TAJE_A_837098.3d (TAJE) 09-10-2013 17:17

Structure of the Mt Isa region from seismic ambient noisetomography

E. SAYGIN1*, H. McQUEEN1, L. J. HUTTON2, B. L. N. KENNETT1 AND G. LISTER1

1Research School of Earth Sciences, The Australian National University, Canberra ACT 0200, Australia.2Geological Survey of Queensland, Department of Natural Resources and Mines, Level 10, 119 Charlotte Street,

Brisbane, QLD 4002, 4000 Australia.

We use seismic tomography, exploiting group velocities derived from ambient noise, to delineate thecrustal structure beneath Mt Isa and the surrounding blocks and basins. The depth extent of the blockscan be traced into the mid-crust and the spatial extent of the associated velocity anomalies mappedover an area of approximately 500 km by 500 km. The Proterozoic Mt Isa block is imaged as a region ofelevated seismic velocities comparable to the Yilgarn craton in Western Australia, while the surroundingbasins have relatively low velocities. Seismic velocity anomalies display correlations with the regionalBouguer gravity data and with high crustal temperatures in the region. There are a number of isolatedlow-velocity anomalies under the Millungera basin that suggest either previously unknown thermalanomalies or zones with high permeability, which can also produce lowered velocities.

KEYWORDS: Mt Isa, Australia, lithosphere, Proterozoic, seismology, ambient seismic noise tomography.

INTRODUCTION

Seismic tomography exploiting ambient noise is a rela-tively new technique, which uses continuously recordedseismic noise propagating through the Earth to map var-iations in crustal seismic velocity associated with struc-tural and thermal contrasts. The main component of theambient noise field of the Earth is generated from ocean-ground coupling (Longuet-Higgins 1950), which spreadsout from ocean storm centres and is continuouslyrecorded at seismic stations; excitation of noise can alsooccur at the edge of the continental shelf. By cross-corre-lating the seismic noise simultaneously recorded at twodifferent stations, it is in general possible to retrieve adeterministic wavelet (e.g. Shapiro & Campillo 2004).This wavelet, or Green’s function, represents the seismicresponse as if one of the stations is acting as a sourceand the other as a receiver. Green’s function betweensource and receiver carries the signature of the seismicwavespeed structure between the stations. With a densenetwork it is possible to extract many Green’s functionsby cross-correlations between all pairs of stations. Theinterstation travel times for waves on the multiple pathscan then be employed in a tomographic inversion to mapthe seismic wavespeed perturbations across thenetwork.

Ambient seismic techniques have the merit that theyallow effective imaging of regional geological structuresusing just passive seismic recordings. Following the pio-neering work of Shapiro & Campillo (2004) and Shapiroet al. (2005) in California, many other studies have been

conducted to image the Earth at various scales by usingseismic ambient noise correlation. The technique hasbeen applied to areas in the USA (Sabra et al. 2005; Liang& Langston 2008; Yang et al. 2011), New Zealand (Linet al. 2007; Behr et al. 2010), Australia (Saygin & Kennett2010, 2012; Arroucau et al. 2010; Young et al. 2011) andChina (Yang et al. 2010; Zhou et al. 2012).

Over the last decade, the seismic structure of the Aus-tralian lithosphere has been investigated with increas-ing resolution owing to the expanding instrumentationcoverage of the continent. Tomographic studies havebeen conducted by using seismic surface waves fromearthquakes (Zielhuis & van der Hilst 1996; Simons et al.1999; Debayle & Kennett 2000a, b; Yoshizawa & Kennett2004; Fishwick et al. 2005, 2008), seismic attenuation(Kennett & Abdullah 2011), seismic anisotropy (Heintz &Kennett 2005, 2006) and seismic body wave tomography(Rawlinson et al. 2006; Rawlinson & Fishwick 2011) havebeen carried out using seismic body waves. Teleseismicstudies using sources at the surrounding plate bound-aries provide good coverage for lithospheric tomography.However, the relatively low seismicity of the Australiancontinent limits the extent of local tomographic studiesusing body waves. For Australia, seismic ambient noisetomography fills an important gap revealing shallowerstructure than conventional methods both at the conti-nental scale (Saygin & Kennett 2010, 2012) and morelocally, as in the work of Young et al. (2011) for Tasmania.The design of a network of portable stations providesdirect control on the distribution of the set of propaga-tion paths between the stations. Along each path, surface

*Corresponding author: erdinc.saygin@anu.edu.au� 2013 Geological Society of AustraliaThe work of Laurie Hutton was conducted within his duties as an employee of the Geological Survey of Queensland and is thereforecopyright � 2013 Crown in the Queensland Government

Australian Journal of Earth Sciences (2013)

60, 707–717, http://dx.doi.org/10.1080/08120099.2013.837098

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:17

wave information from Green’s function gives penetra-tion to a range of depths. Hence, the shallow crust can beimaged without the need for earthquakes as a probingsource.

In this study, we exploit the results from portableseismic instrumentation deployed in the Mt Isa region

between 2009 and 2011, together with permanent seismicstations and other portable stations (Figure 1). Surfacewave dispersion information for 800 paths crossing theMt Isa region was extracted from the cross-correlo-grams of station pairs, and used to image the variationsin seismic wavespeeds. The seismic results identify

Figure 1 Surface geology of the region (Raymond et al. 2010). Seismic networks used in this study are marked with triangles.

