You are on page 1of 16

New Zealand Journal of Geology and Geophysics

ISSN: 0028-8306 (Print) 1175-8791 (Online) Journal homepage: http://www.tandfonline.com/loi/tnzg20

High-resolution view of active tectonic


deformation along the Hikurangi subduction
margin and the Taupo Volcanic Zone, New Zealand
LL Dimitrova, LM Wallace, A J Haines & CA Williams
To cite this article: LL Dimitrova, LM Wallace, A J Haines & CA Williams (2016) High-resolution
view of active tectonic deformation along the Hikurangi subduction margin and the Taupo
Volcanic Zone, New Zealand, New Zealand Journal of Geology and Geophysics, 59:1, 43-57,
DOI: 10.1080/00288306.2015.1127823
To link to this article: http://dx.doi.org/10.1080/00288306.2015.1127823

View supplementary material

Published online: 08 May 2016.

Submit your article to this journal

Article views: 62

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tnzg20
Download by: [149.202.47.181]

Date: 31 May 2016, At: 08:38

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS, 2016


VOL. 59, NO. 1, 4357
http://dx.doi.org/10.1080/00288306.2015.1127823

RESEARCH ARTICLE

High-resolution view of active tectonic deformation along the Hikurangi


subduction margin and the Taupo Volcanic Zone, New Zealand
LL Dimitrovaa, LM Wallacea, AJ Hainesb and CA Williamsc
Institute for Geophysics, University of Texas at Austin, Austin, Texas, USA; bGNS Science, Dunedin, New Zealand; cGNS Science, Lower Hutt,
New Zealand

Downloaded by [149.202.47.181] at 08:38 31 May 2016

ABSTRACT

ARTICLE HISTORY

Understanding the mechanisms and dynamics of deformation at plate boundaries requires


high-resolution images of strain and deformation sources. Vertical derivatives of horizontal
stress (VDoHS) rates are the horizontal-component surface manifestation of all subsurface
deformation sources, and have substantially higher spatial resolution than GPS velocities or
strain rates. We calculate VDoHS rates from GPS data at the Hikurangi subduction margin.
We evaluate our results in the context of interseismic coupling on the subduction interface
and upper plate deformation processes. Instead of the expected rifting signal we nd strong
contraction within the Taupo Volcanic Zone, indicating non-tectonic effects from magmatic
and/or hydrothermal processes. Differences between SHmax directions from historical
seismicity (Townend et al. 2012) and our maximum contraction directions demonstrate that
at the Hikurangi margin historical seismicity and active faulting are strongly controlled by
long-term processes, such as rotation of the forearc, rather than short-term, elastic strain
from interseismic subduction interface locking.

Received 13 June 2015


Accepted 11 November 2015

Introduction
High-resolution measurements of contemporary
crustal deformation to determine strain rates, fault
slip rates and degree of interseismic coupling on faults
are fundamental to understanding the seismic risk at
plate boundaries. GPS velocities are one of the primary
sources of information available to constrain the distribution of active plate boundary deformation and the
dynamics and mechanics of continental plate boundaries. At subduction zones, where GPS networks are
near a potentially locked subduction interface, the
resultant velocities elds are complex and can reect
elastic strain due to interseismic coupling on the
subduction interface, as well as upper plate faulting
processes (e.g. McCaffrey et al. 2000; Wallace et al.
2004).
Two of the most popular ways to interpret complex
interseismic velocity elds at plate boundaries are: (1)
an elastic block modelling approach (e.g. McCaffrey
2002; Wallace et al. 2004; Meade & Hager 2005); and
(2) a strain-rate mapping approach (e.g. Haines 1982;
Haines & Holt 1993; Beavan & Haines 2001). Block
models break a region into a series of elastic tectonic
blocks, and t GPS velocities with a combination of
tectonic block rotation (typically using poles of
rotation) and interseismic locking on block-bounding
faults. However, surface velocities are a long-wavelength response to deformation sources and therefore
provide low-resolution information, which inherently

Forearc extension; GPS


velocity; Hikurangi margin;
North Island New Zealand;
maximum contraction
directions; subduction
interface locking; strain rates;
Taupo Volcanic Zone; VDoHS
rates

limit the ability of block models to use GPS velocities


to uniquely resolve the sources of deformation in
zones of complex deformation and/or closely spaced
faults (e.g. see discussion in Wallace et al. 2007). Strain
rates, being a spatial derivative of velocities, provide
higher-resolution information than GPS velocities
alone, and can provide a more objective description
of the location of plate boundary deformation than
block models. However, current methods to calculate
strain rates (Haines 1982; Haines & Holt 1993; Holt
& Haines 1995; England & Molnar 1997; Flesch &
Kreemer 2010; Parsons & Thatcher 2011; Kreemer
et al. 2012) are subject to misinterpretations due to
requiring purely mathematical and possibly non-physical optimisation and smoothing. Furthermore, they
typically target a subjectively chosen mist-per-observation value and, as a result, when more observation
points are added the match to dense GPS data does
not improve.
Haines et al. (2015) show that vertical derivatives of
horizontal stress (VDoHS) rates, which can be derived
from GPS surface velocities, yield the highest-resolution image of how the surface elastic layer is deformed
by subsurface sources. Evaluating the crustal deformation eld in terms of VDoHS rates may provide a
more objective, higher-resolution alternative to block
and strain-rate models. Mathematically, the VDoHS
rates are spatial derivatives of stress rates, and hence
provide an order of magnitude higher resolution than

CONTACT Lada L Dimitrova


lada@post.harvard.edu
Supplementary data available online at www.tandfonline.com/10.1080/00288306.2015.1127823
2016 The Royal Society of New Zealand

KEYWORDS

Downloaded by [149.202.47.181] at 08:38 31 May 2016

44

LL DIMITROVA ET AL.

strain rates in the same way that strain rates, being a


spatial derivative of velocities, are an order of magnitude higher resolution than velocities. The Bayesian
inversion approach used by Haines et al. (2015) also
removes the need for spatial smoothing and for subjectively chosen mist-per-observation values. When the
high-precision VDoHS rates are integrated to produce
strain rates, they therefore result in a higher-spatialresolution map of strain rates than has been previously
possible.
The North Island and the northern part of the South
Island of New Zealand occupy the boundary zone
between the obliquely converging Australian and Pacic plates. The Pacic Plate underthrusts the North
Island along the Hikurangi subduction zone, and crustal deformation there is strongly affected by interseismic coupling on the subduction interface (Darby &
Beavan 2001; Wallace et al. 2004) and strike-slip faulting in the upper plate (Mouslopoulou et al. 2007a,
2007b; Little et al. 2009, 2010). A dense (520 km average spacing) GPS network has been measured in the
North Island since the mid-1990s (see Beavan et al.
2016), providing an ideal testing ground to apply the
Haines et al. (2015) methodology to a complexly
deforming active subduction margin and compare the
results with previous elastic block and strain-rate
models.
In this paper, we apply the method of Haines et al.
(2015) to averaged GPS velocity data spanning 1995
2013 (Beavan et al. 2016). We derive surface strainrate maps and VDoHS-rate elds for the North Island
and the northern part of the South Island, New Zealand
to identify the major deforming zones. Instead of the
expected rifting signal, we nd strong contraction
within a portion of the normally extensional Taupo
Volcanic Zone (see Figure 1 for locations mentioned
in this paper), indicating that the magmatic and/or
hydrothermal system must strongly inuence crustal
deformation in large portions of the TVZ. We use
the areal strain and VDoHS results to evaluate the
possible distribution of interseismic coupling and aseismic creep on the subduction interface. We also nd
that when the subduction interface locking effect
(determined from block modelling) is removed from
the velocity eld (and resulting strain-rate and
VDoHS-rate elds), SHmax directions from historical
seismicity (Townend et al. 2012) agree well with the
maximum contraction directions from the model. In
contrast, maximum contractions from the uncorrected
(raw) velocity eld disagree strongly with earthquake
SHmax directions and the orientation of upper plate
strike-slip faults. This demonstrates that at the Hikurangi margin the historical seismicity and active faulting
are more strongly modulated by long-term deformation processes rather than the short-term (e.g. hundreds of years) elastic component of strain induced by
subduction interface locking.

