Lithos 189 (2014) 65–76

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r .com/ locate / l i thos

Transportable seismic array tomography in southeast Australia:Illuminating the transition from Proterozoic to Phanerozoic lithosphere

N. Rawlinson ⁎, M. Salmon, B.L.N. KennettResearch School of Earth Sciences, The Australian National University, Canberra ACT 0200, Australia

⁎ Corresponding author.E-mail addresses: nick@rses.anu.edu.au, nicholas.raw

(N. Rawlinson).

0024-4937/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.lithos.2013.06.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 3 February 2013Accepted 3 June 2013Available online 15 June 2013

Keywords:Seismic tomographyBody wavesAustralian lithosphereGondwana evolution

The Phanerozoic Tasmanides of eastern Australia is comprised of a series of orogenic belts that developedalong the east margin of Gondwana following the breakup of the supercontinent Rodinia and subsequent for-mation of the Pacific Ocean. The tectonic complexities of this region have been well studied, but most workhas been confined to evidence collected from the near surface, where extensive Mesozoic and Cenozoic basincover masks large tracts of Palaeozoic basement. We apply teleseismic tomography to distant earthquakedata recorded by WOMBAT – the largest transportable seismic array experiment in the southern hemisphere –

to image P-wavespeed variations in the mantle lithosphere beneath the southern portion of the Tasmanidesin detail. In order to seamlessly suture together the teleseismic datasets from each of the 14 sub-arrays ofWOMBAT, we use P-wavespeeds from the AuSREM model to construct a laterally heterogeneous startingmodel that captures the long wavelength structural variations that would otherwise be lost through the use ofrelative arrival time residuals. Synthetic resolution tests indicate good horizontal resolution of ~50 km withinmost of the array between depths of 50–350 km. A key feature of the 3-D P-wavemodel is a pronounced easterlyhigh velocity salient in the mantle lithosphere beneath the northern limit of the New England Orocline, whichmay indicate the presence of underpinning Proterozoic lithosphere that was instrumental in its formation.Another pronounced high velocity anomaly underlies the Curnamona Province, a large crustal block with astrong Archean provenance, which is clearly separated from the Gawler Craton to the west at upper mantledepths by a low velocity zone beneath the Adelaide Fold Belt.We also estimate the location of the eastern bound-ary of Precambrian Australia at depth, and show that it extends eastward further than previously thought.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Although less than half of the Palaeozoic basement beneath thesouthern Tasmanides of eastern Australia is exposed at the surface,a remarkable tectonic history has been progressively revealed overthe last century, as more sophisticated techniques and methods ofanalysis have been brought to bear on an increasingly larger pool ofdata. Early researchers evidently had a good idea that long timescalesand almost unimaginable forces were at work; for instance Schuchert(1916) wrote of the Tasman Sea “It appears that vast land-masseshave been fractured, broken up and more or less permanently takenpossession of by the oceans”, a statement that by today's standardswould still be considered accurate given what is known of the open-ing of the Tasman Sea and the emplacement of the Lord Howe Rise asa rifted continental fragment from the eastern margin of Australia(Gaina et al., 1998). However, the notion of continental drift had yet toenter the early 20th century mindset, and large scale deformation was

linson@anu.edu.au

rights reserved.

generally attributed to an Earth that periodically shrinks, with parts ofthe crust either rising or sinking rather than moving horizontally.

Today, great strides have been made in understanding how theTasmanides evolved into its current state thanks to advances inmany different fields of Earth Sciences including geophysical imaging.For instance, high resolution gravity (e.g. Leaman and Richardson,1989; Murray et al., 1989; Roach et al., 1993; Spencer, 2004), magnetic(e.g. Glen, 2005; Gunn et al., 1997b; Hill et al., 1997; Mackey et al.,1995; Musgrave and Rawlinson, 2010) and reflection/wide-angleseismic studies (Cayley et al., 2011; Drummond et al., 2000; Finlaysonet al., 1998; Korsch et al., 2002; Rawlinson et al., 2001) have been instru-mental in probing the Palaeozoic crust and defining the main elementsof the southern Tasmanides. Passive seismic imaging has also playedan important role, particularly in differentiating between lithosphereof Phanerozoic, Proterozoic and Archean origin. For example, surfacewave tomography results from continent-wide broadband deploy-ments reveal that the mantle lithosphere beneath cratonic central andwestern Australia has markedly higher S-wavespeeds in comparisonto the Palaeozoic lithosphere beneath the Tasmanides (Debayle, 1999;Debayle and Kennett, 2000, 2003; Fishwick and Rawlinson, 2012;Fishwick et al., 2005, 2008; Simons et al., 1999, 2002; Zielhuis and vander Hilst, 1996), a result also largely mimicked in P-wavespeed and Q

66 N. Rawlinson et al. / Lithos 189 (2014) 65–76

as determined from regional bodywave tomography (Kennett, 2003;Kennett and Abdullah, 2011). This has allowed constraints to beplaced on the boundary between the Tasmanides and cratonic Australia(e.g. Kennett et al., 2004), but there is still considerable uncertainty as tothe nature of the transition (Direen and Crawford, 2003b).

A limitation of the Australian continent-wide broadband dataset isthat station separation is generally of the order of 200–400 km, whichplaces a restriction on the maximum horizontal resolution to around200–250 km. Since 1998, a series of targeted experiments – nowcollectively called WOMBAT – using highly portable short periodrecorders have been carried out in southeast Australia, with one ofthe main objectives being to record distant earthquakes for use inteleseismic tomography. With a station spacing of around 50 km orless, a number of studies have revealed detailed patterns of P-waveperturbations in the upper mantle that have allowed the presenceof a hot-spot, lithospheric boundaries and continental fragments tobe inferred (Clifford et al., 2008; Graeber et al., 2002; Rawlinson andKennett, 2008; Rawlinson and Urvoy, 2006; Rawlinson et al., 2006a,2006b, 2010b). Until 2010, these teleseismic datasets were treatedseparately, but the spatially contiguous nature of the arrays meantthat a joint inversion of the datasets would be far more revealing. Assuch the joint inversion results of Fishwick and Rawlinson (2012),Rawlinson and Fishwick (2012), and Rawlinson et al. (2011) haveallowed detailed inferences about the lithosphere beneath Tasmania,Victoria, eastern South Australia and southern New South Wales tobe made, including relationships between deep seated anomaliesand the distribution of mineral deposits and Cainozoic volcanic cen-tres at the surface.

135˚

135˚

140˚

140˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

Pha

nero

zoic

Pro

tero

zoic

Arc

hean

Mesozoic

Neoproterozoic

Mesoproterozoic

Paleozoic

Paleoproterozoic

RGP SCE HME GOZ CCE MGR

Magnetic inferred crustal boundary

Gravity inferred crustal boundary

CurnamonaProvince

Western S

Del

amer

ian

Oro

gen

Ade

laid

e F

old

Bel

t

Gawler Craton

Tho

Fig. 1. The crustal elements of southeast Australia, as inferred from potential field data (SHME = highly magnetic element; GOZ = geophysically overprinted zone; CCE = covered c1996, for more details).

In this study, we considerably expand on the recent work ofFishwick and Rawlinson (2012), Rawlinson and Fishwick (2012), andRawlinson et al. (2011) by including data from an additional five arraysin northern NSW and southern South Australia, representing a totalof 210 stations or a 53% increase on what was previously available.Furthermore, we also use an improved regional P-wavespeed model(Kennett and Salmon, 2012; Kennett et al., 2013; Salmon et al., 2012)to account for the long wavelength structure that is lost through the useof relative arrival times across multiple arrays. The new P-wavespeedmodel spans a far greater region of the southern Tasmanides and adjacentcratonic region, and allows new constraints to be placed on the tectonicevolution of Phanerozoic Australia.

2. Tectonic setting

The Tasmanides comprise approximately one-third of the Australianland mass, and is constructed from five Palaeozoic orogenic belts thatformed outboard of the Australian section of the east Gondwana —

proto-Pacific convergent margin between the Middle Cambrian andthe Carboniferous (Cawood, 1982; Glen, 2005). The characterisationand spatial extent of each orogen have been considerably assisted byhigh resolution gravity and magnetic datasets that have helped over-come the extensive lack of surface outcrop. As a result, rather thanshowing a map of surface geology, we instead display the distributionof geophysical domains (Fig. 1) as defined by Shaw et al. (1996) basedon gravity, magnetic and crustal element boundaries. This has theadvantage that major structural variations in the upper crust beneathMesozoic and Cainozoic cover are revealed, although it is worth noting

145˚

145˚

150˚

150˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚New England

OOrocline

New EnglandOrogen

Central Subprovince

Eastern Subprovince

Lachlan Orogen

ubprovince

mson Orogen

haw et al., 1996). RGP = relict geophysical pattern; SCE = standard crustal element;rustal element; MGR = sub-element with muted geophysical response (see Shaw et al.,

67N. Rawlinson et al. / Lithos 189 (2014) 65–76

that accuracy will decrease in areas of thick sedimentary cover andsparse data.

