Issuu

Edited by Barry Drummond

THE AUSTRALIAN LITHOSPHERE

Editor

BARRY J. DRUMMOND

Special Publication No. 17 GEOLOGICAL SOCIETY OF AUSTRALIA INCORPORATED

July 1991

1991 THE GEOLOGICAL SOCIETY OF AUSTRALIA INCORPORATED This volume is published in the Special Publications series of the Geological Society of Australia REFERENCES General Reference: DRUMMOND, BARRY (Ed), 1991: The Australian Lithosphere. Geological Society of Australia, Special Publication, 17, 208p. Two forms of reference to specific papers are possible, as follows: GREEN, D.H., 1991: The Earth's lithosphere and asthenosphere - concepts and constraints derived from petrology and high pressure experiments. Geological Society of Australia, Special Publication, 17, 1-22. GREEN, D.H., 1991: The Earth's lithosphere and asthenosphere - concepts and constraints derived from petrology and high pressure experiments. In Drummond, Barry (Ed) The Australian Lithosphere. Geological Society of Australia, Special Publication, 17, 1-22.

Typeset by Metagraphics, Pty Ltd, Brisbane; printed and bound by Watson Ferguson & Co, Brisbane.

Registered in Australia for transmission by post as a book.

Orders for Special Publication No. 17 should be directed to: The Business Manager, Geological Society of Australia, 606 ANA House, 301 George Street, Sydney, NSW, 2000, AUSTRALIA. This book is copyright. Apart from any fair dealing for the purpose of private study, research, criticism or review, as permitted under the Copyright Act, no part may be reproduced by any process without written permission. Inquiries should be made to the publisher. © Geological Society of Australia Incorporated, 1991 National Library of Australia Cataloguing in Publication data The Australian lithosphere. Bibliography. ISBN 0 909869 78 2. ISSN 0072-1085 1. Earth - Crust. 2. Geology - Australia. I. Drummond, B J . (Barry John), 1950-. II. Geological Society of Australia. (Series : Special publication (Geological Society of Australia) ; no. 17). 551.14

Ill

Preface Barry Drummond Bureau of Mineral Resources, Geology and Geophysics, Canberra, ACT., Australia Australia is a continent largely remote from the immediate effects of plate tectonics. With the exception of its offshore northern margin, it does not have active convergent margins along its edges. Within its interior, it contains no active rift systems. Australia is therefore an excellent place to study subcontinental lithosphere that is not being actively modified by a dynamic upper mantle.

from the crust to the mantle. This transition is very important, because it has associated with it one of the largest density contrasts in the Earth (the only larger global density contrast between two layers in the Earth occurs at the core/mantle boundary). As a large density contrast, the crust/mantle boundary exerts considerable influence on many processes, from crustal dynamics to the ascent of magmas.

Parts of the Australian continent are very ancient; its Archaean cratons have yielded the oldest isotopic ages found so far. Large tracts of the continent have Proterozoic crust, and the eastern part, the Tasmanides, are Palaeozoic and Mesozoic. Thus the Australian continent contains the secrets of the evolution of the lithosphere throughout time.

Some of the papers present new data; others present reinterpretations of old data that were undertaken with a view to shedding new light on the definition and nature of the Australian lithosphere. Yet other papers present reviews, short or long, of different aspects of the lithosphere. All papers give valuable lists of references on the Australian lithosphere.

This volume contains a number of papers dealing with the Australian lithosphere. They address the lithosphere from the viewpoints of many disciplines, and present a summary of our knowledge on the subject. The papers are not by any means a comprehensive study by all Australian workers who deal with the lower crust and upper mantle; they do however present many of the key ideas current about half way through the life of the International Lithosphere Project (ILP), an internationally coordinated program whose goal is to focus the attention of many of the world's geoscientists on a single topic - the lithosphere - bringing to bear ideas from as many disciplines as possible. It is worthy of note that even after several years of the ILP, and as noted or intimated in several of the papers, we still do not have a clear definition of the lithosphere, let alone its real influence on plate tectonic processes, mountain building and sedimentary basin formation.

The first paper, by Prof. David Green, was presented as the 'Mawson Lecture' of the Australian Academy of Science in 1982. Its theme was further developed in a paper presented at the 7th Australian Geological Convention in 1984 and in later studies by Prof. Green. It deals with the chemical evolution of the lithosphere. Different lithospheric thicknesses are achieved when different geothermal gradients intersect the solidus, producing partial melting. Different types of magmatism originating in the upper mantle are modelled as different degrees of partial melting, in the presence or absence of volatiles, at different places in pressure-temperature space. Thus the chemical model of the lithosphere and its melts places contraints on thermal history models of the Earth. For example, some Archaean magmas were generated at 200-250°C above the temperatures found at comparable depths in the modem Earth, but other areas of protocontinent had already cooled to modern shield-type pressure/temperature regimes with lithosphere thicknesses of 150 km. Some of these Archaean lithospheric blocks maintained these characteristics at least to the Mesozoic, and probably to the present.

The papers in this volume fall into two broad categories: those that try to define the lithosphere and its transition to the underlying asthensphere these papers deal mainly with the subcrustal lithosphere, and another group that look at the transition

Muirhead & Drummond in their paper used seismic data to try to define the structure of the lithosphere and asthenosphere under present-day Australia. By correlating the base of the lithosphere with a seismic low-velocity zone in the upper mantle, they reached three key conclusions. Firstly, the lithosphere under the shield regions of central Australia is much thicker (?200-250km) than under eastern Australia (110km). Secondly, the base of the lithosphere is not a zone at which the Australian plate decouples from the upper mantle in the plate tectonic process. Rather, they preferred to distinguish between chemical, thermal and mechanical boundary layers, after Jordan, with the zone of decoupling most likely to be defined by the density contrast at the base of the zone of major element depletion. Their third conclusion was based on the ratio of P-wave to S-wave velocities in eastern Australia, and suggests that the upper mantle in the region must be near or above the solidus below about 110km depth. Unfortunately, the S-wave velocity structure of the Australian crust and upper mantle is poorly known, with many of the existing models based on surface wave studies which provide only non- unique estimates of velocities averaged over large travel paths. Thus Denham in his paper was able to show that the S-wave Moho occurs at different depths in different places in Australia, but he was unable to map the Moho undulations in any detail. He did indicate, however, that unlike some P-wave models, S-wave models of the crust usually do not have low velocity zones. Dooley suggests that density contrasts between crustal blocks extends to depths of about 110 km in eastern Australia, and deeper in the shield regions. He bases his arguments on isostatic considerations. Isostasy is nearly complete at the base of the crust in most provinces, but density contrasts are needed in the mantle to achieve complete isostasy. He proposes that during the continent-building process, protocrust forms across a weak layer, with isostatic compensation complete at the base of the crust. This becomes frozen in. Subsequent crustal evolution (orogeny, extension, underplating) upsets the balance, and the depth of compensation moves deeper.