708 E. Saygin et al.

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:17

significant geological boundaries that can be followed todepth. Patches of relatively low seismic wavespeed sug-gest rocks that have relatively low density, because oftheir composition, or because they are porous and/orpermeable sediments. Volumes of fractured rock areunlikely to be resolved at the scale of this experiment.Note that the movement of fluids means that such per-meable zones should have localised high geothermalgradients.

GEOLOGY

The Mt Isa region contains several geological terranesthat were formed at the edge of the Northern AustralianCraton as the result of tectonic activity during early tomid-Proterzoic orogeny (Betts & Giles 2006; Betts et al.2006).

The Mt Isa region is one of the best-exposed Paleo-proterozoic terranes on Earth. The so-called metal orog-eny began with the Barramundi event (1.86 Ga)correlated globally (Hou et al. 2008) with the Svecofen-nian of Europe or the Hudsonian of Northern America(Etheridge et al. 1987). Large-scale crustal extension spo-radically afflicted this terrane, up until ca 1.6 Ga, pro-ducing basins in which much of Australia’s base metalmineral wealth accumulated. These basins wereinverted by approximately east–west compression thatcommenced at ca 1.58 Ga, and produced regionallydeveloped fold and thrust belts as part of the Isan Orog-eny (Betts et al. 2006). Regional low-pressure high-tem-perature metamorphism produced large extents ofmedium- to high-grade crystalline rocks. Later wrenchfaulting dissected these terranes producing the mappatterns we observe today, with three major north–south oriented fold belts.

The Proterozoic Mt Isa inlier is divided into threemajor north–south oriented fold belts (Blake 1987; O’Deaet al. 1997), with the Kalkadoon-Leichardt Belt in the cen-tre, and the Western and Eastern fold belts on eitherside. The Mt Isa inlier is surrounded by two major basinstructures: the Georgina Basin and the Eromanga Basin.The Georgina Basin was formed by disruption of theCentralian Superbasin during the Petermann RangesOrogeny with central uplift, and associated thrusts at600–540 Ma, followed by mid-Carboniferous tectonism(320 Ma) (Walter et al. 1995).

The Eromanga Basin along with the Cooper Basin,host the largest onshore oil and gas fields in Queensland(Reynolds et al. 2006). The Mt Isa domain itself hosts anumber of large mineral deposits including sulfide Pb–Zn–Ag, Cu, Broken Hill-type Ag–Pb–Zn and iron-oxide–Cu–Au type mineralisation (Betts et al. 2006; Blenkinsopet al. 2008), and is one of the world’s leading producers ofZn, Pb, Cu and Ag (Hutton et al. 2012).

DATA

Between 2009 and 2011, the MINQ seismic array wasoperated in the Mt Isa region, spanning from the Geor-gina to Eromanga basins, by the Research School ofEarth Sciences of ANU as an AuScope project, with atotal of 49 stations (see Figure 1). A set of 35 seismic

recorders with Lennartz LE-3Dlite short-period 3 compo-nent sensors were used, and progressively redeployedeastward to increase the spatial coverage. Each stationwas run continuously with a sampling rate of 25 Hz orhigher. Also, nearby stations with short-period andbroadband sensors from the permanent IRIS and Geosci-ence Australia networks and the BILBY temporarydeployment from ANU were used to increase the spatialcoverage of the study.

Before cross-correlating recorded ambient seismicnoise, instrument corrections were applied to everyrecord to equalise the sensor responses. Where available,tabulated instrumental responses were used, but other-wise the nominal response for each sensor type wasemployed. Daily volumes of uncompressed datawere cre-ated for each component recorded at a seismic station.The cross-correlations were carried out in the frequencydomain to reduce the computational cost (see AppendixA). For the cross-correlations between a pair of stations,a 4 h window was used with 1 h overlap. The calculatedcorrelograms for each of the available 4 h window werestacked to create a stable estimate of Green’s functionfor this station pair. This procedure was repeated for allof the other pairs of stations. Cross-correlation of theseismic noise recorded at these various stations pro-duced over 800 individual high quality Green’s func-tions. These Green’s functions were then processed toextract group velocity information for each path as afunction of frequency. The path information was thenused to produce wavespeed variation maps as a functionof frequency using tomographic imaging.

METHOD

The method of cross-correlation of seismic ambientnoise has recently emerged in seismology as an alterna-tive technique to imaging carried out with traditionalsources, e.g. earthquakes, explosions, etc. The control onthe experiment configuration in this method gives free-dom comparable to active source seismic surveys for afraction of the cost (e.g. Young et al. 2011).

Although, in principle, Green’s function wouldinclude all aspects of the seismic wave propagationbetween the pair of stations, the stable part is the contri-bution from seismic surface waves (e.g. Shapiro et al.2005). Rayleigh waves can be extracted from stackedcross-correlations of vertical and radial componentpairs, and often Love waves can also be extracted fromthe cross-correlations of the transverse component ofthe noise field. In this study, we will only use the Ray-leigh wave portion of Green’s function.