North Island and northern South Island


tectonic setting and active faulting
The North Island of New Zealand straddles the boundary zone between the obliquely converging Pacic and
Australian plates (Figure 1). The Pacic Plate subducts
westwards beneath the eastern North Island along the
Hikurangi Trough. The oblique relative plate motion
is partitioned, in that the margin-normal component
largely occurs along the subduction thrust, while the
margin-parallel component is accommodated via dextral strike-slip faulting and clockwise rotation of the
forearc (Beanland & Haines 1998; Nicol & Beavan
2003; Wallace et al. 2004). Rapid clockwise rotation of
the forearc occurs at 34 Ma1 (Walcott 1984; Lamb
1988, 2011; Thornley 1996; Wallace et al. 2004; Nicol
et al. 2007), and leads to a northwards increase in convergence rates from 20 mm a1 at southern Hikurangi to
60 mm a1 at northern Hikurangi (Wallace et al. 2004).
Rifting in the Taupo Volcanic Zone (TVZ) of the central
North Island is also a consequence of forearc rotation. In
the South Island subduction gives way to a strike-slipdominated system, with the Marlborough fault system
and the Alpine fault accommodating most of the AustraliaPacic relative motion (e.g. Van Dissen & Yeats
1991; Holt & Haines 1995; Beavan et al. 1999; Norris
& Cooper 2000).
Interpretation of campaign GPS data (using an
elastic block modelling/backslip approach) in the
North Island has shown that the subduction interface
at the southern Hikurangi margin undergoes interseismic coupling to depths of 3040 km, while there is an
abrupt transition to an aseismic-creep-dominated
interface near c. 40S (Wallace et al. 2004, 2012a)
(Figure 1A). This result is also conrmed by Lamb
and Smith (2013) who use an alternative approach
to estimate Hikurangi locking, and the earlier result
of Darby & Beavan (2001) demonstrating deep
(c. 3040 km depth) locking at southern Hikurangi.
Episodic slow-slip events have been identied and
located at the Hikurangi margin with observations
from the GeoNet continuous GPS network (www.
geonet.org.nz). SSEs are often observed to follow
along the downdip edges of interseismic coupling
(Wallace & Beavan 2010), although there are some
exceptions to this (Wallace et al. 2012b; Wallace &
Eberhart-Phillips 2013).
The primary strike-slip faults in the forearc of the
Hikurangi margin include the North Island Dextral
Fault Belt (comprising the Wellington, Mohaka and
Ruahine faults) and the Wairarapa Fault. The NIDFB
slips at c. 6 mm a1 (Van Dissen & Berryman 1996),
while the Wairarapa Fault slips at 712 mm a1 (Van
Dissen & Berryman 1996; Little et al. 2009). Other
less well-characterised, but lower-slip-rate dextral
strike-slip faults also exist, such as the Ohariu Fault
(Heron et al. 1998) and the Masterton, Carterton and

Downloaded by [149.202.47.181] at 08:38 31 May 2016

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

45

Figure 1. A, Tectonic setting with faults from the GNS active fault catalogue (solid grey lines), grey dashed lines show depth contours to the top of the subduction interface (from Williams et al. 2013) and slow slip event contours (red lines). Long-term convergence rates at the trench (in mm a1; from Wallace et al. 2012a) show motion of the subducting Pacic Plate relative to the forearc.
B, GPS stations and velocity eld (relative to a xed Australian Plate; from Beavan et al. 2016) averaged over the 19952003 time
period used in this study.

Mokonui faults (Zachariasen et al. 2000; Townsend


et al. 2002). The total rate of strike-slip in the upper
plate is thought to decrease from c. 21 mm a1 at the
southern Hikurangi margin to c. 6 mm a1 at the
northern Hikurangi margin (Beanland & Haines
1998; Mouslopoulou et al. 2007a, 2007b). A small
amount of convergence (less than a few millimetres
per year in total) is accommodated along some reverse
structures in the upper plate such as the KapitiManawatu Fault System (offshore from the Kapiti coast), and
the Martinborough fault, among others (Figure 1; Nicol
et al. 2002; Nicol & Beavan 2003; Lamarche et al. 2005;
Litcheld et al. 2007; Nicol & Wallace 2007).
Highly active normal faulting is observed in the
Taupo Volcanic Zone of the central North Island (see
Villamor & Berryman 2001), which undergoes intraarc rifting at rates ranging from a few millimetres per
year in the southern TVZ (Villamor & Berryman
2006) to as high as 15 mm a1 in the northern TVZ
near the Bay of Plenty coastline (Wallace et al. 2004;
Lamarche et al. 2006). Normal faulting is also observed
at a number of locations within the forearc (Cashman &
Kelsey 1990; Chanier et al. 1999; Berryman et al. 2009;
Pettinga 2004), despite the fact that most of the Hikurangi forearc deformation is dominated by transpression.
Most of the forearc extension is localised to the rapidly
uplifting Raukumara Ranges (Berryman et al. 2009) and
in the Maraetotara Plateau area (Cashman & Kelsey
1990; Pettinga 2004) (Figure 1). Extension within the
forearc is thought to be related to gravitational collapse
(e.g. Raukumara Peninsula; Reyners et al. 1999) or largescale, deep-seated landsliding (e.g. Maraetotara Plateau;

Pettinga 2004). The Carterton, Masterton and Mokonui


faults (Figure 1) also appear to have a possible normal
faulting component in addition to dextral strike-slip
(Zachariasen et al. 2000; Townsend et al. 2002; Langridge et al. 2003).
The Marlborough Fault System in the northern South
Island is dominated by four strike-slip faults that take up
most of the relative plate motion (e.g. Van Dissen &
Yeats 1991; Holt & Haines 1995; Figure 1). These faults
strike approximately parallel to the Australia/Pacic
relative plate motion vector, leading to a predominance
of right-lateral strike-slip deformation. Where the Marlborough faults take on a more northerly strike, the
deformation involves a greater convergent component
(e.g. Jordan Thrust; Van Dissen & Yeats 1991). The
Awatere, Clarence and Wairau faults each have slip
rates over the range 48 mm a1 (Benson et al. 2001;
Mason et al. 2006; Zachariasen et al. 2006; Van Dissen
& Nicol 2009), while the Hope Fault is slipping at 13
23 mm a1 or more (Van Dissen & Yeats 1991; Langridge et al. 2003; Langridge & Berryman 2005). Slip
on faults in the Marlborough Fault System is eventually
transferred southwards onto the predominantly dextral
strike-slip Alpine Fault in the central South Island
(Van Dissen & Yeats 1991; Holt & Haines 1995).

Methodology
To investigate the details of crustal deformation in the
North Island and the northern part of the South Island
of New Zealand, we use the methodology of Haines
et al. (2015) to solve the horizontal-component

46

LL DIMITROVA ET AL.

Figure 2. Schematic explanation of A, VDoHS rates, and B,


schematic cross-sectional example of expected VDoHS rate patterns above a locked subduction zone.

observations are made, the interseismic behaviour is


purely elastic; that is, at the surface the rocks are cold
and the rheology is much less complicated than
throughout the rest of the crust. At depth, velocities
are related to total stresses because ductility comes
into play, while at the surface velocities are only related
to rates of change of stress. A simple consequence of
this assumption is that the horizontal-component
manifestations at the surface of any subsurface deformation source can be encapsulated in the form of an
apparent source at the surface that can be deduced
from surface observations of horizontal velocity using
elasticity theory alone. We briey outline the methodology in the at-Earth case; details of the full methodology are covered in Haines et al. (2015).

Downloaded by [149.202.47.181] at 08:38 31 May 2016

Force-balance equations at the Earths surface


force-balance equations at the Earths surface to obtain
the Vertical Derivatives of Horizontal Stress (VDoHS)
rates from the GPS velocities. The VDoHS rates
are dened as the shear traction rates applied at the
base of the innitesimally thin surface elastic layer
divided by the thickness of the innitesimally thin
layer.
The VDoHS rates show in which direction an innitesimally thin surface elastic layer is being deformed by
subsurface sources (Figure 2A), and are the most
tightly resolved horizontal surface expressions of processes at depth. As for GPS velocities the VDoHS
rates are vectors; however, unlike GPS velocities, the
VDoHS rate vectors are tightly localised above the
source and point in the direction of the pull or push
by the sources. For example, we show schematically
the VDoHS-rate vectors above a partially locked subduction zone (Figure 2B). The VDoHS rates above
the locked subduction interface point in the direction
of subduction of the subducting plate, while the
VDoHS rates above the unlocked (slipping) subduction
interface point in the opposite direction. The boundary
where the VDoHS rates reverse directions lies directly
above the boundary between the locked and unlocked
portion of the subduction interface. Another example
is an inating magma chamber; at the surface directly
above the source the VDoHS rate vectors will point
radially away from the centroid of the source (see
Haines et al. 2015).
Another advantage of our new method is that the
corresponding strain rates can then be calculated by
integrating the VDoHS rates. This approach produces much higher-resolution and more-robust
strain rates than other widely used but non-physical
methods of calculating strain rates (e.g. Haines &
Holt 1993; Kreemer et al. 2000, 2003; Beavan &
Haines 2001).
A key assumption in the Haines et al. (2015) method
is that at the surface of the Earth, where GPS

At the Earths surface, the traction rates s xz , s yz are


zero and their vertical derivatives appear in the rateof-change forms of the horizontal-component forcebalance equations:
s xa s ya s za
+
+
= 0 for a = x, y.
x
y
z