The paper by Jackson presents a valuable summary of the theory of seismic energy propagation in cracked (fluid filled) and uncracked media, and makes comparisons with velocities measured in samples in the laboratory and in situ during seismic experiments. It therefore provides an important bridge between seismic models of the crust and upper mantle and their likely petrological significance. The effects of fluid filled cracks on the ratio of P-wave to S-wave velocities means that if the boundary between the lithosphere and asthenosphere is associated with the onset of partial melting, it will be mappable, provided that seismic models with sufficient precision can be erected. The electrical properties of the lithosphere are dealt with in two papers. Lilley provides electrical models for central and southeast Australia which suggest that the lithosphere is thicker under the shield region than in eastern Australia. Constable, in his review of all electrical studies of Australia, shows that the datasets from central and southeast Australia can be (statistically) fitted by the same model; however, the coincidence of Lilley's conclusions with those reached from seismic studies lends powerful support for different electrical thickensses of the lithosphere in central and eastern Australia. The seismic and electrical models of the lithosphere in the shield and non-shield areas of Australia all point to the need to study the thermal state of the lithosphere. Two papers do this. Cull produces geothermal gradients by constructing reasonable geological columns for the crust and upper mantle to which he allocates reasonable physical properties, namely heat production and thermal conductivity, and then combines the physical model with measurements of surface heat flux to estimate temperature as a function of depth. He also places reasonable bounds on the temperature estimates. This is a different approach from the traditional method, in which the distribution of heat producing elements is usually assumed without any a priori geological controls being placed on the models. Cull is able to identify different thickness lithospheres in the Precambrian areas and eastern Australia on the basis of homologous temperature (actual temperature/melting temperature). In areas where the heat flow is <40 mW m"1, the brittle zone must extend to depths of 80± 30 km, whereas in

eastern Australia where heat flow is much higher, the brittle zone may be as shallow as 45 ± 15 km. Thus he also makes a link between partial melting, the strength of the rocks, and the base of the lithosphere. Paterson looks at our knowledge of the rheology of rocks based on laboratory measurements, and urges caution when extrapolating that knowledge to studies in tectonophysics. The nature of the crust/mantle boundary, its coincidence or otherwise with the seismic Moho, and its evolution through time are addressed in a number of papers. O'Reilly & Griffin present a summary of nodules found throughout eastern Australia, and summarise their petrological model of the lower crust and upper mantle. On the basis of the equilibration pressures and temperatures of their samples, they also construct a geothermal gradient for southeast Australia that is consistent with nodules from other provinces in eastern Australia. Rudnick & Taylor studied nodules from Proterozoic and Palaeozoic provinces in northern Australia. They conclude that the lower crust is chemically heterogeneous, with, in northern Australia at least, recent additions of mafic rocks by underthrusting and subsequent metamorphism of supracrustal rocks, basaltic underplating and/or intracrustal melting, all operating at different times since the Proterozoic. McDonough, Rudnick & McCulloch looked at the isotopic signatures of xenoliths from throughout eastern Australia. The lower crust is predominantly mafic. Their studies of the lherzolite and harzburgite xenoliths point to a multistage evolution of the upper mantle, which is now composed of refractory peridotite (depleted in major elements by melt extraction) enriched in incompatible elements. The more refractory rocks are the most enriched in incompatibles. The isotopic sig-

natures of basalts indicate significant isotopic heterogeneity in the lithospheric mantle throughout eastern Australia. The petrological models of the lower crust, and particularly the crust/mantle boundary, all indicate a finely layered lower crust and a transitional rather than sharp crust/mantle boundary. This is consistent with seismic refraction and reflection models from Australia and overseas, summarised by Collins. Each paper in this volume is a work in its own right, but the reader is urged to address each point of interest by looking at what all papers have to say. On some points there is no consensus; the reader will have to make his or her own judgement based on the data and references supplied. Thus this volume presents results from many areas of endeavour. It also indicates where more work needs to be done if we are to increase our knowledge of the lithosphere. We need better seismic models. The data presently available have been interpreted to the limit, and new and more detailed seismic experiments are needed. S-wave data are very important. We need to know how relevant our models based on transient stresses (seismic waves) are to studies of the deformation of the lithosphere, where the stresses act for tens of millions of years. More information of the thermal state of the lithosphere is required, and we need to know how temperature influences lithospheric and asthenospheric rheology. We need more detailed petrological models, because in the end, physical models are only a partial representation of the rocks in the Earth. Barry Drummond Bureau of Mineral Resources Geology and Geophysics GPO Box 378 Canberra ACT 2601 AUSTRALIA

CONTENTS

PREFACE

Page iii

D.H. GREEN: The Earth's Lithosphere and Asthenosphere - Concepts and Constraints Derived from Petrology and High Pressure Experiments

K.J. MUIRHEAD & B J. DRUMMOND: The base of the Lithosphere Under Australia

J.C. DOOLEY: Velocity Variations and Isostatic Compensation in the Australian Region

David DENHAM: Shear Wave Crustal Models for the Australian Continent

C.D.N. COLLINS: The Nature of the Crust-Mantle Boundary Under Australia from Seismic Evidence

Ian JACKSON: The Petrophysical Basis for the Interpretation of Seismological Models for the Continental Lithosphere

M.S. PATERSON: Extrapolation of Laboratory Rheological Behaviour in Tectonophysics

115

P. WELLMAN, A.S. MURRAY & M.W. McMULLAN: Australian Long-Wavelength Magnetic Anomalies

117

S.C. CONSTABLE: Electrical Studies of the Australian Lithosphere

121

F.E.M. LILLEY: The State of Knowledge of the Asthenosphere from Electromagnetic Soundings: Australia and Surrounding Oceans

141

J.P. CULL: Geothermal Gradients in Australia

147

Suzanne Y. O'REILLY & W.L. GRIFFIN: Petrologic Constraints on Geophysical Modelling, Southeastern Australia

157

W.F. MCDONOUGH, R.L. RUDNICK & M.T. MCCULLOCH: The Chemical and Isotopic Composition of the Lower Eastern Australian Lithosphere: A Review

163

Roberta L. RUDNICK & S.R. TAYLOR: Petrology and Geochemistry of Lower Crustal Xenoliths from Northern Queensland and Inferences on Lower Crustal Composition

189

Geol. Soc. Aust. Spec. Publ. 17, 1-22.

The Earth's Lithosphere and Asthenosphere - Concepts and Constraints Derived from Petrology and High Pressure Experiments D.H. Green Geology Department, University of Tasmania, Hobart, Tasmania, Australia. Magmatism in the modern earth can be understood within a thermal regime bounded on the high temperature side by a 'peridotite adiabatic path' passing through 20 kb, 1450°C and a lower bound given by the low temperature limb of the kimberlite xenolith geotherm illustrated by Siberian & South African kimberlite pipes cutting Archaean cratons. Within this P,T envelope, different geothermal gradients intersect the solidus at different depths (i.e. giving different lithospheric thickness) and produce characteristic magmatic suites, e.g. mid-oceanic ridge basalts, intra-plate tholeiite to olivine melilitite suite, kimberlite-olivine lamproite suite. In the Archaean period of Earth history (>2.5b.yrs), the higher temperature boundary for magma genesis was at 200-250°C above that of the modern earth. At the same time, some areas of proto-continent or shield had already cooled to modern shield-type P,T. regimes with lithospheric thickness of >150 km. Some of these Archaean lithospheric blocks maintained these characteristics at least to the Mesozoic and probably to the present and models of earth evolution must accommodate such thermal insulation and extreme contrast to the thermal regimes of intraplate and plate margin volcanism. A model of Archaen peridotitic komatiite pedogenesis is presented which attempts to reconcile these apparently paradoxical observations within the framework of accelerated heat loss from a higher temperature Archaean mantle and small but thick and very mobile proto-continents maintained over the down-flow regions of the Archaean mantle convection patterns.