In Figure 2, we show the estimates of the interstationGreen’s functions extracted from the ambient noise forpropagation paths in two different tectonic domains. Inthe cratonic Mt Isa area, surface waves propagate muchfaster than in the Carpentaria sedimentary basin lead-ing to systematically earlier arrivals (see Figure 2a, b).We exploit the differences in the seismic wave character-istics for different paths, by measuring the group veloci-ties for each path at a number of periods. The groupvelocity of a surface wave is the propagation speed of theenvelope of the wave packet.

Seismic Ambient Noise Tomography of Mt Isa region 709

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:17

AMBIENT SEISMIC NOISE TOMOGRAPHY

After measuring the group velocities at different periodsfor the Rayleigh wave part of Green’s functions betweenavailable stations, a tomographic problem is constructedto map the spatial variations in velocity for everyselected period. We use a similar approach to thatemployed by Saygin & Kennett (2010). We start with theFast Marching Method (FMM; Rawlinson & Sambridge2004) to trace seismic wavefronts propagating in the het-erogeneous media, and calculate the travel times for thewavefronts for a given input model. The second stage ofthe tomographic inversion is to minimise the differencebetween the modelled traveltimes and the observed ones,for which a subspace technique is employed (Kennettet al. 1988).

Seismic tomographic inversion is applied to grouptransit times for the suite of paths at a selection of peri-ods between 1 and 10 s. Surface waves propagating withdifferent periods sample different parts of Earth, and ingeneral, surface waves at longer periods sample deeperparts of Earth compared with shorter periods. We dis-play images of the group wavespeeds as a function ofperiod in Figure 3. Where variations in the phase andgroup wavespeeds are weak, the local dispersion proper-ties as a function of period can be interpreted in terms of1-D structure beneath the point (Woodhouse 1974). It ispossible to perform a similar analysis using phase

velocities of the waves. This would produce a comple-mentary set of images sampling the region in a slightlydifferent ways. However, the approximations used in theextraction of the phase velocity of surface waves breakdown in regions of complex structure such as this, andwe have therefore used only the more robust groupvelocity analysis.

In the Mt Isa region, we have very strong variationsin properties, and unfortunately, the local dispersionrelation cannot be exploited to extract shear wavespeedstructure. However, we can use the approximate corre-spondence of period and depth in kilometres to obtainan indication of the depth extent of features. In areaswith sedimentary cover, the sensitivity is pulled uptowards the surface (see Figure A2), and so the imagesin Figure 3 cannot be interpreted directly as a depthslice.

Because of their finite frequency, the surface wavessample a zone around the nominal propagation path. Aconvenient estimate of the extent of this sampling is pro-vided by the influence zone of Yoshizawa & Kennett(2002), which corresponds to 1/6 of the first Fresnel zone.Figure A3 presents spatial resolution tests for the Mt Isaregion for periods of 3 s and 8 s to indicate the reliabilityof the images. In the central portion with a high densityof stations, structures are well resolved and there is agood recovery of the amplitude of the imposed anoma-lies. Because of the distribution of the available seismic

Figure 2 Extracted Green’s functions from two different geological regions: (a) Mt Isa, (b) Carpentaria Basin. Dashed linesshow the expected arrival time for a reference velocity of 3.2 km/s. Earlier arrivals with respect to these lines indicate fasterseismic velocities. Waveforms are filtered between 0.1 and 0.3 Hz and normalised to unity. Notice the systematic difference inarrival times of Green’s functions in the two different domains.

710 E. Saygin et al.

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

stations, the path density and spatial coverage decreasedrastically towards the edges where only isolated pathsare available. This uneven distribution of available prop-agation paths causes smearing and artefacts in themapped seismic velocity perturbation such as in Fig-ure 3 (south of 21�) for periods from 3 to 8 s. These spuri-ous structures are not included in the interpretation.

RESULTS AND DISCUSSION

On a broad scale, the results from the seismic tomographyshow the expected structure, with a high velocity bodyassociated with the Mt Isa block sandwiched between thelower velocity Georgina, Eromanga and Carpenteria sedi-mentary basins. Within these broad structures, a range ofsmaller-scale features is evident. While increasing periodgenerally corresponds to increasing depth, the relation-ship is not simple. It depends on the local vertical velocitystructure and is complicated by irregular sampling of lat-eral variations. As a result, the conversion is neither sim-ple nor necessarily unique, so the tomographic sections

are presented for surface wave periods ranging from 2 to10 s, which progressively sample deeper into the upper10 km of the crust. The sensitivity kernels contained inthe Figure A2 give an indication of the depth range sam-pled by each period.

Mt Isa block

Our results show a variety of features that are visible inthe group velocity maps. In Figure 4, images at 2, 5 and10 s are compared with the generalised tectonic map ofthe region and a Bouguer gravity anomaly map (Bacchinet al. 2008). In the shallowest image at 2 s, tomographicinversion accurately delineates the borders of the West-ern Fold Belt (WFB), Kalkadoon-Leichhardt Belt (KLB),and Eastern Fold Belt (EFB). In the middle zone, theKLB is marked by elevated group and phase velocities,with a clear north–south trend. In general the WFB haslower velocities compared with the other two belts of theMt Isa domain. Two high-velocity regions in the EFBshown in the 2, 5, and 10 s group velocity images corre-spond to granites in the northeast and southwest. The

Figure 3 Group velocity tomography images for wave periods of 2, 3, 4, 5, 8 and 10 s. Longer periods sense deeper structures.The reference group velocity (U) for every panel is 2.95 km/s.

