Bulletin of the Seismological Society of America, Vol. 104, No. 4, pp. 1964–1975, August 2014, doi: 10.1785/0120130233
Ⓔ
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric along the Denali Fault, Northern Canadian Cordillera by N. Rasendra, M. Bonnin, S. Mazzotti,* and C. Tiberi
Abstract
We analyze crustal and upper-mantle structure near the Denali fault (northern Canadian Cordillera) using 11 broadband seismic stations. Receiver functions at five stations within 5–30 km of the fault trace display a strong P-to-S conversion within the midcrust (10–25 km depth) that systematically varies with back azimuth. Stacking and velocity inversion along complementary northwest–southeast and northeast–southwest back-azimuth ranges yield a strong anisotropy (> 10%) in the midcrust, with complex crustal anisotropic behavior (at least two layers) close to the Denali and Duke River faults junction. Anisotropy occurs in a low-velocity zone with the fast axis parallel to the Denali fault trend. Three additional stations close to the Denali fault (15–30 km), as well as two stations further away (> 100 km) show no significant crustal anisotropy. Shear-wave (SKS) splitting analysis indicates similar upper-mantle anisotropy, with a fast axis parallel to the trend of the Denali fault for all stations, except for station WHY located 125 km away. We relate this crustal and upper-mantle anisotropy to fossilized structural and mineral fabrics due to Eocene transpression on the Denali fault. These results suggest the main transpression phase (∼400 km displacement) was accommodated in a shear zone at least 50–60 km (less than 125 km) wide in the midcrust and upper mantle. Lack of clear anisotropy in the lowermost crust may relate to complex deformation within a detachment layer (orogenic float model). Online Material: Table of SKS splitting information and figures of receiver function analysis.
Introduction Strike-slip plate-boundary faults, such as the San Andreas and Alpine faults, are associated with sharp, localized traces in the upper crust, whereas their geometry and width in the deeper portion of the lithosphere is poorly known. Exposed shear zones exhumed from mid to lower crust levels exhibit widths ranging from a few kilometers to a few tens of kilometers (e.g., West and Hubbard, 2003; Vauchez et al., 2012). Similarly, geophysical evidences point to mid to lower crust deformation zones as wide as 10–50 km below active plate boundaries (Wilson et al., 2004; Wech et al., 2012), although more focused (1–5 km wide) deformation is also proposed on the basis of deep tremor activity (Shelly and Hardebeck, 2010). In the lithospheric mantle, shear-wave splitting and other anisotropy indicators suggest plate-boundary shear zones as large as 200–300 km (e.g., Baldock and Stern, 2005).
*Also at the School of Earth and Ocean Sciences, University of Victoria, 3800 Finnerty Road, Victoria, British Columbia, Canada V8P 5C2.
Using simple mechanical models of ductile shear zones, Platt and Behr (2011) propose that the localized shear zone width depends strongly on temperature, rock composition, hydration, and creep mechanism and can vary from a few meters to over 100 km. Within such simple shear systems, ductile strain localization results in strong vertical foliation and horizontal lineation, and thus in significant anisotropy (Tommasi et al., 1999). In context, lattice-preferred orientation of olivine crystals produces directions of polarization of fast shear waves at low angle to fault trends, commonly associated with seismic anisotropy in the upper mantle (e.g., Lavé et al., 1996; Klosko et al., 1999; Rümpker et al., 2003; Bonnin et al., 2010). In contrast, evidence for mechanical and seismic anisotropy in the crust is sparse, limited mostly to a few receiver function (RF) studies that identify strong anisotropy coincident with major fault trends (e.g., Wilson et al., 2004, on the Marlborough fault zone). In this study, we analyze data from an array of 11 broadband seismic stations located near the Denali fault in
1964
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric
Figure 1. Topography (from 1 arcmin model ETOPO1) and general geodynamic scheme for the study region. Triangles indicate seismic stations used in the present study. The main fault systems are indicated with solid lines. southwest Yukon, Canada, to determine the crustal and upper-mantle structure associated with this large continental strike-slip fault. Present-day tectonic activity along the Denali fault in this region is low (Mazzotti et al., 2012), in contrast with the ∼400 km cumulative motion accommodated in the early Eocene. This provides the opportunity to study inherited tectonic structures and how they affect the lithosphere seismic signature.
Geodynamic Settings Southwest Yukon mostly corresponds to a series of allochthonous terranes accreted to the North America margin, associated with large volumes of syntectonic magmatic and metamorphic sedimentary suites (Gabrielse et al., 1991). In the study area (Fig. 1), the Denali fault marks the limit between two superterranes: to the northeast, the Intermontane superterrane consists of a pericratonic assemblage of the late Paleozoic–mid-Mesozoic continental, oceanic arc, and accretionary complex; to the southwest, the Insular superterrane consists primarily of a distal, Paleozoic oceanic arc assemblage. Accretion of these various terranes between themselves and to the North America margin occurred during several phases of subduction/collision from the early Meso-
1965
zoic to the early Cenozoic and remains a debated issue (e.g., Johnston and Canil, 2007). During, or slightly after, the last phase of early Cenozoic accretion, large-scale dextral strike slip and transpression affected the entire North American Cordillera (Gabrielse et al., 2006). During this episode, nearly 1000 km of total dextral displacement was distributed among several structures, the largest of which, the Tintina and Denali faults (Fig. 1), each accommodated roughly 400 km of dextral motion. Most of this displacement occurred during the early Cenozoic, possibly slightly later for the Denali fault. Local geology underneath each seismic station is obviously more complex than the first-order picture. Along the Denali fault corridor, geological structures comprise a large variety of rocks from lightly metamorphic flysh and turbidites, to highly metamorphic schists, granitoid plutons, mafic lavas, and ophiolites. Numerous faults and shear zones, mostly associated with the late Mesozoic–early Cenozoic accretion and transpression phases, cut through or limit these various formations. Among those, the Duke River fault (Fig. 1) accommodated thrust motion during the mid-Cretaceous and as early as the Pliocene (Cobbett et al., 2009). Recent to present-day tectonics is associated with the collision of the Yakutat Terrane, in the corner of the Gulf of Alaska over the last 5–10 Ma (e.g., Plafker et al., 1994). The oblique convergence is fully partitioned between strike-slip motion on the Fairweather fault and distributed shortening, as evidenced by earthquake focal mechanisms and Global Positioning System data (Leonard et al., 2007; Elliott et al., 2010; Mazzotti et al., 2012). Current slip rate on the Yukon segment of the Denali fault shows a strong gradient, from ∼8 mm=yr near the Alaska border (Matmon et al., 2006) to 0–2 mm=yr at the southern end (Mazzotti et al., 2012).