INTRODUCTION Investigation of the Earth's lithosphere and asthenosphere requires a multidisciplinary approach in which methods of geophysics, geochemistry and geology are applied to regions of the earth which extend to depths of at least 200 km. The approach in this paper uses the data and methods of experimental petrology to constrain models of the upper mantle. An underlying principle throughout the paper is that the solidus or beginning of melting of crustal or mantle rock types marks a very significant and relatively discontinuous change in physical properties, and in particular to properties controlling the deformational and seismic response, i.e. those properties by which the 'lithosphere' and 'asthenosphere' are defined. This change is sufficient to designate subsolidus conditions as characteristic of 'lithosphere' and partially molten conditions as characteristic of 'asthenosphere' (Green and Liebermann, 1976; Wyllie, 1987). This statement places little constraint on the chemical differences, if any, between lithosphere and asthenosphere, and must be qualified for particular aspects of the earth's behaviour - for example, in relative motion between

overlying lithospheric plates and underlying asthenosphere, the uppermost part of the asthenosphere may well be entrained and attached to the overlying plate, the real stationary or return flow region of the mantle lying deeper in the asthenosphere (Green, 1971). Because volatiles, particularly H2O and CO2, play a major role in promoting melting of silicates and in controlling the extent of melting, the paper pays particular attention to the roles of volatile species in the deep earth. It will be argued that the upper mantle (asthenosphere, Transition Zone) is relatively reduced and actively degassing CH4-rich fluids and that redox reactions between oxidized crust and lithosphere rock types and underlying reduced mantle play important roles in stabilizing the lithosphere/asthenosphere boundary in the modern earth. In the final sections of the paper, attention is turned to the past geological record and to the P,T signatures retained in igneous and metamorphic rocks of ancient shield regions. The consequences of higher mantle temperatures in the Archaean and the possibility of lower f02 throughout the Archaean crust and mantle are explored.

The Earth's Lithosphere and Asthenosphere.

REGIONS OF PARTIAL MELTING IN THE DEEP EARTH Crustal Melting Regimes: Without attempting to closely define either the very large range of sedimentary, igneous and metamorphic rock compositions present in the continental crust, or its mean composition, it is nevertheless a valid generalization that melting relations of crustal rocks are dominated by the ubiquitous presence of quartz, plagioclase , potassium feldspar and ferromagnesian phases including biotite , amphibole s (hornblende) and pyroxeness. Melting is typically partial melting in which particular elements are strongly partitioned into the liquid phase, and with movement of the liquid or liquid and crystals as intrusive magmas to shallow depths, the deeper levels of continental crust may become anhydrous, depleted in incompatible elements, and relatively refractory (Wyllie 1977.). ss

In fig. 1, the P,T melting regime for crustal rocks is extended to -60km, i.e. regions of very thick continental crust in continent/continent collision zones. Four distinct solidi are shown (1 to 4) bounded by the anhydrous, fluid-absent solidus and the water-saturated ( X H 0 = 1 ) solidus. The dehydration solidus (2) is defined by melting, in the absence of a fluid phase, of biotite+plagioclase+Kfeldspar+quartz assemblages and will vary over a considerable temperature range depending on the phase compositions. Similarly the solidus (3) for melting with a free fluid phase in which X H O=0.5, i.e. fluid contains CO2 or CH4 or other volatile species, also illustrates the principle rather than being either unique or well- determined. Mantle Melting Regimes: Although, like the lower crust, the Earth's upper mantle is petrographically inhomogeneous, including eclogites, pyroxenites and dunites as minor rock types, the dominant rock type is peridotite, specifically lherzolite with ol>opx>cpx>plagioclase, garnet, spinel or amphibole. Accessory or minor minerals include phlogopite, ilmenite, sulphides and diamond or graphite. With the depth to the V

*Mg #

S Mg+Fe " 1 Q Q M

4-1

M-discontinuity varying from <10km (oceanic crust) to >50km (orogenic regions, continental crust) the phase relationships and melting characteristics of lherzolite are of importance over the pressure range from 2kb to at least lOOkb (top of the Earth's Transition Zone). In fig. 1, solidi for lherzolite are illustrated by the detailed studies on 'Hawaiian pyrolite' (Green, 1973a). This composition has Mg =89*, CaO = 3.1%, AI2O3 = 3.5% and is relatively enriched in Ti02 (0.7%) Na 0 (0.57%) and K 0 (0.13%) in comparison with most natural spinel or garnet lherzolites derived as xenoliths during magma transit through the lithosphere. However, in studies of these xenolith suites and in models ("pyrolite" models) which derive source compositions from mixing of primary magma and residue compositions, the 'primitive' or 'parental' mantle composition is consistently estimated to be lherzolite with olivine -60%, enstatite -23%, diopside -15%, spinel -2%, i.e. lherzolite with Mg -89-90, Si02 -44-45%, AI2O3 -3.5-4%, CaO -3-4%, Na 0 -0.2-0.5% (Green & Ringwood, 1963; Ringwood, 1966; Green et al. 1979; Nickel & Green, 1984; Kurat etal 1980). #

The anhydrous solidus (1) and water-saturated solidus (4) are well determined only to 35kb. Studies of the peridotite solidus to 140kb (Takahashi, 1986) suggest narrowing of the melting interval to approach eutectic melting at -120kb but there is a need to thoroughly evaluate P,T gradients in the apparatus concerned - the major focus in this paper is on the 0-60kb pressure interval. The 'dehydration solidus (2)' is the solidus for pargasite-bearing lherzolite and owes its distinctive shape to the stability of pargasite up to the solidus (1100-1150°C) at pressures <28kb and the higher pressure instability of pargasite at subsolidus conditions, being replaced by olivine+garnet+pyroxenes +phlogopite+ilmenite±melt. Thus, between 28kb and 30kb the solidus of pargasitebearing lherzolite drops by -150°C (Green, 1973a). The distinctive form of the "dehydration solidus" for Hawaiian pyrolite applies for 0.03<H20<0.4% i.e. H20-contents greater than held in phlogopite

Geol. Soc. Aust. Spec. Publ. 17,1 -22. TEMPERATURE

(°C)

Fig. 1 P-T grid illustrating melting conditions for crustal rocks (lightly patterned area) including thickened continental crust to ~60km, and for mantle lherzolite (heavily patterned area). For each generalized source, four solidi are shown: 1-anhydrous; 2- 'dehydration' solidus (see text); 3-fluid-present solidus with X H O ~ 0 . 5 ; and 4-fluid-present solidus with X H 0~1. Model geotherms for mature (100 m.yr) oceanic crust (P-Oioo) overlapping convecting mantle (P-J) and for shield regions (C-R) are shown. The intersection of a geotherm with a lherzolite solidus, for an appropriate anhydrous or hydrated mantle, is considered to mark the base of the lithosphere and top of the asthenosphere. V

The Earth's Lithosphere and Asthenosphere.

