Geophys. J. Int. (2006) 166, 1466–1483 doi: 10.1111/j.1365-246X.2006.03016.xG

JIVol

cano

logy

,ge

othe

rmic

s,flui

dsan

dro

cks

Implications for intraplate volcanism and back-arc deformationin northwestern New Zealand, from joint inversion of receiverfunctions and surface waves

N. A. Horspool,1 M. K. Savage1 and S. Bannister2

1School of Earth Sciences, Victoria University of Wellington, PO Box 600, Wellington, New Zealand. E-mail: nick.horspool@gmail.com2Institute of Geological and Nuclear Sciences, PO Box 30368, Lower Hutt, New Zealand

Accepted 2006 March 20. Received 2006 March 20; in original form 2005 May 2

S U M M A R YWe employ a joint inversion of teleseismic receiver functions and surface wave phase velocitiesto determine the shear wave velocity structure in the crust and upper mantle beneath north-western New Zealand. Receiver functions primarily contain information on velocity contrasts,while surface waves are sensitive to the average shear velocity with depth. By performinga joint inversion we reduce the limitations of each method, resulting in a more robust shearwave model. Inversion results reveal regions of low shear wave velocity of ∼2.8 km s−1 in themid-crust (10–19 km depth) and ∼4.0 km s−1 in the upper mantle (70–90 km depth) beneathQuaternary intraplate basalt fields. We infer that the mid-crustal low-velocity zones (LVZs) arebodies of partial melt, most likely rhyolite intrusions. We suggest that the upper mantle LVZ iscaused by the melt-producing regions of the upper mantle and is a source for the basalts of theAuckland Volcanic Field. This is in agreement with models for a shallow upper mantle sourcerather than a deep-seated mantle plume for the Auckland volcanism.

Average shear velocities for the upper crust are 3.4–3.6 km s−1, increasing to 3.6–4.0 km s−1

in the lower crust. The Moho is interpreted to be 29 ± 1 km deep in the southern end of thearray, shallowing to 26 ± 1 km towards the edge of the continental shelf of northern NewZealand. Low upper mantle shear velocities of 4.2 ± 0.1 km s−1 are observed, and are thoughtto represent a small percentage of partial melt in the upper mantle beneath this back-arcregion. These low velocities extend to greater depth (∼90 km) beneath the southern stations,indicating that mantle deformation associated with the mantle wedge to the south may extendas far north as the southern end of our array. Near-surface low-velocity layers extend to 6 kmdepth beneath a station at the northern tip of New Zealand, likely representing the expression ofa thick sequence of sediments from the Northland Allochthon and the more recent NorthlandBasin.

Key words: Auckland volcanic field, crustal structure, joint inversion, New Zealand, North-land, North Island, receiver functions, surface waves.

1 I N T RO D U C T I O N

Intraplate volcanism occurs throughout northwestern New Zealand

yet its source has remained relatively unexplained. One third of New

Zealand’s population lives in Auckland city, which is situated on the

active Auckland Volcanic Field; the field poses a potential hazard

to the Auckland region and a future eruption would cause major

disruptions to the society. The region occupies a unique tectonic

setting, straddling the transition from back-arc extension to a ‘nor-

mal’ back-arc setting on the continental landmass of New Zealand.

Our knowledge about the volcanism and back-arc deformation pro-

cesses occurring in northwestern New Zealand can be extended

by using seismological methods to investigate the structure of the

region.

Receiver function analysis of teleseismic earthquakes recorded

on three-component seismic stations has become a standard tool for

crust and upper mantle investigators. It has been successfully em-

ployed to invert for the 1-D shear wave velocity structure beneath

stations (Langston 1979; Owen & Zandt 1985; Ammon et al. 1990;

Ammon 1991; Cassidy 1992), to derive elaborate images of upper

mantle discontinuities (Gurrola et al. 1994; Li et al. 2000) and to cal-

culate Moho depth and Vp/Vs ratios for the crust (Zandt & Ammon

1995; Zhu & Kanamori 2000). Receiver function inversion is in-

herently non-unique, caused by a velocity-depth trade-off between

the thickness and the shear wave velocity of a given layer. The

non-uniqueness is apparent when a thick fast layer can cause a sim-

ilar response (in time) to a thin slow layer (Ammon et al. 1990;

Ozalaybey et al. 1997). Ammon et al. (1990) illustrated that

1466 C© 2006 The Authors

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1467

caution must be taken when inverting receiver functions for velocity

structure, as there is a high dependence on the initial starting model

and the use of available a priori constraints should be mandatory. In

recent work, there has been a major drive to combine receiver func-

tion data with additional data sets such as seismic reflection and

refraction data (Eaton & Cassidy 1996; Baumont et al. 2001; Galve

et al. 2002) and surface wave data (Ozalaybey et al. 1997; Julia

et al. 2000; An & Assumpcao 2004; Chang et al. 2004; Lawrence &

Wiens 2004). Surface waves are particularly useful when combined

with receiver functions, as they are also dependent on the shear wave

structure, but they are complementary, in that they are sensitive to the

average shear wave velocity structure with depth rather than to ve-

locity contrasts. In comparison, receiver functions cannot constrain

absolute velocities but they provide a wealth of information on the

fine-scale velocity contrasts (∼1 km). A joint inversion involving

both surface wave dispersion and receiver function data provides a

better-constrained shear wave velocity model. This joint approach

has long been used by active source seismic investigations, where

the seismic refraction data constrain the absolute velocity model,

while seismic reflection data provide the detailed structural image

(Pavlis 2003).

1.1 Geologic and tectonic background

Northwestern New Zealand (Fig. 1) is a region that has been some-

what neglected as a target by crustal scale geophysical experiments,

Figure 1. Location map of northwestern New Zealand. Location of Quaternary intraplate basalt fields shown after Heming (1980), South Auckland Volcanic

Field (SAVF), Auckland Volcanic Field (AVF), Whangarei Volcanic Field (WVF) and the Kaikohe-Bay of Islands Volcanic Field (KBIVF). Isopachs of the

Northland Allochthon after Herzer & Isaac (1992). Station locations used in receiver function and surface wave analysis are shown as triangles for NORD

stations and squares for permanent stations. The dashed line marks the boundary of the Central Volcanic Region (CVR) and the solid line marks the Taupo

Volcanic Zone (TVZ). Inset map shows the plate boundary to the east of New Zealand, with relative plate motion vectors shown as arrows (DeMets et al. 1990).

possibly because at present it is tectonically quiescent, situated in

a behind-arc setting. Nonetheless, northwestern New Zealand has

undergone a complex tectonic history in the past 80 Ma. The region

lies behind the active arc of the Hikurangi Margin on a peninsula

running subperpendicular to the Hikurangi subduction zone (Fig. 1).

The Hikurangi Margin is characterized by oblique subduction of the

Pacific Plate beneath the Australian Plate (Walcott 1978). Relative

plate motion vectors between the Australian and Pacific plates de-

crease from 47 mm yr−1 in the north to 37 mm yr−1 in the south,

reflecting the oblique nature of the New Zealand plate boundary

(DeMets et al. 1990, 1994). In the back-arc of the subduction zone,

the Taupo Volcanic Zone defines the eastern boundary of the Central

Volcanic Region, the zone of active extension in North Island (Stern

1985).

Here, we refer to northwestern New Zealand as the region to

the northwest of the Central Volcanic Region. It is manifested by

extensional faulting inherited during rifting from Gondawana 80

to 60 Ma (Sporli 1989). Basement geology consists of Mesozoic-

age greywacke that is overlain by Cretaceous to Oligocene sedi-

ments, obducted during the Oligocene as the Northland Allochthon

(Brothers & Delaloye 1982; Herzer & Isaac 1992; Isacc 1996; Rait

2000). Miocene-age sediments are present in the southern part of

Northland (Edbrooke 2001). Along the west coast of Northland Ter-

tiary andesite volcanoes are buried, defining the volcanic arc during

the Miocene (Challis 1978; Brothers 1984, 1986; Smith et al. 1989).

During the Miocene, a subduction zone was located off the east coast

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1468 N. A. Horspool, M. K. Savage and S. Bannister

of Northland, oriented parallel to the peninsula with a southwest dip-

ping slab (Hayward 1987; Smith et al. 1989; Herzer 1995; Hayward

et al. 2001). The subduction zone is thought to have regressed and

rotated to the southeast during the Pliocene until it reached its current

position (Kear 1994, 2004). Intraplate basalt fields of Quaternary

age are interspersed throughout northwestern New Zealand (Fig. 1),

the positions of which have remained static during subduction zone

migration, ruling out any ongoing relationship between the subduc-

tion zone and intraplate volcanism (Cole 1986; Smith et al. 1993;

Kear 1994, 2004).

The origin of the intraplate volcanism was originally explained by

a deep seated mantle plume system beneath northern New Zealand

(Heming 1980; Smith et al. 1993). However, more recent geochem-

ical work on the Auckland volcanic field, using U–Th isotopes to

determine dynamic melt generation models, indicates that the field

has a classic HIMU signature, suggesting a shallow mantle source.