Receiver Function Study Data Processing and Analysis We select teleseismic earthquakes recorded from September 2010 to August 2012 at 11 stations of the Canadian National Seismograph Network (see Data and Resources) deployed on both side of the Denali fault (Fig. 1). Five are part of the permanent network (BVCY, DAWY, HYT, PLBC, and WHY); the other six were installed for a temporary seismic experiment specifically targeting the Denali fault in southwest Yukon (YUK1–7). We select 971 earthquakes between 25° and 90° of epicentral distance with a magnitude greater than 5.5 to ensure good signal-to-noise ratio. The spatial distribution of these events shows a gap in the backazimuthal ranges N020°–100°E and N150°–200°E (Fig. 2). We first rotate the three-component seismograms from the geographical system into vertical, radial, and transverse system. We then band-pass filter the signal between 0.3 and 3 s and window the traces 5 s before and 40 s after the first P-wave theoretical arrival time (defined as 0 s time).
1966
N. Rasendra, M. Bonnin, S. Mazzotti, and C. Tiberi 0°
40°
320°
80°
280°
30° 60° 240°
120°
90° 120° 150° 160°
200° 180°
Figure 2. Azimuthal distribution of teleseismic events used for the receiver function (RF) study, with the projection centered on the Yukon seismic network.
RFs are time series composed of P-to-S converted waves at interfaces beneath the seismometer. They show the relative response of the Earth structure near the receiver. They are thus powerful tools to image the layered structure of the medium beneath seismic stations and have been widely used for the last three decades (e.g., Langston, 1977). RFs are obtained by deconvolving the vertical seismic trace from the radial component (e.g., Langston, 1979). In this study, we use the iterative time-domain deconvolution by Ligorria and Ammon (1999). A Gaussian filter of 2 s is used, and we select only the RFs showing a deconvolution variance reduction of more than 90%. The resulting RFs exhibit clear Ps Moho conversions, in most cases around 5 s, with identified first-multiple phases around 14 s (Fig. 3). Although Ps Moho conversions are easily identified, complex intracrustal signals (Px) are clearly visible before 5 s for all the stations. To ease the discussion, we divide the stations into three groups (Fig. 4). For BVCY and YUK1 stations (group 1), a single strong negative peak around 2–3 s is present for events coming from the west (200°–320° N). The transverse RFs also exhibit a strong back-azimuth-dependent signal (Fig. 3 and Ⓔ Fig. S1 in the electronic supplement to this paper). For YUK3, YUK6, and
BVCY - Radial
BVCY - Transverse 4
350
4
350
4
4 3
300
3
300
5
5
9 1 15
200
Px
150
8 3
100
1
50
22
Backazimuth (degree)
Backazimuth (degree)
22 250
9
250
1 15
200
Px
150
8 3
100
1
50
3
0
3
0 0
2
4
6
8 10 12 14 16 18 20 22 24 26 28 30
0
2
4
Times after P arrival (s)
Ps Figure 3.
6
8 10 12 14 16 18 20 22 24 26 28 30
Times after P arrival (s)
Ps
BVCY station RFs: (left) radial RFs and (right) transverse RFs. The RFs are organized by back azimuth and stacked every 20°. The small numbers at the end of the stacked signal denote the number of stacked events. The main P-to-S conversion on the Moho (Ps) is indicated by the dotted line, and the intracrustal conversion phase Px is also indicated.
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric Radial Stacked 1 RF (Baz 200°–320°)
Radial Stacked 2 RF (Baz 0°–200° and 320°–360°)
BVCY YUK1
G1
YUK3 YUK7 YUK6
G2
DAWY HYT PLBC WHY YUK2 YUK5
–5
1967
G3
0
5
10
15
20
25
30
time (s)
–5
0
5
10
15
20
25
30
time (s)
Figure 4. (Left) The stacked radial component for the 11 stations of the study for back azimuths (200–320) and (right) the complementary ones (0–200, 320–360). The stations are gathered into three groups identified within the text (G1, G2, and G3). YUK7 (group 2), the intracrustal signal on the RFs is more complex, with multiple negative peaks between 2 and 4 s (Fig. 4 and Ⓔ Figs. S2–S4). The Ps Moho signature is less pronounced, and the amplitude of the negative peaks varies with back azimuth on the radial component (Fig. 4). For the other stations (YUK2, YUK5, HYT, DAWY, WHY, PLBC, all comprising group 3), the negative peak before 5 s is either absent or very weak and within the RFs noise level (Ⓔ Figs. S5–S10). Back-azimuth variations on both the radial and transverse components associated with a negative P-to-S conversion on radial RFs can result from two possible sources: a dipping interface with negative velocity contrast or a layer with velocity anisotropy (e.g., Jones and Phinney, 1998). In theory, the periodicity of back-azimuth variations in radial and transverse signals can be used to discriminate between these two sources (2π and π periodicity, respectively). This simple test cannot be used for our data due to incomplete back-azimuth coverage and structural complexity that can lead to more complex periodicity (Ozacar and Zandt, 2009). On the basis of the regional tectonic and geodynamic setting, we consider the presence of a systematic, regional-scale intracrustal dipping interface with negative velocity contrast to be unlikely. Thus, we favor the hypothesis of an intracrustal anisotropic layer. Given the lack of back-azimuthal coverage and the complexity of the signal on all stations (Fig. 4 and Ⓔ Figs. S2– S10), inversions to retrieve the full pattern of anisotropy (number of layers, dipping anisotropy, exact azimuth) would have been totally unconstrained. We then focus our analysis on first-order crustal velocity patterns using a stochastic inversion for only two back-azimuthal ranges: a range associated with RFs with strong negative Px phases (representative of a slow crustal velocity) and the complementary range for RFs without negative Px (rapid or normal crustal velocity).
Velocity Structure Inversion We use a neighborhood algorithm (Sambridge, 1999) together with Shibutani et al. (1996) RF modeling. This method calculates the radial component of the RFs from statistically representative bunches of S-wave velocity models (classically tens of thousands) and converges to a solution (or a group of solutions) that explains the whole waveform of the radial RF. The method works on a stacked RF (i.e., no back-azimuthal analysis). We invert only the radial component of the RFs. High amplitudes observed on the transverse component clearly show the presence of energy out of the source-receiver plane (e.g., Fig. 3 and Ⓔ Fig. S2); however, due to the noise level and complexity of the medium, the resulting transverse signal is not coherent enough to be correctly inverted. Our main goal is to characterize crustal velocity structures, particularly anisotropy. Thus, we focus the inversion on the first 5–6 s of the RFs (roughly first 40–60 km depth) by constructing velocity models with six layers, based on previous active and passive seismic analysis in the region and in similar terranes in Alaska and northern British Columbia. For each layer, we specify a thickness range, an S-wave velocity range for its top and bottom, and a range of V P =V S ratio. The first layer represents a potential sedimentary cover (low V S, high V P =V S ). We separate the crust into upper, middle, and lower parts (layers 2, 3, and 4, respectively). We introduce a thin intermediate velocity layer (layer 5) between the lower crust and the mantle, as proposed by Beaudoin et al. (1992) and Lowe and Cassidy (1995). Tests without this intermediate layer return an unrealistic P-wave velocity as low as 7:0 km=s for the whole mantle. Late (post ∼6 s) RF peaks would require more complex models including second-order interfaces that would be too speculative given the data signal-to-noise ratio.