In southeast Australia, the Delamerian Orogen abuts the easternmargin of the Gawler Craton and continues south to include westernTasmania where it is commonly referred to as the Tyennan Orogen(Crawford et al., 2003; Direen and Crawford, 2003a, 2003b; Gibsonet al., 2011; Reed et al., 2002). East of the Gawler Craton, theDelamerian Orogen incorporates the Adelaide Fold Belt, which wasformed by the inversion of the Neoproterozoic–Cambrian AdelaideRift Complex, a thick sequence of mostly marine sediments thatrecords the formation of the eastern margin of Australia prior to con-vergent orogenesis in the Palaeozoic (Foden et al., 1999, 2006; Glen,2005). Outboard of the Adelaide Fold Belt lies the Kanmantoo FoldBelt, an inversion of the marine Kanmantoo Trough, which is com-posed largely of metamorphic sediments intruded by granitoids(Glen, 2005). Convergence along a west dipping subduction zoneagainst the eastern margin of Gondwana between Mid-Late Cambrianand early Ordovician is thought to have produced the DelamerianOrogen (Foden et al., 2006). The underlying basement is thought toconsist mostly of reworked Precambrian continental rocks (Direenand Crawford, 2003b), although there is little exposure in the eastbeneath Victoria and New South Wales. In Tasmania, numerous out-crops of Neoproterozoic rocks have been identified (Crawford et al.,2003), while in the northern region of the Delamerian Orogen, theCurnamona Province hosts the Palaeoproterozoic Willyama Super-group (Webster, 1996).

The Lachlan Orogen formed outboard of the Delamerian Orogenbetween the early Ordovician and early Carboniferous (Kemp et al.,2009). It can be divided into three distinct thrust systems (Foster et al.,2009) that constitute the Western, Central and Eastern subprovinces.The Western subprovince is an east-vergent thrust system with zonesof NW–SE andN–S trending structures; the Central subprovince consistsof a SW-vergent thrust belt with considerable high temperature/lowpressure metamorphism, and is sometimes referred to as the Wagga–Omeo Metamorphic Complex; the Eastern subprovince is characterisedby a north–south structural grain and east-directed faults (Foster et al.,2009), and hosts the Ordovician Macquarie Arc, an intra-oceanicsupra-subduction zone system that contains an abundance of Copper–Gold deposits (Glen et al., 2009, 2012). Glen (2005) identifies a fourthsubprovince – the Southwestern subprovince within the Westernsubprovince – based on the interpretation of potential field data,which appears to reveal a north–south change in geology.

Of the many important questions that still surround the tectonicevolution of the Lachlan Orogen, there are two which are particularlyrelevant to this study. The first regards the origin of the lithospherebeneath the orogen, which has variously been argued as either ocean-ic (Foster and Gray, 2000; Spaggiari et al., 2003, 2004) or mixed oce-anic and continental (Taylor and Cayley, 2000; VandenBerg, 1999;Willman et al., 2002). One possible source of continental crust is thesupercontinent Rodinia, the break-up of which may have set adrift anumber of continental fragments, one or more of which could havebecome incorporated in the convergent margin setting of the LachlanOrogeny. The Selwyn Block idea (Cayley, 2011; Cayley et al., 2002),which argues that western Tasmania and the substrate beneath theMelbourne Zone in Victoria are part of the same Proterozoic crustalfragment is one example of how this process may manifest. Thesecond question relates to the location and nature of the Delamerian–Lachlan boundary. The main candidates for the boundary are theAvoca Fault (Glen et al., 1992) and the Moyston Fault (Cayley andTaylor, 1998; Cayley et al., 2002; VandenBerg, 1999), although seismicreflection data (Cayley, 2011; Korsch et al., 2002) appears to show theMoyston Fault as an east-dipping boundary extending into the deepcrust, with the West dipping Avoca Fault soling into it at shallowerdepth. In an alternative model, Miller et al. (2005) argue that theMoyston Fault overlies a reworked orogenic zone that contains ele-ments of both the Delamerian and Lachlan orogens.

The Thomson Orogen lies to the north of the Lachlan Orogen, withthe boundary between the two obscured by Palaeozoic–Mesozoiccover (see Fig. 1). Potential field datasets reveal the boundary to becurvilinear, with the centre of curvature lying to the north. Reflectionseismic data suggest that the boundary is steeply dipping to the northand separates thick dense crust in the Thomson Orogen from morenormal layered crust in the Lachlan Orogen (Glen et al., 2007). As aresult of extensive sediment cover, little is known about the ThomsonOrogen, although it appears to be underlain by rocks that vary in agefrom Precambrian to Late Devonian (Glen, 2005).

The New England Orogen lies to the east of both the Thomson andLachlan orogens, where its boundary is obscured by Permian–MiddleTriassic sedimentary basins. Here, we only focus on the southern partof the orogen since it lies within the study area. Although the NewEngland Orogen represents the youngest part of the orogenic systemwhich formed the Tasmanides, it appears to be underlain by bothPalaeozoic oceanic and Precambrian rocks, with Re–Os ages, zirconmodel ages and Sm–Nd isochron ages pointing to the presence ofold lithosphere (Glen, 2005). One of the main structural features inthe south is the doubly vergent New England Orocline, which developedbetween 310 and 230 Ma via strong deformation of a pre-Permian arc as-semblage (Cawood et al., 2011; Rosenbaum, 2012). Cawood et al. (2011)explain the formation of this feature by invoking a tectonic model inwhich the arc system is buckled about a vertical axis as the southernpart of the arc moves northwards due to oblique sinistral strike-slip mo-tion between the Palaeo-Pacific and Gondwana plates, with the northernpart pinned relative to cratonic Gondwana. Rosenbaum et al. (2012) sug-gest an alternative model in which a west-dipping subduction zone rollsbackwith variable velocity along-strike, such that the southern end of thearc recedes more quickly and is forced to rotate northwards and eventu-ally fold back on itself, thus forming the orocline.

Following the Palaeozoic formation of the Tasmanides, a numberof significant tectonic events took place that considerably alteredthe nature of the orogen. Of particular relevance to this study is thebreak-up of Australia and Antarctica and the opening of the Tasmansea between 80 and 90 Ma (Gaina et al., 1998); this event resultedin significant lithospheric thinning near the passive margin of south-east Australia, and the formation of the Bass Basin between Victoriaand Tasmania (Gunn et al., 1997a). Another possible result of therifting process is the formation of the Southern Highlands inboardof eastern Australia in Victoria and NSW, which is consistent withthe development of an upper plate passive margin (Lister et al., 1991)in the Cretaceous. However, others (e.g. van der Beek et al., 1999)have suggested that the Southern Highlands are instead an erosionalremnant of a large mountain chain formed during late Palaeozoic oro-genesis. Even more recently, the Flinders Ranges in South Australiahas formed as a result of Neogene intracratonic orogenesis (Sandifordet al., 2004), with some studies indicating that the entire relief (maxi-mum elevation 1200 m) was created in as little as 4 My (Quigleyet al., 2007). Although increased coupling between the Australian andPacific plates during the Miocene may have triggered the event, it isunclear whether thermal or structural anomalies are responsible forthe strain focusing observed in this region. Perhaps the most recentevent to effect the lithosphere and surface geology of southeast Australiais the emplacement of theQuaternaryNewerVolcanics province inwest-ern Victoria, which only ceased some five thousand years ago. Evidentlyarising fromhot-spot volcanism (Price et al., 1997), theNewer Volcanics,together with numerous other Cenozoic eruption centres that populatethe eastern edge of the Australianmainland, has helpedmask older base-ment, and undoubtedly affected the lithosphere through which it hasmigrated.

3. Data and method

Teleseismic data for this study comes from the WOMBAT trans-portable seismic array project which has been operating in southeast

Fig. 3. Distribution of teleseismic sources used to constrain the 3-D P-wave structurebeneath southeast Australia. Dashed black circles are plotted at 30° increments fromthe centre of WOMBAT.

68 N. Rawlinson et al. / Lithos 189 (2014) 65–76

Australia since 1998. It comprises a core array of approximately 50instruments (although this varies depending on availability) that isprogressively moved from one location to the next in order to achievehigh-density coverage over a broad area (see Fig. 2). To date, over 600sites have been deployed as part of 14 separate array movements,with station spacing varying from 15 km in Tasmania to 50 km onthe mainland. Recently, a new array called SQEAL1 has been deployedto the north of EAL3, with SQEAL2 planned for deployment north ofEAL2 in mid 2013. Over the last 15 years, the instrumentation usedfor WOMBAT has gradually evolved; the early arrays used recordingdurations of approximately five months and 1 Hz vertical componentseismometers, but since 2006, recording durations of 9–14 monthshave been achieved using 1 Hz 3-component seismometers. Futuredeployments are expected to use compact broadband sensors. Passiveseismic experiments in southeastern Australia are able to exploit areasonably good distribution of earthquakes across the globe (Fig. 3).However, events to the north and east are far more frequent and somuch more data is available from these directions.