Seismic Ambient Noise Tomography of Mt Isa region 711

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

Figure 4 Comparison of group veloc-ity images with tectonic features (a);Bouguer gravity (Bacchin et al. 2008)(b); and gridded temperature distri-bution at 5 km depth: OZTEMP(Gerner & Holgate 2010) (c). Circlesdenote the actual location of temper-ature measurements. Low seismicvelocity regions that match gravitylows are marked with numerals.Dashed lines show the inferredboundaries of blocks and basin struc-tures at depth, and dashed ellipseshighlight known granite bodies.

712 E. Saygin et al.

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

isolated low-velocity patches at 2 s period marked withDand at 5 s period marked with E also correlate with fea-tures in the gravity image indicating low-densityregions. In general the extent and direction of the anom-aly associated with the KLB appears consistentlythrough increasing period up to 10 s; the surface wavesat this longer period sense deeper material from upperto midcrust and the influence of the surface material isreduced. The images suggest a slight westerly dip for theKLB in depth.

To the north of the Mt Isa domain (north of 19�S),there are no features evident because of lack of path cov-erage traversing this region. Similarly, to the south(21.5�S and below), the resolution decreases drasticallyand smearing effects with elongated features begin toappear, for example in the 3 s images between latitudes142�E and 139�E.

Sedimentary basins

The Georgina Basin to the west of the Mt Isa block ischaracterised by low group velocities with perturba-tions close to –0.8 km/s from the mean in the shallowimages from 2 to 5 s. The southerly extent of the Geor-gina Basin is not captured completely owing to thereduced path density but the northern end of the Geor-gina Basin shows up strongly over a range of depths.The extent of the low-velocity anomalies is in goodagreement with the Bouguer Gravity map featuresmarked with A and B in Figure 4. In Figure 3, the signa-ture of Georgina Basin completely disappears for peri-ods longer than 8 s.

The area west of Mt Isa represents a stratigraphicthickness much greater than other parts of the Geor-gina Basin. The AP Morstone 1 Well, north of theCamooweal to Burketown Road, penetrated 335 m ofmiddle Cambrian limestone (Georgina Basin) overlying427 m of sandstone, siltstone and claystone (ofunknown correlation and age) (Withnall & Hutton2013). The age or affinity of the sandstone/claystonesequence is uncertain but may be equivalent to theSouth Nicholson Group, which outcrops to the north(Withnall & Hutton 2013). If these rocks were equiva-lent to the South Nicholson Group, then they wouldclearly represent a significantly thicker sequence thanoccurs elsewhere. Alternatively the sedimentary rocksmay be Neoproterozoic, and also form an anomalouslythick sequence. The South Nicholson Group is up to3.5 km thick (Sweet pers. comm. 2013), but depth tomagnetic basement modelling in the area west of MtIsa (Geological Survey of Queensland 2011) suggeststhat up to 6 km of sedimentary rock occurs in thisstructure. These depths are consistent with the low-velocity anomaly, visible to about 8 s (features markedwith A and B in Figure 4–2 s). The significance of thisinterpretation is that the area represents an anoma-lously thick sedimentary rock accumulation, which isnot present in other areas under the Georgina Basin.This thickness of sedimentary rocks was deposited in ageodynamic setting that does not occur elsewherebeneath the Georgina Basin.

The Carpentaria Basin is an offshore basin startingin the Gulf of Carpentaria and extending into north

Queensland, with a thickness over 1.5 km (Korsch et al.2012). In the group velocity images (2–4 s) for the shallowcrust, the boundary between the Carpentaria and Ero-manga basins can be inferred from a relatively straightfeature (see dashed red line in 2 s group velocity imagein Figure 4). In between these, localised low-velocity bod-ies appear. For example, to the south of 20�S, the signa-ture of one such localised body marked with F is visibleboth in the Bouguer Gravity anomaly and in all of thegroup velocity images.

MILLUNGERA BASIN

Following deep seismic reflection sounding in northQueensland conducted by Geoscience Australia in 2007,the presence of a new sedimentary basin underneathCarpentaria and Georgina Basin was revealed (Korschet al. 2011, 2012) (see Figure 4a). Subsequent magneto-telluric surveys confirmed the existence of a signifi-cant conductivity contrasts associated with thepresence of the basin. The age of the Millungera Basinis uncertain. Preliminary Rb–Sr dating of illite fromdrill core indicates that the sediments were affected byintermediate (250–300�C) temperature tectonic andhydrothermal events about 1100 Ma (Uysal pers. comm.2013). This suggests that the Millungera Basin is Meso-proterozoic and may correlate with the South Nichol-son Basin (Jell 2013). It has been conjectured that thisrecently discovered basin has the potential to act ablanket above inferred highly radiogenic granites pro-ducing zones of anomalously high subsurface tempera-ture. The outline of the Millungera Basin used inFigure 4a is deduced from seismic reflection imagingand magnetotelluric surveys conducted in the region.Our seismic tomography images using ambient seismicnoise do not show a single consistent velocity anomalyin the vicinity of the Millungera Basin. Instead, a frag-mented structure exists with a number of high andlow-velocity bodies.