1968
N. Rasendra, M. Bonnin, S. Mazzotti, and C. Tiberi B V CY S t ac k ed R F [ 20 0 – 3 2 0]
BVCY Stacked RF [–40–200]
predicted observed
Amplitude
Amplitude
predicted observed
Misfit = 1.520 –5.0
0.0
5.0
10.0
15.0
20.0
25.0
Misfit = 0.718 –5.0
0.0
5.0
10.0
S VELOCITY (km/s) 1.0
2.0
3.0
4.0
0.0
1.0
2.0
3.0
4.0
5.0
20.0
DEPTH (km)
DEPTH (km)
0.0
10.0
20.0
30.0
40.0
50.0
30.0
40.0
50.0
Number of the best 1000 models 1
70.0 1.5
25.0
S VELOCITY (km/s) 5.0
10.0
60.0
20.0
Time (s)
Time (s)
0.0 0.0
15.0
10
2.0
100
60.0
1000
2.5
Number of the best 1000 models 1
3.0
Vp /Vs RATIO
70.0 1.5
10
2.0
100
1000
2.5
3.0
Vp /Vs RATIO
Figure 5. Results from the stochastic inversion for the two stacks of BVCY: (left) forevents between 200° and 320° azimuth and (right) for the other azimuths. The upper panels represents the observed (solid) and the model predicted (dotted) RFs. The bottom panels show velocity models obtained from the inversion. The average model is represented by the thick white line, the best model by the thick dotted line, the best 1000 models are indicated by the gray scale, and all the explored models are indicated by the gray lines on the background. Both V S and V P =V S values are indicated with the horizontal upper and lower axes, respectively. The color version of this figure is available only in the electronic edition. For each station, the stochastic inversion is performed for two sets of RFs based on events stacked according to their back azimuth: southwest–northwest (200°–320° N) and the complementary eastern quadrant (320°–200° N, Fig. 5). For each set, 10,000 velocity models are produced, for which the corresponding synthetic RFs and misfit (chi-square function; Sambridge, 1999) to the observed RFs are estimated (Fig. 5).
SKS Splitting Analysis To complement the crustal structure investigation, we perform a shear-wave splitting analysis on the same stations
using teleseismic core-refracted shear waves (SKS hereafter). This classical technique is used to detect and characterize the presence of seismic anisotropy in the upper mantle (Silver, 1996; Savage, 1999). In order to observe distinct high signalto-noise ratio SKS phases, we systematically select events with magnitude M w > 6 occurring at epicentral distances in the range 85°–120° (60–800 events at each station). Event origin times and locations are from the National Earthquake Information Center–Preliminary Determination of Epicenter (NEIC-PDE) catalog (U.S. Geological Survey). We quantify anisotropy through two parameters: (1) the delay time δt between the two split waves that depends on
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric
(a)
(b)
65°
64°
1969
65°
64°
63°
63°
62°
62°
61°
61°
60°
60°
1s 59° 7° 217° 218° 219° 220° 221° 222° 223° 224° 225° 226° 22
59° 7° 217° 218° 219° 220° 221° 222° 223° 224° 225° 226° 22
Figure 6. (a) Map presenting individual splitting parameters at station location. Directions of the bars are parallel to the fast directions of polarization of the considered events; lengths of the bars are proportional to the delay times. The black circle without radiating lines is station YUK6, where no splitting measurement has been obtained. (b) Map presenting back azimuth of the null events at the station location. Solid lines indicate the main fault systems. The color version of this figure is available only in the electronic edition. the thickness and on the intrinsic anisotropy of the medium, and (2) the azimuth of the fast split wave polarization ϕ, which is related to the orientation of pervasive fabric in the anisotropic structure (foliation and lineation; Nicolas and Christensen, 1987) or to fluid-filled microcracks at upper crustal levels (Crampin, 1984). Data are analyzed using Silver and Chan (1991) minimum energy method, which looks for the (φ, δt) pair that minimizes the energy on the transverse component of the seismogram. We ascribe a quality factor for each measurement (good, fair, or poor) based on several parameters (e.g., signal-to-noise ratio of the initial waveform, correlation between the fast and slow shear wave; see Bonnin et al., 2010, for details). Filtering is manually applied depending on characteristics of each seismogram in order to keep the largest possible amount of signal. Good and fair individual splitting measurements are presented in Figure 6a. Poor measurements are discarded. Null measurements, that is, SKS waves determined as isotropic, are kept to check the relevance of the set of individual splitting parameters measured at each station. Nulls can be caused by (i) an isotropic medium beneath the station, (ii) the presence of pairs of layers of anisotropy with orthogonal fast directions or, more likely, (iii) by shear waves propagating along the fast/slow axis of anisotropy of the medium (apparent isotropy). Back azimuth of the null events are shown on Figure 6b. Counts of the splitting and null measurements for each station are compiled in Ⓔ Table S1.
The characteristics of the good and fair (null and non-nulls) individual measurements are indicated in Ⓔ Table S2.
Results Crustal Structure Models obtained from the inversion of RFs stack 1 (200°–320° N back azimuths) and from stack 2 (320°–200° N back azimuths) are presented in Tables 1 and 2, respectively. Crustal thickness varies between 35 and 40 km at the 11 stations of our network. Stations north and east of the Denali fault are associated with a crustal thickness of ∼36 km, slightly thinner than that of stations south and west of the fault (∼39–40 km). This small difference correlates with the long-wavelength topography (Fig. 1): average elevation of the plateau to the northeast of the Denali fault is 800– 1200 m, whereas the southwest side corresponds to the eastern front of the St. Elias Mountains with mean elevations of 1500–2000 m. The additional 3–4 km of crust fits with the expected crustal root associated with additional 500–1000 m of topography at isostatic equilibrium. First-order crustal structure is also similar at all stations, with an upper–midcrust interface at ∼10–15 km depth and a mid–lower crust interface at ∼25 km depth. Crustal V P =V S ratios are less than 1.75, coherent with active seismic studies in similar terranes in central Alaska (Beaudoin et al, 1992).