For thefirst time,we exploit teleseismic P-wave arrival time residualsfrom all 14 arrays shown in Fig. 2. Previous studies (e.g. Rawlinson et al.,2011; Rawlinson and Fishwick, 2012; Fishwick and Rawlinson, 2012)have only used data up to and including SEAL3 (nine arrays), so theadditional dataset that has become available represents a substantialincrease in coverage. Distant earthquakes are selected from the NEIC cat-alogue based on size, depth, angular distance and phase type. In general,all events at an angular distance greater than 27° with amomentmagni-tude of at least 5.2 are considered; however, any event at a depth inexcess of 150 km is also included if it has a moment magnitude greaterthan 4.6. Large and/or deep focus earthquakes at an angular distanceless than 27° are also examined for the presence of core reflection phases.The data processing schemeused to extract relative arrival time residualsis explained in detail in Rawlinson and Kennett (2008), Rawlinson et al.(2006b), and Rawlinson et al. (2011), so it is only briefly summarisedhere. Traces associated with the arrival of various global phases (e.g. P,ScP, PcP, PKiKP, Pdiff, PP) are windowed and aligned using predictionsfrom the global reference model ak135 (Kennett et al., 1995). Anyremaining lack of alignment can be largely attributed to lateral variationsin structure beneath the array. We use the adaptive stacking procedure

135˚

135˚

140˚

140˚

145˚

145˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

areaStudy

Fig. 2. Coverage of the WOMBAT transportable seismic array experiment in southe

of Rawlinson and Kennett (2004) to achieve final alignment, thus pro-ducing a set of arrival time residual estimates for each source. Themean is then removed to produce a set of relative arrival time residuals,which are less sensitive to errors in source origin time and long wave-length structure outside the model region.

Unlike the recent studies of Fishwick and Rawlinson (2012),Rawlinson and Fishwick (2012), and Rawlinson et al. (2011), we use afully automated version of the adaptive stacking code of Rawlinsonand Kennett (2004) to produce sets of relative arrival time residuals

EVA (2005−2006)

SEAL2 (2007)

SEAL3 (2007−2008)

AUSCOPE−CU (2009)

EAL1 (2009−2010)

EAL2 (2010−2011)

EAL3 (2011−2012)

LF98 (1998)

MB99 (1999)

AF00 (2000)

TIGGER (2001−2002)

SEAL (2004−2005)

AUSCOPE−GW (2008)

SETA (2006−2007)

150˚

150˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

ast Australia, which began in 1998, and comprises a total of 14 arrays to date.

69N. Rawlinson et al. / Lithos 189 (2014) 65–76

from earthquake and phase listings. One of the benefits of the adaptivestacking approach is that picking uncertainty is estimated, which can beused to discard noisy or incoherent phases. Here, we use a threshold of65 ms for picking uncertainty, above which phases are rejected. Theautomated approach has been validated through comparison withthose datasets (SEAL3 and earlier) that were picked using adaptivestacking within a more manual framework. The end result of the dataprocessing is a set of 143,850 relative arrival time residuals producedby a total of 3445 earthquake sources (Fig. 3) and 604 stations. Dueto the uneven distribution of sources, with most events occurringto the north and east of WOMBAT (Fig. 3), we apply data binning (seeRawlinson et al., 2011, for details on this technique) to help reducethe smearing effect that can be produced by clusters of rays originatingfrom the same source region. The final dataset used in the inversiontherefore consists of 54,581 residuals from 1344 sources. An exampleof the coherence of seismic phases across the individual arrays isshown in Fig. 4, where the arrivals are initially aligned using theak135 model of Kennett et al. (1995) and then refined using adaptivestacking. The results illustrate the high quality that is typical of muchof the WOMBAT dataset.

The inversion method used to map relative arrival time residualsas 3-D perturbations in P-wave velocity has been described inRawlinson and Kennett (2008, 2011) and Rawlinson et al. (2006b,2011). Structure is represented by a regular grid of nodes in latitude,longitude and depth, with cubic B-spline functions used to describea smoothly varying and locally controlled velocity continuum. Agrid-based eikonal solver known as the fast marching method orFMM (Rawlinson and Sambridge, 2004) is used to solve the forwardproblem of traveltime prediction, which involves tracking wavefronts

(a) lat=52.2, long=178.1, depth=138 km, M=5.5 (b) lat=6.7, long=123.4, d

P

Fig. 4. Three different phase types recorded by three different sub-arrays of WOMBAT that h(a) P arrival recorded by EAL1; (b) ScP arrival recorded by AuScope-CU; (c) PKiKP arrival relinear and quadratic stack respectively.

from the bottom of a local 3-D model, where traveltimes are defined bythe ak135 model, to the receivers at the surface. Model parameters,which include velocity node values and station terms – the latter usedto account for unresolved structure in the receiver neighbourhood –

are adjusted to satisfy the data using a subspace inversion technique(Kennett et al., 1988), subject to damping and smoothing regularisation.

One challenge that arises from the simultaneous use of differentrelative arrival time datasets is that long wavelength structure, equalto or larger than the horizontal dimensions of the component arrays,will be absent from the final model. Rawlinson and Fishwick (2012),and Rawlinson et al. (2011) address this by using a longwavelength ini-tial model based on results from regional surface wave tomography.However, aside from the assumptions that need to be made to convertfrom S-wave to P-wave velocity, the nature of the regularisation usedto constrain the surface wave model will influence the final P-wavemodel result in a way that is difficult to quantify. Others have used amore direct approach of joint inversion of surface and body waves(Obrebski et al., 2011; West et al., 2004), but in the case of WOMBAT,there is still the problem of S–P-wave velocity conversion. Moreover,it appears to make little sense to try and invert two datasets that differso markedly in coverage and resolution. Here we make use of the re-cently released Australian Seismological Reference Model (AuSREM)to provide a robust long wavelength P-wave velocity model as prior in-formation in the bodywave inversion. Details of the AuSREMmodel andhow it is constructed are provided in detail in Kennett and Salmon(2012), Kennett et al. (2013), and Salmon et al. (2012). The relevantpoints for this study are that it contains an isotropic crustal and mantlecomponent andMoho geometry all at a horizontal sampling of 0.5°withboth P and S velocities available. The crustal component is based on

epth=620 km, M=5.5 (c) lat=−8.1, long=−71.6, depth=581 km, M=6.5

ScP PKiKP

ave been aligned using the adaptive stacking method of Rawlinson and Kennett (2004).corded by EAL2. The top two traces in each case – labelled zssl and zscp – represent the

70 N. Rawlinson et al. / Lithos 189 (2014) 65–76

sediment thickness databases, reflection profiling, refraction and wide-angle reflection experiments, receiver functions and ambient noisetomography. Themantle component is based on surfacewave tomogra-phy, body wave studies and regional tomography. In addition to usingthemantle component of themodel, we also include the crustal compo-nent to help account for near-surface structural variations that cannotbe constrained by the teleseismic dataset. Although these shallow vari-ations are approximate due to the use of a smooth parameterisationwhich is not capable of explicit representation of the Moho, it is none-theless an improvement compared to previous work. According toAuSREM, Moho depth varies between 27 km and 52 km beneath thestudy region. Given that velocity contrasts across the Moho can be upto 2 km/s or more, this variation in crustal thickness may contributesignificantly to the measured traveltime residuals. However, the inclu-sion of a smoothed crustal layer coupled with explicit station terms inthe inversion should help minimise any smearing into the underlyingmantle.

4. Results

The local 3-D model region used for the inversion of WOMBATP-wave arrival time residuals spans a latitude range of 18.7°, a longi-tude range of 23.4° and a depth range of 362 km. The latter value isbased on the estimate that the depth limit of resolution is approximatelyequal to two-thirds the array aperture (Evans and Achauer, 1993);in this case, the aperture of the largest sub-array within WOMBAT.The node spacing of the velocity grid is approximately 20 km in eachdimension, resulting in a total of 209,475 unknowns. An additional604 unknowns in the form of station terms to absorb localised crustalvariations are also constrained during the inversion process. Six itera-tions of the subspace inversion scheme are used to obtain a solution,

135˚

135˚

140˚

140˚

1−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

1−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

7.8 8.0 8.2

(a) (b)

90 km

90 km

Input

Output

Fig. 5. Horizontal depth sections through the synthetic checkerboard model (top) and recovThe smoothly varying background velocities are provided by AuSREM. E–W red and N–Scross-sections shown in Fig. 6.

with traveltimes and Fréchet derivatives updated after each iterationusing FMM. Damping and smoothing regularisation are used to achievean optimum trade-off between satisfying the data and recovering amodel that minimises unwarranted short wavelength features andstrong departures from the initial model (Rawlinson and Kennett,2008; Rawlinson et al., 2006b). Although the initial model is definedby AuSREM, model arrival time residuals are calculated with referenceto the laterally homogeneous ak135 model, which means that residualcontributions arising from lateral velocity variations in AuSREM areaccounted for, i.e. the inversion will only introduce structure that isnot already contained in AuSREM. Before presenting the tomographicmodel from the WOMBAT dataset, we first carry out several syntheticreconstruction tests to provide some insight into the robustness of theinversion results.