In addition, low upwelling rates of melt imply that the source re-

gion for the melt lies in the upper mantle at depths between 80 and

140 km rather than at greater depths (Huang et al. 1997). Helium

isotope data collected at gas discharges throughout New Zealand

have identified regions of recent mantle melting (Hoke & Sutherland

1999); ratios of3He/4He in the crust are three orders of magnitude

less than that from the mantle, with high-3He/4He regions reflect-

ing where mantle melting is occurring. Measurements of3He/4He

from near the Kaikohe-Bay of Islands and Whangarei volcanic fields

(Fig. 1) are the highest (>70 per cent mantle helium) in North Is-

land outside of the Central Volcanic Region. No samples have been

taken near the Auckland volcanic field. Thus, both very low up-

welling rates and helium isotope data suggest that mantle melting

is occurring beneath the Northland peninsula (Huang et al. 1997;

Hoke & Sutherland 1999).

Geophysical evidence is in good agreement with the geochemical

data for the source location of the intraplate basalts. Mooney (1970)

used earthquake traveltimes to identify a low-velocity layer in the

upper mantle, at 75–125 km depth, that coincided with high seismic

attenuation (Q P = 40–80) for the region northwest of the Central

Volcanic Region. More detailed investigations support such a low-

velocity, high-attenuation region (Reyners et al. 2005). However,

it is unclear how far north these low velocities and high seismic

attenuation extend beneath Northland.

The basaltic centres typically have a monogenetic (‘one shot

plumbing’) eruption style, which also indicates that they are sourced

directly from the mantle and not from basaltic magma chambers lo-

cated within the crust (Heming 1980; Smith et al. 1993). There

is some uncertainty regarding the origin of rhyolite centres within

the Kaikohe-Bay of Islands volcanic field (Fig. 1). It is suggested

that these rhyolites formed by partial melting of the lower crust as

basaltic magmas repeatedly migrated to the surface. In addition, the

Kaikohe-Bay of Islands volcanic field (Fig. 1) is the only field with

associated geothermal activity (Ngawha Geothermal System). Such

geothermal activity requires a heat source, which could be provided

by a rhyolite intrusion in the crust (Heming 1980).

Until this study, the crust and upper mantle structure model of

northwestern New Zealand have been solely based on a reconnais-

sance seismic refraction survey, which was designed to provide

a first-order approximation of the crustal structure for the region

(Stern et al. 1987). A crustal thickness of 25 ± 2 km was found.

However, due to large receiver spacings and poor data recovery rates,

the possibility of hidden layers and low-velocity zones (LVZs) could

not be ruled out at the time. P-wave velocities in the upper crust were

found to range from 5.3 to 5.9 km s−1, increasing to ∼6.2 km s−1

in the lower crust. Low-Pn wavespeeds of 7.6 km s−1 were obtained

just below the Moho, increasing to 7.9 km s−1 at 40 km depth (Stern

et al. 1987).

Here, we investigate the shear wave velocity structure to a depth

of 100 km beneath northwestern New Zealand using a joint inversion

of receiver functions and surface wave phase velocity data. The 1-D

structure is determined using data recorded by a linear array of

three broad-band seismometers deployed between 2002 August and

2004 February (MATA, MKAZ, TIKO), referred to as the NORD

array (NORthland Deployment). Additional data recorded on three

permanent broad-band stations (TOZ, WCZ, OUZ) during the same

time period are also used (Fig. 1). Temporary stations used Guralp-

40T sensors and Nanometrics Orion data recorders, while permanent

stations are equipped with Guralp-70T sensors. They initially used

Orion data recorders that were replaced by Quantera data loggers

later in the deployment. We derive and present a model for the shear

wave velocity structure of the crust and upper mantle in northwestern

New Zealand.

2 S U R FA C E WAV E P H A S E V E L O C I T I E S

Surface waves have been extensively used to derive valuable infor-

mation on the shear wave velocity structure with depth (e.g. Brune

& Dorman 1963; Knopoff 1972; Kovach 1978). The dispersive char-

acteristics (i.e. faster velocities for longer period waves) of surface

waves provide an indication of the average shear wave velocity as a

function of depth.

The linear, northwestern orientation of the NORD array offers

a favourable opportunity to conduct a surface wave investigation

of northwestern New Zealand. Calculation of interstation phase ve-

locities requires the event to lie on a similar great circle arc to the

station array. The NORD array benefits from seismicity along the

subduction zones of the western Pacific.

2.1 Data

The standard practice when applying the interstation method to

measure surface wave phase velocities is to use earthquakes that

lie within ∼5◦ of the interstation great circle arc. Recent workers

(e.g. Ritzwoller et al. 2002; Spetzler et al. 2001, 2002) have quan-

titatively expressed lateral sensitivity kernels for surface waves and

have found that events can be included from a wider backazimuth

range. The inclusion of more out-of-line events results in an average

lateral sampling due to the wide lateral region that surface waves

sample. West et al. (2004) determined that by selecting events within

15◦ of the interstation path a larger number of events are used in the

analysis, resulting in an improvement in the rms velocity error. We

search the IRIS-DMC database for earthquakes that lie within 15◦

of the interstation path and have M w > 6. A total of 11 events filled

these criteria and produced dispersive properties (Fig. 2). All but one

event is located to the northwest of the NORD array. The events that

were selected are clustered together at similar epicentral distances,

a direct consequence of the great circle path of the array crossing

the seismicity belt of the western Pacific at a high angle (Fig. 2).

2.2 Slant stacking method

We apply a slant stacking method to determine interstation phase

velocities for multiple station triplets, following the method of

Herrmann & Ammon (2002). This involves slant stacking a series

of waveforms in the frequency domain for a range of slownesses.

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1469

Figure 2. Hypocentres of events used to determine interstation phase ve-

locities (stars) and in receiver function analysis (circles and stars). Note that

most of the events are from the northwest.

We obtain a modulus of the stack, F, which is a function of fre-

quency and slowness. The value of F provides an indication of how

well a given phase stacks for different velocities, with a maximum

of F indicating that significant energy is arriving at that velocity

(Herrmann & Ammon 2002; West et al. 2004). Benefits of this

method include removal of the source spectrum from the analysis,

Figure 3. Final stack functions, F, for each station triplet. The white line represents the resulting dispersion curve for each station triplet. We see good dispersion

between 15 and 50 s for all of the stations. Poor signal at shorter periods could be a result of multipathing due to the wide station spacing.

correction of geometric spreading and separation of the fundamental

and higher modes.

Prior to the slant stacking method, we remove the instrument re-

sponse from the waveforms to minimize distortion of the waveform

due to varying corner frequencies of the two sensors used. Apply-

ing the slant stacking method, we limit the range of velocities to

between 2.5 and 5 km s−1 and display the results as phase velocity

as a function of period (Fig. 3).

We follow the procedure of West et al. (2004), constructing a

rolling bin to measure phase velocities; the slant stack is performed

on a set of traces from three stations. The bin is then rolled along

one station and the stack is performed again. The resulting phase

velocities for each station triplet are representative of the centre

station, since it is sampling the region between the first and third

stations in the triplet. For the NORD array, this results in four station

triplets (TIKO–OUZ–WCZ, OUZ–WCZ–MATA, WCZ–MATA–

MKAZ, MATA–MKAZ–TOZ). Following calculation of the stack

for each event for each station triplet, we perform a second stack

of F for multiple events for each station triplet. In the stack, each

event is summed with its raw amplitude, so that larger events con-

tribute more energy to the stack. This second stack removes effects

of multipathing and spectral gaps that can affect individual stacks.

To determine the final dispersion curve for each station triplet, we

search each second stack for the maximum value of F for periods

between 10 and 50 s, using a sampling interval of 1 s (Fig. 3, white

line).

2.3 Observed phase velocities

Final stack functions for each station triplet show good dispersion

between periods of 15 and 50 s (Fig. 3), indicated by the dark grey

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1470 N. A. Horspool, M. K. Savage and S. Bannister

Figure 4. Phase velocities as a function of distance across the array. Note the lateral variations in phase velocity, indicating some heterogeneity across the

array. A surface was fitted between the phase velocity curves for each station to create the 2-D profile. The surface was created using splines in tension, with a

tension factor of 1. Note that the vertical scale is period (s), not depth.

colour which is the value of F, the stack function. There is little

energy at periods lower than 15 s; this could be due to a lack of

coherency across the array at short periods, due in part to the large

aperture between stations (∼100 km).

The lack of surface wave constraints at the near-surface (i.e. less

than 15 s period) will not influence deeper parts in the model.

Some initial conclusions can be drawn from the raw phase ve-

locities projected onto a 2-D profile (Fig. 4). Lateral variations are

observed across the array, with phase velocities beneath the south-

ern stations 5–10 per cent slower than beneath northern stations

for periods shorter than 40 s. Slower velocities in the south are

in agreement with regional surface wave studies in the southwest

Pacific, which also see the same trend moving from oceanic crust to

the continental crust of New Zealand (Pillet et al. 1999). There are

two possible explanations for the change in phase velocities across

the array, perhaps in combination. If the crust is the same thickness

across the array, then upper mantle velocities must be lower in the

south. However, if the crust is thinner in the north, faster velocities

would be reached at shallower depths than in the south, where the

crust is thicker. Phase velocities derived for the southern end of the

NORD array (Fig. 3) are ∼0.2 km s−1 lower at periods greater than

20 s than phase velocities obtained by Brisbourne & Stuart (1998)

for interstation paths crossing eastern North Island, New Zealand,

which may be affected by the high-velocity, subducting slab.