1970
N. Rasendra, M. Bonnin, S. Mazzotti, and C. Tiberi
Table 1
Table 2
Results from Inversion for Receiver Function Stack 1 (200°–320° Azimuth)
Results from Inversion of Receiver Function Stack 2
Station
BVCY
Layer*
sed u.c. m.c. l.c. Inter. m. YUK1 sed u.c. m.c. l.c. Inter. m. YUK3 sed u.c. m.c. l.c. Inter. m. YUK7 sed u.c. m.c. l.c. Inter. m. DAWY sed u.c. m.c. l.c. Inter. m. HYT sed u.c. m.c. l.c. Inter. m. YUK2 sed u.c. m.c. l.c. Inter. m. YUK5 sed u.c. m.c. l.c. Inter. m.
Thickness (km)
Depth (km)
V S Top (km=s)
V S Bottom (km=s)
V P =V S
1 5.5 18.6 14.8 23.2 8.4 0.2 8.9 11.4 15.9 11.6 14.5 0.2 5.3 14 18.8 11.7 14.4 0.4 17 10.2 10.8 10.1 21.5 0.1 15 12.5 7.2 8.1 21.1 1.7 9.9 9.5 14.9 9.7 7.2 3.4 13.2 13.6 11.3 10.2 19 3.7 11.6 10.6 11.1 9 17.7
1 6.5 25.0 39.8 63.0 71.3 0.2 9.1 20.5 36.3 47.9 62.4 0.2 5.5 19.5 38.3 50.0 64.4 0.4 17.4 27.6 38.4 48.6 70.1 0.1 15.1 27.6 34.8 42.9 63.9 1.7 11.7 21.2 36.1 45.8 53.1 3.4 16.6 30.2 41.5 51.7 70.7 3.7 15.3 25.9 37.0 45.9 63.7
1.5 3.41 2.97 3.48 4.09 4.54 1.5 2.95 3.86 3.22 4.02 4.5 1.36 3.15 3.53 3.17 4.05 4.64 1.35 3.29 3.79 3.97 4.29 4.51 1.36 3.23 3.75 3.72 4.26 4.54 2.95 3.41 3.67 3.36 4.23 4.53 3.44 3.28 3.56 3.65 4.06 4.67 3.43 3.45 2.81 3.86 4.31 4.52
1.4 3.01 2.98 3.96 4.73 4.55 1.1 3.61 2.96 3.81 4.23 4.61 0.9 3 3.16 3.88 4.34 4.63 1.21 3.7 2.98 3.92 4.63 4.61 0.52 3.69 3.61 3.86 4.31 4.52 3 3.58 3.56 3.98 4.28 4.62 2.54 3.83 3.81 3.92 4.33 4.59 2.83 3.31 3.66 3.91 4.55 4.55
2.1 1.65 1.8 1.72 1.74 1.79 2.53 1.78 1.68 1.79 1.88 1.86 2.29 1.69 1.78 1.76 1.88 1.75 2.01 1.78 1.77 1.68 1.79 1.79 2.47 1.71 1.70 1.74 1.75 1.79 1.72 1.74 1.73 1.79 1.86 1.82 1.74 1.75 1.67 1.79 1.86 1.81 1.65 1.80 1.72 1.78 1.87 1.81
*Layers: sed, sedimentary layer; u.c., upper crust; m.c., middle crust; l.c., lower crust; inter., intermediate; and m., mantle.
These values are typical of either a felsic crust or a crust with few fluid intrusions. The intermediate layer (layer 5) at the base of the lower crust is ∼7–15 km thick, with high V P =V S ratio (∼1:8). The P-wave velocity at the top of this layer is on average lower than 8:0 km=s and as low as 7:1 km=s for BVCY and YUK1. The uppermost part of the mantle in the region thus seems to
Station
Layer*
Thickness (km)
Depth (km)
V S Top (km=s)
V S Bottom (km=s)
V P =V S
BVCY
sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m. sed u.c. m.c. l.c. Inter. m.
0 7.8 17.9 9.7 21.8 15 0.2 9.2 14.4 11.9 10.8 24.6 0.1 9.9 12.8 12.8 24.6 21.5 0.4 10.4 12.7 11.4 1.5 9.9 0.1 14.2 10.6 10 6.4 23.3 0.4 17.2 5.5 12.5 2.1 6.7 0 8.7 13.8 18.3 9.1 23.9 4.3 9.5 13.6 5.7 8.9 7.1
0 7.8 25.7 35.4 57.2 72.1 0.2 9.3 23.7 35.6 46.4 71.0 0.1 10.0 22.8 35.6 60.2 81.7 0.4 10.8 23.4 34.8 36.4 46.3 0.1 14.3 24.9 34.9 41.4 64.7 0.4 17.6 23.1 35.6 37.7 44.4 0 8.7 22.5 40.9 50.0 73.9 4.3 13.8 27.3 33.0 41.9 49.0
0.95 2.98 3.68 3.75 4.04 4.51 1.1 3.26 3.79 3.36 4.04 4.65 1.34 3.1 3.09 3.65 4.14 4.63 1.42 3.06 2.98 3.83 4.37 4.54 1.23 3.68 3.64 3.69 4.12 4.50 1.37 3.49 3.34 3.43 4.47 4.53 0 3.09 3.00 3.58 4.15 4.51 3.24 3.73 3.23 3.78 4.02 4.55
1.24 3.46 3.84 3.89 4.56 4.55 0.61 3.45 3.08 3.98 4.82 4.66 0.62 2.76 3.68 3.88 4.44 4.63 0.68 3.27 3.39 3.99 4.56 4.57 1.17 3.49 3.89 3.84 4.76 4.57 2.71 3.30 3.64 3.89 4.61 4.52 0 3.19 3.40 3.80 4.63 4.62 3.05 3.32 3.75 3.82 4.40 4.58
2.01 1.71 1.74 1.73 1.82 1.83 2.91 1.71 1.73 1.71 1.75 1.79 2.2 1.69 1.65 1.75 1.78 1.76 2.56 1.69 1.76 1.77 1.83 1.85 2.37 1.71 1.66 1.77 1.72 1.79 1.79 1.70 1.67 1.79 1.81 1.83 0 1.66 1.74 1.74 1.74 1.78 1.69 1.79 1.72 1.80 1.83 1.76
YUK1
YUK3
YUK7
DAWY
HYT
YUK2
YUK5
*Layers: sed, sedimentary layer; u.c., upper crust; m.c., middle crust; l.c., lower crust; inter., intermediate mantle layer; and m., mantle.
have a lower velocity than classically observed and exhibits a velocity gradient in its first 10 km. This is consistent with low Pn velocities (7:8–7:9 km=s; Levshin et al., 2001) and a lower-crust low P velocity from active seismic data (Beaudoin et al, 1992) observed for similar geological terranes in central Alaska. This layer may be related to the particularly high heat flow and lower crust temperatures observed in the region (> 86 mW=m2 ; Lewis et al., 2003).