4.1. Synthetic tests

Synthetic reconstruction tests are commonly used in seismic to-mography to help assess the scale-length of features that the datasetis capable of recovering (Rawlinson and Sambridge, 2003; Rawlinsonet al., 2010a). The basic idea is to construct a model that might featurea single or multitude of velocity perturbations and then generate asynthetic dataset by using the same source–receiver combinationsand phase types as the observational dataset. Application of the inver-sion scheme to this dataset with the same initial parameters (initialmodel, damping, smoothing) as are usedwith the observational datasetyields a result that can be compared to the true model. The checker-board reconstruction test, in which an alternating pattern of high andlow wavespeeds is used, is by far the most common type of synthetictest found in the published literature. Although the limitations ofsuch tests are well known (Lévêque et al., 1993; Nolet, 2008), their

45˚

145˚

150˚

150˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

45˚

145˚

150˚

150˚

−44˚

−42˚

−40˚

−38˚

−36˚

−34˚

−32˚

−30˚

−28˚

8.4 8.6

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

(c)

Vp (km/s)

250 km

170 km

170 km

Input

Output

Input

Output

250 km

C’ D’

CD

A

B

A’

B’

ered model (bottom) at three different depths (a) 90 km, (b) 170 km and (c) 250 km.blue lines superimposed onto the synthetic model in (c) denote the locations of the

71N. Rawlinson et al. / Lithos 189 (2014) 65–76

simplicity, both in implementation and interpretation, hasmade them avery attractive option.

Two synthetic checkerboard tests have been carried out to deter-mine which regions of the model are well controlled by the teleseismicdata. In the first test the synthetic dataset is computed in the presenceof a checkerboard model which is superimposed on the AuSREM back-ground model. Gaussian noise with a standard deviation of 50 ms isadded to the relative arrival times to simulate the picking uncertaintyassociated with the real data. Figs. 5 and 6 show a comparison betweenthe synthetic model and the reconstructed model after six iterations ofthe tomographic scheme. The checkerboard pattern has been chosensuch that the interface between the individual blocks is representedby a sharp velocity gradient, which is difficult to recover. As shown inboth Figs. 5 and 6, the tendency is to both smooth out the sharp velocitygradient and underestimate the amplitude of the velocity perturbations,which is typical of regularised iterative non-linear schemes. None-the-less, the checkerboard pattern is generally well recovered within thebounds of the receiver array. As is most easily seen in Fig. 5, wherepath coverage is absent, the recovered or output model is unchangedfrom the startingmodel, which is defined by AuSREM. Due to the natureof the incident paths, which converge to the receivers, the horizontalextent of the zone of good resolution tends to increase with depth(cf. Fig. 5a and c).

When raypaths have a dominant direction, there is a tendency tosmear out velocity anomalies in this direction. This can be easilyseen near the edges of the cross-sections in Fig. 6. Furthermore, theheavy concentration of sources to the north of WOMBAT has resultedin a tendency to smear structure at approximately 45° to the verticalin the northerly direction (see Fig. 6b, right). The lack of stations be-neath Bass Strait, which separates Tasmania and Victoria, has resultedin a gap in the recovered checkerboard, that is most easily seen inFig. 6b, bottom right. Although this gap tends to close up as depth in-creases, the relatively small apertures of the two Tasmanian arraysmean that resolution is limited below about 200 km depth. Overall,however, the recovery of the synthetic model is sufficient to feel con-fident that the pattern of anomalies revealed by the inversion ofWOMBAT data will be largely correct, although more caution is re-quired in the interpretation of absolute amplitude. It is worth notingthat at this point absolute velocities are used in Fig. 5, whereas

−300

−200

−100

−300

−200

−100

−300

−200

−100

−300

−200

−100

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

−0.8 −0.6 −0.4 −0.2 0

Longitude

Latitude

Dep

th (

km)

Dep

th (

km)

Dep

th (

km)

Dep

th (

km)

InputA

31.2 S

140.3 E

A’

B B’

C

36.9 S

146.5 E

D

C’

D’

(a)

(b)

Fig. 6. Vertical slices through the synthetic checkerboard model (left) and recovered model (riSlice locations are denoted in Fig. 5c (top). Note that compared to Fig. 5, velocity perturbation

velocity perturbations relative to a horizontally averaged AuSREMmodel are used in Fig. 6. The reason for this is that dominant varia-tions in wavespeed occur with depth, but our interest is in departuresfrom this trend, which for vertical slices are best seen by removingthe depth-dependent component of the velocity field.

The second test is designed to test the sensitivity of the teleseismicdataset to long-wavelength perturbations in wavespeed. Fig. 7 showsthe results of a checkerboard test in which the size of each anomaly isclearlymuch larger than the aperture of the component arrays ofWOM-BAT. The result shown in the right hand panel of Fig. 7 ostensiblyappears to be quite good, with the boundaries between neighbouringanomalies recovered accurately almost without exception. However,given the data density, one would normally expect that the smoothvelocity variation that defines each anomaly (essentially AuSREM witha dc offset) is well recovered, but clearly this is not so, particularly inthe case of the anomaly centred at 142° east and 34° south. In thiscase, the velocity is high in the NW corner, but decreases rapidly tothe south (Fig. 7, right). The reason this happens is that several arraysjoin together within the bounds of this anomaly (see Fig. 7, right), andsince themean is separately removed from the arrival times on a sourceby source basis for each array, the longwavelength featureswill only beproperly recovered if the average velocity of the starting model is cor-rect beneath each array. Consequently, there is a need for an accuratelongwavelength startingmodel for the inversion ofmultiple teleseismicdatasets, although aswe have shownhere, sharp velocity gradientsmaybe correctly recovered in the absence of such information.

4.2. Seismic structure of southeast Australia

For the inversion of the WOMBAT dataset, identical velocity gridspacing, initial model and regularisation to the synthetic test is used.After six iterations of subspace inversion and FMM, the RMS arrivaltime residual is reduced from 260 ms to 147 ms, which correspondsto a variance reduction of 68%. The RMS uncertainty of the WOMBATdataset has been estimated at around 50 ms by adaptive stacking, al-though it is generally a better predictor of relative rather than absoluteuncertainty. Nevertheless, it is clear that the model could be improvedso as to better satisfy the data, but a number of limitations of the meth-odmake this difficult. These include the nonlinear-nature of the inverse

−300

−200

−100

−300

−200

−100

−300

−200

−100

−300

−200

−100

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

.0 0.2 0.4 0.6

Longitude

Latitude

Output

dVp (km/s)

A

B

31.2 S

140.3 E

A’

C’

36.9 S

146.5 E

B’

C

D D’

ght). (a) E–W slices taken at constant latitude; (b) N–S slices taken at constant longitude.s are differential with respect to a 1-D background model.

7.8 8.0 8.2 8.4 8.6Vp (km/s)

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

150 km

Input

150 km

Output

Fig. 7. Horizontal slice through the long wavelength checkerboard model at 150 km depth, showing both the input synthetic model (left) and the output recovered model (right).Station locations from Fig. 2 are superimposed on the recovered model for reference.

72 N. Rawlinson et al. / Lithos 189 (2014) 65–76

problem, which may not be fully addressed by an iterative nonlinearapproach; the use of a regular parameterisation and the application ofsmoothing and damping; and the assumption of high frequency wavepropagation. However, it is worth noting that the RMS residual of thesynthetic test shown in Figs. 5 and 6 is 172 ms, which is 25 ms greaterthan the 147 ms residual achieved with the WOMBAT dataset. In thiscase, much of the misfit is due to the fact that a smooth solutionmodel is required; if the checkerboard pattern is described by a smoothsinusoidal pattern rather than one that approaches a box car function,then the misfit can be reduced to a value much closer to the standarddeviation of the added noise.

Figs. 8 and 9 show a series of horizontal and vertical slices throughthe AuSREM and WOMBAT models, using an identical set of cuts toFigs. 5 and 6, which show the checkerboard reconstruction test re-sults. For each slice, AuSREM is shown in addition to the WOMBATmodel in order to provide a measure of the additional structurerequired by the teleseismic data. Clearly, the general effect of theteleseismic data is to superimpose smaller scale structure on thestarting model. The basic first order trend of the AuSREM models isa gradual decrease in velocity to the southeast; the addition of theteleseismic data does nothing to change this, which is entirelyexpected. However, in addition to the obvious small scale featuresthat are introduced, at least one broad scale structure is enhanced —

the high velocity region that protrudes south at about 142° E, 33°south (Fig. 8b) becomes more pronounced with the addition ofteleseismic body wave data. The vertical cross-sections (Fig. 9) alsoshow a duality in scale with little evidence that the AuSREM modelis anything more than a smoothed version of the combined results.One obvious exception is the B–B′ cross-section at 36.9° south inFig. 9a, where a strong contrast from higher to lower velocities is in-troduced at around 143° E, which is absent from AuSREM.

