Mapping the Moho beneath the Southern Alps continent-continent collision, New Zealand, using wide-angle reflections



[1] A wide-angle stack using all onshore-offshore, and onshore seismic data from the SIGHT'96 experiment provides a direct three-dimensional (3-D) image of the Moho below the continental collision zone through the South Island of New Zealand. A bright reflector sequence (up to 4 s thick), the base of which coincides with the PmP modeled Moho, extends throughout most of the lower crust and bends downward asymmetrically beneath the Southern Alps. The crustal root reaches a maximum depth at ∼15 s (45 km), beneath the regional (−80 mGal) Bouguer gravity low and is flanked east and west by shallower Moho where the average crustal thickness is ∼20 km. The 3-D structure of the crustal root in the central South island is moderately well resolved by the SIGHT'96 experiment.

1. Introduction

[2] The most important compositional and density boundary in the lithosphere is the Moho, which is generally regarded as the boundary that separates intermediate-to-mafic crust (≤6.8 km/s) from ultramafic mantle (7.6–7.8 km/s) [Jarchow and Thompson, 1989]. Isostatic constraints demonstrate that in continent-continent collision zones the crust must thicken and give rise to a crustal root beneath mountain ranges, but the dynamics at the crust-mantle interface are more difficult to determine. For example, in distinguishing between models of strain partitioning and mechanisms for accommodating convergence, many diagnostic predictions of models centre on processes in the lower crust and upper mantle. This is especially true in the initial development of an orogen, when ductile flow is thought to play a major role in accommodating strain in the crustal root and where deflection of the Moho is governed by the ability of the underlying mantle lithosphere to deform [Gerbault et al., 2003]. Geophysical methods are able to obtain Moho geometries and strain indicators (e.g., seismic reflection patterns) in broad convergent orogens such as Tibet [Haines et al., 2003]. However, for narrow orogens such as the Transverse Ranges [Godfrey et al., 2002], the Pyrenees [Choukroune and ECORS Team, 1989], and Taiwan [Wu et al., 1997], root structures are difficult to seismically image where steep flanks lead to complex ray paths and where phases may have double branches.

[3] The South Island of New Zealand provides a simple setting in which models of active convergent mountain belts can be compared to observations. The Australian/Pacific plate boundary through the South Island has progressively formed in the last 45 Ma, transforming an initial Eocene passive margin into the dextral strike-slip Alpine fault, with about 450 km of offset [Carter and Norris, 1976; Sutherland et al., 2000; Wellman and Willett, 1942]. Oblique convergence commenced about 7 Ma ago, and has resulted in approximately 100 km of shortening and uplift (8–10 mm/yr) of the Southern Alps [Beavan and Haines, 2001; DeMets et al., 1990; Walcott, 1998; Wellman, 1979], (Figure 1), although revised Pacific-Australia plate rotations [Cande and Stock, 2004] suggest the average rate of convergence could be 40% smaller. A comprehensive range of geophysical measurements (South Island GeopHysical Transect - SIGHT) were undertaken in 1995/96 to derive a three dimensional structural model for the South Island orogen [Davey et al., 1998]. The principal SIGHT field activity was an integrated onshore and onshore-offshore wide-angle seismic refraction experiment, combined with a marine multi-channel seismic (MCS) component, on two main transects (1 and 2) across the central part of the South Island (Figure 1). These transects were designed to cross the Southern Alps, where the orogen is relatively unaffected by neighboring tectonic regimes, and to provide optimal raypath coverage for 2-D crustal illumination [Okaya et al., 2002]. Seismic velocity models derived from refracted and reflected arrivals along Transects 1 [Van Avendonk et al., 2004] and 2 [Scherwath et al., 2003], and first arrival travel-time tomography of passive seismic data [Eberhart-Phillips and Bannister, 2002] provide an image of a crustal root about 100 km wide, that deepens from approximately 40 km under Transect 1 to a maximum depth of 45 km under Transect 2. The deepest part of the crust lies about 20 km east of the main topographic divide and about 40 km east of the Alpine fault. However, the resolution of velocity models in the region of the crustal root is poor, owing to the limited number of identified lower crustal observations from sources immediately east of the main divide and the steep dip on the root margins.

Figure 1.