(1)

The rst two terms depend on the GPS velocities


and the surface layer rheology only, while the latter
term is the vertical derivative of horizontal stress
(VDoHS) rates:
s az
f a =
for a = x, y.
z

(2)

For an innitesimally thin surface elastic layer


uy
ux
with Lame constants l, m, s zz = l
+l
+
x
y
uz
(l + 2m)
= 0.
z
Using the resulting expression for uz /z, the horizontal-component force-balance equations at the surface become


2m ux
2mn uy
+
x 1 n x 1 n y
(3)
 


ux uy

+
+ f x = 0,
m
+
y
y
x
 


ux uy
+
m
x
y
x
(4)


2mn ux
2m uy

+
+ f y = 0,
+
y 1 n x 1 n y

l
is the Poisson ratio. These
2(l + m)
equations are simply the equations for plane stress
with zero tractions at the surface (e.g. Malvern 1969;
Turcotte & Schubert 2002; Bower 2009) and the
VDoHS-rate term explicitly included.
where n =

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

We have therefore converted the horizontal-component force-balance equations at the Earths surface
into 2D second-order partial differential equations
for the horizontal velocities ux , uy in terms of the
VDoHS rates surface quantities f x , f y . As shown in
Haines et al. (2015), the ux , uy values are much
smoother and more widely distributed spatially than
the f x , f y values, whereas the strain rates


uy
ux
1 ux uy
, exy =
+
, e yy =
have interexx =
2 y
x
x
y
mediate smoothness. In other words, f x , f y values provide the highest-resolution surface expression of
deformation due to subsurface sources.

Downloaded by [149.202.47.181] at 08:38 31 May 2016

Inversion process
We follow Haines et al. (2015) to solve the horizontal
force-balance equations at the Earths surface
(Equations 3, 4). We start by constructing an initial
coarse grid with vertices at the GPS stations. This is
done to ensure that the VDoHS rates will be dependent
on only nearby values of observed GPS velocities. To
avoid singularity of the velocity matrix, if GPS stations
are closer than a minimum spacing, we replace the sites
with a single vertex at the average location of the GPS
stations weighted by the inverse of the velocity variances (e.g. Figure S1). We refer to the resultant vertices
as function sites. For each function site i we create a
VDoHS-rate basis function wi (x), with value 1 at the
function site and value 0 at the other function sites.
Basis functions are typically used in expanding general
functions, and a linear combination of the basis functions represents every function in the function space.
In our case, we create the VDoHS basis functions
based on the Argyris shape functions, and we solve
the force-balance equations in the forwards calculation
for the corresponding velocity functions (see Figures
S2, S3 for an example of a basis VDoHS rate function
and the corresponding velocity response). These velocity functions are in turn used to expand the GPS velocity eld and the inversion aims to nd the
coefcients multiplying the velocity functions that
best t the GPS velocity observations. Note that the velocity functions are longer wavelength compared to the
VDoHS basis functions.
Argyris elements are used in these calculations to
ensure continuous rst derivatives of the VDoHS
rates at element boundaries. Argyris elements have 21
degrees of freedom per element, which are reduced to
one per component of VDoHS rate at the GPS station
nodes (function sites) by using minimisation of integrals of squared magnitudes of derivatives to specify
the other degrees of freedom in terms of the degrees
of freedom we retain, that is, the VDoHS-rate values
at the function sites (Haines et al. 2015). This results

47

in the minimally variable basis functions given the


boundary conditions (see Figure S2). The methodology
allows us to prescribe: reecting boundaries for regions
where deformation continues through the boundaries;
traction-free boundaries for regions near trench edges;
rigid boundaries for regions at the edge of a rigid plate;
and general velocity boundaries.
We rene the original coarse grid using Rupperts
algorithm, resulting in additional vertices or Ruppert
points, and calculate the components ux and uy of
the surface horizontal velocity functions uai (x) corresponding to f a (x) = wi (x) for a = x, y. With the
VDoHS rates expanded as
N
f a (x) =
F a w (x),
(5)
i=1 i i
where N is the number of function sites in our grid, the
general surface horizontal velocity eld is given by

N
F a ua (x)
(6)
u(x) = u0 (x) +
a=x,y
i=1 i i
where u0 (x) is the elastostatic solution for the far-eld
boundary conditions with no subsurface sources of
deformation.
We use a Bayesian inversion (Haines et al. 2015) for
the coefcients Fia that best t the departures of GPS
velocity from the elastostatic solution u0 (x) with no
VDoHS-rate contributions. The Bayesian inversion
methodology (Haines et al. 2015) is based on the maximum entropy principle from information theory and
statistical mechanics (Jaynes 1957a, 1957b) which provides the probability functions used (Jaynes 1968, 1988;
see also Tarantola & Valette 1982; Tarantola 1987).
Quoting Jaynes (1957a), it is the least biased estimate
possible on the given information; i.e., it is maximally
noncommittal with regard to missing information
(p. 620).
Our model vector m consists of the VDoHS rate
values at the basis function sites, and our data vector
d consists of the velocity values at the GPS sites. The
data and model are related according to
d = Am + e,

(7)

where e is a vector of observational errors. Then


1 T
= [AT Wd A + Cm
m
]A Wd d,

(8)

where W d is the inverse of the covariance matrix for


the observational errors and Cm is the a priori covariance matrix for the model, the general least-squares
solution for the model that minimises the cost function
1
(d Am)T Wd (d Am) + mT Cm
m
(Snieder
&
Trampert 1999).
To provide the level of damping required in
the Bayesian inversion for the coefcientsFia , the a
priori probability function has contributions from

2
minimising
both
m2 =
|f | dS
a=x,y

48

 

|f /xb | dS. The


 a
associated weight matrices wi (x, y)wj (x, y) dxdy
 
wi (x, y) wj (x, y)
and
dxdy for a = x, y
b=x,y
xb
xb
are effectively identical to those discussed by Snieder
& Trampert (1999) when the integration is explicitly
at the Earths surface. The rst of these integrals provides a constraint on the VDoHS rates to be as small
as possible while remaining consistent with the observations, while the second provides a constraint on the
short-wavelength variations in the VDoHS rates. The
damping parameters for these two components of the
a priori probability functions are determined by
maximising the marginal probability of the GPS
observations.
The coefcients Fia multiply the VDoHS-rate basis
functions wi (x) to obtain the nal VDoHS rates
(Equation 5). Corresponding strain rates can be
calculated from integrating the VDoHS rates, and velocities from integrating the strain rates. In practice, we
follow the process as described in Table 1 and Haines
et al. (2015). The resultant solution ts the statistically
signicant variations in GPS observations while ignoring the noise-dominated short wavelengths.
and

Downloaded by [149.202.47.181] at 08:38 31 May 2016

LL DIMITROVA ET AL.

m2 =

a=x,y

b=x,y

Synthetic results demonstrating VDoHS rates for


strike-slip and subduction interface faults
To demonstrate what VDoHS rates will look like for
simple faulting cases, we briey summarise the
Table 1. Flow chart of tasks. Note that tasks in groups 1 and 2
need to be performed only once for each set of GPS sites, while
the quicker tasks in group 3 can be repeated (e.g. for inversions
of velocity data from different time periods).
Task

Description

1.1
1.2

Prepare the GPS observations.


Assign VDoHS-rate basis function vertices at GPS sites, identifying
any GPS sites closer together than a minimum separation and
combining those into single vertices.
Add the boundaries and specify boundary types. Perform Delaunay
triangulation on the VDoHS-rate basis function vertices and
boundary points and iteratively add intermediate Ruppert points,
producing a grid of triangles with angles >28.
Calculate integrals needed in specifying the shapes of the VDoHSrate basis functions and also the damping matrices for the Bayesian
inversion.
Use the integrals to determine the shapes of the VDoHS-rate basis
functions and calculate the damping matrices.
Use nite elements to calculate the far-eld boundary condition
solution u0 (x) with no internal sources of deformation, and extract
the u0 (x) values at the GPS sites.
Use nite elements to calculate the VDoHS-rate basis function
solutions uai (x) with the wi (x) basis functions as the internal sources.
Extract the basis function solution uai (x) values at the GPS sites.
Undertake the Bayesian inversion for the best tting coefcients Fia .
Prepare the nal VDoHS rates solution fa (x) using the values of the
coefcients Fia .
Use nite elements to calculate the nal solution for velocity and
strain rate with the far-eld boundary conditions included and the
nal VDoHS rates solution fa (x) as the internal source functions. This
solution is undertaken to avoid having to store the full solutions
from tasks 2.12.3.