–30 –50
–40
Depth (km)
–30 –40 –70
–70
–60
–60
–50
Depth (km)
–20
–20
–10
–10
0
0
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric
2.5
3.5
4.5
VS (km/s)
–0.2
0.0
0.2
(VS1 - VS2)/VS1
Figure 7. Velocity and anisotropic profiles for BVCY station. Left: mean velocity models over the best 1000 ones obtained by the RF inversion. Velocity for the (200–320) azimuth stack events are represented with thick black line, and the other azimuths with dotted gray line. Right: depth-dependent median anisotropy curve with associated standard error (dotted lines). The anisotropy is defined as V S1 –V S2 =V S for the best 1000 velocity models, and the thick black line represent the median of all anisotropic curves. Group 1 stations (BVCY, YUK1) present significantly different velocity models depending on back azimuths (Fig. 5 and Ⓔ Fig. S11), with middle crustal V S lower for western than for eastern events (Fig. 5 and Ⓔ Fig. S11). To quantify this effect, we compute the anisotropy between the two profiles: V S1 − V S2 =V S1 , in which V S1 and V S2 are velocities of the 1000 best models for the western and eastern profile, respectively (Fig. 7 for BVCY). We observe ∼15% of anisotropy localized in the third layer, between about 10 and 25 km depth, and very small to no anisotropy at other depths. For YUK1, the anisotropy is more widely distributed over the mid to lower crust and reaches ∼10%, although it remains below the associated standard error (Ⓔ Fig. S12). In this latter case, the amplitude of the negative peak in the RFs is not fully retrieved by the inversion, and the anisotropy may be both stronger and more localized in the middle crust. For those stations, the low-velocity direction corresponds to events with back azimuth perpendicular to the Denali fault trend. For group 2 stations (YUK3, YUK6, YUK7), which show a highly complex pattern in the radial RFs, inversions result in relatively poorly fitted RF signal (Figs. S13 and S14). The complexity of the RFs could only be modeled through more detailed velocity structures (more layers); however, in that case, the number of unknowns far exceeds the information contained within the data, and the results will not be significant (strongly underdetermined inverse prob-
1971
lem). Group 2 stations are located 5–25 km southwest of the Denali fault within the array of associated faults related to the Insular superterrane accretion (e.g., YUK6 near the Duke River/Denali faults junction). Strong lateral variations in crustal structures (both in terms of interfaces and anisotropy) likely result in the complex RF signal observed at the three stations (Ⓔ Figs. S2–S4). Given the strong azimuth dependency in the radial signal for the two negative peaks (Ⓔ Fig. S3), we expect anisotropy to be present and localized in two different layers with potentially different axes. We estimate the slow axis to correspond to the maximum amplitude of negative Px, with orientations about 200° N and 250° N. With the result of the inversion, we can estimate anisotropic profiles for two of the stations (YUK3 and YUK7; Ⓔ Figs. S15 and S16), but it remains within the uncertainty of our analysis. Inversions of the group 3 stations do not reveal any significant differences within the velocity models (Ⓔ Figs. S17–S20). Back-azimuthal variations in RF signal are not significant enough to be clearly discriminated within velocity models. Averages based on the 1000 best models are very close to each other for the two stacks, indicating that crustal anisotropy beneath those stations is either absent or below the resolution of our analysis (Ⓔ Figs. S21–S24). Upper-Mantle Anisotropy Individual fast directions of polarization for stations close to (< 50 km) the Denali or Tintina faults are nearly parallel to the fault strikes (Fig. 6a), that is, N030°–060°W, depending on the latitude, and consistent with directions of crustal anisotropy inferred from RF analysis. Despite the small number of analyzed events at some stations, this trend is robust as confirmed by the back azimuth of the nulls events (Fig. 6b) that are systematically parallel or perpendicular to the direction of the faults (apparent isotropy). At station WHY, located far from Tintina and Denali faults (> 300 km), fast directions of polarization present a more north–south trend, indicating a source of anisotropy different from stations close to the faults. A high variability of both splitting parameters and nulls directions is observed, caused either by vertical complexities of the anisotropy beneath the receiver (e.g., two layers of anisotropy) or by lateral heterogeneities (e.g., anomalous lithospheric block in the vicinity of the station). Vertical complexities are generally associated with a π=2 periodicity of the splitting parameter according to the back azimuth (Savage and Silver, 1993), whereas lateral heterogeneities are observed in a unique back-azimuthal window. Because of the discontinuous backazimuthal coverage at WHY, we are not able to discriminate between those two scenarios. To ease the interpretations, we determine the averaged splitting parameters for each station (except WHY) by applying, on the events for which shear-wave splitting and nulls are observed, the multichannel analysis approach developed by Chevrot (2000; Fig. 8). We refine the analysis by stacking
1972
N. Rasendra, M. Bonnin, S. Mazzotti, and C. Tiberi
sponding to a 100–150 km thick layer of anisotropy for an averaged 5% of anisotropy in the sampled medium (see figure 9 in Bonnin et al., 2010). For DAWY, PLBC, and YUK7, delay times are smaller (0.5–1 s) and are consistent with a 50–100 km thick layer of anisotropy for 5% anisotropy in the medium.