Although there is some evidence of smearing in the combined so-lution model, particularly in the N–S sections (Fig. 9b), where thepropensity for northerly dipping anomalies should be viewed withcaution, most of the small scale structure within the bounds of thearray is well constrained according to the synthetic test results ofFigs. 5 and 6. It is important, however, to be aware when attemptingto interpret Figs. 8 and 9, that a distinct switch in resolving power,from ~50 km to≥200 km is present near the bounds of the combinedarrays. Another factor to be aware of is the limited reliability of the

model for depths shallower than about 70 km on the mainland and40 km in Tasmania. This is because station spacing largely dictatesthe minimum depth at which rays still cross; above this depth,there is little constraint on structure. The inclusion of a smoothed ver-sion of the AuSREM crustal model and the use of station terms willhelp prevent unresolved shallow structure from being erroneouslymapped to greater depths, but tests carried out by Rawlinson andKennett (2008) indicate that anomalies shallower than 70 km depthon the mainland are not reliable. The presence of the Southern High-lands parallel to the southeast coast of Australia, with maximum ele-vations exceeding 2 km, means that there is likely to be a substantialroot at the base of the crust. Without accounting for this explicitlythrough the inclusion of a Moho interface, it would be dangerous tointerpret velocity variations in the mantle immediately below thecrust in this region.

5. Discussion

Previous studies using teleseismic data from various subsets of theWOMBAT array (Clifford et al., 2008; Fishwick and Rawlinson, 2012;Graeber et al., 2002; Rawlinson and Fishwick, 2012; Rawlinson andKennett, 2008; Rawlinson and Urvoy, 2006; Rawlinson et al., 2006a,2006b, 2010b, 2011) have focused on velocity structure beneath thesouthern half of WOMBAT (Fig. 1), including those associated withthe Newer Volcanics province in western Victoria, the transitionfrom Proterozoic to Palaeozoic lithosphere in Victoria and southernNew South Wales, and the boundary between the Western and East-ern Tasmania terranes. In our interpretation of the new model, wefocus primarily on anomalies beneath central and northern New SouthWales, southernmost Queensland and southern South Australia, asthese have not been previously imaged using WOMBAT data. We alsoexamine how the use of an improved starting model in the form ofAuSREM may have influenced the results in comparison to previousstudies in the south.

Before discussing the results in detail, it is worth briefly sum-marising the relationship between seismic velocities and the physicaland compositional properties of the Earth. Unfortunately, this rela-tionship is highly non-unique, which means that in addition to theuncertainties in the model that are produced due to variations indata coverage and the limitations inherent to seismic tomography,

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

7.8 8.0 8.2 8.4 8.6

(a) (b) (c)

Vp (km/s)

250 km90 km

90 km 170 km

170 km

250 km

A

B B’

A’

AuSREM AuSREM AuSREM

AuSREM + WOMBAT AuSREM + WOMBAT AuSREM + WOMBAT

DC

C’ D’

Fig. 8. Horizontal depth sections through AuSREM (top) and combined model (bottom) obtained by the inversion of teleseismic arrival time residuals. Horizontal red and verticalblue lines superimposed onto the synthetic model in (c) denote the locations of the cross-sections shown in Fig. 9.

73N. Rawlinson et al. / Lithos 189 (2014) 65–76

our ability to make inferences about the underlying geology faces anadditional challenge. Seismic P-wavespeeds are largely dependent ontemperature, composition, the presence of melt, grain size, solid phasetransformations and anisotropy, although in the upper mantle, temper-ature appears to be the dominant factor. For example, Cammarano et al.

−300

−200

−100

−300

−200

−100

−0.8 −0.6 −0.4 −0.2 0

−300

−200

−100

−300

−200

−100

134 136 138 140 142 144 146 148 150 152 154

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

−44 −42 −40 −38 −36 −34 −32 −30 −28

Longitude

Latitude

Dep

th (

km)

Dep

th (

km)

Dep

th (

km)

Dep

th (

km)

AuSREMA

31.2 S

140.3 E

B

C

36.9 S

146.5 E

D

A’

B’

C’

D’

(a)

(b)

Fig. 9. Vertical slices through the AuSREM (left) and combined model (right) obtained by thshown in Fig. 6. See Fig. 8c (top) for location information relative to the horizontal sections

(2003) found that a 100 °C increase in temperature at 200 km depthcan produce a velocity perturbation of as much as 1%. The effects ofcomposition are more difficult to quantify, with estimates of variationsin wavespeed as a result of realistic changes in composition varyingbetween 1 and 5% (Cammarano et al., 2003; Griffin et al., 1998;

−300

−200

−100

−300

−200

−100

−300

−200

−100

.0 0.2 0.4 0.6

−300

−200

−100

134 136 138 140 142 144 146 148 150 152 154

134 136 138 140 142 144 146 148 150 152 154

−44 −42 −40 −38 −36 −34 −32 −30 −28

−44 −42 −40 −38 −36 −34 −32 −30 −28

Longitude

LatitudedVp (km/s)

AuSREM + WOMBATA

B

31.2 S

140.3 E

A’

C’

36.9 S

146.5 E

B’

C

D D’

e inversion of teleseismic arrival time residuals. Slice orientations are identical to that. Note that velocity variations between 0 and 50 km depth are not shown.

74 N. Rawlinson et al. / Lithos 189 (2014) 65–76

Sobolev et al., 1996). In a comparison of laboratory study results withseismic shear wave and attenuation models, Faul and Jackson (2005)found that reasonable temperature differences could explain the veloc-ity variations seen within continental interiors. Thus, the strong changefrom high wavespeeds beneath cratonic Australia to lower speeds be-neath the Tasmanides observed in surface wave tomography may be aresult of older and colder lithosphere juxtaposed against younger andhotter lithosphere. However, a more recent study (Dalton and Faul,2010) shows that while seismic velocity variations in the upper mantleappear to be largely controlled by lateral temperature variations in drymelt-free olivine, outliers do occur. Beneath oceanic regions, these out-liers can be explained by the presence of partial melt, whereas beneathcratonic regions chemically depleted lithosphere (higher Mg#) appearsto play a role in increasing wavespeeds, as reasonable temperaturedecreases alone cannot explain the high velocities that are sometimesobserved above 200 km depth.

Fig. 10 shows a 150 km depth slice through the new WOMBATteleseismic model with crustal element boundaries and magneticand gravity lineaments from Fig. 1 superimposed. In addition, a lineshowing the possible location of the eastern limit of the Precambrianlithosphere is included, based approximately on the 8.1 km/s con-tour. Although the choice of this value as a proxy for the transition be-tween the lithosphere and underlying mantle is somewhat arbitrary,it does correspond to a zone of elevated horizontal velocity gradient.This can also be seen in the E–W cross-sections shown in Fig. 9a,where a distinct lithospheric scale contrast in velocity is clearlypresent. In the study of Rawlinson et al. (2011), a similar contrastin velocity is seen in western Victoria and southern NSW, and isinterpreted as the presence of Delamerian lithosphere of Proterozoiccontinental origin underlying the Stawell Zone of the WesternSubprovince of the Lachlan Orogen. The extension northward of thetomography results in this study clearly shows that the transitionzone is not a simple N–S lineament, but trends northeast, and has anumber of distinct salients and re-entrants.

One of the most distinct salients of the inferred transition zoneshown in Fig. 10 is that which extends beneath the New EnglandOrocline in the northeast (see Fig. 1). The New England Orogen is

135˚

135˚

140˚

140˚

145˚

145˚

150˚

150˚

−44˚ −44˚

−42˚ −42˚

−40˚ −40˚

−38˚ −38˚

−36˚ −36˚

−34˚ −34˚

−32˚ −32˚

−30˚ −30˚

−28˚ −28˚

7.8 8.0 8.2 8.4 8.6Vp (km/s)

150 km

Fig. 10. Horizontal slice at 150 km depth through the combined WOMBAT-AuSREMmodel. The locations of magnetic (black dashed lines) and gravity (red dashed lines)lineaments, plus inferred crustal boundaries from Fig. 1 have been superimposed forreference. The thick dashed magenta line denotes the inferred easterly limit of the Pre-cambrian lithosphere.

clearly observed near the surface by strongly curved magnetic andgravity lineaments, and appears to be one of the most contortedorogens in the world (Rosenbaum, 2012). In the study of Cawoodet al. (2011), the eastern margin of Gondwana is assumed to be rela-tively linear prior to oroclinal folding; this is based on younging direc-tions identified within individual thrust slices that make up theorocline, which point to a progressive imbricate thrusting of oceaniccrust and trench sediments along a linear convergent margin. A keyelement of the reconstruction put forward by Cawood et al. (2011)to explain the formation of the New England Orocline is that southernelements of the subduction system are able to migrate northwardsrelative to northern elements, which are effectively pinned in placeby the presence of the Gondwana craton. Remarkably, the suggestedlocation of the buried tip point of the Gondwana craton in Fig. 9 ofCawood et al. (2011) corresponds almost exactly with the distinctsalient observed in the northeast region of Fig. 10, which appears tounderpin the strongly curved gravity and magnetic lineaments whichcharacterise the New England Orocline. One caveat with our resultsis that the northerly extent of the salient is not well constrained, asresolution drops off rapidly north of about 29° south. Thus, the narrow“finger” that is evidentmay in fact simply be the southerly tip of amuchlarger high velocity region that extends northwards. However, even ifthis is the case, it appears that what is probably Precambrian litho-sphere of continental origin lies beneath the northern part of the NewEngland Orocline; this is also supported by geochemical evidence inthe form of Neoproterozoic Re–Os ages and Neoproterozoic zirconmodel ages (Glen, 2005).