1.3

1.4
1.5
2.1
2.2
2.3
3.1
3.2
3.3

VDoHS rate patterns for strike-slip and low-angle


reverse faults (e.g. a subduction interface fault). The
strike-slip faulting example is similar to that discussed
by Haines et al. (2015). We show results for an east
west striking, vertical, right lateral, innite-length
strike-slip fault slipping at 20 mm a1 and fully locked
(e.g. 0 mm a1 slip rate) between the surface and 10 km
depth (Figure 3A). The VDoHS-rates magnitudes are
characterised by two bands, one on each side of the
fault, with a band of zero VDoHS rate in between.
This zero band coincides with the surface projection
of the bottom of the locked zone, which for a vertical
fault coincides with the surface trace of the fault. The
VDoHS-rate arrows are parallel to the fault trace in
the opposite directions (consistent with the right lateral
sense of motion) on either side of the faults at distances
1030 km away from the fault. When applied to real
cases, the change in VDoHS rate directions will help
to dene the location of any active strike-slip faulting,
while a combination of VDoHS rates and strain rates
can be used to ascertain the fault slip rates and locking
depths (Haines et al. 2015).
For dip-slip faults the pattern is more complicated,
but the primary VDoHS-rate arrows near the fault
are perpendicular to the fault trace and point towards
the fault trace for reverse faults and away from the
fault trace for normal fault. Here we show in Figure
3B results for an eastwest-striking thrust fault dipping
at 12 and slipping at 10 mm a1. The locking depth is
prescribed at 15 km for a 20 km long centre segment
and gradually decreases to 10 km on the anks; the
surface projection of the bottom of the locked zone
(e.g. the downdip transition from locked to creeping)
is shown as a white dashed line. The VDoHS rates
are characterised by a main band, anked by several
smaller bands. The peak magnitude is approximately
centred on the surface projection of the locking depth
line (the white dashed line on Figure 3B). From peak
to peak the VDoHS rates alternate direction but are
always roughly perpendicular to the surface projection
of the locking depth.
Inversion results for the North Island and the
northern South Island
We apply the Haines et al. (2015) methodology to GPS
velocities in the North Island and the northern part of
the South Island (Figure 1B) from campaigns spanning
19952013 (Beavan et al. 2016). We construct a grid
with vertices at the GPS stations (Figure S1). The
southern and northern boundaries are chosen to be
roughly perpendicular to the strike of the Marlborough
Fault System, North Island Dextral Fault Belt and the
Taupo Volcanic Zone and we apply reecting boundary conditions (i.e. deformation continues through
the boundaries as expected for boundaries that cross
orthogonal to major faults; see Haines et al. 2015)

Downloaded by [149.202.47.181] at 08:38 31 May 2016

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

49

Figure 3. VDoHS rates scaled by shear modulus for A, vertical right-lateral strike-slip fault, and B, reverse dip-slip fault dipping at
12 to the top with variable locking depth slipping at 10 mm a1. The horizontal distance scale is kilometres. Surface projection of
the top of the slipping zone is shown in a dashed white line, while solid white lines are where fault planes extend to the surface.
VDoHS-rate magnitudes are plotted on logarithmic scale, while VDoHS-rate vectors are scaled to be proportional to VDoHS-rate
magnitude to the power of 0.25.

(see Figures 46 for locations of these as well as other


locations discussed throughout the paper). The western
boundary of the model space is chosen to lie offshore
the western North Island within the Australian plate
and has zero velocity (i.e. a rigid, boundary condition),
as our GPS velocities are in an Australian-Plate-xed
reference frame. The remaining boundary follows the
Hikurangi trench and is a free boundary with tractions
prescribed to be 0, except on the small NESW-striking
segment south of the termination of the Hikurangi
Trough where a Pacic Plate velocity boundary condition is applied. The damping coefcients are determined by maximising the conditional joint
probability of the damping factors given the GPS
observations (see the section Inversion process). The
velocity residuals in the tted model (i.e. tted values
minus data values) are shown in Figure S3. Overall
the residuals are very small (RMS of c. 1.2 mm a1;
note the difference in scale compared to the velocities)
and are largest near the TVZ.
The VDoHS rate eld vectors are shown in Figure 4
where the arrows are proportional to the VDoHS-rate
magnitude to the power of 0.25. The associated area
and log shear-strain rates are shown in Figures 4 and
5, respectively, with units of 1 strain/year. We also
show the principal contraction directions overlain on
shear strain in Figure 5. Note that all of our GPS data
is onshore (e.g. Figure 1B) and areas where there are
no data show simply an interpolation between areas
with data and the boundary conditions. Along the
east coast of the North Island the VDoHS rates alternate direction from pointing WNW in the direction
of subduction (e.g. near the Wairarapa coast and Gisborne) to pointing trenchwards (e.g. Hawkes Bay

and some portions of the Raukumara Peninsula), indicating a complex pattern of locking and unlocking
along the subduction interface averaged over our
period of observation (see further discussion in Locking along the subduction interface).
The VDoHS rate pattern in the southern North
Island and northern South Island is generally complex,
due to overprinting of many closely spaced tectonic
structures. In areas with GPS coverage however, the
overall large-scale pattern across the Marlborough
Fault System and the southwestern portion of the
Wellington and Wairarapa faults of the VDoHS
arrows are approximately parallel to the faults; vectors
northwest of the faults point in the northeast direction
and vectors to the southeast of the faults point
southwest, consistent with these fault zones being
right-lateral strike-slip. This is consistent with the
large shear-strain rates and directions of principal
contraction in the region (Figure 5). Note that in the
Marlborough Fault System, the NE-trending arrows
begin just north of the Clarence Fault indicating that
most of the strike slip occurs on the Clarence and
Hope Faults, while relatively lesser amounts occur on
the Awatere and Wairau faults. This is consistent
with the cumulative strike-slip rate for the Hope and
Clarence faults being nearly c. 25 mm a1 (Van Dissen
& Yeats 1991; Langridge et al. 2003; Van Dissen & Nicol
2009), while the Wairau and Awatere faults combined
probably only accommodates 810 mm a1 of strike
slip (e.g. Benson et al. 2001; Mason et al. 2006;
Zachariasen et al. 2006).
In the central TVZ at Lake Taupo, the VDoHS-rate
vectors point radially outwards indicating extension
similar to an expanding Mogi source (Haines et al.

Downloaded by [149.202.47.181] at 08:38 31 May 2016

50

LL DIMITROVA ET AL.

Figure 4. VDoHS-rate vectors (106 km1 a) scaled by the VDoHS-rate magnitude to the power of 0.25 from our inversion (raw
model). Background colour is areal strain (in strain/year) with hot colours (pink/tan to red) showing extension and cold colours
(blue) showing contraction. Slow slip contours are shown in red for HB, Hawkes Bay; G, Gisborne; K, Kapiti; M, Manawatu. We indicate possibly locked (L) and unlocked (U) areas of the subduction interface at the Hikurangi margin, according to the VDoHS and
areal strain rates. The contour of 20% locking from Wallace et al. (2012a) is shown in black.

2015; Figure 4). Between Lake Taupo and Rotorua, the


VDoHS-rate vectors point inwards towards the centre
of the rift indicating contraction, while in the Whakatane Graben near the Bay of Plenty coastline the
VDoHS-rate vectors ip directions, again pointing
away from the faults there, indicating extension (Figure
4). The areal strains show a similar pattern with extensional (hot colours) areal strain (c. 0.040.05 strain/
year) in Lake Taupo and Whakatane Graben region
and strong contraction (cold colours) (c. 0.07 strain/
year) in the region between Lake Taupo and Rotorua.
In the southern Taupo Volcanic Zone south of Lake
Taupo VDoHS rates are orthogonal to the NESWtrending faults in the region, while at the southern
termination of the TVZ the VDoHS-rate vectors are
parallel to the EW-trending faults there. The areal

strain rates south of Lake Taupo show a small amount


of extension (c. 0.02 strain/year).
Over much of the northern Raukumara Peninsula,
the VDoHS rates point outwards indicating extension,
which is mirrored in the large extensional areal strain
(Figure 4). In contrast, the VDoHS rates near Gisborne
point radially towards each other indicating contraction; we also observe contractional strain in the region
(Figure 4). The northern part of the Hawkes Bay area
shows a hint of minor extension, transitioning to contraction just west of the Napier region. In the Maraetotara Plateau to the southwest of Hawkes Bay VDoHS
rates point away from and perpendicular to the
mapped normal faults in this area, indicating extension; we observe c. 0.05 strain/year extensional strain
rates here (Figure 4).

Downloaded by [149.202.47.181] at 08:38 31 May 2016

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

51

Figure 5. Shear strain and maximum contraction directions plot. AF, Alpine Fault; AwF, Awatere Fault; CF, Clarence Fault; CK, Cape
Kidnappers; CT, Cape Turnagain; HF, Hope Fault; MFS, Marlborough Fault System; NIDFB, North Island Dextral Fault Belt; WF, Wairau
Fault.