Seismic velocities are influenced by temperature aswell as the mineral composition. Seismic ambient noiseimaging across Australia by Saygin & Kennett (2010,2012) revealed the influence of thick blanketing sedi-ments and high heat generating regions. However, inseismic velocity tomography it is not in general possibleto distinguish thermal and compositional effects con-tributing to velocity anomalies. In the seismic tomogra-phy images shown in Figure 3, the main patches of low-velocity anomalies can be interpreted as a result of thelow-velocity thick sediment combination of the Carpen-taria and Millungera basins in north and the Eromangaand Millungera basins in the south. The isolated low-velocity patches between 140.5 to 142�E and 18.5 to 20�Sare examples of this pattern.

HIGH HEAT GENERATING ZONES

Figure 4c shows a compilation of temperature estimatesat 5 km from the OZTEMP model of Gerner & Holgate(2010) derived from all available borehole temperaturemeasurements across the Australian continent. Some ofthe noticeable low-velocity anomalies located east of141� in the group velocity images are in good agreement

Seismic Ambient Noise Tomography of Mt Isa region 713

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

with the features of the OZTEMP model. However, thesampling in the OZTEMP dataset is not regular and didnot extend across all the anomalies visible in tomo-graphic images. Additional heat flow measurementshave been undertaken from drill holes GSQ Julia Creek1 (Faulkner et al. 2012) (feature H in Figure 4b) andGSQ Dobbyn 2 (Fitzell et al. 2012) (feature C in Fig-ure 4b). Heat flows measured at 113.0 � 2.9 mW/m2

(GSQ Julia Creek 1) and 107.5 � 1.7 mW/m2 (GSQ Dob-byn 2) lie well above the regional average of 65–90 mW/m2. These high heat flows are believed to result fromhigh heat-producing granites underlying the Millun-gera Basin and Carpentaria Basin sedimentary rocksand correspond to low seismic velocity regions, whichare interpreted as being caused by higher heat produc-tion than surrounding rocks. Some of the unmarkedanomalies visible in Figure 4 such as 141�E, 18.5–19�S,may well be associated with previously unrecognisedhigh heat-generating regions.

CONCLUSIONS

The deployment of a set of portable seismic instru-ments in the Mt Isa region has allowed high-resolutionseismic tomography by exploiting passive recording ofambient noise. Cross-correlations of records from pairsof stations allow the extraction of the surface wave por-tion of the seismic response of the structure betweenthe stations, from which group velocities for the pathcan be found as a function of period (frequency). Thefull suite of path observations is employed in tomo-graphic imaging to derive wavespeed maps as a func-tion of period.

The complex geology of the region is reflected in thetomographic images, with distinctive geological featuressuch as the Kalkadoon-Leichhardt, Western and EasternFold Belts readily recognised and tracked into depththrough the proxy of increasing period. However, therecently discovered Millungera basin is not distin-guished as a single feature in the images, perhaps as aresult of multiple succession effects in the formation ofthe overlying basins.

Seismic wavespeeds can be affected by both composi-tional and structural affects, but in addition high tem-peratures can produce significant decreases inwavespeed. Isolated, but distinct, anomalies with low-ered group velocity suggest the presence of local highheat production beneath the Millungera basin. However,there is also the possibility that these features mightarise from zones with high permeability, which also canproduce lowered velocities.

ACKNOWLEDGEMENTS

We would like to thank Dr Michelle Salmon (RSES-ANU)and technical staff at RSES for conducting fieldwork.Generic Mapping Tools (Wessel & Smith 1998) is used forcreating the figures (except Figure 2). The data from per-manent stations were accessed through the IRIS DataManagement Center. E. Saygin was supported by Austra-lian Research Council through ARC Discovery Grants.

The MINQ experiment was supported by Geological Sur-vey of Queensland, AuScope and ANSIR.

REFERENCES

ARROUCAU P., RAWLINSON N. & SAMBRIDGE M. 2010. New insight into Cai-nozoic sedimentary basins and Palaeozoic suture zones in south-east Australia from ambient noise surface wave tomography.Geophysics Research Letters 37, L07303, doi:10.1029/2009GL041974.

BACCHIN M., MILLIGAN P., WYNNE P. & TRACEY R. 2008. Gravity anomalymap of Australian Region (Third Edition), Geoscience Australia,Canberra.

BEHR Y., TOWNSEND J., BANNISTER S. & SAVAGE M. K. 2010. Shear velocitystructure of the Northland Peninsula, New Zealand, inferredfrom ambient noise correlations. Journal of Geophysical Research115, B05309, doi:10.1029/2009JB006737.

BETTS P. G. & GILES D. 2006. The 1800–1100 Ma tectonic evolution ofAustralia. Precambrian Research 144, 92–125.

BETTS P., GILES D., MARK G., LISTER G., GOLEBY B. & AILL�ERES L. 2006. Syn-thesis of the Proterozoic evolution of the Mt Isa Inlier. AustralianJournal of Earth Sciences 53, 187–211.