Location map of the SIGHT onshore and offshore seismic experiment. a) Location of the South Island of New Zealand. The boundary between the Australian and Pacific plates is shown as solid line. b) Shaded relief map of the South Island, showing the SIGHT project's set of multi-channel seismic ship tracks (black lines), ocean bottom seismometers/hydrophones (circles), and onshore-offshore receiver stations (triangles) and shots (stars). The figure also shows the present location of the Alpine fault and eastern South Island coast line and their position at 6 Ma using PAC-AUS finite poles and rotation from [Walcott, 1998].

[4] We present a new three-dimensional (3-D) seismic image of the Moho across the Australian/Pacific plate boundary through the central South Island. Although SIGHT was not specifically designed as a 3-D experiment, the geometry of the shots and receivers used in SIGHT can be used to construct a partial 3-D image of the central South Island. Unlike travel-time tomographic techniques, which require time consuming identification and picking of phases, a common-mid-point image of the lower crust can be created using seismic reflection processing techniques. PmP and other phases, not clearly or confidently identified on individual shot and receiver gathers, are included in our stacks and thus provide additional constraints in determining the shape of the crustal root. The seismic image produced has reasonable signal-to-noise ratio even when the Moho is not visible on coincident offshore MCS reflection data. The main limitation to the coverage and quality of the 3-D image was the linear distribution of land recording stations and the small number of onshore shots.

2. Data Acquisition and Processing

[5] During the SIGHT experiment (see Figure 1) the R/V Maurice Ewing (8600 in3 airgun array) deployed 20 ocean bottom seismometer/hydrophone (OBS/H) instruments and collected 763 km of MCS profiling in the Tasman Sea. Two hundred and nine portable seismic recorders were deployed across the two central transects and along the western coastline to record the airgun signals. Upon completion of this profiling, the Ewing transited to the Pacific Ocean while all instruments were retrieved and their data downloaded. While in the Pacific Ocean the Ewing collected 1286 km of MCS profiling and redeployed the OBS/H instruments. The portable land seismic recorders were again deployed along the same cross-island transects and an eastern coast-array. Previously Transect 1 land refraction profile recorded sixteen explosive shots (300–1200 kg) on 400 stations across the central South Island. On Transect 2, seven shots spaced about 22 km apart were recorded by a similar 400-station array.

[6] Composite wide-angle super-gathers from the SIGHT experiment [Okaya et al., 2002] show clear Pg phases out to distances of 150 km. Strong pre-critical lower-crustal (PiP) and Moho (PmP) reflections at the base of the crust (7–12 s twt beneath shot-points) are observed on individual shot gathers as a train of seismic energy. Strong post-critical reflections of P-waves from the Moho are also seen on onshore-offshore receiver gathers over offset ranges 200–300+ km. Only pre-critical lower crustal reflections have been used in our stacks. At larger offsets reflections from the base of the crust merge with Pg and Pn phases. The data recorded on the OBS/H recorders had considerable reverberation and did not contribute significantly to the image produced in this study.

[7] Details of the processing scheme are included in auxiliary material. Figure 2 shows the distribution of trace midpoints and offset distribution of clean traces. There is a concentration of midpoints along the two main SIGHT transects and midway between the transects that is populated with traces recorded on one profile from sources on the other. Three 2-D CDP lines were extracted along these midpoint concentrations. However, there are gaps in the CDP coverage in the centre of the intermediate line because there are no onshore shot data; each transect was shot and instrumented consecutively, and no onshore shots were placed away from the two main traverses. Stacks created using a single velocity of 6.0 km/s showed the best continuity of lower crustal reflectors, which is consistent with crustal velocities derived independently [Eberhart-Phillips and Bannister, 2002]. Finally, traces were displayed as instantaneous amplitudes (Figure 3).

Figure 2.

a) Midpoint and offset distribution, and b) fold distribution for both the SIGHT'96 onshore-offshore deployment and land reflection/refraction recording. There is a concentration of midpoints along the two main SIGHT transects, that is (dark red colors) and on the profile intermediate between Transect 1 and 2.

Figure 3.

Wide-angle CDP stacked images for a) Transect 1, b) the intermediate profile and c) Transect 2. Superimposed on Transect 2 and 1 are the acoustic synthetic seismogram response of introducing a plane–wave into the top of the velocity models; derived from ray-tracing inversion of observed crustal phases [Scherwath et al., 2003; Van Avendonk et al., 2004]. Part of the MCS reflection line 1w is shown in a) at the same scale as the wide-angle image and placed in the correct model reference frame for comparison.