Discussion
Locking along the subduction interface
At a subduction zone the VDoHS rates above the
locked subduction interface should point in the direction of motion of the subducting plate, while the
VDoHS rates above an unlocked (slipping) subduction
interface point in the opposite direction towards the
trench. The boundary where the VDoHS rates reverse
direction should lie directly above the boundary
between the locked and unlocked portion of the subduction interface (Figures 2B and 3). A locked interface
will also be accompanied by increased contractional
strain above the locked area and changes in the distribution of shear strain.
Along the east coast of the North Island the VDoHS
rates change direction from pointing in the direction of

subduction (e.g. near Wellington and Gisborne) to


pointing in the trenchwards direction (e.g. Hawkes
Bay, northern Raukumara Peninsula), indicating a
complex pattern of locking and unlocking along the
fault zone (Figure 4) similar to that obtained by Wallace et al. (2004, 2012a). In Figure 4, we compare our
locking and unlocking pattern with slow -lip events
(SSEs) occurring between 2002 and 2012, and the contour (thick black line) of 20% locking from Wallace
et al. (2012a). The location of the updip end of the
Kapiti SSEs is in an area beneath the Kapiti Coast
where VDoHS rates begin to point trenchwards. Likewise, in the southern section of the Kapiti slow-slip
events (beneath the Marlborough Sounds of the northern South Island), the VDoHS rates point towards the
trench in the direction of AustraliaPacic plate convergence, consistent with being located above a

Downloaded by [149.202.47.181] at 08:38 31 May 2016

52

LL DIMITROVA ET AL.

Figure 6. Principal directions of contraction from our inversion


before (black) and after (red) the removal of the locking model,
compared with SHmax directions from Townend et al.
(2012) who analysed focal mechanisms of 3242 events in
20042011 with ML> 3 and depth of 0300 km. We show
SHmax from focal mechanisms only with depths less than 60
km. BP, Bay of Plenty; HB, Hawkes Bay; LT, Lake Taupo; MaF,
Martinborough Fault; MCM, Masterton/Carterton/Mokonui
faults; MF, Mohaka Fault; MP, Maraetotara Plateau; N, Napier;
O, Okataina; OF, Ohariu Fault; R, Rotorua; RF, Ruahine Fault;
WaF, Wairarapa Fault; WeF, Wellington Fault; WG, Whakatane
Graben.

creeping subduction interface. VDoHS rates point


towards the trench above the deeper Manawatu SSEs
at the central Hikurangi margin, as well in the region
of SSEs just offshore Cape Kidnappers and Cape Turnagain. This is consistent with an unlocked subduction
zone in these areas, suggesting that SSEs and/or creep
are accommodating the majority of plate motion in
these areas for the duration of the campaign GPS dataset (the last 20 years; e.g. Wallace & Beavan 2010). The
Hawkes Bay region, which hosts shallow slow-slip
events further offshore, is also the site of trenchwards-pointing VDoHS rate vectors, consistent
with a mostly creeping plate interface there.
In contrast, the Gisborne SSEs are just offshore of an
area where the VDoHS-rate vectors point landwards
and areal strain rates are contractional, consistent with
a small portion of the subduction zone being locked
just to the north of Gisborne. A more deeply penetrating
locked patch north of Gisborne is also observed in locking models by Wallace et al. (2004, 2012a), and likely
projects downdip of SSEs observed there (Wallace
et al. 2012b). Wallace et al. (2012b) suggest that the subduction interface offshore Gisborne is characterised by a
complex mosaic of locked, slow-slipping, creeping
patches, possibly related to subduction of seamounts
and thick units of subducted, uid-rich sediments

surrounding the seamounts. This small-scale heterogeneity in locked v. creeping patches is probably why we
dont see a clear unlocking signal that correlates with
SSEs offshore of the Gisborne area in contrast to other
SSE regions of the North Island. North of Gisborne,
the VDoHS rates shift to a trenchwards direction
again, indicating a mostly creeping subduction interface
beneath the northern half of the Raukumara Peninsula.
The trenchwards-pointing VDoHS rates in northern
Raukumara are also likely to be inuenced by upper
plate extensional faults (see discussion in Extensional
collapse of the forearc)
The portion of the southern North Island that exhibits strong interseismic locking in the campaign GPS
data (Figure 4) has not experienced large SSEs in the
last 12 years and the landwards-pointing VDoHS
rates and high rates of contraction (0.040.06 strain/
year) are consistent with the idea that this region of
the plate interface is indeed locked. The accumulating
elastic strain in this region is likely to be relieved in
large earthquakes on the subduction interface, and represents a signicant seismic risk as has been suggested
by previous workers (Darby & Beavan 2001; Wallace
et al. 2004, 2012a; Lamb & Smith 2013). We also
note a patch of high contractional strain and
inwards-pointing VDoHS rates in the southern
Hawkes Bay region, just to the west of the Maraetotara
Plateau. We suggest that this may be a small locked
patch at 2030 km depth on the subduction interface,
just downdip of the southern Hawkes Bay SSEs and
updip of the Manawatu SSEs. This possible locked
patch beneath the southern Hawkes Bay region has
not yet been recognised in previous studies of Hikurangi locking, and likely represents signicant spatial
heterogeneity of locking in the alongstrike transition
from deep locking (south Hikurangi) to shallow locking (north Hikurangi).

Extensional collapse of the forearc


Long-term deformation in the Hikurangi forearc is
dominated by strike-slip (Van Dissen & Berryman
1996; Beanland & Haines 1998; Little et al. 2009) and
contraction (Barnes & Mercier de Lepinay 1997;
Nicol & Beavan 2003; Nicol et al. 2007). However,
extension and normal faulting have also been observed
at a number of locations. Extensional deformation has
been reported in both the Raukumara ranges (Berryman et al. 2009) and the Maraetotara Plateau south
of Hawkes Bay (Cashman & Kelsey 1990; Barnes &
Nicol 2004; Pettinga 2004). The normal faults tend to
be relatively short and are considered to be shallow,
secondary extensional structures formed as a consequence of gravitational collapse, deep-seated landsliding or bending moment faults due to exure
accompanying rapid uplift (Cashman & Kelsey 1990;

Downloaded by [149.202.47.181] at 08:38 31 May 2016

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

Chanier et al. 1999; Pettinga 2004; Berryman et al.


2009).
Our inversions show c. 0.05 strain/year extensional
areal strain in the Maraetorara Plateau and larger (> 0.1
strain/year) extensional areal strain in much of the
Raukumara Peninsula (Figure 4). In both regions the
VDoHS rates point outwards, which is also consistent
with extension (Figure 4). The VDoHS rates in both
regions are also perpendicular to the mapped normal
faults.
The contraction west of the Maraetotara Plateau
could be associated with locking at 20 km depth on
the subduction interface with unlocking signal updip
(to the east) below the extensional area of the Maraetotara Plateau (see discussion in previous section on
Locking along the subduction interface). Alternatively, the VDoHS and strain rates could reect shallow-seated deformation due to westwards motion of
the Maraetotara Plateau relative to the rest of the forearc as part of the extension process, or it could be
related to possible hidden upper plate reverse faulting
west of the Maraetotara Plateau. To be able to differentiate these scenarios we need a much denser GPS coverage with spacing of at most half the depth of the
shallowest possible sources, that is, the order of 12
km spacing (see Haines et al. 2015).
Active faulting in the northern Raukumara Ranges
is characterised by extension (Figure 4). It has previously been suggested that the Raukumara Ranges
are uplifting rapidly as the result of underplating subducted sediment (Walcott 1987). This view is supported by seismic studies that image velocity
anomalies suggestive of a zone of deep subducted sediment accumulation beneath the Raukumara Ranges
(e.g. Eberhart-Phillips & Reyners 2012). Normal faulting in this region has been interpreted as being due to
gravitational collapse as a consequence of rapid uplift
of the ranges (Thornley 1997); the low rate of activity
on the normal faults there supports this hypothesis
(Berryman et al. 2009). Our areal strains match well
the localised areas of extensional faulting observed in
geological studies (Thornley 1997; Berryman et al.
2009).
Consistency of maximum contraction directions
with historical earthquakes and active fault
orientations
The orientations of principal contraction directions
derived from our analysis are shown in Figure 6
(black bars). Above the shallow portion of the Hikurangi subduction interface, the principal contraction
directions are roughly perpendicular to the trench.
Notable exceptions are the northern part of the Raukumara Peninsula and the Maraetotara Plateau (see previous section on Extensional collapse of the forearc).
In the southern North Island, along the Wellington