alone and up to the maximum which can be accommodated in the pargasite. A similar study for K-enriched spinel lherzolite (phlogopite pargasite lherzolite) composition (Mengel and Green, 1987) confirms the stability of pargasitic amphibole to >1100°C and to 27- 29kb and the distinctive solidus shape. It is also important to note that this solidus applies for f 0 2 conditions close to MW. This conclusion derives from the reduced character of the starting composition and the low f 0 2 (graphitewater approximately) of the furnace assembly of the piston cylinder apparatus used. The solidus (3) for a mixed H2O-CO2 fluid in which X V h 2 o~0.5~X V co 2 is schematic (Wyllie, 1979, 1987). It is sensitively dependent on fC>2 as this will determine whether CO2 or CH4 is the second major component of the fluid and at pressures above the carbonation reactions in peridotite, CH4 will lead to increase in solidus temperatures whereas CO2 will maintain low solidus temperatures (Olaffson & Eggler, 1983; Brey etaL, 1983; Wyllie, 1987; Taylor & Green, 1987; Green et al 1987). Solidi for mixed H2O-CO2 fluids have been inferred by Wyllie (1979, 1987) and have a pronounced cusp or maximum on the solidus at ~25kb which results in part from amphibole breakdown at T<1000°C at 10-15kb. This will occur in the Ca0-Mg0-Si02-C02-H20 system but the pargasitic amphibole stable in lherzolite compositions is stable to T> 1100°C in this pressure range (Green, 1973a; Mengel & Green, 1989). Nevertheless because of the relationships between amphibole stability and both rock composition and fluid compositions and the appearance of stable carbonates (dolomite and magnesite) at higher pressures, the

solidus for pyrolite* with X V h 2 O~0.5~X V co 2 is n o t well constrained. In the following discussion, solidi 1,2 and 4 are emphasized as defining melting for dry conditions, melting of the mantle containing accessory or minor pargasitic amphibole and melting in the mantle in the presence of water-rich fluid.

TEMPERATURE DISTRIBUTION IN THE LITHOSPHERE AND ASTHENOSPHERE: A variety of methods are used to estimate temperature distribution in the deep earth. The methods rely on models of heat production within the earth, knowledge of heat transfer by conductive, radiative and convective means and measurements of heat flow in different regions of the earth. Early models based on fixist models of continent/ocean relationships have been replaced by models in which the convective heat loss associated with sea-floor spreading and plate tectonics hypotheses have been incorporated (Clark and Ringwood, 1964; Parker and Oldenberg, 1973; Peltier and Jarvis, 1982). In fig. 1, the low geothermal gradients beneath stable shield regions of the continental crust are contrasted with the steeper geothermal gradients beneath oceanic and some young continental regions. Mid-ocean ridges are loci of adiabatic upwelling from the deeper mantle and the thermal perturbation of the lithosphere/asthenosphere boundaiy extends to crust -lOOm.yr old on either side of the ridge crest. The lOOm.yr crust geotherm (Parker & Oldenberg, 1973) therefore represents a mature or representative oceanic

Increasing sophistication in experimental study of peridotite melting has emphasised the importance of mantle oxidation state in determining the conditions for melting and the nature of melting products in peridotite -C-H-O. Most recent work on the calibration of mantle fC>2 sensors (Ballhaus, Berry & Green 1990) has established f02 conditions from near IW+1 log unit (i.e. H2O+CH4 fluids, graphite stable) to IW+3,4 log units (i.e. H2O+CO2 fluids, graphite unstable) for various spinel lherzolite xenoliths and primitive magmas. Experimental studies under oxidized conditions (i.e. peridotite +H2O+CO2) have established conditions at P>22kb where sodic dolomitic carbonatite melt occurs at temperatures below the peridotite+IfeO solidus (Wallace & Green 1988, Falloon & Green 1989,1990; Green 1990). These studies have importance for more oxidized regions of the lithosphere and asthenosphere, particularly in convergent margin regimes (where excess oxygen is inferred to be carried into the mantle) and in sub-continental lithosphere. In the context of the discussion in this paper of mantle evolution from Archaean to present, the possibility of progressive oxidation of the upper mantle, or parts of it, through geological time must be considered together with concepts of change of ambient surface oxidation conditions and the role of subduction in cycling excess oxygen and oxidized carbon into the upper mantle. Mantle melting, mantle metasomatism and magma genesis have taken place in the remarkably sensitive and complex peridotite-C-H-0 system.

Geol. Soc. Aust. Spec. Publ. 17,1-22.

lithosphere geotherm. This geotherm intersects the water-saturated lherzolite solidus at about 70km and the dehydration solidus at about 90km. Because of the shape of the dehydration solidus, cooling of the geotherm by 100-150°C at 90-100km, would only marginally increase the depth of intersection of solidus and geotherm. Intersection of the peridotite solidus is interpreted to mark the boundary between lithosphere (subsolidus) and asthenosphere (above-solidus). Thus the combination of peridotite phase petrology and estimation of temperature distribution in the earth predicts the thickness for the oceanic lithosphere of about 90km. This conclusion applies if the upper mantle contains real but minor (<~0.4wt%) water contents and contrasts with the conclusion that the oceanic geothermal gradient illustrated would not intersect the solidus at any depth less than -200km if the mantle is anhydrous. Referring to fig. 1, the estimated geothermal gradient for shield regions does not intersect the solidus at any depth, even for water- saturated conditions, i.e. the lithosphere extends to depths >170km. The geotherms illustrated are conductive geotherms representing heat transfer through the lithosphere within plate locations beneath oceans and continental shields respectively. However, in the plate tectonics model of the earth, lithospheric plates are in motion over the asthenosphere. Plate margins are of constructive (sea-floor spreading), convergent (subduction) or conservative (transform) type and plate motions may be seen as responsive to and determined by deeper convective motions in the Asthenosphere and Transition Zone of the mantle. The temperature distribution within this convective layer will vary from 'hot' regions of adiabatic upwelling to cool regions of downflow, presumably represented by subduction of lithospheric slabs to depths in excess of 700km. In fig. 1, the curve P-J is an estimate of the temperature distribution in the adiabatic upwelling regions (Peltier and Jarvis, 1982). This curve lies within the region of small degrees of melting (~l-2%) for pyrolite mantle containing very small water contents (Green & Liebermann 1976). Upwelling, partially molten, mantle diapirs fed from such a convecting asthenosphere would rise with very little temperature drop until the anhydrous solidus (1)

is crossed, i.e. at P=25-30kb, T=1450°C. At this depth the degree of partial melting would increase rapidly as diapirs moved to shallower depths and temperature would drop more rapidly due to latent heat of melting and heat transfer by convective movement of magma batches (Mid Ocean Ridge picrites and basalts) to the surface. Petrological Evidence for Mantle P,T Regimes from Modern Magmatism Experimental petrology contributes to our knowledge of mantle characteristics by defining the conditions for segregation of mantle-derived magmas from their source rocks. In fig. 2, a summary diagram from such petrogenetic studies is presented. Magmatic processes in the modern earth can be encompassed within an envelope (fig. 3) defined by an adiabat passing through 20kb, ~1450°C. The upper boundary reflects magma genesis at mid-ocean ridges with primary tholeiitic picrites separating at 17-20kb, 1400-1450°C by 20-25% melting of upwelling mantle lherzolite. Magmas reaching the surface are commonly olivine tholeiites with eruption temperatures <1270°C reflecting cooling, crystallization and loss of olivine (+minor chromian spinel) from primary picrites [Green et al., 1979; Jaques & Green, 1980; Stolper, 1980; Elthon and Scarfe, 1984; Green et al., 1987; Falloon & Green, 1987 (a,b)]. Other mantle-derived magmas are formed at lower temperatures within this envelope and their conditions of genesis reflect the important roles for volatiles and geothermal gradients in determining specific magma characteristics. From studies of Hawaiian volcanism, primary olivine tholeiite magmas, containing <0.5% H2O, are generated at ~40-50km by -25-30% melting of incompatible element enriched lherzolite source leaving residual harzburgite (Green and Ringwood, 1967). Temperatures of magma genesis are ~1350°C. Hawaiian alkali olivine basalts represent lower degrees of melting and slightly deeper sources. The more silica- undersaturated olivine-basanites, olivine nephelinites and olivine melilitites form by small degrees (2-7%) melting of garnet lherzolite under the influence of enhanced C-H-O volatiles (Green, 1971,1973b). In particular, the presence of dissolved (C03) = is essential in generating Ca-rich,

The Earth's Lithosphere and Asthenosphere.