64°
63°
Discussion
62°
61°
60°
1s 59° ° –134° –142° –141° –140° –139° –138° –137° –136° –135
Figure 8. Averaged splitting parameters at station locations. Black bars are splitting parameters determined from SKS splitting measurements; white bars represent the direction orthogonal to the strongest negative peaks (Px) observed in RFs (compare with Fig. 4). Long solid white bars (BVCY, YUK1) are for stations at which strong crustal anisotropy is inferred from RFs; the short solid white bar (YUK3) is for weak crustal anisotropy; dashed white bars are for stations where Px are observed but do not allow for a good characterization of the anisotropy (noise, few events…). For WHY, the item is a schematic representation of the individual splitting parameters (see SKS Splitting Analysis section for details). The black circle without radiating lines accounts for station YUK6, where no splitting measurement has been obtained. Solid black lines indicate the main fault systems. individual splitting intensity measurements in back-azimuthal bins and using Wiener filtering (Monteiller and Chevrot, 2010). This technique is particularly relevant for cases with a single layer of anisotropy, as it avoids issues caused by the averaging of directions of polarization (in case of east–west or north–south dominant φ) and takes into account null measurements. Puzzling results at station YUK5 (90° discrepancy between averaged and individual splitting measurements) are probably caused by the small number of measurements (four) and by low signal-to-noise ratio. Thus, for YUK5 we present the arithmetic average of the splitting parameters, and we assign to WHY a range of splitting parameter fitting the overall individual splitting parameters. Ⓔ Table S3 presents averaged splitting measurements for all stations (except WHY and YUK6). Delay times observed in the central part of the network (BVCY, HYT, and YUK1-5) are strong (1–1.5 s), corre-
Crustal anisotropy deduced from RF signal is strong at most stations near the Denali fault and localized within the midcrust (10–25 km depth). Its fast axis is parallel to the fault trend. Stations further from the Denali fault (WHY, DAWY) and some stations close by (HYT, YUK5, YUK2) show no significant crustal anisotropy, although low-intracrustal signal may also be present. SKS splitting analysis evidences similar direction for mantle anisotropy (fast axis N030°– 060°W), parallel to the fault trend with an amplitude of ∼1 s. Hanna and Long (2012), using stations from U.S. national network throughout Alaska, also evidence an SKS splitting pattern parallel to the plate boundary in the vicinity of the southern Denali fault, which they associate with the plate boundary dynamics. Hereafter, we investigate potential sources that could explain this crustal and mantle anisotropy, which is nearly orthogonal to, thus not related to, the North America absolute plate motion. One potential source for anisotropy comes from regional tectonic and structural heritage (e.g., Barruol et al., 2008; Ozacar and Zandt, 2009). In southwest Yukon, Paleozoic and Mesozoic assemblages display a first-order northwest–southeast trend (Johnston and Canil 2007), which may correspond to the anisotropy fast axis recorded by both RFs and SKS. Mesozoic accretion of terranes located southwest of the Denali fault and between the Denali and Tintina faults may have created a frozen-in anisotropy in a general northwest–southeast direction (Figs. 1 and 8). However, the seismic stations are located on a variety of bedrock (metamorphic, plutons, schists, ophiolites) with locally complex tectonic history. No clear correlation can be found between bedrock geology and the presence or absence of crustal and upper-mantle anisotropy; for example, YUK3, YUK6, and YUK7 are located within the Insular superterrane accretion system, but so is YUK2, which does not display any significant midcrust anisotropy. Fossilized tectonic fabrics may participate in (and complicate) the crustal anisotropy signal but should not produce a strong coherent signal throughout the whole region. A more likely explanation is structural anisotropy associated with the Eocene Denali fault activity. The observed midcrust and upper-mantle anisotropy direction is parallel to the Denali fault trend and consistent with strike-slip or simple-shear transpression. Under this hypothesis, uppermantle anisotropy relates to lattice-preferred orientation of olivine crystals within the shear zone, as proposed for other large strike-slip faults (e.g., Klosko et al., 1999; Rümpker et al., 2003; Herquel and Tapponnier, 2005; Bonnin et al., 2010). Exposed plutons and metamorphic bedrock indicate
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric significant erosion in our study region since the early Mesozoic (5–15 km). Thus, anisotropy observed in the presentday midcrust level finds its origin in deeper mid to lower crust shear zones and could be due to highly deformed oriented fabrics. Foliated felsic rocks, such as gneiss or schists, are known to be highly birefringent (Barruol and Mainprice, 1993) and can explain both > 10% of mid to lower crust anisotropy and low V P =V S ratio (< 1:75). This explanation implies a shear zone associated with the Denali fault at least 50 km wide in both the midlower crust and upper mantle. The only constraint on the maximum width comes from the absence of RFs and SKS anisotropy at WHY, ∼125 km away from the Denali fault trace. A few stations closer to the fault (HYT, YUK2, YUK5) show no significant crustal anisotropy but display a strong upper-mantle anisotropy. Absence of midcrust signal at these stations may relate to complex crustal structure where the Denali fault shear zone is overprinted by noncollinear fabrics. RF analysis suggests the lowermost crust does not show significant fault-parallel anisotropy. Present-day lower crust would correspond to deeper levels (35–50 km depth) during the Eocene transpression phase. Several explanations can be proposed for the lack of anisotropy with this depth range. Strong lower crust rheology can result in highly localized shear (< 1 km; Platt and Behr, 2011) and would not be seen at our stations. Alternatively, the lower crust may have acted as a kinematic detachment level between the upper mantle and upper crust, that is, an orogenic float system (Oldow et al., 1990). In this model, the ductile lower crust can act as a partial accommodation zone allowing for partitioning deformation between upper and lower layers with independent kinematics (but still driven by similar boundary conditions). Thus, within this lower crust detachment level, deformation is highly heterogeneous and generates less anisotropy on the frequency range at which seismological tools are sensitive. Although likely associated with the early Cenozoic main activity phase of the Denali fault, this orogenic float process is akin to present-day tectonics, with coeval deformation along strike-slip and thrust faults merging on a common deep-crust detachment (Mazzotti and Hyndman, 2002), favored by the high temperature at the base of the crust (Lewis et al., 2003). A similar system is proposed for the San Andreas fault system to explain vertical partitioning of current deformation (Teyssier and Tikoff, 1998).
1973
both studies only provide an upper limit on the shear zone width (< 125–150 km). This possible shear-zone width contrasts with indications of deep, midcrust deformation on present-day strikeslip faults from tremor activity. Locations of seismic tremors below the San Andreas and Alpine faults suggest that the main shear zone may be only 1–5 km wide at 15–25 km depth (Shelly and Hardebeck, 2010; Wech et al., 2012). This difference might indicate a deformation gradient within the main shear zone, where the core (a few kilometers) accommodates most of the displacement but significant deformation extends over a much larger zone (> 50 km), where it is recorded in birefrengent rock fabrics. Although still tentative, these results suggest a strong potential for combined RF and shear-wave splitting analysis on dense cross-fault arrays to better characterize and understand the mechanics of large-scale strike-slip faults from the surface to the bottom of the lithosphere.
Data and Resources Seismograms used in this study were collected as part of the Canadian National Seismic Network (http://www .earthquakescanada.nrcan.gc.ca/stndon/CNSN‑RNSC/stnbook‑ cahierstn/index‑eng.php; last accessed September 2012). Data can be obtained from the Incorporated Research Institutions for Seismology (IRIS) Data Management Center at www. iris.edu (last accessed September 2012). Event origin times and locations are taken from the National Earthquake Information Center–Preliminary Determination of Epicenter (NEIC-PDE) catalog (U.S. Geological Survey; (http:// earthquake.usgs.gov/data/pde.php; last accessed September 2012). SKS splitting measurement were performed using SplitLab software (Wüstefeld et al., 2008), available at http:// www.gm.univ-montp2.fr/splitting/ (last accessed June 2012). Maps from Figures 1, 2, 6, and 8 were made using the Generic Mapping Tools version 4.5.9 (www.soest.hawaii.edu/gmt; last accessed August 2013; Wessel and Smith, 1998).
Acknowledgments Station information and data provided by Earthquakes Canada (Geological Survey of Canada). We thank two anonymous reviewers and the editor for their constructive comments. This work was supported by the French Agence National pour la Recherche Grant ANR-12-CHEX-0004-01.