53

and Wairarapa Fault System, the maximum contraction directions are oriented mostly perpendicular to
the faults. Further north along the North Island Dextral
Fault Belt, the principal contractions are oriented more
northeasterly and oblique to the strike-slip faults, as
expected for right-lateral strike-slip faulting there.
It is likely that the dominantly margin-perpendicular contraction directions from the GPS velocities are
strongly inuenced by interseismic locking on the subduction thrust. To address this, we remove the elastic
component of the velocity eld due to subduction
interface coupling using the model of Wallace et al.
(2012a), and obtain VDoHS rates and strain rates
from the corrected velocity eld. The orientation of
the principal contraction directions for the corrected
solution (i.e. with subduction locking removed) is
shown as red bars on Figure 6. The contraction directions in the models with subduction locking removed
are far more compatible with the styles of faulting in
the southern North Island, as they are rotated in a
more northeasterly direction compared to the raw
(margin perpendicular; black bars) contraction directions. This indicates that stresses arising from interseismic coupling on the subduction thrust are a temporary
phenomenon, and do not exert a strong inuence on
the long-term upper plate deformation patterns. We
also note that there is little difference in the contraction
directions between the corrected and uncorrected
models in the southern North Island and northern
South Island. This likely indicates incomplete removal
of the locking signal by the model of Wallace et al.
(2012a).
We also compare our raw and corrected principal
contraction directions with the axis of maximum horizontal compressive stress (SHmax) derived from earthquake focal mechanisms (Townend et al. 2012; Figure
6; SHmax are cyan bars). If the contemporary deformation processes as measured by seismological and
GPS data are the same, we expect the SHmax directions
and our principal contraction directions to agree.
There is good agreement between the seismological
SHmax and the raw principal contraction directions in
parts of the TVZ, the Whakatane Graben, along the
central and northern portion of the North Island Dextral Fault Belt, the northern part of the Raukumara
Peninsula and the northern part of the Maraetotara
Plateau. There are large angular differences in the western portion of the North Island (e.g. Taranaki and
Waikato regions); however, the strain rates there are
small and the maximum contraction directions derived
from GPS are therefore not well resolved.
There are large angular differences between the
SHmax directions and our maximum contraction directions from our raw (uncorrected) inversion in the forearc of the southern Hikurangi margin. Principal
contraction directions from our corrected model,
after subtracting the locking model (see section on

Downloaded by [149.202.47.181] at 08:38 31 May 2016

54

LL DIMITROVA ET AL.

Inversion results for the North Island and the northern


South Island; red bars on Figure 6), provide an
improved t to the SHmax directions along the southern
Hikurangi margin. This demonstrates that recent seismicity is strongly controlled by stress directions related
to long-term deformation patterns (related to oblique
Pacic/Australia convergence and long-term motion
of the forearc relative to the Australian Plate), rather
than the short-term, elastic component of strain
induced by subduction interface locking. The exception
to this is the far southern end of the North Island
(south of 41), where neither the raw nor the corrected
principal directions match the observed SHmax directions very well. This possibly indicates that our locking
model may underestimate the amount of locking
beneath the far southern North Island.
Townend & Zoback (2006) found a similar discrepancy in the orientations between the SHmax directions and the axis of maximum contractional strain
rate from raw GPS data in Central Japan. They suggest
that this difference reects the effects of interseismic
subduction zone coupling on the GPS velocity eld
and that the state of stress in the upper plate is largely
governed by motion of northeastern Honshu with
respect to the Amurian Plate, while GPS strain rates
are controlled by the interseismically locked Nankai
subduction thrust. Following that argument, Townend et al. (2012) interpreted the discrepancy between
SHmax and principal contraction directions from
strain rates (uncorrected for locking) by Beavan &
Haines (2001) is due to interseismic locking on the
subduction interface. Our contraction directions
from the uncorrected (black) and corrected (red)
(Figure 6) GPS velocities are largely consistent with
this interpretation.
Strain across the TVZ
In the Whakatane Graben region the VDoHS-rate vectors point away from the faults (Figure 4) indicating
extension, consistent with the extensional areal strain.
Extensional strain rates of 0.08 strain/year (0.04
strain/year of areal strain) over a 100 km wide zone
suggest 8 mm a1 extensional slip rate in the Whakatane Graben, which is higher than the rate reported
by Nicol & Wallace (2007) but is within the range of
Wallace et al. (2004). In the central TVZ at Lake
Taupo, the VDoHS-rate vectors point radially outwards indicating extension. This may be related to
magma migration patterns and/or changes in the
hydrothermal system (e.g. Peltier et al. 2009). However,
instead of the expected rifting signal between Lake
Taupo and Rotorua, we nd the VDoHS-rate vectors
point inwards towards the centre of the rift and the
faults therein, indicating contraction. We also observe
high contractional strain rates (c. 0.07 strain/year of
areal strain). This indicates that non-tectonic effects

related to magmatic and/or hydrothermal system processes must strongly inuence crustal deformation in
large portions of the TVZ. Holden et al. (2015) also
observe a similar zone of contraction from GPS data
in the Okataina region, and Hamling et al. (2015)
observe a large amount of subsidence in this area (up
to 2 cm a1). Hamling et al. (2015) use a deating sill
model for the TVZ to t GPS and InSAR displacement
data. Their model predicts c. 1820 mm a1 contraction in the central part of the TVZ, where our areal
strains from the horizontal GPS are also contractional.
Hamling et al. (2015) argue that the majority of the
contraction may be due to volume loss due to
magma cooling at 410 km depth.

Conclusions
We have analysed interseismic GPS velocities from
campaigns spanning 19952013 (Beavan et al. 2016)
using a new method to estimate high-resolution strain
rates and vertical derivatives of horizontal stress
(VDoHS) rates. Our new higher-resolution strain-rate
maps reveal zones of extension within the forearc, in
the northern Raukumara Ranges and Maraetotara Plateau, which may represent gravitational collapse of the
forearc in regions of high uplift rates. One of the most
striking results are the high rates of contraction
throughout the central Taupo Volcanic Zone, which
is in contrast to long-term extension observed there
(e.g. Villamor & Berryman 2001). Contemporary contraction in the TVZ may be due to the presence of a
shallow cooling magma body beneath the central
TVZ (Hamling et al. 2015).
Our VDoHS rate map provides a high-resolution
image of possible locked (and unlocked) regions
along the Hikurangi subduction margin that agrees
well with previous studies (Darby & Beavan 2001; Wallace et al. 2004, 2012a; Lamb & Smith 2013). The
VDoHS rates and strain rates are consistent with a deeply locked southern Hikurangi subduction interface, as
well as small locked patches near Gisborne. Our results
also highlight (for the rst time) a possible deeper
locked patch beneath the southern Hawkes Bay region,
west of the Maraetotara Plateau. The remaining segments on the Hikurangi subduction interface beneath
north and central Hawkes Bay and the Raukumara
Peninsula appear to be unlocked during the c. 20year period of this study.
We compare principal contraction directions from
our models (before and after removing a locking
model for the Hikurangi subduction margin) with
onshore faults and SHmax directions from historical seismicity (Townend et al. 2012). In most of the North
Island, there is much better agreement between the principal contraction directions that are corrected for locking and the SHmax directions and fault orientations.
This indicates that historical seismicity is more strongly

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

modulated by stresses arising from long-term processes


(such as rotation of the Hikurangi forearc relative to the
Australian Plate) rather than the short-term, elastic
component of strain induced by locking on the subduction interface between large earthquakes.

Supplementary data
Figure S1. Grid used in this study. GPS sites are shown in red
circles, while function sites are shown in black squares.
Figure S2. Examples of 2D VDoHS-rate basis functions
scaled by shear modulus wi (x) for the mesh in Figure S1.
Basis function sites are shown as black dots, and the
basis functions are unit-less and have values 1 at the corresponding function sites and 0 at other function sites.
Within the triangles in Figure 5 Argyris elements are
used to interpolate the basis functions.

Downloaded by [149.202.47.181] at 08:38 31 May 2016

Figure S3. The uxi (x) (left panels) and ui (x) (right panels) velocity functions determined from the 2D VDoHS-rate basis
function wi (x) in Figure S2. Basis function sites are shown
as black dots. These velocity functions are solved for using
nite elements with a much ner mesh than in Figure S1.
Figure S4. GPS velocity from our inversion (blue) and
residual (red). Note the difference in scale.

Acknowledgements
We thank Andy Nicol and two anonymous reviewers for
their helpful comments. The GPS velocities used here are
the outcome of decades of GPS data acquisition led by the
late John Beavan. His drive and vision in establishing geodetic research in New Zealand will be greatly missed.
Guest Editor: Dr Ian Hamling.

Disclosure statement
No potential conict of interest was reported by the authors.

Funding
This work was funded by grant 2012-GNS-02-NHRP.