MANTLE CRUSTAL

MELTS

PHANEROZOIC VOLCANISM

MELTS

TEMPERATURE (°C) 800 1200

1600 \MORB PERIDOTITIC KOMA TIITES

BONINITESV

TROODOS

U.P.L.

- 35km HA WAIIAN OL. THOL. ALK. OL.

BASALT

\ MID-OCEAN \RIDGE 3 PICRITE S

LU CC D

OL.

BASANITE

oc Q.

OL.

NEPHELINITE

OL.

MELILITITE

CO CO LU

~100km

KIMBERLITE

Fig. 2 The P,T conditions deduced for magma segregation in the mantle of the modern Earth. Solidi and geothermal gradients are from fig.l. The horizontally shaded areas illustrate examples of P-T conditions for crystallization of mantle xenolith suites from the localities shown (taken from Nickel & Green 1985). The conditions of magma segregation for a variety of modern primary basaltic magamas are illustrated (diagonally shaded areas). These are confined within an envelope defined by adiabatic upwelling ('olivine adiabat') through 1450-1500°C, -18kb, i.e. midocean ridge picrite segregation, and a lower temperature boundary defined by the South African shield geotherm deduced from garnet peridotite xenoliths from diamond bearing kimberlites. E1-E2 and VI-V2 show isobaric cooling paths deduced for deep continental crust (3.1 b.yrs to 2.5 b.yrs) in Enderby Land, Antarctica and for mantle xenolith suites from Victoria, S.E. Australia, in each case after disturbance of the geotherm by a major magmatic and thermal event.

Geol. Soc. Aust. Spec. Publ. 17,1-22.

ARCHAEAN VOLCANISM PHANEROZOIC VOLCANISM TEMPERATURE 800

(°C) 1200

1600

Hl-Mg THOLEIITES KOMA TIITIC \VOLCANISM

35km

\MID-OCEAN V-z. RIDGE VOLCANISM

-100km INTRA-PLATE BASALTIC VOLCANISM

SHIELD VOLCANISM

Fig. 3 A simplification of fig.2 emphasizing that different areas of the modern Earth have very different lithosphere thicknesses and the lithosphere overlies a boundary layer (heavily shaded areas) linking the conductively cooling lithosphere with the convectively cooling asthenosphere with temperatures in the upwelling regions passing through 1450-1500°C, 18kb approximately. These boundary layers give rise to different magmatic characteristics in different areas, i.e. midocean ridge, intra-plate and shield (kimberlitic to lamproitic) magma suites. In contrast to the modern Earth, the Archaean Earth experienced peridotitic komatiites with extrusion temperatures of ~1650°C (32%MgO liquids). Extrapolation of such extrusion temperatures to the Archaean mantle could use either a liquid adiabat or an olivine adiabat. The 'olivine adiabat' is used to suggest minimum temperatures in the Archaean mantle and this intersects the anhydrous peridotite solidus at depths of 150km approx. The labels 3.1 b.yrs and 2.5 b.yrs on the higher temperature geotherms suggest an association of komatiitic volcanism, including peridotitic komatiites, with extremely high temperature, anhydrous crustal metamorphism as is evident in Enderby Land, Antarctica. By 2.5 b.yrs, isobaric cooling to lower temperature granulite facies metamorphism was associated with tholeiitic dykes implying thicker lithosphere.

The Earth's Lithosphere and Asthenosphere.

A1 -poor, and extremely undersaturated olivine d i t i ° of magma aggregamelilitite magmas with residual garnet lherzolite Cr+Al '^ (Brey & Green, 1976; Eggler, 1978; Wyllie, 1979; tion have been shown to be relatively low pressure Wendlandt & Mysen, 1980). Olivine lamproites and temperatures of 1200-1300°C. These condiand olivine leucitites require source compositions tions lie within the MOR picrite to basalt envelope enriched in K2O but also either are (H2O + CO2) because their anhydrous liquidus temperatures are enriched (Ryabchikov & Green, 1978) or with lowered by appreciable dissolved water [2-4% for reduced C-H-0 fluids and significant dissolved boninites, 0.5-1% for Troodos Upper Pillow fluorine (Foley et al, 1986; Foley, 1989). More Lavas]. Genesis of these magmas requires refracolivine-rich liquids (diamond-bearing olivine tory source rocks which have previously lost a large lamproites) require pressures in excess of 40kb melt fraction (i.e. they are second or third stage (Foley, 1989). Kimberlite magmas contain xenolith melts) and melting is aided by access of H20.rich and xenocryst suites indicating pressures of origin fluid, possibly released from the subducted oceanic >45kb and the initially liquid component of kim- slab (Duncan and Green 1987). berlite is inferred to require both very high presConvergent margin regions are technically sures and significant dissolved (C03) in the melt complex interplay of subduction, major trans[Wyllie, 1978, 1980; Eggler & Baker, 1982; Brey form and with transcurrent faults, and back arc spreadet al., 1983]. This group of "within plate" primary ing centres. In this environment the chemical magmas, ranging from Hawaiian olivine tholeiites of potential magma source rocks at mantle to kimberlites, olivine melilitites and olivine diversity depths is very great and opportunity exists for meltlamproites represent magma segregation and ex- ing of subducted crustal rocks, including intertraction (possibly but not necessarily following leaved sediments and volcanics, and for melting of diapirism) along geothermal gradients at lower refractory residual peridotite (including originally temperatures than those in the Mid-Ocean Ridge serpentinized harzburgite) within subducted lithosenvironment. Referring to fig. 3, they represent pheric slabs. Extensive literature on island arc liquids formed along geotherms lying between the petrogenesis is reviewed elsewhere (Gill, 1981; maximum of the 'adiabat' leading to MOR picrites Ringwood, 1975) and current work provides and the minimum illustrated by the coolest conduc- petrological and geochemical evidence for diverse tive geotherm defined by xenolith suites from kim- magmatic sources and for processes of magma berlite pipes on Archaean shields (see next section). mixing. Transient, irregular geotherms, some with very low Large temperature contrasts may exist to deep temperatures at intermediate depths, occur in vertical profiles through subduction zones - these are levels in the asthenosphere across transform faults considered qualitatively in a later section but are which juxtapose very cool subducted lithosphere against 'normal' oceanic geotherms or against treated as if they are relatively rapidly smoothed oceanic lithosphere penetrated by younger 'hotover time scales of -lOOm.yr. spot' volcanic centres, eg. the northern end of the arc or southern end of the Marianas arc. In As well as "constructive margin" and "within Tonganlocations, thermal erosion of the cool subplate" basalts, primary, mantle-derived magmas these ducted slab may yield magmas from melting of occur in : (i) convergent margins; (ii) at convergent hydrated, highly refractory lithospheric lherzolite margin/transform fault intersections; and (iii) at and harzburgite, e.g. the distinctive lavas from the secondary ("back-arc basin") spreading near con- northern end of the Tongan arc (Falloon & Green, vergent margins. The conditions of magma 1986; Falloon et al., 1987; Green et al., 1987). In segregation for two magma types -'boninites' and addition to the potential for various source rock high-Mg, olivine tholeiites to picrites of "Troodos chemical compositions and opportunities for type" are illustrated in Fig. 3 (Tatsumi, 1981; Jen- magma mixing in convergent margin environner, 1983; Duncan & Green, 1987). These magmas ments, the subduction of oceanic crust and lithoindicate refractory sources by their highly sphere carries H2O, CO2, CI, and SO2 to deep forsteritic olivines and liquidus spinels with mantle levels as hydrates, carbonates, sulphates >

con

Geol. Soc. Aust. Spec. Publ. 17,1-22.