Conclusion
References
Our results present tantalizing indications of strong crustal anisotropy associated with strike-slip (or simple shear) deformation along a large-scale continental fault. A similar signal is observed along the Marlborough fault system in northern South Island, New Zealand (Wilson et al., 2004). In both cases, strong midcrust anisotropy is observed in a 50–60 km wide zone, possibly the fault shear zone that accommodated several hundred kilometers of displacement;
Baldock, G., and T. Stern (2005). Width of mantle deformation across a continental transform: Evidence from upper mantle (Pn) seismic anisotropy measurements, Geology 33, 741–744, doi: 10.1130/G21605.1. Barruol, G., and D. Mainprice (1993). A quantitative evaluation of the contribution of crustal rocks to the shear wave splitting of teleseismic SKS waves, Phys. Earth Planet. Int. 78, 281–300, doi: 10.1016/0031-9201 (93)90161-2. Barruol, G., A. Deschamps, J. Deverchère, V. V. Mordvinona, M. Ulziibat, J. Perrot, A. A. Artemiev, T. Dugarmaa, and G. H. R. Bokelmann (2008). Upper mantle flow beneath and around the Hangray dome, central
1974 Mongolia, Earth Planet. Sci. Lett. 274, 221–233, doi: 10.1016/ J.epsl.2008.07.027. Beaudoin, B. C., G. S. Fuis, W. D. Mooney, W. J. Nockleberg, and N. I. Christensen (1992). Thin, low-velocity crust beneath the southern Yukon-Tanana terrane, east central Alaska: Results from trans-Alaska crustal transect refraction/wide angle reflection data, J. Geophys. Res. 97, 1921–1942, doi: 10.1029/91JB02881. Bonnin, M., G. Barruol, and G. H. R. Bokelmann (2010). Upper mantle deformation beneath the North American-Pacific plate boundary in California from SKS splitting, J. Geophys. Res. 115, no. B04307, doi: 10.1029/2009JB006438. Chevrot, S. (2000). Multichannel analysis of shear wave splitting, J. Geophys. Res. 105, 21,579–21,590, doi: 10.1029/2000JB900199. Cobbett, R., S. Israel, and J. K. Mortensen (2009). The Duke River fault, southwest Yukon: Preliminary examination of the relationship between Wrangellia and the Alexander terrane, in Yukon Exploration and Geology, K. E. Macfarlane, L. H. Weston, and L. R. Blackburn (Editors), Yukon Geological Survey, Whitehorse, Yukon, Canada, 143–158. Crampin, S. (1984). Effective anisotropic elastic constants for wave propagation through cracked solids, Geophys. J. Roy. Astron. Soc. 76, 135– 145, doi: 10.1111/j.1365-246X.1984.tb05029.x. Elliott, J., C. Larsen, J. T. Freymueller, and R. Motika (2010). Tectonic block motion and glacial isostatic adjustement in southeast Alaska and adjacent Canada constrained by GPS measurements, J. Geophys. Res. 115, 2156–2202, doi: 10.1029/2009JB007139. Gabrielse, H., J. W. H. Monger, J. O. Wheeler, and C. J. Yorath (1991). Part A: Morphogeological belts, tectonic assemblages, and terranes, in Geology of the Cordilleran Orogen in Canada, Geology of Canada Series 4, H. Gabrielse and C. J. Yorath (Editors), Geological Survey Canada, Ottawa, Ontario, Canada, 15–28. Gabrielse, H., D. C. Murphy, and J. K. Mortensen (2006). Cretaceous and Cenozoic dextral orogen-parallel displacements, magmatism and paleogeography, north-central Canadian Cordillera, in Paleogeography of the North American Cordillera: Evidence For and Against Large-Scale Displacement, Geological Association of Canada Special Paper 46, J. W. Haggart, J. W. H. Monger, and R. J. Enkin (Editors), 255–276. Hanna, J., and M. D. Long (2012). SKS splitting beneath Alaska: Regional variability and implications for subduction processes at slab edge, Tectonophysics 530/531, 272–285, doi: 10.1016/j.tecto.2012.01.003. Herquel, G., and P. Tapponnier (2005). Seismic anisotropy in western Tibet, Geophys. Res. Lett. 32, doi: 10.1029/2005GL023561. Johnston, T., and D. Canil (2007). Crustal architecture of SW Yukon, northern Cordillera: Implication for crustal growth in a convergent margin orogen, Tectonics 26, TC1006, doi: 10.1029/2006TC001950. Jones, C. H., and R. A. Phinney (1998). Seismic structure of the lithosphere from teleseismic converted arrivals observed at small arrays in the southern Sierra Nevada and vicinity, California, J. Geophys. Res. 103, 10,065–10,090, doi: 10.1029/97JB03540. Klosko, E. R., F. T. Wu, H. J. Anderson, D. Eberhart-Phillips, T. V. McEvilly, E. Audoine, M. K. Savage, and K. R. Gledhill (1999). Upper mantle anisotropy in the New Zealand region, Geophys. Res. Lett. 26, 1497–1500, doi: 10.1029/1999GL900273. Langston, C. A. (1977). The effect of planar dipping structure on source and receiver responses for constant ray parameter, Bull. Seismol. Soc. Am. 67, 1029–1050. Langston, C. A. (1979). Structure under Mount Rainier, Washington, inferred from teleseismic body waves, J. Geophys. Res. 84, 4749– 4762, doi: 10.1029/JB084iB09p04749. Lavé, J., J. P. Avouac, R. Lacassin, P. Tapponnier, and J.-P. Montagner (1996). Seismic anisotropy beneath Tibet: Evidence for eastward extrusion of the Tibetan lithosphere? Earth Planet. Sci. Lett. 140, 83–96, doi: 10.1016/0012-821X(96)00045-3. Leonard, L. J., R. D. Hyndman, S. Mazzotti, L. Nykolaishen, M. Schmidt, and S. Hippchen (2007). Current deformation in the northern Canadian Cordillera inferred from GPS measurements, J. Geophys. Res. 112, no. B11401, doi: 10.1029/2007JB005061.