References
Barnes PM, Mercier de Lepinay BM. 1997. Rates and mechanics of rapid frontal accretion along the very obliquely
convergent southern Hikurangi margin, New Zealand. J
Geophys Res. 102:2493124952.
Barnes PM, Nicol A. 2004. Formation of an active thrust triangle zone associated with structural inversion in a subduction setting, eastern New Zealand. Tectonics. 23.
doi:10.1029/2002TC001449.
Beanland S, Haines J. 1998. The kinematics of active deformation in the North Island, New Zealand, determined
from geological strain rates. New Zeal J Geol Geophys.
41:311323.
Beavan J, Haines J. 2001. Contemporary horizontal velocity
and strain rate elds of the 1015 Pacic-Australian plate
boundary zone through New Zealand. J Geophys Tes.
106:741770. doi:10.1029/2000JB900302.

55

Beavan J, Moore M, Pearson C, Henderson C, Parsons B,


Bourne S, England P, Walcott D, Blick G, Darby D,
Hodgkinson D. 1999. Crustal deformation during 1994
1998 due to 1019 oblique continental collision in the central Southern Alps, New Zealand, and implications for
1020 seismic potential of the Alpine Fault. J Geophys
Res. 104:2523325255.
Beavan J, Wallace LM, Palmer N, Denys P, Ellis S, Fournier
N, Hreinsdottir S, Pearson C, Denham M. 2016. New
Zealand GPS velocity eld: 19952013. New Zeal J Geol
Geophys. doi:10.1080/00288306.2015.1112817.
Benson AM, Little TA, Van Dissen RJ, Hill N, Townsend DB.
2001. Late Quaternary paleoseismic history and surface
rupture characteristics of the eastern Awatere strike-slip
fault, New Zealand. Geol Soc Am Bull. 113:10791091.
doi:10.1130/0016-7606.
Berryman K, Marden M, Palmer A, Litcheld N. 2009.
Holocene rupture of the Repongaere Fault, Gisborne:
implications for Raukumara Peninsula deformation and
impact on the Waipaoa Sedimentary System. New Zeal J
Geol Geophys. 52:335347.
Bower A. 2009. Applied mechanics of solids. Boca Raton, FL:
CRC Press.
Cashman SM, Kelsey HM. 1990. Forearc uplift and extension, southern Hawkes Bay, New Zealand: mid-pleistocene to present. Tectonics. 9:2344.
Chanier F, Ferrire J, Angelier J. 1999. Extensional deformation across an active margin, relations with subsidence,
uplift, and rotations: the Hikurangi subduction, New
Zealand. Tectonics. 18. doi:10.1029/1999TC900028.
Darby D, Beavan J. 2001. Evidence from GPS measurements
for contemporary interplate coupling on the southern
Hikurangi subduction thrust and for partitioning of strain
in the upper plate. J Geophys Res. 106. doi:10.1029/
2000JB000023.
Eberhart-Phillips D, Reyners M. 2012. Imaging the
Hikurangi plate interface region, with improved localearthquake tomography. Geophys J Intl. 190:12211242.
doi:10.1111/j.1365-246X.2012.05553.x.
England P, Molnar P. 1997. The eld of crustal velocity in
Asia calculated from Quaternary rates of slip on faults.
Geophys J Int. 130:551582.
Flesch LM, Kreemer C. 2010. Gravitational potential energy
and regional stress and strain rate elds for continental
plateaus: examples form the central Andes and Colorado
Plateau. Tectonophysics. 482:182192.
Haines AJ. 1982. Calculating velocity elds across plate
boundaries from observed shear rates. Geophys J Roy
Astr Soc. 68:203209.
Haines AJ, Dimitrova L, Wallace L, Williams C. 2015.
Enhanced surface imaging of crustal deformation: obtaining tectonic force elds using GPS data. Germany:
Springer.
Haines AJ, Holt WE. 1993. A procedure for obtaining the
complete horizontal motions within zones of distributed
deformation from the inversion of strain rate data. J
Geophys Res. 98:1205712082.
Hamling IJ, Hreinsdottir S, Fournier N. 2015. The ups and
downs of the TVZ: geodetic observations of deformation
around the Taupo Volcanic Zone, New Zealand. J
Geophys Res 120:46674679.
Heron DW, Van Dissen RJ, Sawa M. 1998. Late Quaternary
movement on the Ohariu Fault, Tongue point to MacKays
crossing, North Island, New Zealand. New Zeal J Geol
Geophys. 41:419439.
Holden L, Wallace LM, Beavan J, Fournier N, Cas R, Ailleres
L, Silcock D. 2015. Contemporary ground deformation at

Downloaded by [149.202.47.181] at 08:38 31 May 2016

56

LL DIMITROVA ET AL.

Okataina Volcanic Centre from 1998 to 2011 measured


using GPS. Geophys J Int. 202:20822105.
Holt WE, Haines AJ. 1995. The kinematics of northern
South Island New Zealand determined from geologic
strain rates. J Geophys Res. 100:1799118010.
Jaynes E. 1957a. Information theory and statistical mechanics.
Phys Rev. 106:620630. doi:10.1103/PhysRev.106.620.
Jaynes E. 1957b. Information theory and statistical mechanics II. Phys Rev. 108:171190. doi:10.1103/PhysRev.
108.171.
Jaynes E. 1968. Prior probabilities. IEEE Trans Syst Sci
Cybern. 4:227241. doi:10.1109/TSSC.1968.300117.
Jaynes E. 1988. The relation of Bayesian and maximum
entropy methods. In: Erickson GJ, Smith CR, editors.
Maximum-entropy and Bayesian methods in science
and engineering, Vol 1. New York, NY: Kluwer
Academic Publishers.
Kreemer C, Haines AJ, Holt WE, Blewitt G, Lavallee D. 2000.
On the determination of a global strain rate model. Earth
Planets and Space. 52:765770.
Kreemer C, Hammond WC, Blewitt G, Holland AA, Bennett
RA. 2012. A geodetic strain rate model for the PacicNorth American plate boundary, western United States.
Nevada Bureau of Mines and Geology map 178, scale
1:1,500,000. Reno: University of Nevada.
Kreemer C, Holt WE, Haines AJ. 2003. An integrated global
model of present-day plate motions and plate boundary
deformation. Geophys J Int. 154:834.
Lamarche G, Barnes PM, Bull JM. 2006. Faulting and extension rate over the last 20,000 years in the offshore
Whakatane Graben, New Zealand continental shelf.
Tectonics. 25:TC4005. doi:10.1029/2005TC001886.
Lamarche G, Proust J-N, Nodder SD. 2005. Long-term slip
rates and fault interactions under low contractional strain,
Wanganui Basin, New Zealand. Tectonics. 24:TC4004.
doi:10.1029/2004TC001699.
Lamb S. 1988. Tectonic rotations about vertical axes during
the last 4 main part of the New Zealand plate boundary
zone. J Struct Geol. 10:875893. doi:10.1016/0191-8141
(88)90101-0.
Lamb S. 2011. Cenozoic tectonic evolution of the New
Zealand plate-boundary zone: a review. Tectonophysics.
509. doi:10.1016/j.tecto.2011.06.005.
Lamb S, Smith E. 2013. The nature of the plate interface and
driving force of interseismic deformation in the New
Zealand plate boundary zone, revealed by the continuous
GPS velocity eld. J Geophys Res. 118. doi:10.1002/hgrb.
50221.
Langridge R, Berryman KR. 2005. Morphology and slip rate
of the Hurunui section of the hope Fault, South Island,
New Zealand. New Zeal J Geol Geophys. 48:4357.
Langridge RM, Campbell J, Hill NL, Pere V, Pope J, Pettinga
J, Estrada B, Berryman KR. 2003. Paleoseismology and
slip rate of the conway segment of the hope Fault at
Greenburn stream, South Island, New Zealand. Ann
Geophy. 46(5):11191139.
Litcheld N, Van Dissen R, Nicol A. 2007. Re-assessment of
slip rate and implications for surface rupture hazard of the
Martinborough Fault, South Wairarapa, New Zealand.
New Zeal J Geol Geophys. 50:239243.
Little TA, Van Dissen R, Rieser U, Smith EGC, Langridge
RM. 2010. Coseismic strike slip at a point during the
last four earthquakes on the Wellington fault near
Wellington, New Zealand. J Geophys Res. 115:B05403.
doi:10.1029/2009JB006589.
Little TA, Van Dissen R, Schermer E, Carne R. 2009. Late
Holocene surface ruptures on the southern Wairarapa