3 R E C E I V E R F U N C T I O N S

3.1 Data

We searched the IRIS-DMC catalogue for earthquakes that lie within

25◦–100◦ epicentral distance of the NORD array and have M w >

6, and found 118 events which filled these criteria that were further

used in the receiver function analysis (Fig. 2). The epicentral range

was chosen to include events that arrive near vertically beneath the

station but to avoid core phases for events at larger epicentral dis-

tances. We use the term backazimuth to describe the angle from

north between the station and event. The majority of events had

backazimuths ranging from 270◦–020◦, originating in subduction

zones in the western Pacific (Fig. 2). Few events lie outside this

backazimuth range; events that do are usually at the high end of

the epicentral distance range and have low magnitudes, which re-

sults in poor-quality receiver functions. Events from the northwest

quadrant, however, have good epicentral coverage. Not all events are

recorded on every station due to noise levels on individual stations

and technical difficulties.

3.2 Multiple-taper correlation method

The P-wave receiver function method has become a popular tool

to investigate the shear wave structure beneath broad-band stations

using converted phases from interfaces (Langston 1979; Owen &

Zandt 1985; Ammon et al. 1990). The method assumes that a seismic

waveform is a product of three factors: source side, travel path and

receiver side effects. It is assumed that source and travel path effects

are removed from the waveform by rotating the components into the

Z–R–T coordinate system followed by deconvolution of the verti-

cal component from the horizontal components. This leaves only

the receiver response, characterized as a time-series that contains

phases that result from P-to-S conversions (Ps) at seismic disconti-

nuities beneath the station. In addition to primary Ps phases, we also

record secondary reflections (multiples) that are generated by rever-

berations from the free surface or another interface. Multiples arrive

later in the time-series and can mask deeper Ps phases. In terms of

primary Ps phases, the time-delay between the initial P wave and

a given Ps phase is a function of the depth of the interface and the

difference of P- and S-wave slownesses. The amplitude of the phase

is dependent on the velocity contrast across the interface and the

incident angle of the incoming wave. Positive phases are generated

by a velocity increase with depth (e.g. Moho) and a negative phase

is caused by a velocity decrease with depth (i.e. the top of a low-

velocity layer). Note that some multiples exhibit negative arrivals in

spite of increasing velocity. We use the radial receiver function to

determine the shear velocity structure beneath the station. Signifi-

cant coherent energy on the transverse receiver functions is caused

by out-of-plane energy, indicative of seismic anisotropy or dipping

interfaces (Cassidy 1992; Savage 1998).

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1471

We compute teleseismic P-wave receiver functions using the

multiple-taper correlation (MTC) method of Park & Levin (2000)

as an alternative to popular methods such as time domain decon-

volution (Gurrola et al. 1995; Abers 1998) and spectral division

(Ammon 1991). The MTC method computes receiver functions in

the frequency domain, avoiding numerical instabilities often associ-

ated with spectral division. The MTC method distinguishes between

coherent and incoherent scattering to reduce unwanted seismic en-

ergy. Once the receiver functions are computed, they are binned as

a function of backazimuth. Each bin is separated by 5◦ with a 10◦

overlap between bins so that each trace influences each adjacent bin.

An advantage of using the MTC method is that each receiver func-

tion is weighted by the inverse of its variance during bin formation

or stacking of multiple events. This positively biases towards the

best receiver functions and results in an improvement in multiple

event stacks, both for receiver function gathers and for formulating

summed traces that are used to represent stations (Park & Levin

2000). We compute receiver functions for all events recorded on

each station, applying a low-pass filter of 1 Hz during analysis.

4 J O I N T I N V E R S I O N F O R S H E A R

WAV E S T RU C T U R E

We determine the shear wave velocity as a function of depth using a

linearized joint inversion of radial receiver functions and Rayleigh

wave phase velocities for northwestern New Zealand, following the

method of Julia et al. (2000, 2003). A solution is obtained by a

damped least-squares method that smoothes velocities in adjacent

layers based on a priori constraints. An influence factor, p (0 ≤ p ≤1) is used to specify the weighting of receiver function and surface

wave data, where p = 0 is receiver function data only and p = 1 is

surface wave dispersion data only. We tested the influence of p and

determined that a value of p = 0.5 was robust and provided the best

fit to both data sets.

We base our starting velocity model on the velocity model ob-

tained by Stern et al. (1987) for northwestern New Zealand, which

consisted of a two-layer crust and a two-layer mantle. We invert for

the shear wave structure down to 100 km, as the vertical sensitivity

kernels of the surface waves sample to a depth of ∼200 km and our

receiver functions are sensitive to 100 km depth. Layer thicknesses

are set to 1 km in the upper 6 km and 2 km for the rest of the model.

We solve for Vs and determine Vp by keeping Vp/Vs fixed at 1.75

for the crust and at 1.8 for the mantle. Parts of the region have low

Pn wave speeds, indicating partial melt in the uppermost mantle

(Stern et al. 1987; Stratford & Stern 2005). This will subsequently

increase the true value of Vp/Vs and result in underestimation of

the Vp values.

Input for the joint inversion uses a stacked receiver function ob-

tained by summing all events from the northwest quadrant. We lack

a range of backazimuths to complete a thorough investigation of the

presence of anisotropy and dipping layers, thus we take a first-order

approach by determining the average velocity structure beneath each

station by stacking good quality events with backazimuths in the

northwest quadrant. This approach will tend to smooth the inter-

preted structure, since the receiver functions will be smoothed by

arrivals at slightly different times across the backazimuth range. The

corresponding surface wave data consist of dispersion curves from

each station triplet, with a sampling interval of 1 s between points

on the dispersion curve. The two edge stations are represented by

using the dispersion curves of the adjacent stations. We assume that

the structure is 1-D and isotropic, which is a reasonable assumption,

given that there is minimal shear wave splitting observed across the

array from SKS phases (Duclos 2005) and that there is only minor

energy observed on the transverse receiver functions (Figs 5–10).

The strength in the joint inversion lies in that fact that surface wave

data at the near-surface will not influence the rest of the model which

typically happens during forward modelling where the user takes a

down-model approach to fitting the waveform.

It is difficult to estimate errors during the inversion procedure

using a linear damped least-squares method (see Julia et al. 2000).

Therefore, we estimate errors by examining errors in surface wave

phase velocities and timing of receiver function phases. We first

invert for the shear structure by using three dispersion curves: the

observed dispersion and two obtained by adding and subtracting

the phase velocity error of 0.09 km s−1. The shear velocities from

the resulting three models vary on average by 0.12 km s−1. An er-

ror representing the misfit in timing of receiver function phases is

estimated to be ±0.05 s. We investigate the error in the depth of an

interface by using the relationship between the timing of a Ps phase

and the depth to the interface (Gurrola et al. 1994):

H = T Ps(V s−2 − p2

r

)0.5 − (V p−2 − p2

r

)0.5, (1)

where H is the depth to the interface, TPs is the delay time of the Ps,

Vs and Vp are the shear and compressional velocities of the layer.

The ray parameter p r takes into account variations from vertical ray

paths. We calculate the depth to the interface, using extreme values

of Vs (±0.12 km s−1) and TPs (±0.05 s) based on errors mentioned

previously. From these crude error calculations, shear velocities in

a given layer vary by ±0.12 km s−1 and the depth to an interface

varies by ±1 km.

4.1 Station TOZ

Station TOZ is located just west of the Central Volcanic Region,

the area of active extension in the central North Island. TOZ has

the largest data set of all of the stations, though the majority of

events still come from the northwest quadrant (Fig. 5). A strong

Ps phase arrives at 3.5 s delay time, which is interpreted to be

from the Moho. The timing indicates a shallow Moho, around 25–

30 km depth. There is also some coherent energy on the transverse

component for backazimuths 300◦–340◦ at 2–3 s delay time.

Fig. 5 shows the inversion results for station TOZ. We plot the

observed receiver functions and surface wave dispersion data as grey

lines. The predicted data are shown as a black line. In general, we

see a good fit between the two data sets; we note that the predicted

dispersion is smoothed compared to our observed data. Receiver

functions fit well but diverge slightly at delay times between 14 and

16 s. Low shear velocities of 3.2 km s−1 are observed to 4 km depth,

but sharply increase in the mid-crust to faster velocities of 3.5–

3.7 km s−1, which extend to 15 km depth. Lower velocities occur

from here to the Moho. A sharp Moho is seen at ∼29 km, where

shear wave velocities of 4.2 km s−1 are reached. This matches the Psphase at 3.5 s delay time well. Low upper mantle velocities extend

down to 100 km depth. The velocity model is remarkably similar to

shear velocity models found ∼50 km to the southeast, for stations

CAMV and MAMV, just inside the western boundary of the Central

Volcanic Region (Bannister et al. 2004).

4.2 Station MKAZ

MKAZ is located in the Hunua Ranges, a region where greywacke

basement outcrops. The Auckland Volcanic Field lies 30 km to the

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1472 N. A. Horspool, M. K. Savage and S. Bannister

Figure 5. Receiver function gathers and joint inversion analysis for station TOZ. Radial (R) and transverse (T) receiver functions are plotted as a function of

backazimuth. Backazimuth bins are 5◦ wide, with each receiver function influencing the adjacent bin. Amplitudes scales are the same for radial and transverse

components. Panel A: the best-fitting shear wave velocity model, with 1-km-thick layers for the top 6 km, then 2 km layers to 100 km depth. Panel B: synthetic

(black) and observed (grey) receiver functions with backazimuth range shown. Panel C: synthetic (black) and observed (grey) surface wave phase velocities.

northwest of MKAZ. The receiver function gather (Fig. 6) shows

a coherent Ps phase arriving at ∼3 s delay time, interpreted to be

from a shallow Moho. Some variability is noted for phases arriv-

ing later in the receiver functions but a positive–negative phase

with a large amplitude arriving at 9 s delay time is seen on mul-

tiple events. When plotted against epicentral distance (not shown

here), this phase exhibits normal moveout, indicating that it is a

primary conversion and not a multiple. There are some coher-

ent arrivals on the transverse component at 0–2 and 6–8 s delay

time.

