Seismic Body Wave Attenuation Tomography of the Australian Continent Annual Report 2005

Home , Alpide belt

Seismic Body Wave Attenuation Tomography of the Australian Continent

Agus Abdulah

Research School of Earth Sciences The Australian National University

Annual Report 2005

February 2006

Table of Contents

Abstract...... 2

Chapter 1 ...... 4 Introduction...... 4

Chapter 2 ...... 10 Australian Setting and Previous Seismic Studies of Australia ...... 10 2.1 Australian Setting...... 10 2.1.1 The Australian Continent ...... 10 2.1.2 Surrounding Regions ...... 13 2.2 Previous seismic Tomography and Seismic Attenuation Studies of Australia...... 15 2.2.1 Tomography Studies of Australia ...... 16 2.2.2 Seismic Attenuation Studies of Australia ...... 21

Chapter 3 ...... 24 Seismic Attenuation Tomography...... 24 3.1 Seismic Attenuation Measurement Theory...... 24 3.2 Method of Attenuation Measurements...... 26 3.2.1 The Station Ratio Method ...... 27 3.2.2 The Wave Ratio Method...... 28 3.3 Attenuation Measurement Technique ...... 29 3.4 Robust Attenuation Measurements of Australian Data ...... 33 3.4.1 Seismic Data ...... 33 * 3.4.2 Differential Attenuation (∆t SP) Representations ...... 33 3.5 Seismic Attenuation Tomographic System of Equations and Model Parameterizations ...... 33 3.6 Checkerboard Test ...... 40

Chapter 4 ...... 42 Australian Seismic Wave Speeds, Attenuation, and Anisotropy Models ...... 42 4.1 Seismic Wave Speeds Models ...... 42 4.2 Seismic Attenuation Models...... 43 4.3 Seismic Attenuation Anisotropy Model ...... 46

Future Work...... 48

References ...... 49

1 Abstract

The strategic position of the Australian continent in the middle of the seismicity belt which extends from Indonesia, through New Guinea to Fiji, Tonga and

New Zealand with the extensive deployment of portable broadband seismic stations across the Australian Continent and Tasmania since 1993 offers robust seismological data with a dense coverage at distances from 5° to 45°. P and S wave seismic traveltimes from nearly 4050 three-component seismic dataset from the record have been hand picked. The wave ratio method is then applied to estimate the spectral ratio between shear (SH and SV) and compressional waves. Seismic spectra are estimated using the Multitaper method with 512 points in window range of 30s to 45s and frequency range of 0.25Hz to 1.00Hz.

Three-dimensional P and S wave speed tomography is conducted by inverting a kernel matrix obtained from a quasi three dimensional ray tracing which respect to P and S wave seismic traveltime residuals from the ak135 model. The study area from latitude 22° N to 65S° and longitude 78° to 189° and 0-1240km depth is discretized into 11100 cells with a cell size 3°x3° and depth range of 35 to 200km. Both P and S- wave speed information from the seismic wave speed tomography are then utilized as data input for the three-dimensional seismic attenuation tomography. In this inversion, it is assumed that QP=2.3QS. The seismic attenuation anisotropy in term of the ratio between seismic attenuation derived from SV and SH component is also presented.

The major feature that is revealed from the both seismic wave and seismic attenuation studies is a strong contrast in deep structure between central Australia and the eastern seaboard. The Archaean and the Proterozoic rocks in the west and in the middle of the continent are associated with a high seismic wave speed anomaly and

2 low seismic attenuation and the Phanerozoic rocks and the presence of recent volcanism and region of high heat flow in the east are associated with low seismic wave speed anomaly and high seismic attenuation. The representation of seismic attenuation anisotropy suggests that in the region where seismic coverage is good, transverse component (SH) wave is less attenuated than radial component (SV).

3 Chapter 1

Introduction

The strategic position of the Australian continent in the middle of seismicity belts to the north and east of the continent: the world’s greatest earthquake belt, the circum-Pacific belt (also called “The Ring of Fire”) which extends from Japan,

Philippine Islands, New Guinea, the island groups of the Southwest Pacific, and to

New Zealand and the Alpide belt which extends from Java to Sumatra and the mid- ocean ridge to the south of the continent provides a wealth of events at suitable distances to be used as probes into the seismic structure of the upper mantle. The extensive deployments of portable broadband seismic stations across the Australian

Continent and Tasmania since 1993 offers robust seismological data with a dense coverage at distances from 5° to 45°. Over the last two decades, a wide range of studies has been used to gain information even on one-dimensional and three- dimensional structure in the mantle which exploits different aspects of seismograms.

Studies on seismic tomography in which utilize thousands of seismic traveltimes picked from high quality body wave seismograms and utilize seismic waveform of large amplitude surfaces waves in the later part of the seismogram travel nearly horizontally has been conducted and successfully to delineate the major features of the geological structure beneath the Australian Continent. Study on seismic travel time tomography for the Australian region was pioneered by

Widiyantoro & van der Hilst (1996) who produced tomographic images in a zone covering the southern Philippines, Malaysia, Indonesia, Papua New Guinea and northern Australia by linearized inversion of travel-time data of direct P phases and

4 the surface-reflected depth phases pP and pwP. The radially stratified iasp91 model was used as global reference for the seismic velocities and for the tracing of the ray paths. The images suggest that the lithospheric slab beneath the Sunda island arc penetrates to a depth of at least 1500 kilometers. The slab was imaged as a continuous feature from the surface to the lower mantle below Java, with a local deflection where the slab continues into the lower mantle. Subsequently, Gorbatov and Kennett (2001) have implemented a tomographic inversion with a parameterization which has smaller cells near the earth’s surface and larger cells at depth. Blocks of 50x50x50 km have been used in the upper mantle, and the cell size increased to 100x100x100 km blocks in the middle mantle. An iterative development with a 3-D ray tracing algorithm which exploits the 3-D structures recovered during the course of inversion have been used. It is reveals that, the tomographic image for the southwestern Pacific region in which the subduction zones below New Caledonia, the Solomon Islands, and Papua New Guinea shows seismic speed contrast with their surroundings.

Effort in exploring the features of geological structure beneath the Australian

Continent and surrounding regions is continued in the implementation of surface wave tomography to an abundant of seismic data recorded by temporary broadband seismometers which have been deployed in a numerous number of projects conducted by Research School of Earth Sciences, the Australian National University: SKIPPY

[continent-wide,1993-1996], KIMBA [Kimberley Block, 1997,1998], QUOLL

[southeast Australia, 1999], WACRATON [Western Australia, 2000-2001, 2002-

2003], TIGGER [Tasmania, 2001-2002], and TASMAL experiment [Gulf of

Carpentaria to the southern Australia, 2003-]. Zielhuis and van der Hilst (1996) have implemented the “Partitioned Waveform Inversion” (PWI) in which has two stages of

5 procedure. The first step generates a shear wavespeed model for each path and the second combines the path dependent information in to a 3-D model. They presented the first model based on the analysis of the Rayleigh wave data from the stations in the eastern Australia. This already indicated the presence of a very major contrast in the shear wavespeed in the mantle component of the lithosphere between central