and in phases such as apatite, scapolite and other silicates. With increasing temperature in the subducted slab, these volatile components act to depress solidi and to flux melting within the slab or on the overlying mantle wedge. In addition, if the earth's upper mantle has low oxygen fugacity (fC>2 near to iron-wustite + 1 to 2 log units, Ryabchikov et al, 1981; Taylor, 1987; Taylor & Green, 1987,1988; Green, Falloon & Taylor, 1987; Wyllie, 1987) and is degassing reduced CH4-rich (CH >H20,H2) fluids, then subduction of oceanic crust and lithosphere (at fC>2>FMQ to MW) will carry 'excess' oxygen into thp upper mantle and redox interaction between oxidized lithosphere and CH4-rich fluids will produce H20-rich fluids in the vicinity of subducted lithospheric slabs (Green etal., 1987). Viewed on a global scale, given (a) that the earth's deep interior is degassing reduced fluids dominated by CH4, (b) that hydrogen is lost from the earth's upper atmosphere, and (c) that the earth's atmosphere has become increasingly oxidized through time, the lithosphere/asthenosphere region of the earth can be interpreted as a zone of redox interaction and the process of subduction be seen as the principal process introducing oxygen (high f02 mineral assemblages, including Fe2C>3 in silicates, carbonates and sulphates) to depths in excess of 700km, i.e. permitting progressive oxidation of the lithosphere/asthenosphere/Transition Zone through geological time, and maintaining the upper mantle as a region of inhomogeneity in f02. It is interesting to speculate that other terrestrial planets, particularly Mars, may possess an oxidized surface layer but, in the absence of subduction, the interior could remain highly reduced to very shallow levels. Small planetary bodies, such as Earth's Moon, fail to retain an atmosphere and degassing of reduced gases is reflected in very low fC>2 mineral assemblages at the planetary surface.

Petrological Evidence for Mantle P,T Regimes from Mantle Xenoliths The use of xenoliths in kimberlite and other magmas to define P,T conditions within the wall rocks to the host magma, is now well established. In particular the division of lherzolite xenoliths into granular and sheared types, with the latter being higher temperature, higher pressure types, and the former being of lower temperature, lower pressure origin, has yielded characteristic 'kinked geotherms' for many kimberlitic xenolith suites from stable shield regions (e.g. Nixon and Boyd, 1973; Boyd & Nixon, 1978; Boyd, 1984, 1987). Combinations of experimental and thermodynamic methods have been used to calibrate various geothermometers and geobarometers applicable to lherzolite minerals. Nickel & Green (1985) experimentally investigated the influence of chromium on garnet and orthopyroxene solid solutions and the results of application of their empirical geobarometer, coupled with the two-pyroxene geothermometer of Wells (1977), to several xenolith suites are illustrated in Fig. 2 (from Nickel & Green 1985). Particular attention is directed to the conclusion that the steep limb of the 'kinked geotherm' (Boyd & Nixon, 1978) for South African kimberlite xenoliths is an isobaric (48-52kb) temperature step from T~1000°C to T~1400°C (Nickel & Green, 1985). A temperature 'discontinuity' of 400°C over a very small depth interval must represent a transient feature of local or more regional character. Conductive or convective heat transfer will occur to smooth such a discontinuity resulting in heat transfer to overlying cooler regions.

Referring to fig.3, the complexities of magma genesis within convergent margins with emphasis on the roles of volatiles in depressing mantle solidi, do not contravene a major generalization, i.e. that volcanism and magma generation within the modern earth occurs within a P,T regime bounded by a convecting mantle adiabat passing through 20kb ~1450°C, i.e. restricting possible magmas to those with liquidus temperatures (at 1 bar) of less

A local thermal perturbation may be due to diapirism from deeper (> 150km) asthenosphere and heat dissipation may occur locally with little perturbation of the regional very cool lithospheric geotherm defined by the suite of granular xenoliths. A regional discontinuity of this character could arise if shield regions, with subjacent thick lithosphere, are fragmented by the shifting pattern of deep mantle convection and begin to move over

than about 1450°C. A later section summarizes the evidence that at 2.5-3.5b yrs ago this upper bound was 200-250°C higher.

The Earth's Lithosphere and Asthenosphere.

regions of deep mantle with temperatures appropriate to the upwelling limb rather than downwelling regions of deep mantle convection. In this model, there will be a steepening of the conductive limb of the shield geotherm with increased temperatures in the 100km - 150km depth interval such that the intersection of the geotherm and peridotite-C-H-0 solidus would progressively move to shallower levels. Recognizing that the diamond/graphite reaction boundary intersects the low temperature limb of the South African xenolith geotherm at ~40kb, 900°C and 50kb, 1250°C, it is clear that this type of thermal erosion of the lithosphere would progressively eliminate the 'diamond window' in which diamond would be stable at subsolidus and near-solidus conditions. The model predicts that those shield regions and subjacent lithosphere which move from long term stability over cool, downflow regions of deep mantle convection to overlie warmer regions will show a sequence of magmatism from early diamondiferous kimberlite or olivine lamproite, through younger diamond-free kimberlites and olivine lamproites, to younger volcanism with olivine melilitite and olivine leucitite, characteristic of lithospheric thickness of80-100km and geothermal gradients similar to those of oceanic intraplate character. The process of thermal erosion of very old lithosphere beneath shield regions would be expected to produce magma batches with radiogenic isotopic signatures and incompatible element contents reflecting inhomogeneities created by frozen-in magma batches or peridotite diapirs, by fluid metasomatism and redox interactions. Evidence that some South African diamonds are >3 b.yrs old (Richardson, 1986) emphasises the very long times available for such lithospheric inhomogeneity to develop and also that some regions of the earth show extra-ordinary stability of a very cool geothermal gradient. It is implied, if the South African diamond ages of ~3 b yrs are correct, that the South African shield retained a cool, non-inflected geotherm (similar to that illustrated by Siberian kimberlite xenoliths, Sobolev, 1977,; Boyd, 1984; Nickel and Green, 1985) for ~3 b yrs but continental break-up in the Mesozoic led to distinctive and extensive basaltic volcanism ranging from tholeiitic (e.g. Karoo magmatism) near newly rifted margins through variable Rift Valley-

type volcanism to kimberlitic volcanism in residual shields. Fig. 3 illustrates the envelope of modern earth geothermal gradients bounded by the low temperature conductive geotherm defined by xenolith suites beneath South African and Siberian Shields which may, for a stable, long-lived scenario, pass smoothly into down-welling regions of deep mantle convention. The upper boundary of the envelope represents adiabatic upwelling along constructive plate margins. Also illustrated in Fig. 3 , are geotherms representing the consequences of plate tectonics in which movement of lithospheric plates is decoupled from deep mantle convection and transient steep geothermal gradients develop when old, very cool lithosphere over-rides upwelling regions (resulting in characteristic magmatism). The three transient geotherms illustrate conditions which will lead to convective heat loss and 'smoothing' of geotherms by means of: (i)

kimberlite or olivine lamproite eruption on stable shield regions,

(ii) intraplate basaltic volcanism as in S.E. Australia, African Rift Valley etc., characterized by olivine melilitite to olivine tholeiite volcanism, inferred to be derived from the upper part of the asthenosphere/lowermost lithosphere (Green, 1971; Brey and Green, 1978; Green and Liebermann, 1976).