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1473

Figure 6. Receiver function gathers and joint inversion analysis for station MKAZ. Details as in Fig. 5. An LVZ is present in the mantle causing the large

arrival on the receiver function at 9 s time.

Inversion results show low velocities of 2.5–3.0 km s−1 in the top

1–2 km (Fig. 6). Below this, a steady positive gradient in shear ve-

locities extends to 20 km depth. Upper mantle velocities are reached

at 27–29 km depth. On the receiver function, the negative–positive

arrival at 9 s delay time (Fig. 6) is modelled by an LVZ in the upper

mantle at 70–90 km depth with shear velocities as low as 4.0 km s−1.

These velocities are 15 per cent lower than typical continental upper

mantle velocities (Christensen & Mooney 1995). The positive phase

is generated from the bottom of the LVZ, while the negative phase

at 9 s derives from the top of the LVZ.

4.3 Station MATA

MATA is located in a sedimentary basin of Miocene age consisting

of alternating sandstones and mudstones. There is some variation

in the radial receiver functions with backazimuth for station MATA

(Fig. 7), with Ps phases arriving at 3–4 s delay time and some earlier

phases arriving at 0–2 s time over a varying range of backazimuths.

The transverse component is quite noisy with a semicoherent arrival

at 5–6 s time; this varies in amplitude, period and timing as a function

of backazimuth. The presence of anisotropy or dipping structures

could be causing the heterogeneity in the receiver functions.

Inversion results indicate a model with velocities decreasing from

3.6 to 3.2 km s−1 in the near-surface to 8 km depth (Fig. 7). Beneath

this a steady increase in velocities from 3.6 to 3.9 km s−1 is observed,

with a jump from crust to upper mantle velocities at 26–29 km depth.

Another step in upper mantle velocities is seen from 4.3 to 4.5 km s−1

at 45 km depth, with velocities subsequently gradually increasing

to 4.7 km s−1 at the base of the model. The observed and predicted

data fit well and again we see smoothing of the predicted dispersion

curve.

4.4 Stations WCZ and OUZ

Figs 8 and 9 show the receiver function gathers for stations WCZ and

OUZ. These two permanent stations are located near the Whangarei

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1474 N. A. Horspool, M. K. Savage and S. Bannister

Figure 7. Receiver function gathers and joint inversion analysis for station MATA. Details as in Fig. 5.

and Kaikohe-Bay of Islands Volcanic Fields, which consist of in-

traplate basaltic vents and rhyolite centres of Quaternary age. The

radial receiver functions for the stations are remarkably similar.

They both have a large Ps phase at ∼3.5 s delay time, interpreted

to be from the Moho. This phase is preceded by positive phase at

2–3 s delay time and a large amplitude negative phase at 1–2 s

delay time. The initial P wave exhibits no delay, indicating that

near-surface layers are not causing this strong negative arrival. It is

thought that the two arrivals before the Moho Ps originate from a

mid-crustal discontinuity; relative arrival times are consistent with

conversions from the top and bottom of an LVZ. For station WCZ,

there is also some energy arriving at 2 s delay time on the transverse

component.

The inversion results for stations WCZ and OUZ (Figs 8 and

9) are quite similar. WCZ is located near the Whangarei Volcanic

Field. The best-fitting velocity model for WCZ shows low velocities

of 3.0 km s−1 in the upper 1–2 km, which increase to 3.5 km s−1 at

∼11 km depth. Beneath this is an LVZ between 12 and 18 km depth,

with shear velocities as low as 3.0 km s−1. The LVZ is required to

be present to fit the large-amplitude negative arrival in the receiver

functions at 1–2 s delay time. The predicted receiver function un-

derestimates the magnitude of the negative arrival at 1–2 s time. As

a consequence, the velocities in the LVZ are probably overestimated

and are perhaps lower than 3.0 km s−1. Beneath the LVZ, velocities

increase and at 26–28 km depth jump to upper mantle velocities.

A gentle gradient is observed in the upper mantle, with velocities

reaching 4.5–4.6 km s−1 at 100 km depth.

The inversion results for OUZ show velocities of ∼3.0 km s−1 for

the near-surface increasing to 3.5–3.8 km s−1 down to 9 km depth

(Fig. 9). A large negative arrival on the receiver function at 1–2 s

time requires the model to have an LVZ between 10 and 19 km

depth with velocities of 2.9–3.0 km s−1, a 20–25 per cent reduction

from normal mid-crustal velocities. As with the WCZ model, the

best-fitting model for OUZ overestimates the velocities in the LVZ,

not quite fitting the full amplitude of the negative phase. Beneath

the LVZ a steady increase in shear velocities is observed down to

24–25 km depth. A jump to upper mantle velocities occurs at 26 km

depth. Upper mantle shear velocities then increase to a depth of

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1475

Figure 8. Receiver function gathers and joint inversion analysis for station WCZ. Details as in Fig. 5. An LVZ in the mid-crust is needed to fit the negative

arrival at 1–2 s time on the receiver function.

55 km where a slight decrease of 0.3 km s−1 occurs at 55–70 km

depth. We see good fit between both data sets for both stations.

4.5 Station TIKO

Station TIKO is located at the northern tip of New Zealand. The

station is situated on a thick sequence (∼1000 m) of obducted

oceanic sediments that form the Northland Allochthon (Herzer

& Isaac 1992) and also more recent sediments that make up the

Northland Basin (Hayward 1993). We observe a strong coherent Psarrival at 3 s delay time (Fig. 10). An earlier negative arrival at 2 s

time is also seen. However, unlike stations OUZ and WCZ, the ini-

tial P wave exhibits interference with a near-surface conversion that

causes it to broaden and be slightly delayed. This indicates that the

negative conversion is most likely from a multiple from an interface

between basement and overlying sedimentary rocks.

The initial P delay and the negative phase arriving at 2 s delay

time require the model to have low velocities of 2.6–2.9 km s−1 in

the upper 7 km (Fig. 10). Below this, velocities strongly increase to

10 km depth, to mid-crustal wave speeds of 3.5 km s−1. A large step

to upper mantle velocities occurs at 25–26 km depth, below which

velocities increase to 4.5–4.7 km s−1. A good fit is seen between

both data sets.

5 D I S C U S S I O N

5.1 Crust and upper mantle structure

We create a 2-D shear wave velocity model by interpolating between

individual 1-D profiles for each station (Fig. 11). A surface is created

using splines in tension with a tension factor of 1. We follow the

method of Bannister et al. (2004) by investigating the derivative of

the shear wave velocity with depth. Receiver functions are primarily

sensitive to velocity contrasts, so by observing the areas where the

shear velocity changes the most rapidly we can identify where the

receiver functions are most sensitive (Fig. 11). In general, our crustal

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1476 N. A. Horspool, M. K. Savage and S. Bannister

Figure 9. Receiver function gathers and joint inversion analysis for station OUZ. Details as in Fig. 5. An LVZ in the mid-crust is needed to fit the negative

arrival at 1–2 s time on the receiver function.

structure model varies with similar trends to the P-wave model of

Stern et al. (1987). Although their model involved a simple two-

layer crust, we observe similar lateral variations in the upper crust.

In the central region of our array (between OUZ and MATA), shear

velocities in the upper crust are 3.4–3.6 km s−1, while at the north-

ern and southern end velocities are lower at <3.2 km s−1 (Fig. 11).

Stern et al. (1987) modelled P-wave velocities of 5.9 km s−1 for

the central portion of our array and 5.3 km s−1 and 5.4 km s−1 for

the north and south end, respectively. The lower crust is remarkably

consistent across the profile with average shear wave velocities of

3.6–4.0 km s−1. This equates to a Vp/Vs of 1.72–1.76 for the upper

crust and 1.70–1.75 for the lower crust, using the P-wave velocities

of Stern et al. (1987). Low shear wave velocities of 4.2 km s−1 are

present in the uppermost mantle across the profile but extend to

greater depths (∼90 km) beneath TOZ. Upper mantle velocities of

4.2 km s−1 are ∼10 per cent less than the global average (Christensen

& Mooney 1995) and indicate a disturbed upper mantle. Velocities

are also lower than the 4.45 km s−1 obtained for the northwestern

North Island by Haines (1979) for Sn phases using earthquake travel-

times. It must be noted that the value Haines (1979) found was based

on very few measurements and that the ray paths most likely sample

slightly deeper mantle. A Vp/Vs for the uppermost mantle is 1.81 ±0.03 using the Pn velocities of Stern et al. (1987).