Australia and eastern Australia. The 3-D model indicated the presence of lowered shear wavespeeds beneath the east coast of Australia at depths around 140 km and confirmed the inferences from surface wave dispersion. The location of the zones of low wavespeed has a strong correlation with Neogene volcanism. By using the same technique, van der Hilst et al. (1998) and Simons et al. (1998) have developed 3-D models as more data have been incorporated. The models suggest that the areas of exposed Precambrian rocks largely correspond to high seismic wavespeeds but there are zones of enhanced wavespeed extending to 150 km or more lying to the east of the conventional Tasman line. The Precambrian regions show the presence of significant internal structure with an indication of the separation of the major cratonic blocks, especially at shallower depths. A three-stage inversion technique for surface wave tomography is applied to the Australian region [Yoshizawa and Kennett, 2004]. In the first stage, path-specific one-dimensional (1-D) shear velocity profiles are derived from multimode waveform inversion to provide dispersion information. The information from all paths is then combined to produce multimode phase speed maps as a function of frequency. The second stage, the 2-D phase speed maps are updated by including ray tracing and finite frequency effects through the influence zone around the surface wave paths over which the phase is coherent. The third stage, the

3-D shear wave speed distribution is reconstructed from the set of updated multimode phase speed maps. The final 3-D model of the Australian region shows the faster

6 wave speed anomaly in the center of the continent down to 300 km depth. The higher- velocity anomalies down to depths of 200-250 km beneath the middle and western part of Australia, corresponding to the continental lithosphere of the Australian

Continent. In the region beneath the Proterozoic blocks in the central Australia, the continental lithosphere seems to reach 300 km. The model shows improvement in the regions with high gradients in shear velocity, such as near tectonic boundaries, especially in the eastern Australia. The reliability of the tomographic method has been improved by the development of two new techniques [Fishwick, 2005]: multiple starting models are used in the waveform inversion procedure and introducing a multi-scale component, where the final model is damped towards the large scale features of the data rather than a global reference model. These new techniques have been implemented to robust seismic data from all Australian National University temporary deployments, and additional data from permanent stations and a temporary deployment in New Zealand. The tomographic models for the Australasian region reveal at 75 km depth a broad-scale correlation between the age of the sea floor and the shear velocity. Within the continent, a zone of low wavespeeds, a typical for a shield region, is observed in central Australia.

Previous seismic attenuation studies of the Australasian region in term of quantifying the slope of the spectral ratio between S and P windows on the same seismograms to extract a quantitative estimate of differential attenuation have been initiated by Gudmundsson et al. (1994) for the analysis of 22 seismograms from the

WRA broadband instrument and a further four from portable broadband instruments deployed near Warramunga. They found that the measurements are consistent with the average quality factor for S waves in the lid of the order of Qs=1400 (Qp=2800) on the top of a highly attenuative asthenosphere of 200km thickness with Qs=100

7 (Qp=200), which is underlain by a transition zone with Qs=600 (Qp=1200). A recent study of attenuation in the Australian region is carried out by Cheng (2000). He studied the Australian attenuation structure by estimating spectral ratio of nearly 2000 three-component seismograms from SKIPPY project [1993-1996]. The spectra of the

P, SV, and SH waves and their accompanying noise were estimated by using FFT of single real function. The differential attenuation of each ray path is measured by estimating the slope of spectral ratio of S to P waves. The differential attenuation data are organized into azimuthal corridors and inverted by using the Neighbourhood

Algorithm (NA) to produce a set of 1-D Q profiles. A 3-D Q model at fixed frequency is then constructed by combining these 1-D Q profiles weighted by ray density. His measurements clearly delineate major variations in attenuation between cratonic structures in the centre and west and the eastern part of Australia, with much stronger attenuation in the east. The differential attenuation correlates with the velocity structure. Weak attenuation of S waves was found in central and western Australia where the S velocity is high. However, strong attenuation of S waves is found in eastern Australia and the Coral Sea where the S wave velocity is low.

The genuine seismic tomography technique for exploiting thousand of seismograms from previous and recent projects is carried out to give a better understanding in the interpretation of three-dimensional structure and temperature distribution beneath the Australian continent and its surrounding regions. The major feature that is revealed from this study is a strong contrast in deep structure between central Australia and the eastern seaboard. The Archaean and the Proterozoic rocks in the west and in the middle of the continent are associated with a high seismic wave speed anomaly and low seismic attenuation and the Phanerozoic rocks and the presence of recent volcanism and region of high heat flow in the east are associated

8 with low seismic wave speed anomaly and high seismic attenuation. With three- component recording from which rotated transverse (SH) and radial (SV) components can be deduced from it, the three-dimensional seismic attenuation anisotropy is also presented. The representation of seismic attenuation anisotropy model suggests that in the region where seismic coverage is good, transverse component (SH) wave is less attenuated than radial component (SV).

9 Chapter 2

Australian Setting and Previous Seismic Studies of Australia

In the Geophysical Corner published by American Association of Petroleum

Geologists (2005), Alistair R. Brown said: “Everyone is a product of their own experience. Hence geophysicists tend to favor geophysical methods and geologists tend to favor geological methods. It’s only natural”. Since what seismologists observed from the earth is ‘signal’ and the signal is then processed and interpreted for understanding the geological features of the earth, the balance between geophysics and geology knowledge is important. Therefore, basic knowledge of geological and tectonic features of Australasia is needed to have a better understanding in the interpretation of seismic image representations of this region.

2.1 Australian Setting

2.1.1 The Australian Continent

Based on studies of rocks at the surface, in general the basement structure of the Australian continent is divided into three divisions which major stages in the evolution of the crust: the Archaean (older than 2500 Ma), the Proterozoic (2500-550

Ma) and the Phanerozoic (younger than 550 Ma). The Major Archaean outcrop lies in the west and the Proterozoic rocks are found in the east of the continent (Figure 2.1).

The Australian Continent is characterized by several cratons: Pilbara Craton, Yilgarn

10 Craton, Gawler Craton, Altjawarra Craton, Central Craton, and Curnamona Craton.

The Yilgarn Craton is characterized by granite-greenstone belts bounded by major strike-slip faults and shear zones, the south-western gneiss terranes and the gneissic

Narryer terrane in the northwestern part of the craton. Based on seismic reflection data over the eastern Goldfields Province, it suggested that there is three-layered crust

[Drummond et al., 2000]. The uppermost layer is formed by greenstones belts in the central and eastern parts of the section. The mid-crustal section (10-20km) consists of an east-dipping, west vergent duplex system, whereas the lower crust is characterized by shallowly dipping reflectors that are interpreted to represent ductile deformation

[Drummond et al., 2000]. The seismic reflection data also shows that the crustal thickness of the Yilgarn and Pilbara Cratons is between 30 and 35 km thick [Betts et al., 2002].

Figure 2.1: Geological map of the Australian plate showing major Archaean and Proterozoic terranes and Paleozoic Tasmanides [courtesy of Betts et al., 2002].

11 The Pilbara Craton is located in the south of Yilgarn Craton comprises a central granite-grenstone belt characterized by a outcrop pattern dominated by 50-100 km diameter domal granitoid complexes, separated by synformal greenstones belts

[Oliver & Cawood, 2001]. The Pilbara Craton consists of a series of early to mid-

Archaean to Paleoproterozoic strata of the Hamersley Basin [Hickman, 1983].