The Curnamona Province (Fig. 1) lies within the Delamerian Orogen,and was initially regarded as a piece of craton (Thomson, 1970) owingto its apparent immunity to tectonic alteration since the Neoproterozoic.However, ample evidence for Middle to Late Cambrian deformation andmetamorphism associated with the Delamerian Orogeny towards theedges of the crustal block has resulted in the label “Curnamona Craton”being dropped (Conor and Preiss, 2008). Much of the CurnamonaProvince is buried, with aeromagnetic data revealing its full 50,000 km2

extent beneath Mesozoic–Quaternary cover (see Fig. 1). Regional sur-face wave tomography has hinted at its presence (e.g. Fishwick andRawlinson, 2012) by way of a high velocity zone that extends to depthsof at least 200 km – testament to its Archean origins – but does notshow any separation between it and the Gawler Craton to the west.This is also true of the AuSREMmantlemodel, which exhibits a southerlysalient in this region (Fig. 8b, top). In both Figs. 10 and 8b, bottom, amoredistinct high velocity region can be observed beneath the CurnamonaProvince, which terminates to the west at the Adelaide Fold Belt (seeFig. 1). To the north, the high velocity zone appears to bend around thenortherly limit of the Adelaide Fold Belt and connect with the main cra-tonic elements of Precambrian Australia. This is consistent with the ideathat the Curnamona Province was once an easterly continuation ofthe Gawler Craton prior to the Neoproterozoic (Belousova et al., 2006;Hand et al., 2008; Wade et al., 2012).

In a previous study using WOMBAT teleseismic data, Fishwick andRawlinson (2012) converted a regional S-wave model into a P-wavemodel – assuming a constant Vp/Vs ratio – and used it as a startingmodel to invert for P-wave structure south of about 32° S. A compar-ison of the P-wave model of Fishwick and Rawlinson (2012) with theone produced in this study shows good general agreement wherethey overlap — for instance the location of the Lachlan–Delamerianboundary and the geometry of the low velocity zone associatedwith the hot spot beneath the Newer Volcanics province. However,there are some notable differences, such as the magnitude of thevelocity decrease towards the passive margin, which is less in Fig. 11(bottom left) of Fishwick and Rawlinson (2012) than it is in Fig. 10 ofthis paper. Another difference is that the velocities in the deep litho-sphere (150+ kmdepth) beneath themore cratonic region of Australiaare higher in Fishwick and Rawlinson (2012) (~8.6 km/s) than they arehere (8.3–8.4 km/s). The higher velocities of the previous model seem

75N. Rawlinson et al. / Lithos 189 (2014) 65–76

less likely to be correct, and are probably due to the very approximatemethod of constructing the initial P-wave model. As the WOMBATdataset gradually expands with the deployment of more arrays andmethods of analysis improve, variations in the results such as theseare to be expected. The hybrid approach of combining a long wave-length initialmodelwith accurate relative arrival time residuals appearsto be robust, with the only real weakness appearing to be a lack of goodconstraint on absolute velocity; however, this is a problem which af-flicts most forms of traveltime tomography due to the reliance onregularised inversion.

6. Conclusions

A new P-wave model of the lithosphere beneath southeast Australiahas been constructed by combining long wavelength constraints fromthe newly constructed Australian Seismological Reference Earth Model(AuSREM) with shorter-wavelength constraints from teleseismic datacollected by 14 sub-arrays of the WOMBAT transportable array experi-ment. In the mantle lithosphere, the first-order pattern of velocity vari-ation sees higher velocities beneath cratonic regions in South Australia,and lower velocities beneath the central and eastern Tasmanides, whichlikely corresponds to a transition from Precambrian lithosphere of con-tinental origin, to thinner Palaeozoic lithosphere which appears to havea mixed oceanic and continental ancestry. Towards the passive marginof the continent velocities in the uppermost mantle decrease even fur-ther, which may be due to the combined effect of a thinner lithosphereand thicker crust beneath the Southern Highlands. On a smaller scale,the main features of interest that differentiate this model from thosepreviously published in the literature are the presence of a high velocitysalient beneath the New England Orocline, which may reflect the pres-ence of a piece of cratonic Gondwana that forms the pinning point nec-essary for the northward buckling of the pre-existing linear arc system;the presence of a distinct high velocity zone in the mantle beneath theCurnamona Province, which demonstrates a separation at depth withthe Gawler Craton to thewest; and the northeast extension of an appar-ent boundary between thicker and thinner lithosphere, which suggeststhat basement of Proterozoic origin may be present beneath much ofthe Tasmanides, including the Thomson Orogen. Interestingly, the lith-osphere beneath western Tasmania, which outcrops at the surface asProterozoic rocks and appears to have compositional affinities withthe mainland Delamerian Orogen, is characterised by relatively low ve-locities, although thismay be a function of limited resolution in AuSREMand a lack of constraints at depth in the teleseismic dataset due to therelatively small apertures of the two Tasmanian arrays.

Acknowledgements

Thisworkwas supported by the Australian Research Council Discov-ery Grant DP120103673. G. Randy Keller and an anonymous reviewerare thanked for their constructive comments which significantly im-proved the paper.

References

Belousova, E., Preiss, W., Schwarz,M., Griffin,W., 2006. Tectonic affinities of the HoughtonInlier, South Australia: U–Pb and Hf-isotope data from zircons in modern streamsediments. Australian Journal of Earth Sciences 53, 971–989.

Cammarano, F., Goes, S., Vacher, P., Giardini, D., 2003. Inferring upper-mantle temper-atures from seismic velocities. Physics of the Earth and Planetary Interiors 138,197–222.

Cawood, P.A., 1982. Terra Australia orogen: Rodinia breakup and development of thePacific and Iapetus margins of Gondwana during the Neoproterozoic and Paleozoic.Geophysical Journal of the Royal Astronomical Society 69, 339–354.

Cawood, P.A., Pisarevsky, S.A., Leitch, E.C., 2011. Unraveling the New England orocline, eastGondwana accretionarymargin. Tectonics 30. http://dx.doi.org/10.1029/2011TC002864.

Cayley, R., 2011. Exotic crustal block accretion to the eastern Gondwana margin in theLate Cambrian–Tasmania, the Selwyn Block, and implications for the Cambrian–Silurian evolution of the Ross, Delamerian and Lachlan orogens. Gondwana Research19, 628–649.

Cayley, R.A., Taylor, D.H., 1998. The Lachlan margin Victoria: the Moyston Fault, anewly recognised terrane boundary. Geological Society Australia Abstracts 49, 73.

Cayley, R., Taylor, D.H., VandenBerg, A.H.M., Moore, D.H., 2002. Proterozoic–EarlyPalaeozoic rocks and the Tyennan Orogeny in central Victoria: the Selwyn Blockand its tectonic implications. Australian Journal of Earth Sciences 49, 225–254.

Cayley, R., Korsch, R.J., Moore, D.H., Costello, R.D., Nakamura, A., Willman, C.E., Rawling,T.J., Morand, V.J., Skladzien, P.B., O'Shea, P.J., 2011. Crustal architecture of CentralVictoria: results from the 2006 deep crustal reflection seismic survey. AustralianJournal of Earth Sciences 58, 113–156.

Clifford, P., Greenhalgh, S., Houseman, G., Graeber, F., 2008. 3-d seismic tomography ofthe Adelaide fold belt. Geophysical Journal International 172, 167–186.

Conor, C.H.H., Preiss, W.V., 2008. Understanding the 1720–1640 Ma PalaeoproterozoicWillyama Supergroup, Curnamona Province, Southeastern Australia: implicationsfor tectonics, basin evolution and ore genesis. Precambrian Research 166, 297–317.

Crawford, A.J., Meffre, S., Symonds, P.A., 2003. 120 to 0 Ma tectonic evolution of thesouthwest Pacific and analogous geological evolution of the 600 to 220 Ma TasmanFold Belt System. Geological Society of Australia Special Publication 22 andGeologicalSociety of America Special Publication, 372, pp. 383–404.

Dalton, C.A., Faul, U.H., 2010. The oceanic and cratonic upper mantle: clues from jointinterpretation of global velocity and attenuation models. Lithos 120, 160–172.

Debayle, E., 1999. SV-wave azimuthal anisotropy in the Australian upper mantle:preliminary results from automated Rayleigh waveform inversion. GeophysicalJournal International 137, 747–754.

Debayle, E., Kennett, B.L.N., 2000. The Australian continental upper mantle: structureand deformation inferred from surface waves. Journal of Geophysical Research 105,25,423–25,450.

Debayle, E., Kennett, B.L.N., 2003. Surface wave studies of the Australian region. In:Hillis, R.R., Müller, R.D. (Eds.), The Evolution and Dynamics of the Australian Plate,pp. 25–40.

Direen, N.G., Crawford, A.J., 2003a. Fossil seaward-dipping reflector sequences pre-served in southeastern Australia: a 600 Ma volcanic passive margin in easternGondwanaland. Journal of the Geological Society 160, 985–990.