Localized features are present along the array. Beneath TIKO,

low-velocity layers of 2.5–2.9 km s−1 in the upper 6–8 km most

likely reflect a thick sequence of sediments, comprised of 1–2 km

of material from the Northland Allochthon (Herzer & Isaac 1992)

and the remainder from sediments in the Northland Basin. Tertiary

sediments associated with the Northland Basin, located 30 km south-

west of Cape Reinga, have been inferred to be up to 5 km thick from

gravity modelling (Woodward 1993). An LVZ is present beneath

OUZ and WCZ in the mid-crust at depths of 10–19 km, with shear

velocities as low as 2.8 km s−1. The top and bottom of the LVZ are

apparent in the velocity gradient plot (Fig. 11). The seismic refrac-

tion line of Stern et al. (1987) did not identify any LVZ in Northland;

however, their receiver spacing was very large (20–50 km) and it is

doubtful that they could identify such layers if present. In the upper

mantle beneath MKAZ, another LVZ is present at 70–90 km depth

and velocities drop to ∼4.0 km s−1, ∼15 per cent lower than normal

upper mantle velocities (Christensen & Mooney 1995).

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1477

Figure 10. Receiver function gathers and joint inversion analysis for station TIKO. Details as in Fig. 5. A thick sequence of low-velocity layers extends to

∼6 km depth.

5.2 Crustal thickness

There is often heated debate as to the depth of the Moho in tec-

tonically active regions of the Earth (Christensen & Mooney 1995).

Investigations using the same data set have even resulted in different

depths to the Moho in the Central Volcanic Region, New Zealand

(Harrison & White 2004; Stratford & Stern 2005) and the Sierra

Nevada, California (Jones et al. 1994; Savage et al. 1994). The un-

derlying question fuelling the debate is ‘what velocities signify that

the upper mantle has been reached?’ This seemingly simple ques-

tion is complicated by the fact that the upper mantle in regions that

have undergone extension or are adjacent to plate boundaries is usu-

ally hot with small amounts of partial melt (Stern 2002; Wiens &

Smith 2003). The presence of high temperatures and partial melt

decrease seismic velocities, with shear velocities being affected to

a larger degree than compressional velocities (Sato & Sacks 1989;

Hammond & Humphreys 2000). P and S velocities in the upper

mantle beneath northwestern New Zealand have been shown to be

lower than that would be expected for typical continental settings

(Stern et al. 1987; Bannister et al. 2004; Harrison & White 2004;

Stratford & Stern 2005).

Instead of assuming an upper mantle velocity and inferring that

the Moho lies where this velocity is reached, we determine the Moho

by examining the shear velocity gradient (Fig. 11), and interpret

the Moho to lie just below where the shear velocity changes the

most. This occurs at 29 ± 1 km beneath TOZ at the southern end

of the array and at 26 ± 1 km beneath TIKO at the northern end

(Fig. 11). The Moho depth ranges between these two values along

the profile, although the vertical gradient is not as strong beneath

MKAZ. Projecting the Moho from the gradient plot onto the velocity

profile, we obtain a velocity of 4.2 ± 0.1 km s−1 for the upper mantle

(Fig. 11, dotted line). The only other crustal scale investigations that

correspond directly to our array obtained a depth of 25 ± 2 km for the

Moho (Stern et al. 1987) for a seismic refraction line that ran along

the Northland Peninsula. However, their crustal thickness estimate is

a minimum because the presence of hidden high-velocity layers may

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1478 N. A. Horspool, M. K. Savage and S. Bannister

Figure 11. Top panel: shear wave velocity profile across the array formed by interpolation between each 1-D profile using splines in tension (tension factor = 1).

Note that horizontal sampling is sparse. Locations of intraplate basalt fields (black bars) are projected onto the profile at top: Kaikohe-Bay of Islands (KBIVF),

Whangarei (WVF) and Auckland (AVF). The triangles represent the locations of the broad-band stations. Key features include an LVZ in the mid-crust beneath

OUZ and WCZ and a broader LVZ in the upper mantle beneath MKAZ. Low upper mantle velocities extend to greater depths beneath TOZ. There is a sequence

of low-velocity materials beneath TIKO. The blacked dotted line indicates the position of the Moho based on the velocity gradient plot. Bottom panel: shear

wave velocity gradient with depth, down to 50 km depth. Blue colours represent an increase in shear velocity with depth, as would be expected at the Moho.

We can identify a Moho where the gradient is the highest along the array. This ranges from 29 km beneath TOZ to 26 km beneath TIKO. The LVZs beneath

OUZ and WCZ are also apparent by the presence of a negative velocity gradient at the top of the LVZ.

increase this value. Other studies located 30–50 km north of TIKO on

the Reinga Ridge, a continuation of Cape Reinga, have determined a

crustal thickness of 25 km based on modelling of marine gravity data

(Zhu & Symonds 1994; Herzer et al. 1997). A number of studies

have been conducted across the Central Volcanic Region, which

has its western margin only 30–40 km to the southeast of TOZ.

There is debate over the crustal thickness in the Taupo Volcanic

Zone, the active part of the Central Volcanic Region, with estimates

varying from 15–21 km (Stratford & Stern 2004, 2005) to 25–30 km

(Bannister et al. 2004; Harrison & White 2004). In the western part

of the Central Volcanic Region, values for the Moho range from

25 km from wide-angle seismic data and earthquake traveltimes

(Stern et al. 1987; Harrison & White 2004; Stratford & Stern 2005)

to 26–30 km from receiver function inversion (Bannister et al. 2004).

All of these estimates are similar to the 26–29 km obtained from

our joint inversion model.

5.3 Low-velocity zones

Joint inversion results reveal LVZs in the mid-crust and upper mantle

beneath northern New Zealand.

The mid-crustal LVZ is located beneath WCZ and OUZ, two sta-

tions that lie on or adjacent to intraplate basalt fields of Quaternary

age. A source region for the basalts is thought to lie in the man-

tle (Heming 1980; Smith et al. 1993). However, the presence of

rhyolite centres and an active geothermal system in the Kaikohe-

Bay of Islands Volcanic Field indicate that melting of the crust is

occurring and that there is the possibility of a magma body heating

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1479

Figure 12. Testing of MKAZ LVZ. Forward modelling is used to determine if the 9-s arrival is a multiple or a primary Ps phase. The 9-s arrival is causing

the inversion to return an LVZ in the upper mantle. Panel A: forward model of the final MKAZ velocity model without an LVZ at 70 km depth (black model)

compared to the inversion result (grey model). Grey trace is observed receiver function and black is synthetic from forward modelling. For the model without

the LVZ (black), the multiple from the Moho would arrive at ∼11 s, later than the 9-s phase, indicating it is not a multiple. Note the negative phase at ∼4 s on

the black trace is not present in the other trace (grey). This phase could be a Ps conversion generated by a Psp phase. This would arrive ∼2 s later than a given

Ps phase and hence complicating the waveform and explaining the arrival. Panel B: assuming a two-layer crust and mantle, a 21-km-thick crust is required

to match the 9-s phase as a multiple from the Moho. This is much thinner than that expected in this region. Based on these two forward models, we can be

confident of that the 9-s phase is a primary phase from a discontinuity in the upper mantle.

the geothermal field (Heming 1980). The LVZ is located at 10–

19 km depth, with shear velocities as low as 2.9 km s−1 (Fig. 11),

a ∼23 per cent reduction on normal crustal velocities. Large re-

ductions in shear velocities are usually caused by the presence of

fluids or melt (Sato & Sacks 1989; Hammond & Humphreys 2000).

LVZs have been identified in the mid-crust by teleseismic receiver

functions and been interpreted as melt bodies in Socorro (Sheetz

& Shlue 1992), the Andes (Chmielowski et al. 1999) and in the

Taupo Volcanic Zone (Bannister et al. 2004). Therefore, our LVZ

is interpreted to represent a body of partial melt in the crust. The

partial melt could be generated by a batch of basaltic magmas that

have migrated from the mantle into the crust, melting the lower crust

and forming alkalic rhyolite magma, thus explaining the presence

of rhyolite centres in the Kaikohe-Bay of Islands Volcanic Field.

A rhyolite intrusion could also be the heat source for the Ngawha

geothermal system near OUZ, as it is thought that the basalt magmas

do not reside in the crust for long enough to create a heat source. The

interpretation of a zone of partial melt in the mid-crust is consistent

with patterns of volcanism at the surface. Regional Bouguer gravity

data do not have any significant anomalies that would indicate par-

tial melt in the crust; however, the resolution of the data set is too

low to identify such bodies.

Beneath station MKAZ, an LVZ is present in the final model

at a depth of 70–90 km in the upper mantle (Figs 6 and 11), with

velocities of 4.0 km s−1. This LVZ is inferred from a large negative–

positive Ps arrival at 9 s delay time on the receiver functions (Figs 6

and 12). This time-delay is near where we would expect multiples

from the Moho to arrive. Therefore, we test whether this arrival is

a Ps arrival or instead just a multiple by using reflectivity forward

modelling of synthetic receiver functions (Fig. 12). We begin by

removing the LVZ from the final velocity model for MKAZ. By

doing this, we will observe when the Moho multiples should arrive

(Fig. 12a). The Ps phase from the Moho arrives at ∼3 s time but

the multiple (PpPms) from the Moho arrives at 11–12 s delay time,

indicating that the 9-s phase is not a multiple from the Moho in the

inversion model. Secondly, we determine the thickness of the crust

required to generate a multiple arriving at the same time as our 9-s

Ps phase. We use a simple two-layer model with a shear velocity in

the crust of 3.6 km s−1, and 4.3 km s−1 in the mantle (Fig. 12b). We

then modify the crustal thickness until we fit the Moho Ps phase at

3 s time and the 9-s phase. We need a 21-km-thick crust in order to

obtain a Moho multiple arriving at 9 s time for the simple velocity

model used. Based on these models, we can determine that this LVZ

is real, unless the crust has a thickness of 21 km which is much

thinner than that previously observed. It would also mean that the

crust would thin from 29 km beneath TOZ and 28 km beneath MATA

to 21 km beneath MKAZ, an unlikely scenario.

