 1A and the proto-cataclasite in Borehole 1B. In both cases, stick–slip occurred at slip velocities of 1 µm/s or lower. In a similar fashion to coefficient of friction values, a−b does not clearly correlate with variations in lithology, physical sample state, or slip velocity.

Fig. 7
figure 7

Velocity-dependent friction parameter a−b with recovery depth for samples of major lithologic units from DFDP Boreholes 1A and 1B. All experiments in this figure were conducted water saturated at 100 MPa and 160 °C. SS indicates stick–slip behavior, for the gravel in Borehole 1A and the (proto) cataclasites at 106.6 m in Borehole 1B

Full size image
Table 3 Velocity-dependent friction parameters for DFDP-1A and DFDP-1B core samples at 100 MPa and 160 °C
Full size table

Discussion

Observed frictional behavior

In general, our experimental results compare favorably with other recent work using Alpine Fault DFDP core samples. For example, Ikari et al. (2014) measured friction coefficients of 0.56–0.65 for dry ultramylonite samples at \(\sigma_{n}^{{\prime }} = 40{-} 100\,{\text{MPa}}\); which compares favorably with our values of µ = 0.55–0.7 at 50–150 MPa, even though our experiments were conducted at higher temperatures. They also observed that water saturation tends to reduce friction coefficients for samples from Borehole 1A, which we also observe (Fig. 4), and to promote velocity-strengthening behavior. While an increase in a−b for wet samples is not as clear in our ultramylonite dataset, we do observe that only dry samples exhibit stick–slip behavior, while wet samples do not, suggesting a stabilizing effect of pore fluid. Using disaggregated samples of the PSZ from Borehole 1A, Boulton et al. (2014) measured a friction coefficient of 0.62 at 94 MPa and 210 °C, which matches our measured value of 0.67 at 100 MPa and 160 °C for the gouge sample adjacent to the PSZ. They also reported strictly positive a−b values for the 1A-PSZ at lower pressure and temperature (<62 MPa, ≤140 °C) and velocity weakening at higher pressure and temperature (94 MPa, 210 °C). This is also consistent with our measurements, considering that we observe both velocity strengthening and velocity weakening at conditions within this range.

The strength of the gouge sample near the PSZ of µ = 0.67 is notably high in comparison with drilling samples from other major fault zones tested at comparable effective normal stresses, for example the Nankai Trough offshore Japan (µ < 0.4 at 80–120 MPa, Saffer et al. 2012) and the San Andreas Fault near Parkfield, California (µ = 0.1–0.15 at 40–200 MPa, Lockner et al. 2011; Carpenter et al. 2012). These studies also reported only velocity-strengthening behavior; however, their experiments were conducted at room temperature. It has been demonstrated that elevated temperatures can cause both an increase in friction (to µ ≥ 0.6) and velocity-weakening behavior to occur in materials that are velocity strengthening at room temperature, for both clay-rich and clay-poor samples (e.g., Blanpied et al. 1998; den Hartog and Spiers 2013). We suggest that this is the case for the Alpine Fault, where lower friction (µ = 0.45) and velocity-strengthening behavior has been reported at low temperatures (Ikari et al. 2014) but higher friction and velocity-weakening behavior is common at higher pressure and temperature (Boulton et al. 2014).

We do not observe a significant difference in frictional properties between the three samples tested as intact cylinders compared to disaggregated samples (Figs. 6, 7) which is somewhat surprising considering that rock fabric can have a significant effect depending on the intensity of the foliation and whether phyllosilicate minerals are present (e.g., Collettini et al. 2009; Ikari et al. 2011). For our samples, the predominant rock fabric dips 35°–65° based on visual core description, consistent with earlier field observations indicating that the foliation of Alpine Fault-related rocks dips ~40°–50° (Sibson et al. 1981). Therefore, we suggest that the fabric is unfavorably oriented for weakening with respect to the shear plane (which is perpendicular to the core axis in our experiments) and thus does not play role in these experiments. Furthermore, fabric and cementation that may have developed in situ at shallow depths may be lost when the samples are subjected to higher pressure and temperature.

Implications for the slip behavior of the Alpine Fault

The conditions we used for our selection of core samples, 100 MPa and 160 °C, correspond to a depth range of ~4–7 km, assuming an effective stress gradient of 15–20 MPa/km and a temperature gradient of ~23–40 °C/km. However, larger temperature gradients on the Alpine Fault may also be appropriate due to rapid uplift (e.g., Koons 1987), consistent with borehole temperature measurements indicating a gradient of ~60 °C/km, at least near the surface (Sutherland et al. 2012). In this case, the shallow end of our applicable depth range may be 2.5–3 km. Therefore, our observations of partially velocity-weakening behavior for all major Alpine Fault lithologies suggest that, at in situ conditions representative of a few km depth, unstable slip should be expected. This is consistent with earthquake hypocenters which are distributed from 8 to 12 km up to the near surface on the Alpine Fault (Leitner et al. 2001) and GPS measurements which indicate that the central Alpine Fault is not creeping and the upper 5–8 km is likely locked and accumulating strain (Beavan et al. 1999). This evidence suggests that the Alpine Fault is frictionally unstable throughout the shallow crust, a unique characteristic that distinguishes the Alpine Fault from other plate boundary fault zones which typically exhibit a creeping “aseismic” zone in the shallowest few km (e.g., Byrne et al. 1988; Marone and Scholz 1988; Hyndman et al. 1997). Rapid exhumation of high-grade metamorphic fault rock is an attractive explanation (Koons 1987; Little et al. 2005).