Direen, N.G., Crawford, A.J., 2003b. The Tasman Line: where is it, what is it, and is itAustralia's Rodinian breakup boundary? Australian Journal of Earth Sciences 50,491–502.

Drummond, B.J., Barton, T.J., Korsch, R.J., Rawlinson, N., Yeates, A.N., Collins, C.D.N.,Brown, A.V., 2000. Evidence for crustal extension and inversion in eastern Tasmania,Australia, during the Neoproterozoic and Early Palaeozoic. Tectonophysics 329, 1–21.

Evans, J.R., Achauer, U., 1993. Teleseismic tomography using the ach method: theoryand application to continental scale studies. In: Iyer, H.M., Hirahara, K. (Eds.), SeismicTomography: Theory and Practice. Chapman & Hall, London, pp. 319–360.

Faul, U.H., Jackson, I., 2005. The seismological signature of temperature and grain sizevariations in the upper mantle. Earth and Planetary Science Letters 234, 119–134.

Finlayson, D.M., Collins, C.D.N., Lukaszyk, I., Chudyk, E.C., 1998. A transect acrossAustralia's southern margin in the Otway Basin region: crustal architecture andthe nature of rifting from wide-angle seismic profiling. Tectonics 288, 177–189.

Fishwick, S., Rawlinson, N., 2012. 3-D structure of the Australian lithosphere fromevolving seismic datasets. Australian Journal of Earth Sciences 59, 809–826.

Fishwick, S., Kennett, B.L.N., Reading, A.M., 2005. Contrasts in lithospheric structurewithin the Australian craton — insights from surface wave tomography. Earthand Planetary Science Letters 231, 163–176.

Fishwick, S., Heintz, M., Kennett, B.L.N., Reading, A.M., Yoshizawa, K., 2008. Stepsin lithospheric thickness within eastern Australia, evidence from surface wavetomography. Tectonics 27. http://dx.doi.org/10.1029/2007TC002116.

Foden, J., Sandiford, M., Dougherty-Page, J., Williams, I., 1999. Geochemistry and geo-chronology of the Rathjen Gneiss: implications for the early tectonic evolution ofthe Delamerian Orogen. Australian Journal of Earth Sciences 46, 377–389.

Foden, J., Elburg, M.A., Dougherty-Page, J., Burtt, A., 2006. The timing and duration ofthe Delamerian Orogeny: correlation with the Ross Orogen and implications forGondwana assembly. Journal of Geology 114, 189–210.

Foster, D.A., Gray, D.R., 2000. Evolution and structure of the Lachlan Fold Belt (Orogen)of eastern Australia. Annual Review of Earth and Planetary Sciences 28, 47–80.

Foster, D.A., Gray, D.R., Spaggiari, C., Kamenov, G., Bierlein, F.P., 2009. Palaeozoic Lachlanorogen, Australia; accretion and construction of continental crust in a marginal oceansetting: isotopic evidence from Cambrian metavolcanic rocks. Geological Society,London, Special Publications 318, 329–349.

Gaina, C., Müller, D., Royer, J.Y., Stock, J., Hardebeck, J., Symonds, P., 1998. The tectonichistory of the Tasman Sea: a puzzle with 13 pieces. Journal of Geophysical Research103, 12,413–12,433.

Gibson, G.M., Morse, M.P., Ireland, T.R., Nayak, G.K., 2011. Arc–continent collision andorogenesis in western Tasmanides: insights from reactivated basement structuresand formation of an ocean–continent transform boundary off western Tasmania.Gondwana Research 19, 608–627.

Glen, R.A., 2005. The Tasmanides of eastern Australia. In: Vaughan, A.P.M., Leat, P.T.,Pankhurst, R.J. (Eds.), Terrane Processes at the Margins of Gondwana. GeologicalSociety, London, pp. 23–96.

Glen, R.A., Scheibner, E., Vandenberg, A.H.M., 1992. Paleozoic intraplate escape tecton-ics in Gondwanaland and major strike-slip duplication in the Lachlan Orogen ofsouth-eastern Australia. Geology 20, 795–798.

Glen, R.A., Poudjom-Djomani, Y., Korsch, R.J., Costello, R.D., Disk, S., 2007. Thomson–Lachlan Seismic Project — Results and Implications. In: Lewis, P.C. (Ed.), AustralianInstitute of Geoscientists, Bulletin, volume 46, pp. 73–78.

Glen, R.A., Percival, I.G., Quinn, C.D., 2009. Ordovician continental margin terranesin the Lachlan Orogen, Australia: implications for tectonics in an accretionaryorogen along the east Gondwana margin. Tectonics 28. http://dx.doi.org/10.1029/2009TC002446.

76 N. Rawlinson et al. / Lithos 189 (2014) 65–76

Glen, R.A., Quinn, C.D., Cooke, D.R., 2012. The Macquarie Arc, Lachlan Orogen, NewSouth Wales; its evolution, tectonic setting and mineral deposits. Episodes 35,177–186.

Graeber, F.M., Houseman, G.A., Greenhalgh, S.A., 2002. Regional teleseismic tomogra-phy of the western Lachlan Orogen and the Newer Volcanic Province, southeastAustralia. Geophysical Journal International 149, 249–266.

Griffin, W.L., O'Reilly, S.Y., Ryan, C.G., Gaul, O., Ionov, D.A., 1998. Secular variation in thecomposition of subcontinental lithospheric mantle: geophysical and geodynamicimplications. In: Braun, J., Dooley, J., Goleby, B., van der Hilst, R., Kllotwijk, C.(Eds.), Structure and Evolution of the Australian Continent. American GeophysicalUnion Geodynamic Series. , volume 26, pp. 1–26.

Gunn, P.J., Mackey, T.E., Yeates, A.N., Richardson, R.G., Seymour, D.B., McClenaghan,M.P., Calver, C.R., Roach, M.J., 1997a. The basement elements of Tasmania. Explora-tion Geophysics 28, 225–231.

Gunn, P.J., Mitchell, J., Meixner, A., 1997b. The structure and evolution of the bass basinas delineated by aeromagnetic data. Exploration Geophysics 28, 214–219.

Hand, M., Reid, A., Szpunar, M., Direen, N., Wade, B., Payne, j, Barovich, K., 2008. Crustalarchitecture during the early Mesoproterozoic Hiltaba-related mineralisation event:are the Gawler Range Volcanics a foreland basin fill? MESA Journal 51, 19–24.

Hill, P.J., Meixner, A.J., Moore, A.G., Exon, N.F., 1997. Structure and development of thewest Tasmanian offshore sedimentary basins: results of recent marine and aero-magnetic surveys. Australian Journal of Earth Sciences 44, 579–596.

Kemp, A.I.S., Hawkesworth, C.J., Collins, W.J., Gray, C.M., Blevin, P.L., EIMF, 2009. Isotopicevidence for rapid continental growth in an extensional accretionary orogen: theTasmanides, eastern Australia. Earth and Planetary Science Letters 284, 455–466.

Kennett, B.L.N., 2003. Seismic structure in the mantle beneath Australia. In: Hillis, R.R.,Müller, R.D. (Eds.), The Evolution and Dynamics of the Australian Plate, pp. 7–23.

Kennett, B.L.N., Abdullah, A., 2011. Seismic wave attenuation beneath the Australasianregion. Australian Journal of Earth Sciences 58, 285–295.

Kennett, B.L.N., Salmon, M., 2012. AuSREM: Australian seismological reference model.Australian Journal of Earth Sciences 59, 1091–1103.

Kennett, B.L.N., Sambridge, M.S., Williamson, P.R., 1988. Subspace methods for largescale inverse problems involving multiple parameter classes. Geophysical Journal94, 237–247.

Kennett, B.L.N., Engdahl, E.R., Buland, R., 1995. Constraints on seismic velocities in theearth from travel times. Geophysical Journal International 122, 108–124.

Kennett, B.L.N., Fishwick, S., Reading, A.M., Rawlinson, N., 2004. Contrasts in mantlestructure beneath Australia: relation to Tasman Lines? Australian Journal of EarthSciences 51, 563–569.

Kennett, B.L.N., Fichtner, A., Fishwick, S., Yoshizawa, K., 2013. Australian SeismologicalReference Model (AuSREM): mantle component. Geophysical Journal International192, 871–887.

Korsch, R.J., Barton, T.J., Gray, D.R., Owen, A.J., Foster, D.A., 2002. Geological interpretationof a deep seismic-reflection transect across the boundary between the Delamerianand Lachlan Orogens, in the vicinity of the Grampians, western Victoria. AustralianJournal of Earth Sciences 49, 1057–1075.

Leaman, D., Richardson, R.G., 1989. Production of a residual gravity fieldmap for Tasmaniaand some implications. Exploration Geophysics 20, 181–184.

Lévêque, J.J., Rivera, L., Wittlinger, G., 1993. On the use of the checker-board test to as-sess the resolution of tomographic inversions. Geophysical Journal International115, 313–318.

Lister, G.S., Ethridge, M.A., Symonds, P.A., 1991. Detachment models for the formationof passive continental margins. Tectonics 10, 1038–1064.

Mackey, T.E.,Milligan, P.R., Richardson, R.G., 1995. Totalmagnetic intensitymapof Tasmania,1:500 000. Technical Report. Australian Geological Survey Organisation.