As a consequence of the incoming waves arriving at an angle

beneath the station, the receiver functions actually sample an area

surrounding the station. We backproject all of the receiver functions

recorded from the northwest of MKAZ to observe their piercing

points at 90 km depth, the base of the LVZ (Fig. 13). We plot the

piercing points for receiver functions as crosses if they have the 9-s

phase, or as circles if they do not (Fig. 6). The majority of the receiver

functions with the 9-s Ps phase pierce in or around the boundary

of the Auckland Volcanic Field (Sporli & Eastwood 1997). This

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1480 N. A. Horspool, M. K. Savage and S. Bannister

Figure 13. Piercing points at 90 km depth (base of LVZ) for receiver functions from the northwest quadrant at station MKAZ. The crosses indicate the

corresponding receiver function has an arrival at 9 s delay time (that is generating the LVZ), while the circles are corresponding receiver functions that do not

have the arrival. The elliptical boundary of the Auckland Volcanic Field (AVF) is depicted as a dashed line after Sporli & Eastwood (1997).

indicates that the LVZ is located not beneath MKAZ, but beneath

the Auckland Volcanic Field. Recent work suggests that tempera-

ture variations and melt content can cause velocity reductions in

the upper mantle; thus, the LVZ is thought to represent a melt-

producing region in the upper mantle (Hammond & Humphreys

2000).

In light of this work, we propose that our LVZ at 70–90 km depth

represents the source region for the Auckland Volcanic Field. The

location of the melt-producing zone has implications for the location

of future volcanism. Receiver functions that contain the Ps phase

from the LVZ mostly pierce directly beneath the Auckland Volcanic

Field (Fig. 13). However, a small number of receiver functions pierce

the base of the LVZ ∼10 km north of the present extent of volcan-

ism (Sporli & Eastwood 1997). If we assume that volcanism occurs

directly above the source region, then this implies that future vol-

canism could occur further north than the present boundary of the

field.

5.4 Limits on the lateral extent of mantle wedge

deformation below the Central Volcanic Region

We observe a vertical boundary between station TOZ and MKAZ

that marks a change in upper mantle velocities (Fig. 11). This is

consistent with several lines of evidence that indicate that a bound-

ary marking the extent of mantle deformation in the mantle wedge

lies around or near the western boundary of the Central Volcanic

Region. Pulford & Stern (2004) calculated buoyancy forces in the

mantle beneath North Island, based on rock-uplift histories. A buoy-

ancy force per unit area of 65 MPa is found beneath the Central

Volcanic Region and 35 MPa beneath northwestern North Island.

Stratford & Stern (2005) expressed the buoyancy force as density

and showed that there is a density anomaly of −70 km m−3 to a depth

of 80–100 km below the Central Volcanic Region and a −40 kg m−3

anomaly below northwestern North Island. They attributed half of

the anomaly to temperature differences and the remainder to partial

melt. Using Pn velocities for northwestern North Island (Stern et al.1987) and the Central Volcanic Region (Stratford & Stern 2005),

they estimated 1 and 2 per cent melt, respectively, based on estimat-

ing melt percentage from P-wave speed reductions (Hammond &

Humphreys 2000). Similarly relating our upper mantle shear wave

velocities to partial melt, we obtain 1 per cent beneath the array and

1–2 per cent beneath TOZ. However, the relation between percent

melt and geophysical properties is an area of active investigation,

and these values will change if new values are determined.

A swath of seismometers deployed across much of the central

North Island have been used to image the bulk structure of the

subduction zone beneath North Island by body wave tomography

(Reyners et al. 2005). The mantle wedge is identified as a region

with low Vp (7.5–7.9 km s−1) and high Vp/Vs (1.87) that extends to

a depth of ∼100 km. It is unclear how far north this zone extends

as the model has lower resolution north of the Central Volcanic

Region (Reyners et al. 2005). Converting the Vp found in the man-

tle back-arc wedge to Vs, we obtain a value of 4.0–4.25 km s−1 for

Vs in the mantle wedge. This is similar to the upper mantle shear ve-

locities we obtain beneath TOZ down to a depth of 90 km, indicating

that the processes causing the low velocities in the mantle back-arc

wedge are still occurring beneath TOZ. Since there is a change in

upper mantle velocities between TOZ and MKAZ, we propose that

the change in velocities at 40–90 km depth marks a boundary to the

extent of mantle deformation in the mantle wedge, approximately

100 km northwest of the active volcanic front. Lateral velocity

boundaries have been observed from body wave traveltime tomog-

raphy in other subduction zone settings, such as Japan (Zhao et al.1992) and Tonga (Zhao et al. 1997). It is difficult to compare the

extent of velocity anomalies in the mantle wedge between different

subduction systems. Variations in dip of the subducting slab and age

of the subduction zone contribute to the size of the mantle back-arc

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1481

wedge, making it difficult to suggest a standard model for the extent

of horizontal deformation.

Seismic anisotropy has been successfully used to infer in situ de-

formation in the mantle and crust by means of shear wave splitting

(Savage 1999). Fast orientations from SKS splitting are used to infer

the direction of mantle flow. Shear wave splitting from SKS phases

has been carried out on a number of stations in North Island (Marson-

Pidgeon et al. 1999; Hoffman 2002; Audoine et al. 2004; Duclos

2005). In North Island, nearly all stations exhibit trench-parallel

fast orientations (Marson-Pidgeon et al. 1999; Hoffman 2002;

Audoine et al. 2004), indicating trench-parallel flow above and

below the subducting slab (Savage 1999). Recent SKS splitting

results on the same stations used in this study show somewhat

different results (Duclos 2005). TOZ is similar to other stations

in North Island, with a NE–SW fast orientation, suggesting simi-

lar mantle deformation as in the mantle back-arc wedge. However,

the other stations exhibit E–W fast orientations, indicating that the

trench-parallel flow is not occurring beneath them. No anisotropy

is observed beneath MKAZ. This change in fast orientation implies

that a boundary to the trench-parallel mantle flow occurs near TOZ

(Duclos 2005).

The change in the SKS fast orientation (Duclos 2005), the pres-

ence of low shear velocities that extend to greater depth beneath TOZ

than the rest of northwestern North Island (Fig. 11) and a density

difference between the Central Volcanic Region and northwestern

North Island (Pulford & Stern 2004; Stratford & Stern 2005) all

indicate that a boundary exists near station TOZ, which marks the

extent of deformation associated with the mantle back-arc wedge to

the south.

6 C O N C L U S I O N S

We successfully constrain the shear wave velocity structure to 100

km depth beneath northwestern New Zealand by performing a joint

inversion of receiver functions and surface waves. A joint inversion

removes the limitations of each method and results in a more robust

velocity model.

Crustal thickness ranges from 26 ± 1 to 29 ± 1 km, thinning

towards the northern tip of New Zealand, near the boundary be-

tween oceanic and continental crust. The inferred thicknesses are in

agreement with previous seismic and marine gravity investigations.

Two stations located on or adjacent to Quaternary intraplate basalt

fields exhibit large-amplitude negative arrivals on the receiver func-

tions at 1–2 s delay time. Inversion results attribute these phases to

an LVZ at 10–19 km depth, with a 23 per cent reduction in shear

velocities. We infer that the LVZ represents partial melt in the mid-

crust, consistent with surface volcanic and geothermal features.

An LVZ is also present in the upper mantle at 70–90 km depth

beneath the Auckland Volcanic Field. Shear velocity reductions in-

dicate the LVZ may represent a melt-producing region in the upper

mantle. Such a zone is interpreted to represent the source region for

the intraplate basalts of the Auckland Field. This is the first time

such a source region has been directly imaged beneath the field but

is in agreement with hypothetical source zones based on geochemi-

cal and other geophysical data, and is in agreement with models for

a shallow upper mantle source rather than a deep seated plume.

Lateral changes in upper mantle shear velocities are observed

beneath the array. A boundary that marks a change from low to

normal upper mantle velocities ∼100 km northwest of the active

volcanic front may represent the northwestern extent of mantle de-

formation which occurs in the mantle wedge beneath the Central

Volcanic Region. This would imply that the deformation extends

50 km northwest of the western Boundary of the Central Volcanic

Region. A change in the mantle properties further to the northwest

is also supported by changes in seismic anisotropy fast orientations.

This first-order determination of the shear wave velocity structure

beneath northwestern New Zealand could be tested by deploying a

more dense network of receivers around the stations. This would

also allow the magnitude of anisotropy and dipping layers that are

affecting our models to be determined.

A C K N O W L E D G M E N T S

We are grateful to Chuck Ammon and Robert Herrmann for the use

of their COMPUTER PROGRAMS IN SEISMOLOGY suite and Jeff Park

and Vadim Levin for their MTC program. Figures were generated

using Generic Mapping Tools (GMT) of Wessel & Smith (1991).

Mathieu Duclos and Ken Gledhill helped with field work and initial

data processing. GeoNet is acknowledged for providing data from

their permanent broad-band seismometer stations. Helpful discus-

sions with Tim Stern, Mathieu Duclos, Rick Herzer and Audrey

Galve improved this study. Funding for the instruments was pro-

vided by the Planet Earth Fund, New Zealand Lotteries Board and

the Royal Society’s Marsden Grant. Additional financial support was

provided by the Foundation for Research, Science and Technology

(FRST).