(iii) constructive margin volcanism as in Red Sea, Gulf of California, Iceland etc. - characterized by tholeiitic picrite to olivine tholeiite volcanism inferred to be derived by upwelling from the lower part (LREE-depleted) of the asthenosphere (Green, 1971; Green and Liebermann, 1976). Referring to fig. 2, the data on peridotite xenolith suites from various magma hosts and localities confirm that different tectonic regimes sampled by Tertiary to Quaternary volcanism, show the spread of lithospheric geotherms illustrated in figs.2 and 3 (see also Boyd and Nixon, 1978; Boyd 1987, O'Reilly and Griffin, 1985; Nickel and Green, 1985).

Geol. Soc. Aust. Spec. Publ. 17,1-22.

In some cases xenolith suites preserve textures showing cooling, possibly from 'perturbed' geotherms associated with mantle diapirism and magma genesis, to a lower temperature continental geotherm (fig.2). For example, lherzolite nodule suites in Quaternary volcanics in Western Victoria record a temperature range of ~850°C to ~1075°C with evidence of cooling and exsolution particularly in the lower temperature amphibole-bearing lherzolites (Nickel and Green 1984). These xenolith suites sample lithosphere possibly developed in convergent margin tectonics throughout the Lower Palaeozoic [S.W. Pacific type with island arcs, back-arc basins and continental accretion (Crawford & Keays, 1978)] but since stabilization in Upper Palaeozoic, the area has experienced intraplate basaltic volcanism in Jurassic, Tertiary and Quaternary Periods. It is not yet possible to assign particular mineral equilibria and mineral textures to a time frame so that it is not possible to distinguish between assemblages reflecting ambient lithosphere P,T conditions at the time of detachment of xenoliths, from P,T conditions which are frozen-in and relict from early diapirism and magmatism.

LITHOSPHERIC EVOLUTION THROUGH GEOLOGICAL TIME: We may use the petrogenetic grid illustrated in fig.2 to place some constraints on the thermal evolution of the Earth's mantle and particularly of the lithosphere. Apart from the general conception of an initial high temperature for the mantle, during or following accretion and core formation (Ringwood, 1975) there is evidence in the geological record that mantle temperatures at >2.5 b yrs ago were higher than at present. The eruption temperatures of peridotitic komatiites were from 1520°C (24% MgO liquids) to 1650°C (33% MgO liquids) (Green 1975, 1981; Arndt 1977) and liquids with >30% MgO appear to be restricted to pre-2.5b yrs. With eruption temperatures of 1650°C, projection back to source P,T conditions requires assumptions on petrogenesis, knowledge of latent heat of melting and knowledge of the adiabatic gradient for basaltic and peridotitic liquids. The most conservative model (in terms of requiring minimum temperatures in the deep

mantle) would take the adiabatic gradient for olivine (~l°C/kb) and seek a model for peridotitic komatiite genesis which postulates an upwelling centre in which diapirism and an increasingly refractory conduit walled by refractory residue, creates a temperature distribution beneath the peridotitic komatiite centre which matches the olivine adiabat (l°C/kb) from 1 bar, 1650°C to intersect the (extrapolated) anhydrous peridotite solidus (or peridotite-C-H-0 solidus with (CH4>H20)-rich fluids) at about 160km (Green 1975,1981). Figs. 4,5 illustrate an attempt to define such a model. The model presented on figs. 4,5 relies on a quantitative knowledge of melting temperatures of 'fertile' and refractory peridotite to dunite compositions [measured to ~40kb, extrapolated to higher pressure with acknowledgement of the pioneering studies to 140kb by Takahashi, 1986], a quantitative knowledge of peridotite melting behaviour in the 0-40kb pressure interval and a fragmentary and interpretative understanding of Archaean geology. As type examples of contrasted Archaean terranes, the deep levels (25-35 km) of sialic shields are illustrated by Enderby Land, Antarctica and the shallow levels of Archaean shields are illustrated by granite-greenstone terranes such as occur in Western Australia. The model is presented, to the extent of illustrating geotherms for different regions, without the benefit of numerical geophysical input except in a schematic way (figs. 4,5). This unconventional approach is justified because of the uncertainties and latitude permitted in the model, particularly in the absence of a horizontal scale! We do not know the dimensions of Archaean protocontinents, or of 'ocean basin' convective overturn regions nor do we know the Archaean rates of spreading and subduction. These uncertainties are sufficient to accommodate the large unknowns in the thermal and rheological properties of the crystalline and liquid components of the model. The model however uses a major geophysical constraint in that it maximizes convective heat transport from the deep interior by creating a refractory crystalline conduit system feeding liquids at T>1600°C to eruption at the surface. Secondly it presents a distribution of subsolidus and partially molten regions with reasonably long-term stability, i.e. no regions in which the degree of melting exceeds

The Earth's Lithosphere and Asthenosphere.

- 3 0 % (destroying the high effective viscosity of a grain-supported crystal liquid mix) and which approaches or attains a gravitationally stable, densitystratified depth section. The constraints imposed on Archaean tectonics by information on melting relationships and by knowledge of Archaean geology, are very major constraints and suggest that geophysical modelling may be guided as much by these data as by knowledge of thermal and rheological properties of crust and mantle materials.

Geological information which has influenced the model of fig.4 includes the following:(a) Peridotitic komatiites occur within Archaean greenstone belts, apparently as major volcanic centres which may be flanked by and interleaved with volcanics of 'normal' type (basaltic, andesitic etc.). (b) Peridotitic komatiite flow sequences may occur several times in a stratigraphic sequence in a

Fig. 4 Cartoon illustrating a model (see text) which seeks to reconcile the extremely high temperatures and repetitive peridotitic komatiite extrusions from centres in close proximity and interleaved with high-magnesium and tholeiitic basalts with the existence of other regions of stabilized crust and cool lithosphere extending to at least 150 kms. The model envisages small scale convection in which 'convection fingers or plumes' develop very refractory feeder chimneys by continuous extraction of melt from adiabatically upwelling mantle. The left-side columns show the character of a 30% melt fraction moving at velocity VL through an increasingly refractory lherzolite (F089) to dunite (F094) residue ascending at velocity Vc and continuously reacting with and precipitating liquidus phases on the conduit walls. Peripherally to the peridotitic komatiite centres magmatism from the convective cycling of asthenosphere is picritic to basaltic. A subduction process which is also continuously accreting and cooling the proto-continental nucleus overlying the down-welling part of the convective cycle appears to be necessary in view of the evidence for the existence of ancient, thick (150 kms+) cool lithosphere. This is illustrated in the central part of the figure where residues of F094 dunite mark former peridotitic komatiite volcanic centres within a dominant lherzolite to harzburgite (~Fo9i).

Geol. Soc. Aust. Spec. Publ. 17, 1-22.

TEMPERATURE

(°C)

Fig. 5 Schematic temperature distributions within the model (sections 1 to 5 as shown) of fig.4. These temperature distributions prevent thermal erosion of the 'proto-shield', and predict reasonable stability for the komatiitic centres. Note that for the deep, primitive mantle the pyrolite + CH4,H20 solidus is applicable but near-surface oxidation and hydration may make the pyrolite + H2O solidus relevant to the subduction environment (lightly shaded area of fig. 4). However, if the Archaean atmosphere has low fC>2 then the subduction environment may retain high fcH4> fH20 and melting may approach the anhydrous solidus (or the peridotite-CH4>H20 solidus (Taylor & Green, 1988)).