Between the Pilbara and Yilgarn Cratons there is the Carpicorn Orogen which is characterized by regional metamorphism and plutonism with the main activity ending at 1840 Ma [Myers et al.,1996]. According to teleseismic receiver functions study of Reading & Kennett (2003) with three-component temporary stations were deployed in a line running southwards across Pilbara Craton, Capricorn and Yilgarn

Craton, it is suggested that the crust-mantle boundary under the Pilbara Craton to be shallow, at 30 km, with a sharp Moho and high-velocity crust beneath the exposed

Pilbara granite-greenstone terrane. The Yilgarn Craton which extends beneath the basins exposed on the surface is deeper at 40 km.

In the south Australian part there is the Gawler Craton which is part of a complex collage of Archean to Mesoproterozoic metasedimentary and metaigneous terranes in southern Australia [Dirren et al., 2005]. The core of the Gawler Craton is comprised of Archean gneissic complexes, whose protoliths may be as old as 2.98 Ga [Dawson et al., 2002]. In the eastern part of Australian Continent, there are three Orogens named Lachlan, New England, and Thompson Orogens. The Lachlan Orogen is a turbidite-dominated orogen that forms the central part of composite Palaeozoic

Tasman Orogen [Coney et al., 1990] along the eastern margin of Australia.

Successive cratonisation from west to east included the Early Palaeozoic Delamerian

Orogen (550–470Ma), the Middle Palaeozoic Lachlan Orogen (450–340 Ma) and the

Late Palaeozoic to Early Mesozoic New England Orogen (310–210 Ma), with their

12 respective peak deformations of Late Cambrian – Early Ordovician, Late Ordovician

– Silurian and Permian–Triassic age [Foster & Gray, 2000]. The Lachlan part is a

Middle Palaeozoic orogen with a 200 million years history that occupies ~50% of the present outcrop of the Tasman Orogen [Gray & Foster, 2004]. The Lachlan and New

England Orogens belong to an orogenic system that extended approximately 20,000 km along the eastern margin of Gondwana between the northern Andes and eastern

Australia [Gray & Foster, 1998].

2.1.2 Surrounding Regions

In the surrounding regions of the Australian Continent there are several geological and tectonic features such as the Tasman Sea, the Coral Sea, the New

Caledonia Basin, the Norfolk Ridge, the Australian Antarctic Discordance, the South

East Indian Ridge, the West Australian Basin, the Lord Howe Rise, the Macquarie

Ridge, and the Indian-Antarctic Ridge (Figure 2.2).

The Tasman Sea is located to the east of the Australian Continent, it is an ocean basin which is bounded by the Lord Howe Rise and New Zealand to the east and to the south by major discordance that separates it from younger oceanic crust generated at the South East Indian Ridge and the extinct the Macquarie Spreading center. In the northern part, it contains an elongated segment of continental crust, the

Dampier Ridge, separated from the Lord Howe Rise by two small basins: the Lord

Howe and the Middleton basins. Based on tectonic lineaments visible in the gravity grid and interpreted as strike-slip faults, by magnetic anomaly, bathymetry, and seismic data, Gaina et al. (1998) identified 13 tectonic units. These 13 tectonic blocks and the Australian continent gradually separated due to either extensional or strike- slip movements, generating the Tasman Sea, the Lord Howe and the Middleton

13 basins, and several failed rifts. The opening of the northern Tasman Sea is pretty explained by the gradual stepwise separation of four Dampier Ridge tectonic fragments plus the Chesterfield Plateau and Australia in the frame work of a northward propagating rift.

Figure 2.2: Geographic and Tectonic features in the surrounding regions of the Australian Continent [digital data courtesy of NOAA].

Another major oceanic basin border the Australian margin to the east is the

Coral Sea. The Coral Sea Basin is located northeast of Australia and is rimmed by several submarine plateaus such as the Queensland Plateau to the southwest, to the northwest by the Eastern Plateau, the Papuan Plateau to the North, the Louisiade

Plateau to the northeast, and the Mellish Rise to the southeast. The Coral Sea basin is formed by oceanic crust as Early Eocene [op cit. Gaina et al., 1999]. According to magnetic anomaly interpretation and fracture zone data revealed from satellite-derived gravity anomalies, Gaina et al. (1999) derived finite rotations for the Coral Sea opening. These rotations are combined with the previous work of Gaina et al. (1998) in order to interpret the relative motion between the Louisiade Plateau, and the

14 Mellish Rise. They revealed the existence of a triple junction between the Australian

Plate, Lousiade Plateau and several small plateaus attached to the northern Lord Howe

Rise.

To the south of the Australian Continent, there is the Australian Antarctic

Discordance which is a zone of subdued ridge morphology and a series of north-south trending fracture zones [Weissel & Hayes, 1974]. The Australian Antarctic

Discordance is the deepest portion of the mid-ocean ridge system with 600km long segment of the South East Indian Ridge. It has anomalous in terms of bathymetry and is characterized by unusual sea-floor morphology, isotope geochemistry, petrology, and seismic structure [Gurnis and Müller, 2003]. The Australian Antarctic

Discordance is associated to the location of former Mesozoic position of long-lived seduction on the Pacific margin of Australia. Seismic tomographic images suggest that beneath this former Mesozoic margin is a linear north-south high velocity seismic anomaly within the lower mantle and a high velocity anomaly within the transition zone, as originally predicted by dynamic models [Gurnis and Müller, 2003].

2.2 Previous Seismic Tomography and Seismic Attenuation Studies of Australia

Studies on seismic tomography of the Australasian region have been extensively conducted since data availability from deployments of broadband instruments. In the SKIPPY experiment from 1993 to 1996 [van der Hilst et al.,

1994], recorders were deployed across the whole continent at approximately 400 km spacing. Subsequently in 1997 and 1998 a denser array of broadband instrument

(KIMBA) were deployed in northwestern Australia. In 1999 additional recorders were installed in southeastern Australia (QUOLL). In 2000-2001, Western Australia was

15 revisited with a broadly spread array to supplement the SKIPPY stations which had had technical problems. In less than a decade there has been a very thorough coverage of the Australian continent with broadband seismic stations. The high quality seismic data availability at a suitable distance from events in surrounding region with a dense stations spacing is ideal to explore the deep structure beneath the continent using seismic tomography technique.

2.2.1 Tomography Studies of Australia

Studies on surface wave tomography of the Australian Continent have been initiated by Zielhuis and van der Hilst (1996) by analyzing Rayleigh wave data from the stations in eastern Australia deployed in SKIPPY experiment. The model suggests the presence of a major contrast in the shear wavespeed in the mantle component of the lithosphere between central Australia and eastern Australia. The model revealed the presence of lowered shear wavespeeds beneath the east coast of Australia at depths around 140km which has a strong correlation with Neogene volcanism. The

Australian surface wave tomographic models have been improved as more data have been incorporated by van der Hilst et al. (1998) and Simons et al. (1999). They implemented the “Partitioned Waveform Inversion” (PWI) on approximately 2000

Rayleigh wave paths recorded in SKIPPY experiment. The shear wavespeed distribution for SV waves at 200 km depths derived from this inversion is presented in

Figure 2.3 below.

Figure 2.3: Distribution of radial (SV) waves of the Australasian region at 200 km depth derived from 2000 Rayleigh waves.