3. Results

[8] Images from all three wide-angle stacks reveal a highly reflective lower crust up to 4 s thick that extends throughout most of the mid-crust on the three profiles and bends downward beneath the Southern Alps. The accuracy of this image and interpretation can be verified by forward modelling of the velocity models for Transects 1 and 2. The compressional velocity models for Transect 1 [Van Avendonk et al., 2004] and Transect 2 [Scherwath et al., 2003] were used to compute normal-incidence wavefields using acoustic finite differences, and then superimposed on the CDP image (Figures 3a and 3c).

[9] On Transect 2 the Moho in the velocity models (as defined by PmP) coincides with the base of reflectivity on wide-angle stacked images. The top of the lower crust in the velocity model, determined from wide-angle reflections (PiP), is identified as the onset of reflectivity (Figure 3c). In places along Transect 2, for example at model km 380, a thin lower crust has resulted in enhanced crustal reflectivity. On Transect 1 the Pacific plate lower crustal reflectivity appears to coincide with interpreted PiP reflections between CDP's 850 to 1000 (near km 300) but matches PmP between CDP's 1200 to 1300 (near km 390). On the Australian plate the strong reflectivity is 1 s deeper than the modeled PmP off the West Coast but is consistent with the location of mantle reflections observed on the MCS lines 1w (Figure 3a insert) and 3w (A. Melhuish et al., Crustal and upper mantle seismic structure of the Australian plate, South Island, New Zealand, submitted to Tectonophysics, 2004, hereinafter referred to as Melhuish et al., submitted manuscript, 2004). However, these prominent mantle reflections dip to the north and are not observed where Transect 2 intersects 3w (Melhuish et al., submitted manuscript, 2004).

[10] The Australian plate Moho appears to display a sharp jump of 2 s east of the surface trace of the Alpine fault on all three profiles. East of the Alpine fault and beneath the South Island, the synthetic wavefields and wide-angle stacks are marked by a classic bow-tie; a function of the steep sided flanks of the crustal root between distance 250 to 300 km. Our image for Transect 1 suggests that both the eastern and western side of the root are steeper than the ∼25° dip modeled by [Van Avendonk et al., 2004] but are less steep than the 45° dip modeled by [Scherwath et al., 2003] on Transect 2. By using the synthetic image of Transect 2 as a guide, the pattern of lower crustal reflectivity on the intermediate profile (Figure 3b) is best matched using dips of at least 40° and 25° for the western and eastern flanks of the crustal root. A comparison of the modeled arrivals and wide-angle images also confirms that the Moho ranges in depth from 8 s (∼20 km depth), in the east beneath the Canterbury Basin and 10 s (∼25 km depth) beneath the West Coast basin. The crustal root is shallower in the north, where the base of lower crustal reflectivity is at 12 s (∼40 km depth) beneath the Alps on Transect 1, and reaches a maximum depth at ∼15 s (45 km) on Transect 2.

4. Discussion

[11] The 3-D structure of the crustal root is moderately well resolved in the central South Island by the SIGHT'96 experiment where the seismic images along the three profiles presented here confirm the presence of a pronounced root 80 km wide that reaches its maximum thickness in the south along Transect 2 (45 km thick). The orientation of the root follows the trend previously identified in 3-D velocity models [Eberhart-Phillips and Bannister, 2002] and the Bouguer gravity field (Figure 1). In addition, the stacked images reveal a pronounced asymmetry to the shape of the root with the western flank dipping at ∼45°. Numerical modelling of the Southern Alps orogen [Gerbault et al., 2002, 2003] successfully predicts the 3-D root structure by invoking ductile flow in the lower crust, both perpendicular to and along the plate boundary.

[12] The additional constraints provided by our direct image of the crust and new finite rotation poles [Cande and Stock, 2004] can now be used to reassess the 3-D mass balance. The volume of material in the crustal root is estimated to be 150,000 km3, 30% of the total crustal rocks to have entered the orogen in the last 5 Myr (∼500,000 km3 for a 25 km thick undeformed Pacific crust). After accounting for the present Southern Alps topography, this implies that the amount of Pliocene sediment eroded and deposited is approximately 300,000 km3, which is similar to the estimate of 350,000 km3 [Walcott, 1998].


[13] Funding for this joint USA-New Zealand project was provided by the U.S. National Science Foundation Continental Dynamics Program (EAR-9418530) and the New Zealand Foundation for Research Science and Technology (C05X0203).