Miller, J.M., Phillips, D., Wilson, C.J.L., Dugdale, L.J., 2005. Evolution of a reworked orogeniczone: the boundary between the Delamerian and Lachlan fold belts, southeasternAustralia. Australian Journal of Earth Sciences 52, 921–940.

Murray, C.G., Scheibner, E., Walker, R.N., 1989. Regional geological interpretation of adigital coloured residual Bouguer gravity image of eastern Australia with a wave-length cutoff of 250 km. Australian Journal of Earth Sciences 36, 423–449.

Musgrave, R., Rawlinson, N., 2010. Linking the upper crust to the uppermantle: comparisonof teleseismic tomography with long-wavelength features of the gravity and magneticfields of southeastern Australia. Exploration Geophysics 41, 155–162.

Nolet, G., 2008. A Breviary of Seismic Tomography: Imaging the Interior of the Earthand Sun. Cambridge University Press, Cambridge.

Obrebski, M., Allen, R.M., Pollitz, F., Hung, S.H., 2011. Lithosphere–asthenosphere inter-action beneath the western United States from the joint inversion of body-wavetraveltimes and surface-wave phase velocities. Geophysical Journal International185, 1003–1021.

Price, R.C., Gray, C.M., Frey, F.A., 1997. Strontium isotopic and trace element heteroge-neity in the plains basalts of the Newer Volcanic Province, Victoria, Australia.Geochimica et Cosmochimica Acta 61, 171–192.

Quigley, M., Sandiford, M., Fifield, K., Alimanovic, A., 2007. Bedrock erosion and reliefproduction in the northern Flinders Ranges, Australia. Earth Surface Processesand Landforms 32. http://dx.doi.org/10.1002/esp.1459.

Rawlinson, N., Fishwick, S., 2012. Seismic structure of the southeast Australian litho-sphere from surface and body wave tomography. Tectonophysics 572, 111–122.

Rawlinson, N., Kennett, B.L.N., 2004. Rapid estimation of relative and absolute delaytimes across a network by adaptive stacking. Geophysical Journal International157, 332–340.

Rawlinson, N., Kennett, B.L.N., 2008. Teleseismic tomography of the upper mantlebeneath the southern Lachan Orogen, Australia. Physics of the Earth and PlanetaryInteriors 167, 84–97.

Rawlinson, N., Sambridge, M., 2003. Irregular interface parameterization in 3-D wide-angle seismic traveltime tomography. Geophysical Journal International 155, 79–92.

Rawlinson, N., Sambridge, M., 2004. Wavefront evolution in strongly heterogeneouslayered media using the fast marching method. Geophysical Journal International156, 631–647.

Rawlinson, N., Urvoy, M., 2006. Simultaneous inversion of active and passive sourcedatasets for 3-D seismic structure with application to Tasmania. GeophysicalResearch Letters 33. http://dx.doi.org/10.1029/2006GL028105.

Rawlinson, N., Houseman, G.A., Collins, C.D.N., Drummond, B.J., 2001. New evidence ofTasmania's tectonic history from a novel seismic experiment. Geophysical ResearchLetters 28, 3337–3340.

Rawlinson, N., Kennett, B.L.N., Heintz, M., 2006a. Insights into the structure of theupper mantle beneath the Murray Basin from 3D teleseismic tomography. AustralianJournal of Earth Sciences 53, 595–604.

Rawlinson, N., Reading, A.M., Kennett, B.L.N., 2006b. Lithospheric structure of Tasmaniafrom a novel form of teleseismic tomography. Journal of Geophysical Research 111.http://dx.doi.org/10.1029/2005JB003803.

Rawlinson, N., Pozgay, S., Fishwick, S., 2010a. Seismic tomography: a window into deepEarth. Physics of the Earth and Planetary Interiors 178, 101–135.

Rawlinson, N., Tkalčić, H., Reading, A.M., 2010b. Structure of the Tasmanian lithospherefrom 3-D seismic tomography. Australian Journal of Earth Sciences 57, 381–394.

Rawlinson, N., Kennett, B., Vanacore, E., Glen, R., Fishwick, S., 2011. The structure of theupper mantle beneath the Delamerian and Lachlan orogens from simultaneousinversion of multiple teleseismic datasets. Gondwana Research 19, 788–799.

Reed, A.R., Calver, C., Bottrill, R.S., 2002. Palaeozoic suturing of eastern and westernTasmania in the west Tamar region: implications for the tectonic evolution ofsoutheast Australia. Australian Journal of Earth Sciences 49, 809–830.

Roach, M.J., Leaman, D.E., Richardson, R.G., 1993. A comparison of regional-residualseparation techniques for gravity surveys. Exploration Geophysics 24, 779–784.

Rosenbaum, G., 2012. Oroclines of the southern New England Orogen, eastern Australia.Episodes 35, 187–194.

Rosenbaum, G., Pengfei, L., Rubatto, D., 2012. The contorted New England Orogen(eastern Australia): new evidence from U–Pb geochronology of early Permiangranitoids. Tectonics 31. http://dx.doi.org/10.1029/2011TC002960.

Salmon, K., Kennett, B.L.N., Saygin, E., 2012. Australian Seismological Reference Model(AuSREM): crustal component. Geophysical Journal International 192, 190–206.

Sandiford, M., Wallace, M., Coblentz, D., 2004. Origin of the in situ stress field in south-eastern Australia. Basin Research 16, 325–338.

Schuchert, C., 1916. The problem of continental fracturing and diastrophism inoceanica. Proceedings of the National Academy of Science 2, 407–413.

Shaw, R.D.,Wellman, P., Gunn, P., Whitaker, A.J., Tarlowski, C., Morse, M., 1996. AustralianCrustal Elements based on the distribution of geophysical domains (1:5 000 000 scalemap; version 2.4, ArcGIS dataset). Geoscience Australia, Canberra.

Simons, F., Zielhuis, A., van der Hilst, R.D., 1999. The deep structure of the Australiancontinent from surface wave tomography. Lithos 48, 17–43.

Simons, F.J., van der Hilst, R.D., Montagner, J.P., Zielhuis, A., 2002. Multimode Rayleighwave inversion for heterogeneity and azimuthal anisotropy of the Australian uppermantle. Geophysical Journal International 151, 738–754.

Sobolev, S.V., Zeyen, H., Stoll, G., Werling, F., Altherr, R., Fuchs, K., 1996. Upper mantletemperatures from teleseismic tomography of French Massif Central includingeffects of composition, mineral reactions, anharmonicity, anelasticity and partialmelt. Earth and Planetary Science Letters 139, 147–163.

Spaggiari, C.V., Gray, D.R., Foster, D.A., McKnight, S., 2003. Evolution of the boundarybetween thewestern and central Lachlan Orogen: implications for Tasmanide tectonics.Australian Journal of Earth Sciences 50, 725–749.

Spaggiari, C.V., Gray, D.R., Foster, D.A., 2004. Lachlan Orogen subduction–accretion sys-tematics revisited. Australian Journal of Earth Sciences 51, 549–553.

Spencer, R., 2004. Gravitymap of New SouthWales. Scale 1:2500000. Geological Survey ofNew South Wales, Sydney.

Taylor, D.H., Cayley, R.A., 2000. Character and kinematics of faults within the turbidite-dominated Lachlan Orogen: implications for tectonic evolution of eastern Australia:discussion. Journal of Structural Geology 22, 523–528.

Thomson, B.P., 1970. A review of the Precambrian and lower Palaeozoic tectonics ofSouth Australia. Transactions of the Royal Society of South Australia 94, 193–221.

van der Beek, P.A., Braun, J., Lambeck, K., 1999. Post-Palaeozoic uplift history of south-eastern Australia revisited: results from a process-based model of landscape evolu-tion. Australian Journal of Earth Sciences 46, 157–172.

VandenBerg, A.H.M., 1999. Timing of orogenic events in the Lachlan Orogen. AustralianJournal of Earth Sciences 46, 691–701.

Wade, C.E., Reid, A.J., Wingate, M.T.D, Jagodzinski, E.A, Barovich, K., 2012. Geochemistryand geochronology of the c. 1585 ma Benagerie Volcanic Suite, southern Australia:relationship to the Gawler Range Volcanics and implications for the petrogenesisof a Mesoproterozoic silicic large igneous province. Precambrian Research 206,17–35.

Webster, A.E., 1996. Delamerian refolding of the Palaeoproterozoic Broken Hill Block.Australian Journal of Earth Sciences 43, 85–89.

West,M., Gao,W., Grand, S., 2004. A simple approach to the joint inversion of seismic bodyand surface waves applied to the southwest U.S. Geophysical Research Letters 31.

Willman, C.E., VandenBerg, A.H.M., Morand, V.J., 2002. Evolution of the southeasternLachlan Fold Belt in Victoria. Australian Journal of Earth Sciences 49, 271–289.

Zielhuis, A., van der Hilst, R.D., 1996. Upper-mantle shear velocity beneath eastern Australiafrom inversion of waveforms from SKIPPY portable arrays. Geophysical Journal Interna-tional 127, 1–16.