The relatively high strength of the Alpine Fault, based on our sample near the PSZ combined with other high pressure–temperature data using PSZ material (Boulton et al. 2014), suggests that high resolved shear stresses may accumulate on the fault before failure. Because we also observe friction coefficients as low as µ = 0.43 (excluding the gravel, which is not expected to be present at greater depth), it is possible that the Alpine Fault may fail in other lithologic units. However, our observations of velocity weakening in most of our samples suggest that earthquake nucleation is also possible in each of the major lithological units. Earthquake nucleation requires not only velocity-weakening frictional behavior but also sufficient compliance in the wall rocks surrounding the fault zone (Scholz 1998, 2002). This is formulated as the critical stiffness criterion, in which the stiffness of the fault surroundings K (stress/length) must be lower than a critical value K c defined by frictional properties and the effective normal stress:

$$K < K_{c} = \frac{{ - (a - b)\,\sigma_{n}^{{\prime }} }}{{D_{c} }}$$
(5)

From our experimental data, we may calculate the critical value K c necessary for slip instability. If we consider our gouge sample from the DFDP-1A borehole near the PSZ to be most appropriate for the fault zone, appropriate minimum values are a−b = −0.001 and D c  = 10 µm (considering D c1, see Table 3). For an effective normal stress of 100 MPa, K c is then approximately 0.01 MPa/µm, or 10 GPa/m. This value, in conjunction with the Young’s modulus E and Poisson’s Ratio ν, define a minimum patch length 2L c :

$$2L_{c} = \frac{E}{{(1 - \nu^{2} )\,K_{c} }}$$
(6)

(where L c is the patch half-length) for the nucleation of unstable slip, whereas slip nucleating on patches smaller than 2L c will be stable (Scholz 1998, 2002).

We use laboratory measurements of elastic properties on DFDP-1A and DFDP-1B core samples (Carpenter et al. 2014) of E = 30 GPa and ν = 0.3 for hanging wall cataclasite samples at isotropic confining pressures of 63.5 MPa, which results in a critical patch length 2L c  = 3.3 m. This is a very small value which corresponds to a seismic moment M o  = ~4 × 1010 Nm based on the scaling relation M o  = 109(2L c )3 (Ohnaka 2000), or an earthquake of moment magnitude ~1. If values of E = 16 GPa and ν = 0.37 are used, which were obtained for a cataclasite very near the PSZ at ~90 m depth in Borehole 1A, the calculated value of 2L c decreases further to ~2 m. The small critical nucleation patch size is significant because it suggests that unstable slip on the Alpine Fault is easily achieved, without requiring a long phase of quasi-stable slip during nucleation.

It also suggests that any slip event on the fault is likely to be an earthquake, rather than occurring as various types of discrete slow or transient slip events (Ide et al. 2007; Peng and Gomberg 2010). Ide et al. (2007) showed that the seismic moment (and moment magnitude) for slow earthquakes of all forms scales consistently with their characteristic duration in a manner that clearly distinguishes them from ordinary earthquakes. This was demonstrated for slow events with M o ranging from ~1011 to 1021 Nm and M w from ~1 to 8. However, they also show that the relations for slow and normal earthquakes converge at the low end of the scale, which coincides with the M o and M w calculated for a minimum nucleation patch length on the Alpine Fault. Because our frictional and elastic property estimates for the Alpine Fault indicate that any slip event with M o  > ~4 × 1010 and M w  > ~1 are likely to be unstable, we suggest that slow slip events should not be expected in the depth range of ~3–7 km and that any nucleating slip event will result in an ordinary earthquake. Slow fault slip has also been recently observed on the Alpine Fault in the form of tectonic tremor (Wech et al. 2012) and low-frequency earthquakes (Chamberlain et al. 2014), but these events are located deeper than 20–25 km, well below the inferred lower limit of the seismogenic zone on the Alpine Fault.

Conclusions

Based on laboratory shearing experiments using a sample of ultramylonite from borehole DFDP-1A, we show that the friction coefficient µ and the velocity dependence of friction, a−b, can vary substantially depending on the presence of pore fluid, effective normal stress, and temperature. Under conditions appropriate for several km depth on the Alpine Fault (100 MPa, 160 °C, fluid-saturated), the friction coefficient µ varies between 0.43 and 0.73, with a gouge sample located very near to the principal slip zone (PSZ) exhibiting µ = 0.67. This relatively high value suggests the capacity for large shear stresses at depth and that slip may also occur in other lithological units, e.g., ultramylonite or cataclasite, where the friction is lower. Observations of velocity-weakening friction in all lithologies tested from both boreholes suggest that earthquake nucleation is possible regardless of where the slip nucleates. Calculations using experimentally obtained values of a−b and D c , combined with known values of the Young’s modulus and Poisson’s ratio, indicate a minimum critical nucleation patch size of ~3 m. This minimum value suggests that slip events with a seismic moment M o of at least ~4 × 1010 or an M w of at least ~1, should be expected to be unstable and thus result in earthquakes, not slow slip events.