fault, New Zealand: link between earthquakes and the


uplifting of beach ridges on a rocky coast. Lithosphere.
1:428. doi:10.1130/L7.1.
Malvern L. 1969. Introduction to the mechanics of a continuous medium. Eaglewood Cliffs, NJ: Prentice-Hall.
Mason DPM, Little TA, Van Dissen RJ. 2006. Rates of active
faulting during late Quaternary uvial terrace formation
at Saxton River, Awatere Fault. NZ Geol Soc Am Bull.
118:14311446. doi:10.1130/B25961.1.
McCaffrey R. 2002. Crustal block rotations and plate coupling. In: Stein S, Freymueller J, editors. Plate boundary
zones. AGU Geodynamics Series. 30; p. 100122.
McCaffrey R, Long MD, Goldnger C, Zwick PC, Nabelek
JL, Johnson CK, Smith C. 2000. Rotation and plate locking
at the southern Cascadia subduction zone. Geophys Res
Lett. 27:31173120. doi:10.1029/2000GL011768.
Mouslopoulou V, Nicol A, Little TA, Walsh JJ. 2007a.
Terminations of large strike-slip faults: an alternative
model from New Zealand. In: Cunningham WD, Mann
P, editors. Tectonics of strike-slip restraining and releasing bends. Geol Soc Lon. 290:387415.
Mouslopoulou V, Nicol A, Little TA, Walsh JJ. 2007b.
Displacement transfer between intersecting strike-slip
and extensional fault systems. J Struc Geol. 29:100116.
Nicol A, Beavan J. 2003. Shortening of an overriding plate
and its implications for slip on a subduction thrust, central
Hikurangi margin, New Zealand. Tectonics. 22. doi:10.
1029/2003TC001521.
Nicol A, Mazengarb C, Chanier F, Raitt G, Uruski C, Wallace
L. 2007. Tectonic evolution of the active Hikurangi subduction margin, New Zealand, since the Oligocene.
Tectonics. 26. doi:10.1029/2006TC002090.
Nicol A, Van Dissen RJ, Vella P, Alloway B, Melhuish A.
2002. Growth of contractional structures during the last
10 m.y. at the southern end of the emergent Hikurangi
forearc basin, New Zealand. New Zeal J Geol Geophys.
45:365385.
Nicol A, Wallace LM. 2007. Temporal stability of deformation rates: comparison of geological and geodetic
observations, Hikurangi subduction margin, New
Zealand. Earth Planet Sci Lett. doi:10.1016/j.epsl.2007.
03.039.
Norris RJ, Cooper AF. 2000. Late Quaternary slip rates and
slip partitioning on the Alpine Fault, New Zealand. J
Struct Geol. 23:507520.
Parsons T, Thatcher W. 2011. Diffuse Pacic North
American plate boundary: 1000 km of dextral shear
inferred from modeling GPS data. Geology V. 39:943
946. doi:10.1130/G32176.
Peltier A, Hurst T, Scott B, Cayol V. 2009. Structures
involved in the vertical deformation at Lake Taupo (new
Zealand) between 1979 and 2007: new insights from
numerical modeling. J Volc Geothermal Res. 181. doi:10.
1016/j.jvolgeores.2009.01.017.
Pettinga J. 2004. Three-stage massive gravitational collapse
of the emergent imbricate frontal wedge, Hikurangi subduction zone, New Zealand. New Zeal J Geol Geophys.
47:399414.
Reyners M, Eberhart-Phillips D, Stuart G. 1999. A three
dimensional image of shallow subduction: crustal structure of the Raukumara Peninsula, New Zealand. Geophy
J Int. 137:873890.
Snieder R, Trampert J. 1999. Inverse problems in geophysics.
In: Wirgin A, editor. Waveeld inversion. New York, NY:
Springer Verlag.
Tarantola A. 1987. Inverse problem theory. Amsterdam:
Elsevier.

Downloaded by [149.202.47.181] at 08:38 31 May 2016

NEW ZEALAND JOURNAL OF GEOLOGY AND GEOPHYSICS

Tarantola A, Valette B. 1982. Inverse problems = quest for


information. J Geophys. 50:159170.
Thornley S. 1996. Neogene tectonics of Raukumara
Peninsula, northern Hikurangi margin, New Zealand
[Unpublished PhD thesis]. Wellington: Victoria
University of Wellington.
Thornley S. 1997. Neogene tectonics of Raukumara Peninsula,
northern Hikurangi margin, New Zealand [PhD thesis].
Wellington: Victoria University of Wellington.
Townsend D, Begg J, Villamor P, Lukovic B. 2002. Late
Quaternary displacement of the Mokonui Fault,
Wairarapa, New Zealand: a preliminary assessment of
earthquake generating potential. Client Report 2002/58.
Lower Hutt: Institute of Geological and Nuclear Sciences.
Townend J, Sherburn S, Arnold R, Boese C, Woods L. 2012.
Three-dimensional variations in present-day tectonic
stress along the AustraliaPacic plate boundary in New
Zealand. Earth Planetary Sci Lett. 353354:4759.
Townend J, Zoback MD. 2006. Stress, strain, and mountain
building in central Japan. J Geophys Res. 111:B03411.
doi:10.1029/2005JB003759.
Turcotte D, Schubert, G. 2002. Geodynamics. Cambridge:
Cambridge University Press.
Van Dissen R, Nicol A. 2009. Mid-holocene paleoseismicity
of the eastern Clarence Fault, Marlborough, New Zealand.
New Zeal J Geol Geophys. 52:195208. doi:10.1080/00
288300909509886.
Van Dissen R, Yeats RS. 1991. Hope Fault, Jordan thrust, and
uplift of the Seaward Kaikoura range, New Zealand. Geol.
19:393396.
Van Dissen RJ, Berryman KR. 1996. Surface rupture earthquakes over the last c. 1000 years in the Wellington region,
New Zealand, and implications for ground shaking
hazard. J Geophys Res. Solid Earth. 101:59996019.
Villamor P, Berryman KR. 2001. A late Quaternary extension rate in the Taupo Volcanic Zone, New Zealand,
derived from fault slip data. New Zeal J Geol Geophys.
44:243269.
Villamor P, Berryman KR. 2006. Late Quaternary geometry
and kinematics of faults at the southern termination of the
Taupo Volcanic Zone, New Zealand. New Zeal J Geol
Geophys. 49:121.
Walcott RI. 1984. The kinematics of the plate boundary
zone through New Zealand: a comparison of shortand long-term deformations. Geophys J R Astron Soc.
79:613633.

57

Walcott RI. 1987. Geodetic strain and the deformational history of the North Island of New Zealand during the late
Cainozoic. Philos Trans R Soc London Ser A. 321:163
181. doi:10.1098/rsta.1987.0009.
Wallace LM, Barnes P, Beavan J, Van Dissen R, Litcheld N,
Mountjoy J, Langridge R, Lamarche G, Pondard N. 2012a.
The kinematics of a transition from subduction to strikeslip: an example from the central New Zealand plate boundary. J Geophysl Res. 117. doi:10.1029/2011JB008640.
Wallace LM, Beavan J. 2010. Diverse slow slip behavior at the
Hikurangi subduction margin, New Zealand. J Geophys
Res. 115:B12402. doi:10.1029/2010JB007717.
Wallace LM, Beavan J, Bannister S, Williams C. 2012b.
Simultaneous long-term and short-term slow slip events
at the Hikurangi subduction margin, New Zealand: implications for processes that control slow slip event occurrence, duration, and migration. J Geophys Res. 117:
B11402. doi:10.1029/2012JB009489.
Wallace LM, Beavan J, McCaffrey R, Darby D. 2004.
Subduction zone coupling and tectonic block rotation in
the North Island, New Zealand. J Geophys Res. 109.
doi:10.1029/2004JB003241.
Wallace LM, Beavan RJ, McCaffrey R, Berryman KR, Denys
P. 2007. Balancing the plate motion budget in the South
Island, New Zealand using GPS, geological and seismological data. Geophys J Int. 168. doi:10.1111/j.1365-246X.
2006.03183.x.
Wallace LM, Eberhart-Phillips D. 2013. Newly observed,
deep slow slip events at the central Hikurangi margin,
New Zealand: implications for downdip variability of
slow slip and tremor, and relationship to seismic structure. Geophys Res Lett. 40:53935398. doi:10.1002/
2013GL057682.
Williams CA, Eberhart-Phillips D, Bannister S, Barker DHN,
Henrys S, Reyners M, Sutherland R. 2013. Revised interface geometry for the Hikurangi subduction zone, New
Zealand. Seismol Res Lett. 84:10661073. doi:10.1785/
0220130035.
Zachariasen J, Berryman K, Langridge R, Prentice C, Rymer
M, Stirling M, Villamor P. 2006. Timing of late Holocene
surface rupture of the Wairau Fault, Marlborough, New
Zealand. New Zeal J Geol Geophys. 49:159174.
Zachariasen J, Villamor P, Lee J, Lukovic B, Begg J. 2000.
Late Quaternary faulting of the Masterton and Carterton
Faults, Wairarapa, New Zealand. Client Report 2000/70.
Lower Hutt: Institute of Geological and Nuclear Sciences.

You might also like