BLAKE D. H. 1987. Geology of the Mount Isa Inlier and environs,Queensland and Northern Territory. Bureau of Mineral Resour-ces, Geology and Geophysics, Australia, Bulletin 225, 83 pp.

BLENKINSOP T. G., HUDDLESTONE-HOLMES C. R., FOSTER D. R. W., EDMISTON

M. A., LEPONG P., MARK G., AUSTIN J. R., MURPHY F. C., FORD A. &RUBENACH M. J. 2008. The crustal scale architecture of the EasternSuccession, Mount Isa: The influence of inversion. PrecambrianResearch 163(1–2), 31–49.

DEBAYLE E. & KENNETT B. L. N. 2000a. The Australian continentalupper mantle: Structure and deformation inferred from surfacewaves. Journal of Geophysical Research 105(B11), 25,423–25,450,doi:10.1029/2000JB900212.

DEBAYLE E. & KENNETT B. L. N. 2000b. Anisotropy in the Australianupper mantle from Love and Rayleigh waveform inversion. EarthPlanetary Science Letters 184, 339–351.

ETHERIDGE M. A., RUTLAND R. W. R. & WYBORN L. A. I. 1987. Orogenesisand tectonic process in the early to Middle Proterozoic of north-ern Australia. American Geophysical Union Geodynamics Series17, 131–147.

FAULKNER S. P., MAXWELL M., O’CONNOR L. K., SARGENT S. N. & TALEBI B.2012. Coastal Geothermal Energy Initiative GSQ Julia Creek 1:Well completion report and heat flow modelling results. Queens-land Geological Record, 2012/05.

FISHWICK S., HEINTZ M., KENNETT B. L. N., READING A. M. & YOSHIZAWA K.2008. Steps in lithospheric thickness within eastern Australia,evidence from surface wave tomography. Tectonics 27, TC4009,doi:10.1029/2007TC002116.

FISHWICK S., KENNETT B. L. N. & READING A. M. 2005. Contrasts in litho-spheric structure within the Australian Craton Insights fromsurface wave tomography. Earth Planetary Science Letters 231,163–176.

FITZELL M. J., MAXWELL M., O’CONNOR L. K., SARGENT S. N. & TALEBI B.2012. Coastal Geothermal Energy Initiative GSQ Dobbyn 2: Wellcompletion report and heat flow modelling results. QueenslandGeological Record, 2012/04.

GEOLOGICAL SURVEY OF QUEENSLAND 2011. North-west Queensland Min-eral and Energy Province report. Queensland Department ofEmployment, Economic Development and Innovation, Brisbane.

GERNER E. J. & HOLGATE F. L. 2010. OZTemp-interpreted temperature at5 km depth image. Geoscience Australia, Canberra.

HEINTZ M. & KENNETT B. L. N. 2005. Seismic anisotropy beneath theAustralian continent inferred from shear-wave splitting analysis.Earth Planetary Science Letters 236, 106–119.

HEINTZ M. & KENNETT B. L. N. 2006. The apparently isotropic Austra-lian upper mantle. Geophysics Research Letters 33, L15319,doi:10.1029/2006GL026401.

HOU G., SANTOSH M., QIAN X., LISTER G. & LI J. 2008. Configuration ofthe Late Paleoproterozoic Supercontinent Columbia: Insightsfrom radiating mafic dyke swarms. Gondwana Research 14, 395–409.

HUTTON L. J., DENARO T. J., DHNARAM C. & DERRICK G. M. 2012. MineralSystems in the Mount Isa Inlier. Episodes 35(1), 120–130.

714 E. Saygin et al.

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

JELL P. A. 2013. 2.5 Platform Cover. In: Jell P. A. ed. Geology of Queens-land, pp. 85–86. Geological Survey of Queensland, Department ofNatural Resources & Mines, Brisbane.

KENNETT B. L. N. & ABDULLAH A. 2011. Seismic wave attenuationbeneath the Australasian region. Australian Journal of Earth Sci-ences 58, 285–295.

KENNETT B. L. N., SAMBRIDGE M. S. & WILLIAMSON P. R. 1988. Subspacemethods for inverse problems with multiple parameter classes.Geophysics Journal International 94, 237–247.

KORSCH R. J., HUSTON D. L., HENDERSON R. A., BLEWETT R. S., WITHNALL I.W., FERGUSSON C. L., COLLINS W. J., SAYGIN E., KOSITCIN N., MEIXNER A.J., CHOPPING R., HENSON P. A., CHAMPION D. C., HUTTON L. J., WORMALD

R., HOLZSCHUH J. & COSTELLOE R. D. 2012. Crustal architecture andgeodynamics of North Queensland, Australia: Insights from deepseismic reflection profiling. Tectonophysics 572–573, 76–99, doi:10.1016/j.tecto.2012.02.022.

KORSCH R. J., STRUCKMEYER H. I. M., CARR L. K., HUTTON L.J., CHOPPING R.,HOFFMANN K., KIRKBY A., ROY I. G., FITZELL M., TOTTERDELL J. M. &TALEBI B. 2011. Energy potential of the Millungera Basin: A newlydiscovered basin in north Queensland. The APPEA Journal 51,295–332.