Shear wave velocity contrasts between eastern Australia and central

Australia (between 140ºE and 145ºE) is clearly pronounced at 200 depths which may be associated with the conventional Tasman line. The zones of exposed Precambrian rocks are associated with high seismic wavespeeds. The Precambrian regions show the presence of significant internal structure with an indication of the separation of the major cratonic blocks, especially at shallower depths.

The availability of three-component seismic data allows the exploitation of polarization anisotropy beneath the Australian Continent. Debayle & Kennett (2003) presented radial anisotropy in term of the ratio between tangential (SH) waves velocity and radial (SV) velocity. As can be seen from Figure 2.4 in the continental region the value of the ratio is higher than 1 which means polarization anisotropy of

SH is faster than SV. The vertical cross section at -25ºS suggests at central and eastern

17 Australia SH is faster than SV down to 200-250km. The area where SH is faster than

SV is associated with horizontal flow dominates in the upper mantle.

Figure 2.4: Distribution of polarization anisotropy model at slice 125km depth and vertical cross section at -25ºS.

The recent work on surface wave tomography has been conducted by Fishwick

(2005) as more data from all Australian National University temporary deployments and additional data from permanent stations and temporary deployment in New

Zealand are incorporated. He has also improved the reliability of the tomographic method by the development of two new techniques: the first: Multiple starting models are used in the waveform inversion procedure, improving both the data available and

18 the reliability of the 1D models used within the tomographic inversion. The second: the tomographic inversion has been improved by introducing a multi-scale component, where the final model is now damped towards the large scale features of the data rather than a global reference model.

Figure 2.5: Correlations among age and shear wave speeds anomaly in Australasian

Figure 2.5 above shows the shallowest layer of the tomographic model at slice

75km depth. The model suggests correlations between shear wavespeeds and geological structures in the oceans surrounding Australia. Evidence from the depth of the sea floor and the reduced heat flow away from mid-oceanic ridges suggests that the oceanic lithosphere cools with increasing age. The image suggests that shear wavespeeds is increased as increasing age. Beneath the old West Australian Basin, fast shear velocity perturbations are observed, and beneath the young Southeast

Indian, and Indian-Antarctic ridge system much lower wavespeeds are imaged. To the east of Australia, the fastest wavespeeds are observed beneath the Tasman and Coral

Seas and appear to be related to the coherent area of sea floor spreading.

Figure 2.6: Correlations between Tasman Line concept and surface wave tomography at 150km depth.

Slice at 150km depth of shear wavespeeds derived from surface wave tomography is presented in Figure 2.6 along with proposed Tasman Lines locations.

At this depth the transition from slower to faster velocities is not a simple linear feature, and appear to somewhat complex boundary between 138ºE and 143ºE

[Fishwick, 2005].

The seismic structure beneath the Australasian regions also has been analyzed using seismic traveltimes tomography. van der Hilst et al. (1998) presented compressional (P) and shear (S) waves models derived from data set contains 544690

P-wave and 393866 S-wave ray paths. All arrival times were selected from the catalog which have source or receiver located within the zone of study (95ºE to 190ºE; 50ºS to 10ºN). Since calculations of 3D ray paths for a large amount of data are very time consuming they combined the information from event clusters in a 2º x 2º x 50 km volume and station clusters in a 2º x 2º region into a single summary ray path for the

20 region outside of the zone under study. The residual time assigned to the summary ray was the median of all the relevant data selected for summary ray path. Each summary ray was composed of at least three individual rays. The resulting data set contains

544690 P-wave and 393866 S-wave ray paths. The region of interest was parameterized by irregular cells from 0.5ºx0.5º to 1ºx1º and 19 layers down to the depth of 1600 km. The whole Earth mantle was parameterized by cells of 5ºx5º and

16 layers. The reference 1D Earth’s velocity model was ak135 of Kennett et al.

(1995). The extended pseudobending ray tracing algorithm of Koketsu and Sekine

(1998) is implemented in their tomographic code. The tomographic models for P and

S waves at 210-270km depth are presented in Figure 2.7.

Figure 2.7: P and S wave tomographic models derived from traveltimes tomography. The colorbar is presented in percent relative to ak135 model.

2.2.2 Seismic Attenuation Studies of Australia

The studies of attenuation of the Australian region were initiated by Clements

[1982] who studied the intrinsic attenuation and frequency dependence by utilizing seismograms at the Warramunga array in the Northern Territory of Australia.

21 Subsequently, Clements [1992] revealed that the values of Q for P and S waves are frequency dependent.

An attenuation profile of the upper mantle beneath the north of the Australian continent was presented by Gudmundsson et al. [1994] from the analysis of 22 seismograms from the WRA broadband instrument and a further four from portable broadband instruments deployed near Warramunga. They found that the measurements are consistent with the average quality factor for S waves in the lid of the order of Qs=1400 (Qp=2800) on the top of a highly attenuative asthenosphere of

200km thickness with Qs=100 (Qp=200), which is underlain by a transition zone with

Qs=600 (Qp=1200).

A recent study of attenuation in the Australian region is carried out by Cheng

[2000]. They studied the Australian attenuation structure and frequency dependence of attenuation by estimating spectral ratio of nearly 2000 three-component seismograms from SKIPPY project [1993-1996]. The spectra of the P, SV, and SH waves and their accompanying noise were estimated by using FFT of single real function. The differential attenuation of each raypath is measured by estimating the slope of spectral ratio of S to P waves. The differential attenuation data are organized into azimuthal corridors and inverted by using neighbourhood algorithm (NA) to produce a set of 1-D Q profiles. Then a 3-D Q model at fixed frequency was constructed by combining these 1-D Q profiles weighted by ray density. Slice of attenuation and α at 77.5-120km depth is presented in Figure 2.8. His measurements clearly delineate major variations in attenuation between cratonic structures in the centre and west and the eastern part of Australia, with much stronger attenuation in the east. The differential attenuation correlates well with the velocity structure. Weak attenuation of S waves was found in central and western Australia where the S

22 velocity is high. However, strong attenuation of S waves is found in eastern Australia and the Coral Sea where the S wave velocity is low.

Figure 2.8: Slice of attenuation (Q) and frequency dependency (α) at 77.5-120km depth.

The frequency dependence parameter (α) is also estimated directly from the spectral ratio. The raypaths covering the north-west part of the Australian continent show α close to zero with a small error in α; so that the frequency dependence of Q in this area is relatively weak. In the eastern part of Australia and Coral Sea area, there is a mixture of paths with small and larger α so that the frequency dependence in those areas is more complex and depends on the depth of penetration of the waves.

23 Chapter 3

Seismic Attenuation Tomography

3.1 Seismic Attenuation Measurement Theory

Seismic attenuation which is described by a quality factor Q-1 has become an important property for imaging the earth structure. The effective measurement of the intrinsic attenuation is one of the most difficult problems in global seismology. This is because the amplitude and frequency content of seismic waves are affected by errors in the determination of the magnitudes and source mechanisms, and by variations in elastic structure of the earth [Selby, 2003]. Thus, methodologies to measure seismic attenuation are designed to cancel or at least to minimize these errors, although they are still based on simplifying assumptions.

To give detailed explanation of seismic attenuation measurements, we follow the comprehensive review which is provided by Båth [1974], also we refer to Pujol

[2003], Teng [1968], and Gudmundsson et al. [2001].