R E F E R E N C E S

Abers, G.A., 1998. Array measurements of phases used in receiver function

calculations: importance of scattering, Bull. seism. Soc. Am., 87, 313–318.

Ammon, C., Randall, G. & Zandt, G., 1990. On the nonuniqueness of receiver

function inversions, J. geophys. Res., 95, 15 303–15 318.

Ammon, C., 1991. The isolation of receiver effects from teleseismic P wave-

forms, Bull. seism. Soc. Am., 81, 2504–2510.

An, M. & Assumpcao, M.S., 2004. Multi-objective inversion of surface

waves and receiver functions by competent genetic algorithm applied to

the crustal structure of the Parana Basin, SE Brazil, Geophys. Res. Lett.,31, doi:10.1029/2003GL019179.

Audoine, E., Savage, M.K. & Gledhill, K.R., 2004. Anisotropic structure

under a back-arc spreading region, the Taupo Volcanic Zone, New Zealand,

J. geophys. Res., 109, doi:10.1029/2003JB002932.

Ballance, P.F. & Sporli, K.B., 1979. Northland Allochthon, J. R. Soc. N.Z.,9, 259–275.

Bannister, S., Bryan, C.J. & Bibby, H.M., 2004. Shear wave velocity variation

across the Taupo Volcanic Zone, New Zealand from receiver function

inversion, Geophys. J. Int., 159, 291–310.

Baumont, D., Paul, A., Zandt, G. & Beck, S.L., 2001. Inversion of Pn travel

times for lateral variations of Moho geometry beneath the Central Andes

and comparison with the receiver functions, Geophys. Res. Lett., 28, 1663–

1666.

Brothers, R.N., 1982. Subduction regression and oceanward migration of

volcanism, North Island, New Zealand, Nature, 309, 688–670.

Brothers, R.N., 1986. Upper Tertiary and Quaternary volcanism and sub-

duction zone regression, North Island, New Zealand, J. R. Soc. N.Z., 16,275–298.

Brothers, R.N. & Delaloye, M., 1982. Obducted ophiolites of North Island,

New Zealand: origin, age, emplacement and tectonic implications for

Tertiary and Quaternary volcanicity, N.Z. J. Geol. Geophys., 25, 257–274.

Brisbourne, A.M. & Stuart, G.W., 1998. Shear-wave velocity structure be-

neath North Island, New Zealand, from Rayleigh-wave interstation phase

velocities, Geophys. J. Int., 133, 175–184.

Brune, J. & Dorman, J., 1963. Seismic waves and Earth structure in the

Canadian shield, Bull. seism. Soc. Am., 53, 167–210.

Cassidy, J.F., 1992. Numerical experiments in broadband receiver function

analysis, Bull. seism. Soc. Am., 82, 1453–1474.

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

1482 N. A. Horspool, M. K. Savage and S. Bannister

Challis, G.A., 1978. Volcanism and volcanic trends in the geological his-

tory of New Zealand, in The Geology of New Zealand, pp. 664–667, ed.

Suggate, R.P., Stevens, G.R. & Te Punga, M.T.

Chang, S.J., Baag, C.E. & Langston, C.A., 2004. Joint analysis of tele-

seismic receiver functions and surface wave dispersion using the genetic

algorithm, Bull. seism. Soc. Am., 94, 691–701.

Chmielowski, J., Zandt, G. & Haberland, C., 1999. The Central Andean

Altiplano-Puna magma body, Geophys. Res. Lett., 26, 783–786.

Christensen, N.I. & Mooney, W.D., 1995. Seismic velocity structure and

composition of the continental crust: A global view, J. geophys. Res., 100,9761–9788.

Cole, J., 1986. Distribution and tectonic setting of late Cenozoic volcanism

in New Zealand, in Late Cenozoic volcanism in New Zealand, R. Soc. N.Z.Bull., 23, pp. 7–20, ed. Smith, I.E.M.

DeMets, C.G., Gordon, R.G., Argus, D.F. & Stein, S., 1990. Current plate

motions, Geophys. J. Int., 101, 425–478.

DeMets, C.G., Gordon, R.G., Argus, D.F. & Stein, S., 1994. Effect of recent

revisions to the geomagnetic time scale on estimates of current plate

motions, Geophys. Res. Lett., 21, 2191–2194.

Duclos, M., 2005. Mantle tectonics of a plate boundary in New Zealand,

PhD thesis, Victoria University of Wellington, Wellington, New Zealand.

Eaton, D.W. & Cassidy, J.F., 1996. A relic Proterozoic subduction zone

in western Canada: new evidence from seismic reflection and receiver

function data, Geophys. Res. Lett., 23, 3791–3794.

Edbrooke, S., 2001. Geology of the Auckland area 1:250 000 geological

map 3, Institute of Geological and Nuclear Sciences, Lower Hutt, NewZealand, 1 sheet + 74p.

Galve, A., Sapin, M., Hirn, A., Diaz, J., Lepine, J.-C., Laigle, M. & Gallart,

J., 2002. Complex images of Moho and variation of Vp/Vs across the

Himalaya and South Tibet, from a joint receiver-function and wide-angel-

reflection approach, Geophys. Res. Lett., 29, doi:10.1029/2002GL015611.

Gurrola, H., Minster, J.B. & Owens, T., 1994. The use of velocity spectrum

for stacking receiver functions and imaging upper mantle discontinuities,

Geophys. J. Int., 117, 427–440.

Gurrola, H., Baker, F.G. & Minster, J.B., 1995. Simultaneous time-domain

deconvolution with application to the computation of receiver functions,

Geophys. J. Int., 120, 537–543.

Haines, A.J., 1979. Seismic wave velocities in the uppermost mantle beneath

New Zealand, N.Z. J. Geol. Geophys., 22, 245–257.

Hammond, W.C. & Humphreys, E.D., 2000. Upper mantle seismic wave

velocity: effects of realistic partial melt geometries, J. geophys. Res., 105,10 975–10 986.

Harrison, A.J. & White, R.S., 2004. Crustal structure of the Taupo Volcanic

Zone, New Zealand: stretching and igneous intrusion, Geophys. Res. Lett.,31, doi:10.1029/2004GL019885.

Hayward, B.W., 1987. Identity of igneous bodies beneath the continental

shelf of west Northland, New Zealand, N.Z. J. Geol. Geophys., 30, 93–95.

Hayward, B.W., 1993. The tempestuous 10 million year life of a double arc

and intra-arc basin - New Zealand’s Northland Basin in the early Miocene,

in South Pacific sedimentary basins, pp. 113–142, ed. Ballance, P.F.

Hayward, B.W. et al., 2001. K-Ar ages of early Miocene arc-type volcanoes

in northern New Zealand, N.Z. J. Geol. Geophys., 44, 285–311.

Heming, R.F., 1980. Patterns of Quaternary basaltic volcanism in northern

North Island, New Zealand, N.Z. J. Geol. Geophys., 23, 335–344.

Herrmann, R.B. & Ammon, C., 2002. Computer Programs in Seismology,Version 3.20, Department of Earth and Atmospheric Sciences, Saint Louis

University, Missouri.

Herzer, R.H., 1995. Seismic stratigraphy of a buried volcanic arc, Northland,

New Zealand and its implications for Neogene subduction, Mar. Petr.Geol., 12, 511–531.

Herzer, R.H. & Isaac, M.J., 1992. The size of the Northland Allochthon,

N.Z. Geol. Surv. Rec., 44, 37–42.

Herzer, R.H. et al., 1997. Seismic stratigraphy and structural history of the

Reinga Basin and it margins, southern Norfolk Ridge system, N.Z. J. Geol.Geophys., 40, 425–451.

Hoffman, S.D., 2002. Seismic anisotropy in the crust and mantle: a study at

the western edge of Central Volcanic Region, New Zealand, MSc thesis,Victoria University of Wellington, Wellington, New Zealand.

Hoke, L. & Sutherland, R., 1999. Mantle melting beneath New Zealand

revealed by helium isotopes in gas discharges, scale 1:2,000,000, version

1999.1, Institute of Geological and Nuclear Sciences Folio Series, 4.

Huang, Y., Hawkesworth, C., van Calsteren, P., Smith, I.E.M. & Black,

P., 1997. Melt generation models for the Auckland volcanic field, New

Zealand: constraints from U-Th isotopes, Earth planet. Sci. Lett., 149,67–64.

Isacc, S., 1996. Geology of the Kaitaia area 1:250 000 geological map 3,

Institute of Geological and Nuclear Sciences, Lower Hutt, New Zealand,

1 sheet + 44p.

Jones, C.H., Kanamori, H. & Roecker, S.W., 1994. Missing roots and mantle

drips: regional Pn and teleseismic arrival times in the southern Sierra

Nevada and vicinity, California, J. geophys. Res., 99, 4567–4601.

Julia, J., Ammon, C., Herrmann, R. & Correig, A., 2000. Joint inversion of

receiver function and surface wave dispersion observations, Geophys. J.Int., 143, 99–112.

Julia, J., Ammon, C. & Herrmann, R., 2003. Lithospheric structure of the

Arabian Shield from the joint inversion of receiver functions and surface-

wave group velocities, Tectonophysics, 371, 1–21.

Kear, D., 1994. A ‘least complex’ dynamic model for late Cenozoic volcan-

ism in the North Island, N.Z. J. Geol. Geophys., 37, 223–236.