LIANG C. & LANGSTON C. A. 2008. Ambient seismic noise tomographyand structure of eastern North America. Journal of GeophysicalResearch 113, B03309, doi:10.1029/2007JB005350.

LIN F., RITZWOLLER M. H., TOWNSEND J., SAVAGE M. & BANNISTER S. 2007.Ambient noise Rayleigh wave tomography of New Zealand.Geophysical Journal International 170(2), 649–666, doi:10.1111/j.1365-246X.2007.03414.x.

LONGUET-HIGGINS M. S. 1950. A theory of the origin of microseisms.Philosophical Transactions Royal Society of LondonA 243, 1–35.

O’DEA M. G., LISTER G. S., BETTS P. G. & POUND K. S. 1997. A shortenedintraplate rift system in the Proterozoic Mount Isa terrane, NWQueensland, Australia. Tectonics 16(3), 425–441, doi:10.1029/96TC03276.

RAWLINSON N. & FISHWICK S. 2011. Seismic structure of the southeastAustralian lithosphere from surface and body wave tomography.Tectonophysics 572, 111–122, doi:10.1016/j.tecto.2011.11.016.

RAWLINSON N. & SAMBRIDGE M. 2004. Wavefront evolution in stronglyheterogeneous layered media using the fast marching method.Geophysical Journal International 156, 631–647.

RAWLINSON N., READING A. M. & KENNETT B. L. N. 2006. Lithosphericstructure of Tasmania from a novel form of teleseismic tomogra-phy. Journal of Geophysical Research 111, B02301, doi:10.1029/2005JB003803.

RAYMOND O. L., GALLAGHER R., SHAW R. D., YEATES A. N., DOUTCH H. F.,PALFREYMAN D., BLAKE D. H. eds. 2010. Surface geology of Australia1:2 500 000 scale, 2010 edition [Digital Dataset]. Geoscience Aus-tralia, Commonwealth of Australia, Canberra.

REYNOLDS S. D., MILDREN S. D., HILLIS R. R. &MEYER J. J. 2006. Constrain-ing stress magnitudes using petroleum exploration data in theCooper Eromanga Basins, Australia. Tectonophysics 415, 123–140,doi:10.1016/j.tecto.2005.12.005.

SABRA K. G., GERSTOFT P., ROUX P., KUPERMAN W. A. & FEHLER M. C. 2005.Extracting time-domain Green’s function estimates from ambi-ent seismic noise. Geophysics Research Letters 32, L03310,doi:10.1029/2004GL021862.

SAYGIN E. & KENNETT B. L. N. 2010. Ambient seismic noise tomographyof Australian continent. Tectonophysics 481, 116–125, doi:10.1016/j.tecto.2008.11.013.

SAYGIN E. & KENNETT B. L. N. 2012. Crustal structure of Australia fromambient seismic noise tomography. Journal of GeophysicalResearch 117, B01304, doi:10.1029/2011JB008403.

SHAPIRO N. M. & CAMPILLO M. 2004. Emergence of broadband Rayleighwaves from correlations of the ambient seismic noise. GeophysicsResearch Letters 31, L07614, doi:10.1029/2004GL019491.

SHAPIRO N. M., CAMPILLO M., STEHLY L. & RITZWOLLER M. H. 2005. Highresolution surface wave tomography from ambient seismic noise.Science 307, 1615–1618, doi:10.1126/science.1108339

SIMONS F. J., ZIELHUIS A. & VAN DER HILST R. D. 1999. The deep structureof the Australian continent from surface-wave tomography.Lithos 48, 17–43, doi:10.1016/S0024-4937(99)00041-9.

WALTER M. R., VEEVERS J. J., CALVER C. R. & GREY K. 1995. Neoprotero-zoic stratigraphy of the Centralian Superbasin, Australia. Pre-cambrian Research 73, 173–195, doi:10.1016/0301-9268(94)00077-5.

WESSEL P. & SMITH W. H. F. 1998. New, improved version of the GenericMapping Tools Released. EOS Transactions AGU 79, 579.

WITHNALL I. W. & HUTTON L. J. 2013. Proterozoic – North AustralianCraton. In: Jell P. A. ed. Geology of Queensland, pp. 23–112. Geolog-ical Survey of Queensland, Department of Natural Resources &Mines, Brisbane.

WOODHOUSE J. H. 1974. Surface waves in a laterally varying layeredstructure. Geophysical Journal Royal Astronomical Society 37,461–490.

YANG Y., RITZWOLLER M. H. & JONES C. H. 2011. Crustal structure deter-mined from ambient noise tomography near the magmatic cen-ters of the Coso region, southeastern California. Geochemistry,Geophysics, Geosystems 12, Q02009, doi:10.1029/2010GC003362.

YANG Y., ZHENG Y., CHEN J., ZHOU S., CEYLAN S., SANDVOL E., TILMANN F.,PRIESTLEY K., HEARN T. M. & NI J. F. 2010. Rayleigh wave phasevelocity maps of Tibet and the surrounding regions from ambi-ent seismic noise tomography. Geochemistry, Geophysics, Geosys-tems 11, Q08010, doi:10.1029/2010GC003119.