Consider a seismic wave that propagates from source to receiver. During that propagation, the seismic wave is affected by three main factors [Båth, 1974]: The first is the seismic source properties, which includes the source time function and the source space function (source mechanism and focal depth), magnitude, coupling, the surrounding geology, stress field, reverberations etc in the source crust. The second is the path properties, such as absorption and scattering, reflection, refraction and diffraction, dispersion, interference and geometrical spreading. The third is the receiver properties which include reverberation and other wave interactions, as a signal with noise, resonance effects, in the receiver crust, also the receiver represents

24 a filter, constituting the last factor before the record is obtained. Thus, by applying this representation to seismic body waves, we can write the observed amplitude at frequency ω as:

A(ω,r )= S ()()ω B θ Cs ()ω M (ω,r )G(r)Cr (ω,r)I(ω) (3.1) where:

A()ω,r = receiver spectrum at distance r

S()ω = source spectrum, corresponding to source time function

B()θ = source space function, where θ stands for direction from source

Cs ()ω = source crust effect on spectrum

M ()ω,r = mantle effect on spectrum

G()r = amplitude function in propagation

Cr ()ω,r = receiver crust effect on spectrum

I()ω = instrumental response

If we transform equation 3.1 into its amplitude and phase forms, we have:

A()ω,r = S()ω B()θ Cs ()ω M (ω,r)G(r)Cr (ω,r) I(ω) (3.2a) and:

Φ X = Φ S + Φ p + Φ I + 2mπ (3.2b) where ФS and ФP stand for the source and path effects, respectively.

Equation 3.2 forms the basis for procedures to suppress specific parts of the response.

The term G()r is a function of geometrical spreading and the attenuation effect

[Kurita, 1968]:

 ω dr  G()r = g ()r exp− ∫  (3.3)  2 Qa ()()ω,r Va r 

25 where g()r is the geometrical spreading and the second term on the right-side is the effect of attenuation.

3.2 Methods for Attenuation Measurements

Let us consider an amplitude ratio between two body waves labeled a and b.

We employ the ratio of the two cases of equation 3.2a in a form that emphasis attenuation effects (equation 3.3):

 ω dr  S a ()ω Ba ()θ Csa ()ω g a ()r Cra (ω,r )I a ()ω exp−  A ()ω,r 2 ∫ Q ()()ω,r V r a =  a a  (3.4) A ()ω,r  ω dr  b S ω B θ C ω g r C ω,r I ω exp − b () b () sb ()b ()rb ( ) b ()  ∫   2 Qb ()()ω,r Vb r 

Here we let the mantle effect M appear through attenuation; if in addition reflection or refraction enters the propagation, the corresponding coefficient must be included on the right-hand side [Båth,1974].

According to the way in which the ratio is evaluated, Båth (1974) suggests three methods for attenuation measurements based on reviews of attenuation studies in the 1960s and 1970s: (1) the frequency-ratio method: the method is mainly concerned with the slope of spectra but can be enlarged to include spectral shapes, then comparison between spectral shapes of a particular wave at a particular station can be made; (2) the station-ratio method: this method estimates the ratio between two spectral recorded at two stations or at one stations resulted by two epicenters; and

(3) the wave-ratio method: this method estimates spectral ratio between two types of waves e.g. P and S waves which come from one event and recorded at one station.

These methods still form the mainstay of attenuation studies.

26 In this chapter it is discussed the second and third attenuation measurement methods i.e. the station-ratio method and the wave-ratio method which are used for the analysis of Australasian attenuation.

3.1.1 The Station-Ratio Method

Consider two amplitudes spectra of the same seismic phase for a single event

recorded at two stations as Aa (ω) and Ab (ω) respectively. Under the assumption that these two recorders are identical or after instrumental correction, taking the natural logarithm and simplify equation 3.4, we obtain:

A()ω a G Ca ()ω ω  dr dr  ln = ln a − ∫∫−  (3.5a) A()ω b Gb Cb ()ω 2  abQV QV  or:

A()ω a Cb ()ω Ga ω * * ln = ln − (ta − tb ) (3.5b) A()ω b Ca ()ω Gb 2 where t* = ∫ dr QV

The ratio on the left-hand side is called reduced spectral ratio and the difference

* * ta − tb is called differential attenuation [Båth, 1974].

The equation 3.4 could be further simplified by considering an average Q between two stations, so we have:

A()ω a G C ()ω ω()t − t G C (ω) ω(r − r ) ln = ln a a + b a = ln a a + b a (3.6) A()ω b Gb Cb ()ω 2Q Gb Cb ()ω 2QV

If the amplitude of seismic wave is corrected for geometrical spreading and crustal effects, we obtain:

27 A()ω G C ()ω ω()r − r ln a a = a (3.7) A()ω a G C()ω 2QV

The equation 3.6 is suitable for application to only two stations, when the first term of the right-hand side is a constant. By using equation 3.6, we estimate the differential attenuation between two P waves and two S waves recorded at different stations. For this purpose, we plot ln A(ω)a A(ω)b versus ω. Then, in the range of seismic frequency band, linear regression is applied to obtain a slope and yield a value of Q. In this method, the estimated value of Q is the relative to the reference.

Alternatively we can use equation 3.5b directly to estimate the differential

* * * t = (ta − tb ) from the slope of the corrected spectral ratio as a function of frequency.

3.1.2 The Wave-Ratio Method

This method has been applied on Australian seismic data to estimate differential attenuation between two body waves P and S [see e.g. Gudmundsson et al., 1994; Cheng H.-X & Kennett, 2002]. This method has advantage of canceling unpredictable, common, frequency-dependent factors by means of the spectral ratio and avoids the geometrical effects on absolute amplitudes by measuring the natural logarithm frequency-derivative of the spectral ratio [Gudmundsson et al., 1994]. The spectral ratio between two body waves a and b from the same event recorded at the same receiver cancels most effects of source and instrument function. By assuming that these two waves have a similar path, the effect of geometrical spreading and radiation pattern can be eliminated as well. Based on these assumptions, recalling equation 3.4 to obtain:

28 A()ω b ω  dr dr  ln = − − + ln g r − ln g r (3.8) ∫∫ b ()a () A()ω a 2  baQV QV 

Since the geometrical spreading is frequency independence, the only remaining frequency-dependent contribution is the effect of attenuation. Rewrite equation 3.8:

A()ω b ω ln = − (t * − t * ) + c = −πf (t * − t * ) + c (3.9) A()ω a 2 b a b a

In the linear system of amplitude ratio in the left-hand side against frequency f, the

* * * slope is (tb − ta ) , which is defined asδtba .

3.3 Attenuation Measurement Technique

Both the Multitaper and Fast Fourier Transform (FFT) Methods are applied to estimate the spectra of the seismic data. I have evaluated the both methods on the

Australian seismic data and we found that both the two methods have advantages and disadvantages. The spectra estimated using the Multitaper method are smoother and have a lower variance than the FFT; in this case, the Multitaper method is suitable for calculating spectral ratio in which linear regression is applied. However, the

Multitaper method has a drawback in that the spectral estimates are limited to frequencies 0.25Hz where as the FFT method that can resolve spectra down to 0.1Hz.