N. Rasendra, M. Bonnin, S. Mazzotti, and C. Tiberi Levshin, A. L., M. H. Ritzwoller, M. P. Barmin, A. Villaseñor, and C. A. Padgett (2001). New constraints on the arctic crust and uppermost mantle: Surface wave group velocities, Pn and Sn, Phys. Earth Planet. Int. 123, 185–204, doi: 10.1016/S0031-9201(00)00209-0. Lewis, T. J., R. D. Hyndman, and P. Flück (2003). Heat flow, heat generation, and crustal temperatures in the northern Canadian Cordillera: Thermal control of tectonics, J. Geophys. Res. 108, doi: 10.1029/ 2002JB002090. Ligorria, J. P., and C. J. Ammon (1999). Iterative deconvolution and receiver-function estimation, Bull. Seismol. Soc. Am. 89, 1395–1400. Lowe, C., and J. F. Cassidy (1995). Geophysical evidence for crustal thickness variations between the Denali and Tintina fault systems in westcentral Yukon, Tectonics 14, 909–917, doi: 10.1029/95TC00087. Matmon, A., D. P. Schwartz, P. J. Haeussler, R. Finkel, J. J. Lienkaemper, H. D. Stenner, and T. E. Dawson (2006). Denali fault slip rates and Holocene–late Pleistocene kinematics of central Alaska, Geology 34, 645–648, doi: 10.1130/G22361.1. Mazzotti, S., and R. D. Hyndman (2002). Yakutat collision and strain transfer across the northern Canadian Cordillera, Geology 30, 495–498, doi: 10.1130/0091-7613(2002)0302.0.CO;2. Mazzotti, S., A. Maréchal, J. Elliot, J. T. Freymueller, and M. Schmidt (2012). Present-day strain partitioning and strain transfer across the Fairweather and Denali faults in SW Yukon–SE Alaska, Eos Trans. AGU (Fall Meet. Suppl.), San Francisco, California, Abstract T11A–2545. Monteiller, V., and S. Chevrot (2010). How to make robust splitting measurements for single-station analysis and three-dimensional imaging of seismic anisotropy, Geophys. J. Int. 182, 311–328, doi: 10.1111/ j.1365-246X.2010.04608.x. Nicolas, A., and N. I. Christensen (1987). Formation of anisotropy in upper mantle peridotites: A review, in Composition, Structure and Dynamics of the Lithosphere-Asthenosphere System, K. Fuchs and C. Froidevaux (Editors), American Geophysical Union, Washington, D.C., 111–123. Oldow, J. S., A. W. Bally, and H. G. Avé-Lallemant (1990). Transpression, orogenic float, and lithospheric balance, Geology 18, 991–994, doi: 10.1130/0091-7613(1990)0182.3.CO;2. Ozacar, A. A., and G. Zandt (2009). Crustal structure and seismic anisotropy near the San Andreas fault at Parkfield, California, Geophys. J. Int. 178, 1098–1104, doi: 10.1111/j.1365-246X.2009.04198.x. Plafker, G., J. C. Moore, and G. R. Winkler (1994). Geology of the southern Alaska margin, in The Geology of Alaska G. Plafker and H. C. Berg (Editors), Geological Society of America, Boulder, Colorado, 389–449. Platt, J. P., and W. Behr (2011). Deep structure of lithospheric fault zones, Geophys. Res. Lett. 38, L24308, doi: 10.1029/2011GL049719. Rümpker, G., T. Ryberg, G. Bock, and Desert Seismology Group (2003). Boundary-layer mantle flow under the Dead Sea Transform fault inferred from seismic anisotropy, Nature 425, 497–501, doi: 10.1038/ nature01982. Sambridge, M. (1999). Geophysical inversion with a neighbourhood algorithm—I. Searching a parameter space, Geophys. J. Int. 138, 479–494, doi: 10.1046/j.1365-246X.1999.00876.x. Savage, M. K. (1999). Seismic anisotropy and mantle deformation: What have we learned from shear wave splitting? Rev. Geophys. 37, 65–106, doi: 10.1029/98RG02075. Savage, M. K., and P. G. Silver (1993). Mantle deformation and tectonics: Constraints from seismic anisotropy in western United States, Phys. Earth Planet. Int. 78, 207–227, doi: 10.1016/0031-9201(93)90156-4. Shelly, D. R., and J. L. Hardebeck (2010). Precise tremor locations and amplitude variations along the lower-crustal central San Andreas fault, Geophys. Res. Lett. 37, L14301, doi: 10.1029/2010GL043672. Shibutani, T., M. Sambridge, and B. L. N. Kennett (1996). Genetic algorithm inversion for receiver functions with applications to crust and uppermost mantle structure beneath eastern Australia, Geophys. Res. Lett. 23, 1829–1832, doi: 10.1029/96GL01671. Silver, P. G. (1996). Seismic anisotropy beneath the continents: Probing the depths of geology, Annu. Rev. Earth Planet. Sci. 24, 385–432, doi: 10.1146/annurev.earth.24.1.385.
Crustal and Upper-Mantle Anisotropy Related to Fossilized Transpression Fabric Silver, P. G., and W. W. Chan (1991). Shear wave splitting and subcontinental mantle deformation, J. Geophys. Res. 96, 16,429–16,454, doi: 10.1029/91JB00899. Teyssier, C., and B. Tikoff (1998). Strike-slip partitioned transpression of the San Andreas fault system: A lithospheric-scale approach, Geol. Soc. Lond. Spec. Publ. 135, 143–158, doi: 10.1144/GSL.SP.1998.135.01.10. Tommasi, A., B. Tikoff, and A. Vauchez (1999). Upper mantle tectonics: Three-dimensionnal deformation, olivine crystallographic fabrics and seismic properties, Earth Planet. Sci. Lett. 168, 173–186, doi: 10.1016/S0012-821X(99)00046-1. Vauchez, A., A. Tommasi, and D. Mainprice (2012). Faults (shear zones) in the Earth’s mantle, Tectonophysics 558/559, 1–27, doi: 10.1016/ j.tecto.2012.06.006. Wech, A. G., C. M. Boese, T. A. Stern, and J. Townend (2012). Tectonic tremor and deep slow slip on the Alpine fault, Geophys. Res. Lett. 39, L10303, doi: 10.1029/2012GL051751. Wessel, P., and H. F. Smith (1998). New, improved version of the Generic Mapping Tools released, Eos Trans. AGU 79, 579, doi: 10.1029/ 98EO00426. West, D. P., and M. S. Hubbard (2003). Progressive localization of deformation during exhumation of a major strike-slip shear zone: Norumbega fault zone, south-central Maine, USA, Tectonophysics 273, 185–201, doi: 10.1016/S0040-1951(96)00306-X. Wilson, C. K., C. H. Jones, P. Molnar, A. F. Sheehan, and O. S. Boyd (2004). Distributed deformation in the lower crust and upper mantle beneath a
1975
continental strike-slip fault zone: Marlborough fault system, South Island, New Zealand, Geology 32, 837–840, doi: 10.1130/G20657.1. Wüstefeld, A., G. H. R. Bokelmann, C. Zaroli, and G. Barruol (2008). SplitLab: A shear-wave splitting environnement in MATLAB, Comput. Geosci. 34, 515–528, doi: 10.1016/j.cageo.2007.08.002.
Géosciences Montpellier UMR5243 université Montpellier 2 CNRS, Place E. Bataillon F34095 Montpellier Cedex 5, France (N.R., S.M., C.T.)
Géoazur, UMR7329 université de Nice Sophia-Antipolis CNRS, observatoire de la Côte d’Azur 250 rue A. Einstein F06560 Valbonne, France (M.B.) Manuscript received 2 September 2013; Published Online 8 July 2014