YOSHIZAWA K. & KENNETT B. L. N. 2002. Determination of the influencezone for surface wave paths. Geophysical Journal International149, 440–453.

YOSHIZAWA K. & KENNETT B. L. N. 2004. Multimode surface wave tomog-raphy for the Australian region using a three-stage approachincorporating finite frequency effects. Journal of GeophysicalResearch 109, B02310, doi:10.1029/2002JB002254.

YOUNG M., RAWLINSON N., ARROUCAU P., READING A. M. & TKAL�CI�C H. 2011.High-frequency ambient noise tomography of southeast Aus-tralia: New constraints on Tasmania’s tectonic past. GeophysicsResearch Letters 38, L13313, doi:10.1029/2011GL047971.

ZHOU L., XIE J., SHEN W., ZHENG Y., YANG Y., SHI H. & RITZWOLLER M. H.2012. The structure of the crust and uppermost mantle beneathSouth China from ambient noise and earthquake tomography.Geophysical Journal International 189(3), 1565–1583, doi: 10.1111/j.1365-246X.2012.05423.x.

ZIELHUIS A. & VAN DER HILST R. D. 1996. Mantle structure beneath theeastern Australian region from Partitioned Waveform Inversion.Geophysical Journal International 127, 1–16.

Received 9 April 2013; accepted 16 August 2013

APPENDIX A: GROUP VELOCITY ESTIMATIONFROMGREEN’S FUNCTIONS

Group velocity

For two seismic stations A and B, the cross-correlation ofsimultaneously recorded ambient noise can be repre-sented in the frequency domain as

GABðvÞ ¼Z1

�1dvUAðvÞU�

BðvÞexpð�ivtÞ

where U is the recorded displacement at a station,GAB isGreen’s function between A and B, and � denotes thecomplex conjugate.

Surface waves are subject to dispersion when pass-ing through Earth owing to the influence of the increas-ing seismic wavespeed with depth that is preferentiallysampled by longer period waves. The dispersion has amodulating effect on the shape and speed of the wave-form, which can be used to probe structure. For exam-ple, lower frequencies tend to penetrate deeper and

Seismic Ambient Noise Tomography of Mt Isa region 715

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

therefore sample deeper structures. The group velocity,corresponding to the velocity of energy propagation,can be measured for a set of periods by applying a nar-row-band Gaussian filter to a single Green’s function(Figure A1). The maximum of the envelope shows thearrival of a packet and the velocity can be estimatedusing the distance between two stations and measuredarrival time. In group velocity filtering, the measure-ment can be described as

G0ABðvcÞ ¼ HðvcÞGABðvÞ

where H(vc) is narrow-band Gaussian filter kernel andvc is the chosen centre frequency for narrow band

filtering. A detailed account of the processing steps isgiven in Saygin & Kennett (2010).

APPENDIX B: RESOLUTION ISSUES

Depth sensitivity kernels

The depth resolution attainable with seismic surfacewaves can be described by sensitivity kernels in depth,which require knowledge of the seismic velocity model.In Figure A2, two sensitivity kernel plots for Rayleighwave group velocity are given for cratonic and sedimen-tary environments. Each peak of the normalised ampli-tude of the sensitivity curves indicates the dominantdepths that surface wave will sample. Note the strong

Figure A1 Group velocity measure-ment procedure for Green’s func-tion of a station pair located in theEromanga basin. (a) Time–groupvelocity relation. Hot colours indi-cate localised energy. (b) Same anal-ysis for different frequency bands.The grey curve indicates the enve-lope of the corresponding Green’sfunction. The later arrival of wave-lets with increasing frequenciesindicates the dispersion effect.

Figure A2 Rayleigh wave groupvelocity sensitivity kernels for twodifferent models. The peak ampli-tude of each curve shows the domi-nant depth that the surface wave issensing. (a) Fast crustal velocitymodel representative of the Mt Isadomain. (b) Same model with a2.5 km thick low-velocity cover attop, characteristic of surroundingsedimentary basin areas.

716 E. Saygin et al.

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13

TAJE_A_837098.3d (TAJE) 09-10-2013 17:18

influence of the shallow structure on the shape of thesensitivity kernels.

Spatial resolution

In Figure A3 we present representative tests of spatialresolution based on the recovery of a pattern ofimposed wavespeed perturbations. We have used a two-dimensional sinusoidal form. We have not used the more

conventional ‘chequerboard’ style of perturbationbecause this is not appropriate to the measurementsthat we are making. As can be seen from Figure A3,which shows results at both 3 s and 8 s periods, in thecentral portion of the areawith the dense station spacingrecovery of both the geometry and amplitude of theanomalies is good. But, away from this zone, the geome-try of available paths leads to smearing and streakingand poor amplitude recovery.

Figure A3 Synthetic tests for an input model of two partially superposed sinusoids. (a) Input model. (b) Recovered model for3 s. (c) Recovered model for 8 s. The inset maps show the raypath distribution of travel times used in this study.

Seismic Ambient Noise Tomography of Mt Isa region 717

Dow

nloa

ded

by [

Aus

tral

ian

Nat

iona

l Uni

vers

ity]

at 1

5:09

07

Nov

embe

r 20

13