In practices, I also do a first quick look at the waves spectra and visibly comparing the frequency content of the seismic waves and classified the records into

4 groups as used previously by Cheng and Kennett [2000]:

A : similar frequency for both a and b waves

B : b slightly lower frequency than a

C : b clearly lower frequency than a

D : b low frequency compared to a

29 Then, the Multitaper method is applied on class A, B and C, while the FFT is only applied to class D.

The three-component seismic data in which contains Vertical (BHZ), East-

West (BHE) and North-West (BHN) components is decomposed into Radial (R),

Transverse (T), and Horizontal (H) components by following scheme of Kennett

(1991).

R(t)=BHN(t)* COS(α)+BHE(t)* SIN(α),

T(t)=-BHN(t)* SIN(α)+BHE(t)* COS(α), and

H(f)=[R(f)2+T(f)2]0.5

Where t is time domain, f is frequency domain, and α is azimuth.

This decomposition has no effect on the shape of the amplitude spectra, but slightly reduces the relative level noise which is incoherent in time across components.

* For calculation of differential attenuation between S and P waves (∆t SP), spectra of P wave is estimated from Vertical component (BHZ) and S waves spectra from T, R, and H with a vary time window between 25 and 45s. The contribution of noise to the spectra is removed upon the assumption that the data and the noise are independent in phase. The noise spectra are estimated in the range of 25 to 45s before

P and S first arrivals. Smoothing is not applied to the noise and signal spectra. To remove the noise contribution, I apply:

Z P ( f ) = Z P ( f ) − Z nP ( f ) , RS ( f ) = RS ( f ) − RnS ( f ) ,

TS ( f ) = TS ( f ) − TnS ( f ) , H S ( f ) = H S ( f ) − H nS ( f ) , (3.10)

The differential attenuation between S and P wave spectra can be obtained from the slope of the linear regression of the amplitude spectra ratio against frequency at the seismic frequency band of 0.32 to 1.0Hz. I also only calculate the amplitude

30 spectra of the seismic signal which has a value above the noise level. The differential attenuation of each component is defined by:

ln()RS ( f ) Z P ( f ) /π , ln()TS ( f ) Z P ( f ) /π

ln()H S ( f ) Z P ( f ) /π (3.11)

Figure 3.1 shows an implementation of the spectral ratio explained above to unfiltered Australian seismic data 2004.177.2.TL13 with epicentral distance is 24º and 70km depth and mb 6.1. Time windows of signal and background noise for both P and S waves are 40s. All spectra are estimated using the Multitaper method. It can be seen that S wave spectra is more decayed rapidly than P wave spectra. The differential

* attenuation (∆t SP) is quantified as the slope of natural logarithm of spectral ratio between P and S waves spectra in the range of frequency 0.25 to 1.00Hz

The effect of background noise into the slope of spectral ratio is also demonstrated: Figure 3.1 (a) spectral ratio estimation with noise and (b) without noise. Figure 3.1 (b) shows the noise spectra (plotted in cyan open circles) is well separated from signal spectra which in this case, the noise slightly changes the value of the slope of spectral ratio.

(a)

(b)

Figure 3.1: Implementation of the spectral ratio method to unfiltered seismic signals recorded at the TL13 station (a) with background noise and (b) without noise, background and the signal spectra is estimated using the Multitaper method, differential attenuation is estimated as logarithmic slope of S to P spectral ratio in frequency range of 0.25 to 1.00Hz.

32 3.4 Robust Attenuation Measurement of Australian Data

3.4.1 Seismic Data

Seismic events which are lied across Indonesia, Philippines, New Guinea,

New Caledonia, New Zealand, and spreading center to the south of Australia offers robust data with suitable distances between 5 and 40º and depth between 3 and 700km to be used for exploring seismic structure beneath the Australian continent and surrounding regions. Figure 3.2(a) shows seismic events location used in this study occurs from 1993 to 2005 with depth between 3km and 700km and magnitude (mb) between 5.0 and 7.0, when information of magnitude is not provided it is marked as

0.0. As can be seen from the figure, the majority of events are located to the north of and the east Australia, on the other hand from the south, although there is a few of events, it still provide valuable information. Thousands of seismic events from that regions has been archived since 1992 by research School of Earth Sciences,

Australian National University in a major program of deployment of portable broadband seismic stations across Australia in experiments primarily designed to improve knowledge of the 3-D structure of the region. The SKIPPY experiment (1993-1996) with a sequence of 6 progressive deployments across the continent with an inter- station spacing of around 400km, provided the first coverage of the whole of Australia and its surroundings. Subsequent experiments have been on a range of scales, including detailed studies of the Kimberley region, SE Australia and Tasmanian and broader scale coverage of Western Australia and the surroundings of the Tasman line marking the edge of Precambrian outcrop (Figure 3.2 (b)). Recorders used in Skippy and Kimba projects are Refraction Technology units with 24-bit Analogue-to-Digital

Conversion (ADC) and an internal Global Positioning System (GPS) corrected clock.

Data are continuously recorded using a Digital Audio Tape (DAT) drive. The DAT

33 drive was replaced by external disc drives for Kimba and latter projects. The sensors in the field are three-component broadband Güralp CMG-3ESP seismometers that are flat in velocity response from 30mHz to 30Hz. Four seismometers are flat to 0.016Hz

(62.5s period) (type 1) while the remaining eight are flat to 0.033Hz (30 s period)

(type 2). Data is digitized at 25 samples per second and with the flat velocity response of the sensors and 24-bit ADC ensures very high data quality. Data are extracted for selected earthquakes (based on simple magnitude-distance criteria) from the

Preliminary Determinations of Epicenters (PDE) catalogue, published by NEIC.

Figure 3.2(a) continued

(b)

Figure 3.2: (a) Seismic events distribution in the range of Mb 5.0-7.0 recorded at portable broadband stations across Australia. When magnitude is not provided, it is marked as Mb 0.0 (b) Portable broadband stations location used in this study.

* 3.4.2 Differential Attenuation (∆t SP) Representations

The wave ratio method that estimates spectral ratio between two body waves P and S from the same event recorded at the same receiver is implemented to nearly

4050 seismic data set recorded at seismic stations across the Australian continent. By using estimation procedure described in section 3.2 above, differential attenuation paths of the Australian continent is presented. It is founded that there is a correlation

* * between values of the ∆t SP and lithological feature. Figure 3.3 shows the ∆t SP

* estimation for TL01 station, it can be seen that the ∆t SP from eastern Australia

(Figure 3.3 (a) and (b)) which associated with oceanic region is higher than from western (Figure 3.3 (c) and (d)) which associated with cratonic region. All of these seismic paths are plotted in Figure 3.4.

* Figure 3.3: The differential attenuation (∆t SP) estimation for TL01 station, P wave spectra is estimated from vertical component (red) and S wave spectra from horizontal component (blue). Signal spectra is plotted as stars and its background is plotted as dots, magenta dots is spectra of P wave background and green dots is spectra of S wave background. (a) and (b) are paths from the east of Australia and (c) and (d) from the west.

Figure 3.4: Seismic paths from four events discussed in Figure 3.3 above recorded at TL01 station which is located around Tasman Line location.

* Figure 3.5: Summary ∆t SP estimated from 4032 dataset. The raypaths are color * scaled by the value of ∆t SP with line thickness inversely proportional to the inverse of * the errors in the ∆t SP estimation.