Kear, D., 2004. Reassessment of Neogene tectonism and volcanism in North

Island, New Zealand, N.Z. J. Geol. Geophys., 47, 361–374.

Knopoff, L., 1972. Observation and inversion of surface wave dispersion,

Tectonophysics, 13, 497–519.

Kovach, R.L., 1978. Seismic surface waves and crust and upper mantle

structure, Rev. Geophys. Space Phys., 16, 1–13.

Langston, C.A., 1979. Structure under Mount Rainer, Washington, inferred

from teleseismic body waves, J. geophys. Res., 84, 4759–4762.

Lawrence, J.F. & Wiens, D.A., 2004. Combined receiver function and sur-

face wave phase-velocity inversion using a niching genetic algorithm:

application to Patagonia, Bull. seism. Soc. Am., 94, 977–987.

Li, X., Sobolev, S.V., Kind, R., Yuan, X. & Estabrook, Ch., 2000. A detailed

receiver function image of the upper mantle discontinuities in the Japan

subduction zone, Earth planet. Sci. Lett., 183, 527–541.

Malpas, J., Sporli, K.B., Black, P.M. & Smith, I.E.M., 1994. Northland ophi-

olite, New Zealand, and its implications for plate tectonic evolution of the

southwest Pacific, Geology, 20, 149–152.

Marson-Pidgeon, K., Savage, M.K., Gledhill, K.R. & Stuart, G., 1999. Seis-

mic anisotropy beneath the lower half of the North Island, New Zealand,

J. geophys. Res., 104, 20 277–20 286.

Mooney, H.M., 1970. Upper mantle inhomogeneity beneath New Zealand:

seismic evidence, J. geophys. Res., 75, 285–309.

Owen, T. & Zandt, G., 1985. The response of the continental crust-mantle

boundary observed on broadband teleseismic receiver functions, Geophys.Res. Lett., 12, 705–708.

Ozalaybey, S., Savage, M.K., Sheehan, A.F., Louie, J.N. & Brune, J.N., 1997.

Shear-wave velocity structure in the northern Basin and Range Province

from combined analysis of receiver functions and surface waves, Bull.seism. Soc. Am., 87, 183–199.

Park, J. & Levin, V., 2000. Receiver functions from multiple-taper spectral

correlation estimates, Bull. seism. Soc. Am., 90, 1507–1520.

Pavlis, G.L., 2003. Imaging the earth with passive seismic arrays, The Lead-ing Edge, 22, 224–231.

Pillet, R., Rouland, D., Roult, G. & Wiens, D.A., 1999. Crust and upper

mantle heterogeneities in the southwest Pacific from surface wave phase

velocities, Phys. Earth planet Inter., 110, 211–234.

Pulford, A. & Stern, T.A., 2004. Pliocene exhumation and landscape evolu-

tion of central North Island, New Zealand: the role of the upper mantle,

J. geophys. Res., 109, doi:10.1029/2003JB000046.

Rait, G.J., 2000. Thrust transport directions in the Northland Allochthon,

New Zealand, NZ J. Geol. Geophys., 43, 271–288.

Reyners, M., Eberhart-Phillips, D., Stuart, G. & Nishimura, Y., 2006. Imag-

ing subduction from the trench to 300 km depth beneath the central North

Island, New Zealand, with Vp and Vp/Vs, Geophys. J. Int., 165, 565–583.

Ritzwoller, M.H., Shapiro, N.M., Barmin, M.P. & Levshin, A.L., 2002.

Global surface wave diffraction tomography, J. geophys. Res., 107,doi:10.1029/2002JB001777.

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from

Lithospheric Structure in NW New Zealand 1483

Salmon, M.L., Bannister, S., Bibby, H., Savage, M.K. & Stern, T.A., 2003.

Attenuation and electrical resistivity in an asymmetric back-arc exten-

sional environment, Eos Trans. AGU, Fall Meet. Suppl., 84, Abstract

S22B-0450.

Sato, H. & Sacks, I.S., 1989. The use of laboratory velocity data for es-

timating temperature and partial melt fraction in the low velocity zone:

comparison with heat flow and electrical conductivity studies, J. geophys.Res., 94, 5689–5704.

Savage, M.K., 1998. Lower crustal anisotropy or dipping boundaries? Effects

of receiver functions and a case study in New Zealand, J. geophys. Res.,103, 15 069–15 087.

Savage, M.K., 1999. Seismic anisotropy and mantle deformation: what have

we learned from shear wave splitting, Rev. Geophys., 37, 65–106.

Savage, M.K., Li, L., Eaton, J.P., Jones, C.H. & Brune, J.N., 1994. Earthquake

refraction profiles of the root of the Sierra Nevada, Tectonics, 13, 803–817.

Sheetz, K.E. & Shlue, J.W., 1992. Inferences for the Socorro magma

body from teleseismic receiver functions, Geophys. Res. Lett., 19, 1867–

1870.

Smith, I.E.M., Ruddock, R. & Day, R., 1989. Miocene arc-type vol-

canic/plutonic complexes of the Northland Peninsula, New Zealand, in

Geology of Northland: accretion, allochthons and arcs at the edge of theNew Zealand micro-continent, Vol. 26, pp. 205–213, eds Sporli, K.B. &

Kear, D., R. Soc. N.Z. Bull.

Smith, I.E.M., Okada, T., Itaya, T. & Black, P.M., 1993. Age relationships and

tectonic implications of late Cenozoic basaltic volcanism in Northland,

New Zealand, N.Z. J. Geol. Geophys., 36, 385–393.

Spetzler, J., Trampert, J. & Sneider, R., 2001. Are we exceeding the limits

of the great circle approximation in global surface wave tomography?,

Geophys. Res. Lett., 28, 2341–2344.

Spetzler, J., Trampert, J. & Sneider, R., 2002. The effect of scattering in

surface wave tomography, Geophys. J. Int., 149, 755–767.

Sporli, K.B., 1989. Tectonic framework of Northland, New Zealand, in Geol-ogy of Northland: accretion, allochthons and arcs at the edge of the NewZealand micro-continent, Vol. 26, pp. 3–14, eds Sporli, K.B. & Kear, D.,

R. Soc. N.Z. Bull.

Sporli, K.B. & Eastwood, V., 1997. Elliptical boundary of an intraplate vol-

canic field, Auckland, New Zealand, J. Volc. Geotherm. Res., 79, 169–179.

Stern, R.J., 2002. Subduction Zones, Rev. Geophys., 40, doi:10.1029/

2001RG000108.

Stern, T.A., 1985. A back-arc basin within continental lithosphere; the

Central Volcanic Region of New Zealand, Tectonophysics, 112, 385–

409.

Stern, T.A., Smith, E.G.C., Davey, F.J. & Muirhead, K.J., 1987. Crustal and

upper mantle structure of northwestern North Island, New Zealand, from

seismic refraction data, Geophys. J. R. astr. Soc., 91, 913–936.

Stratford, W.R. & Stern, T.A., 2004. Strong seismic reflections and melts

in the mantle of a continental back-arc basin, Geophys. Res. Lett., 31,doi:10.1029/2004GL019232.

Stratford, W.R. & Stern, T.A., 2006. Crust and mantle structure of a conti-

nental back-arc: central North Island, New Zealand, doi:10.1111/j.1365-

246X.2006.02967.x.

Walcott, R.I., 1978. Tectonics and late Cenozoic evlolution of New Zealand,

Geophys. J. R. astr. Soc., 52, 137–164.

Wessel, P. & Smith, W.H.F., 1991. Free software helps map and display data,

EOS, Trans. Am. geophys. Un., 72, 445–446.

West, M., Ni, J., Baldridge, S., Wilson, D., Aster, R., Gao, W. & Grand,

S., 2004. Crust and upper mantle shear wave structure of the southwest

United States: Implications for rifting and support for high elevation, J.geophys. Res., 109, doi:10.1029/2003JB002575.

Wiens, D.A. & Smith, G.P., 2003. Seismological constraints on the

structure and flow patterns within the mantle wedge, in Inside theSubduction Factory, Vol. 138, ed. Eiler, J., Geophysical Monograph,

doi:10.1029/138GM05,

Woodward, D.J., 1970. Interpretation of gravity and magnetic anomalies at

Hokianga, Northland, N.Z. J. Geol. Geophys., 13, 364–369.

Zandt, G. & Ammon, C.J., 1995. Continental crust composition constrained

by measurements of crustal Poissons ratio, Nature, 374, 152–154.

Zhao, D.A., Hasegawa, A. & Horiuchi, S., 1992. Tomographic imaging of

P and S wave velocity structure beneath northeastern Japan, J. geophys.Res., 97, 19 909–19 928.

Zhao, D.A., Xu, Y., Wiens, D.A., Dorman, L., Hildebrand, J. & Webb, S.,

1997. Depth extent of the Lau back-arc spreading center and its relation-

ship to the subduction process, Science, 278, 254–257.

Zhu, H. & Symonds, P.A., 1994. Seismic interpretation, gravity modelling

and petroleum potential of the southern Lord Howe Rise region, N.Z.Petroleum. Conf. Prec., 83, 93–100.

Zhu, L. & Kanamori, H., 2000. Moho depth variation in southern California

from teleseismic receiver functions, J. geophys. Res., 105, 2969–2980.

C© 2006 The Authors, GJI, 166, 1466–1483

Journal compilation C© 2006 RAS

by guest on March 29, 2013

http://gji.oxfordjournals.org/D

ownloaded from