* The behavior of the differential attenuation (∆t SP) raypaths of the Australian continent can be seen in Figure 3.5 above. The raypaths are classified into three groups of epicentral distances i.e. ∆:5-20°, ∆:18-30° and ∆:28-40° and are color

* scaled by the value of ∆t SP with line thickness inversely proportional to the inverse of

* the errors in the ∆t SP estimation. The figure suggests somewhat distinctive contrast in

* ∆t SP between eastern and central Australia especially for epicentral distances ∆:5-20°, the contrast still can be seen for epicentral distances ∆:18-30°, meanwhile for epicentral distances ∆:28-40°, there is no significant contrast.

37 3.5 Seismic Attenuation Tomographic System of Equation and Model Parameterizations.

The seismic attenuation tomographic system of equation used in this study is described as follow:

γ .dl  Q V  ∆t * = ; γ = 1− S . S  (3.12) SP ∫   path QSVS  QP VP 

* Where ∆t SP is differential attenuation between S and P waves, dl is seismic raypath segment in each tomographic cell, QS is S wave attenuation, QP is P wave attenuation,

* VS is S wave speed and VP is P wave speed. For the calculation of ∆t SP it self, the information of P and S wave speed are required. Thus in this study, seismic body waves speed tomography for the Australian region are also conducted. The seismic wave speed tomography is conducted by following system of equation [Rawlinson &

Sambridge, 2003]:

T −1 −1 T −1 T −1 δm = []G Cd G + εCm +ηD D G Cd δd (3.13)

Where δm is a slowness perturbation matrix, G is a matrix kernel containing raypaths segments in each cell, Cd and Cm is a data covariance matrix and an a priori model covariance matrix respectively, ε is referred to as the damping factor, η is smoothing parameter, D is a second derivative operator and δd is a relative traveltimes matrix to a reference model, T and -1 is a notation for transpose and inverse matrix operation.

The three dimensional study area with latitude 22° N to 65S° and longitude

78° to 189° and 0-1240km depth is discretized into 11100 cells with the cell size is

3°x3° and depth range of 35 to 200km (see Table 3.1 for depth interval and seismic velocity used as initial model). It is assumed that each cell has a constant value of seismic speed and attenuation. A quasi three dimensional ray tracing which respect to

Snell’s law is carried out to obtain the kernel matrix. Figure 3.6 shows seismic path

38 coverage and paths density map at slices 35-120km, 120-220km, 220-320km, and

320-410km.

Table 3.1: Depth Interval and P and S waves speed used as initial model Layer Depth Interval (km) P-Wave Speed (km/s) S-Wave Speed (km/s) 1 0-35 6.1200 3.6383 2 35-120 8.0450 4.4899 3 120-220 8.1892 4.5112 4 220-320 8.5190 4.6266 5 320-410 8.8659 4.7918 6 410-535 9.5698 5.2124 7 535-660 9.9902 5.4776 8 660-860 11.0302 6.1655 9 860-1060 11.3910 6.3517 10 1060-1260 11.7070 6.4846

Figure 3.6(a) continued

(b)

Figure 3.6: (a) Seismic path coverage for the Australian continent and regions, 4032 raypaths are drawn in blue lines; seismic events are presented by red circle and seismic stations by red triangle. (b) Paths density distribution with cell size 3º x 3º at slices 35-120km, 120-220km, 220-320km, and 320-410km.

3.6 Checkerboard Test

Before producing tomographic images from the inversion of the Australian data set, the checkerboard test is carried out to evaluate the feasibility of raytracing method being developed, model parameterizations, and inversion method and to see how well wave-speed anomalies are recovered upon inversion. In this test, a set of synthetic model is divided into alternating regions of high and low velocity with a scale is categorized into two classes: coarse scale (12°x12°) and intermediate scale

(6°x6°) and varies in depth. Figure 3.7 below shows the checkerboard test result of S waves by inverting 4032 synthetic dataset using LSQR algorithm [Paige & Saunders,

1982]

Figure 3.8: Input and recovered checkerboard models at slices 35-120km, 120- 220km, 220-320km, and 320-410km obtained by inverting synthetic dataset (a) coarse scale (12°x12°) and (b) intermediate scale (6°x6°). Color scale indicates S wave velocity perturbation in percent relative to ak135.

41 Chapter 4

Australian Seismic Wave Speed, Attenuation and Anisotropy Models

Representations of tomographic images of earth physical properties such as seismic velocity and attenuation are important to have a better understanding in the interpretation of earth structure and temperature distributions. Previous studies on seismic tomography of Australian continent are mainly presented in term of seismic speed property which is successfully to delineate geological structure beneath the continent. Study on attenuation tomography is carried out to give another perspective in the interpretation of geological structure and temperature distribution in the

Australian continent and the region.

4.1 Seismic Wave Speeds Models

Australian seismic wave speeds models has been derived by inverting P and

S seismic wave traveltime residuals from the ak135 model from 4032 seismic dataset.

Figure 3.9 show horizontal slices at 35-120km, 120-220km, 220-320km, and 320-

410km of both P and S waves seismic speed across the Australian continent and its regions. Seismic images of P-wave speed at slices 35-120km and 120-220km depth show somewhat patchy pattern which may be associated with limitation of seismic sampling and low velocity effect near the earth surface. The images are color scaled by values of seismic wave speed perturbation relative to ak135 model. S wave speed has variation almost three times of P wave speed.

(a)

(b)

Figure 3.9: Horizontal slices through three-dimensional solution model obtained by inversion of seismic traveltimes residuals of Australian seismic data. (a) P-wave speed perturbation (b) S-wave speed perturbation. Zero hit of seismic path is presented as white blocks in the study area.

43 Due to limitation of seismic sampling from the south of Australia, such a

‘smearing’ effect clearly can bee seen in the region from -30°S to -40°S and 140°E to 145°E. This effect is shown in both the checkerboard test and P and S wave speed images.

Both Compressional and Shear wave velocity contrasts between eastern

Australia and central Australia (between 140ºE and 145ºE) is clearly pronounced at all slices except for P wave speed at 35-120km and 120-220km. This seismic contrast may be associated with the conventional Tasman line. The zones of exposed

Precambrian rocks in the west and in the middle of the continent are associated with high seismic wave speeds and the Phanerozoic rocks and the presence of recent volcanism and region of high heat flow in the east are associated with low seismic wave speed anomaly

4.2 Seismic Attenuation Models

Both P and S-waves speed information from the seismic wave speed tomography above are then utilized as data input for the three-dimensional seismic attenuation tomography. By using the same model parameterization as seismic wave

* speed inversion, the Australian differential attenuation (∆t SP) presented in 3.4.2 is inverted by the LSQR algorithm. In this inversion, it is assumed that QP=2.3QS. The result from the inversion is presented in Figure 10. The Figure shows the absolute value of Q-1 for both Radial (SV) and Transverse (SH) components.

The seismic attenuation (Q-1) of Australia and its regions varies from 0.0 to

6.5x 10-3. Overall it is suggested that eastern part of Australia has high attenuation and western and central Australia has low attenuation. Attenuation contrast at around

140-145 probably is associated with Tasman Line.

(a)

(b)

Figure 3.10: S wave attenuation images of (a) Radial component and (b) Transverse component at slices 35-120km, 120-220km, 220-320km, and 320-410km. Zero hit of seismic path is presented as white blocks in the study area.

45 The seismic Attenuation (Q-1) of the Australian continent derived from horizontal component is also presented in Figure 3.11. The values of Q-1 are color scaled to relative reference of 0.0035. The strong contrast in attenuation is clearly pronounced between 135°E and 145°E.

Figure 3.11: S-Wave seismic attenuation (Q-1) derived from horizontal component at at slices 35-120km, 120-220km, 220-320km, and 320-410km. Zero hit of seismic path is presented as white blocks in the study area. The values of Q-1 are color scaled to relative reference of 0.0035.

4.3 Seismic Attenuation Anisotropy Model

The availability of three-component seismic data allows the exploitation of attenuation anisotropy beneath the Australian Continent. The attenuation anisotropy

(ξ) of the Australian continent is derived by calculating ratio between seismic

-1 -1 attenuation derived from SV and SH component (ξ = Q SH/ Q SV).

Figure 3.12: Seismic attenuation anisotropy at slices 35-120km, 120-220km, 220- 320km, and 320-410km. Zero hit of seismic path is presented as white blocks in the study area.

Figure 3.12 above is the representation of seismic attenuation anisotropy al slices 35-120km, 120-220km, 220-320km, and 320-410km. The Figure suggests that in the region where seismic coverage is good, transverse component (SH) wave is less attenuated than radial component (SV). If it is assumed that seismic attention has an inverse correlation with seismic velocity, the attenuation anisotropy of Australian continent agrees with Debayle and Kennett work’s [2003] which revealed that in the

Australian continent SH waves is faster than SV waves.

47 Future Work

The ultimate goal in the future work is how to improve the image of seismic wave speed, seismic attenuation, and seismic anisotropy. To achieve this purpose there are some works that I have to do such as:

• Three-dimensional ray tracing using the Fast Marching Method FMM of

Sambridge et.al.

• Traveltime picking and Attenuation measurement for seismic data with

epicentral distance greater than 40°.

• Model parameterizations refinements i.e. using node model rather than block

model.

• Inversion using multi-stage method rather than ak135 model as initial model.

48 References

Betts P. G., D. Giles, G.S. Lister and L. R. Frick. 2002. Evolution of the Australian lithosphere, Australian Journal of Earth Sciences, 49, 661–695

Cheng, H.-X. & Kennett, B. L. N. (2002) Frequency dependence of seismic wave attenuation in the upper mantle beneath the Australian region. Geophysical Journal International 150 (1), 45-57. doi: 10.1046/ j.1365-246X.2002.01677.x

Cheng, H.-X., 2000, Seismic Body Wave Attenuation in the Upper Mantle beneath the Australian Continent, Ph.D. thesis, Research School of Earth Sciences, The Australian National University.

Coney P. J., Edwards A., Hine R., Morrison F. & Windrum D. 1990. The regional tectonics of the Tasman Orogenic system, eastern Australia. Journal of Structural Geology 125, 19–43.

Dawson, G.C., Krapez, B., Fletcher, I.R., McNaughton, N.S., Rasmussen, B., 2002. Did late Paleoproterozoic assembly of proto- Australia involve collision between the Pilbara,Yilgarn and Gawler Cratons? Geochronological evidence from the Mount Barren Group in the Albany-Fraser Orogen ofWestern Australia. Precambrian Res. 118, 195–220.

Direen N.G., Allan G. Cadd, Patrick Lyons , Jonathan P. Teasdale, Architecture of Proterozoic shear zones in the Christie Domain, western Gawler Craton, Australia: Geophysical appraisal of a poorly exposed orogenic terrane, Precambrian Research 142, 28–44

Drummond B. J. Goleby B. R. & Swanger C. P. 2000. Crustal signature of late Archaean tectonic episodes in the Yilgarn Craton, Western Australia: evidence from deep seismic sounding, Tectonophysics, 329, 193–221.

Fishwick S. 2005. The Australasian Upper Mantle–Insights from Surface Wave Tomography, PhD Thesis, Research School of Earth Sciences, ANU.

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

Gaina C., Müller, R.D., Royer, J.Y., Stock, J.M., Hardebeck, J., & Symonds, P., 1998. The tectonic history of the Tasman Sea: a puzzle with thrirteen pieces, J.Geophys. Res., 103, 12413-12433.

Gaina, C. and Müller, R.D., 1999. Evolution of the Louisiade triple junction, J. Geophys. Res., 104, 12927-12939.

49 Gorbatov A. & Kennett B.L.N. 2001. P wave tomographic imaging of the southeast Pacific, Annual Report of Research School of Earth Sciences, ANU, for 2000, 140- 141. Gray & Foster, 2004, Tectonic evolution of the Lachlan Orogen, southeast Australia: historical review, data synthesis and modern perspectives, Australian Journal of Earth Sciences,51, 773–817

Gray D. R. & Foster D. A. 1998. Character and kinematics of faults within the turbidite-dominated Lachlan Orogen: implications for tectonic evolution of eastern Australia. Journal of Structural Geology 20, 1691–1720.

Gudmundson, O., Kennett, B.L.N. and Goody, A., 1994, Broadband observations of upper mantle seismic phases in northern Australia and the attenuation structure in the upper mantle, Phys. Earth Planet. Inter., 84, 207-226.

Gurnis M. and Müller, R.D.,2003. Origin of the Australian Antarctic Discordance from an ancient slab and mantle wedge. Geol. Soc. Australia Spec. Publ. 22 and Geol. Soc. America Spec. Pap. 372, 417-429.

Hickman A. H. 1983. Geology of the Pilbara Block and its environs. Geological Survey of Western Australia Bulletin 127.

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

Myers J.S. Shaw R.D. & Tyler I. M. 1996. Tectonic evolution of Proterozoic Australia. Tectonics 15, 1431–1446.

Oliver N.H.S. & Cawood P. A. 2001. Early tectonic dewatering and brecciation on the overturned sequence at Marble Bar, Pilbara Craton, Western Australia: dome-related or not? Precambrian Research 105, 1–15.

Paige, C. C. and M. A. Saunders, "LSQR: An Algorithm for Sparse Linear Equations And Sparse Least Squares," ACM Trans. Math. Soft., Vol.8, 1982, pp. 43-71.

Reading A. M. & B.L.N. Kennett, 2003,Lithospheric structure of the Pilbara Craton, Capricorn Orogen and northern Yilgarn Craton, Western Australia, from teleseismic receiver functions, Australian Journal of Earth Sciences, 50, 439–445 van der Hilst R. D., Kennett B.L.N., Criestie D. & Grant J. 1994. Project SKIPPY explores the mantle and lithosphere beneath Auatralia. EOS 75, 177-181.

Weissel, J.K. & Hayes, D.E., 1974. The Australian Antarctic Discordance: new results and implications, J. Geophys. Res., 79, 2579-2587.

Widiyantoro S. & van der Hilst R. 1996. Structure and evolution of lithospheric slab beneath the Sunda arc, Indonesia, Science, 271, 1566-1570.

50 Yoshizawa K. & Kennett B.L.N. 2004. Multimode surface wave tomography for the Australian region using a three-stage approach incorporating finite frequency effects. Journal of Geophys. Res.,109, B02310, doi:10.1029/2002JB002254.

Zielhuis A. and van der Hilst R. 1996. Upper mantle shear velocity beneath eastern Australia from inversion of waveforms from Skippy portable arrays, Geophys. J. Int.,127, 1-16.