The Earth's Lithosphere and Asthenosphere.

given area, separated by basaltic sequences and by sediment horizons. The feeder system for peridotitic komatiites has sufficient long term stability to yield individual flows separated by sedimentary horizons.

in some sequences of komatiitic and high-Mg basalts, which show geochemical and penological evidence of assimilation of and contamination by crustal rocks, accompanied by olivine precipitation (i.e. evolution from parental peridotitic komatiite magmas), is evidence that feeder systems in some cases act as open systems with respect to their crustal wall rocks.

(c) Peridotitic komatiite and komatiite sequences are part of synclinal greenstone belts between and underlain by 'basement gneiss terranes'. The respective roles of deformation and original emplacement in determining this juxtaposition remain debatable but komatiite eruptive centres and proto- continental nucleii appear to form relatively small-scale 'domains' in the earliest stages of crustal evolution.

Figs.6,7,8 and 9 are cartoons summarizing a model for the crustal evolution of Archaean shields drawn from the example of Enderby Land, Antarctica [(E) of figs.6-9 and figs.2 and 3]. In this area, the studies of metamorphism, crustal magmatism and mafic dyke emplacement (Sheraton & Collerson 1984, Ellis 1983, Harley 1983) provide evidence of isobaric cooling of 30-35km continental crust from T~1050° at 3.1 b.yrs to 650-700°C at 2.5 b.yrs, followed by remarkable crustal stability to clOOm.yrs. The mafic dykes range from distinctive high Mg, high S1O2 dykes at 2.5 b.yrs, through suites ranging from undersaturated alkali-enriched varieties to fractionated, Fe-rich olivine tholeiites at 1.4 to 1.1 b.yrsLto olivine leucitites at a modern

(d) In spite of the extremely high temperatures and thus the very large capacity for assimilation of and contamination by crustal wall rocks, the arrival at the surface of liquids of peridotitic komatiite character free of crustal contamination and unfractionated by accompanying crystallization, requires remarkably high temperature and geochemically isolated feeder systems. Conversely, the presence

LITHOSPHERE

(DISCONTINUOUS CONDUCTIVE

LAYER)

T ~ 900- 1 1 00° C BOUNDARY LA YER 7" ~ 1 500- 1 700 C

PERIDOTITIC MEL TS

^ ^

KOMATIITIC

MELTS

ASTHENOSPHERE

(CONVECTIVE LAYER! (UPPER MANTLE OR POSSIBLY WHOLE MANTLE)

PICRITIC MEL TS

PERIDOTITE — 1 5 0 KM

+ MELT + (CH +Hy) a

fluid

+ PICRITIC

— KOMATIITIC

(CH +H )fluid A

MELTS

Fig. 6 Cartoon illustrating concepts of Archaean tectonics invoking higher temperatures in a rapidly convecting asthenosphere, a reduced (fC>2~IW to IW+1) upper mantle and reduced atmosphere, a proto-crust with a very shallow boundary layer between convective and conductive regimes would lie at 30-40 kms.

Geol. Soc. Aust. Spec. Publ.

bilized and cooled to intersect the diamond stability field by 3.0b.yrs ago. In contrast, dating of komatiitic basalts at Kambalda, W.A., established a 2.7b.yr age of extrusion and of intrusion into sialic crust. (Compston et. ai 1986).

volcano, Gaussberg. The petrogenetic grid of fig.2 is used to illustrate these magma types in terms of temperatures and depths of source regions, i.e. lithosphere thickening with time. Figs.6-9 also illustrate an increasing depth and changing temperature contrast across the 'boundary layer' marking the base of lithospheric plates over-riding convective motions of the asthenosphere.

The recognition that the Archaean earth (at least by b.yrs ago) contained thick, 'cold' lithospheric slabs of >150km thickness while activity elsewhere was yielding peridotitic komatiite lavas with 2532%MgO provides a paradox in that the temperature contrasts of boundary layers illustrated in fig.3 and figs.6-9 would be short-lived leading to thermal erosion of the base of the lithosphere. To avoid such thermal erosion, the Archaean proto-continents must always remain positioned directly over downflow regions of deeper mantle convention. This effect is illustrated in fig.4 where small scale

Although figs.6-9 may illustrate appropriate broad-brush models for the evolution of Enderby Land and subjacent lithosphere, they clearly cannot explain other equally important but very different geological histories in areas which immediately prior to the Mesozoic were neighbouring regions of the Gondwana super continent. The work on dating of mineral inclusions in diamond (Richardson, 1986) indicates that some cratons had already sta-

loss

PERIDOTITE (ALK

2.5

r ~

+MELT

BASALT

PICRITE

17,1-22.

b.yr

1300-1500'C

COMPOSITION]

ASTHENOSPHERE (CONVEC

TIVE

LAYER)

I 5 0 KM

CH + H 4

Fig. 7 Later evolutionary stage based on sequence of events in Enderby Land [E2] with thickened lithosphere and boundary layer between convective and conductive regimes lying at depths of 60-70 km. The P,T distribution is such that alkali olivine basalts to high Mg-tholeiites could appear as dykes or extrusive in this regime (ref. fig. 2,3). Changing atmospheric composition and the interaction of this environment with the mantle via the subduction process will introduce increasing hydration into the mantle and magmatic and metamorphic processes will reflect this.

The Earth's Lithosphere and Asthenosphere.

(in comparison with modern earth super continents) proto-continents are cooled (figs.4,5) by rapid subduction of picritic and basaltic crust containing local centres of komatiitic to peridotitic komatiite volcanism. The corollary of the paradox posed by the ancient, cold and deep lithosphere, is the existence of extremely efficient, high temperature feeder systems for peridotitic komatiite volcanism. These are illustrated in figs.4,5) as refractory chimneys of harzburgite to dunite, becoming increasingly forsteritic at shallower depths.

It is emphasized that the schematic model of figs.4,5 is an attempt at deducing the most conservative model in terms of the high temperature and source depth required to generate peridotitic komatiite liquids extruding at the earth's surface at ~1650°C. Recent models have been based on ex-

INTRACRA MOBILE

IONIC ZONE

OLIVINE NEPHELINITES

perimental work by Takahashi (1986) and Takahashi and Scarfe (1985) which present experimental data with liquidus and solidus of peridotite converging at 120-140kb. Several authors have suggested that peridotitic komatiites could be melts moving directly from -400km depth (120-140kb) to surface extrusion. A major difficulty with this model is that such liquids would move from source to surface adiabatically or with some cooling. The adiabatic gradient for peridotitic liquid is unknown but will be steeper (dT/dP) than the l°C/kb used for olivine c.f. fig.3.

CONCLUSIONS Knowledge of the melting behaviour of peridotite is necessary for understanding mantle

1.0 ALK OL BASALTS OL THOLEIITES

Fe-RICH

THOLEIITE

DYKES

CRUSTAL

b.yr

SHIELD

UNDERPLATING

LITHOSPHERE

GARNET PERIDOTITE (OLIVINE MELILITITE OLIVINE NEPHELINITE

+MELT TO COMPOSITION)

^ T~ 1050- 1 150"C BOUNDARY LAYER T ~ 1200- 1500 "C »/>/n

ASTHENOSPHERE (CONVECTIVE

LAYER)

Fig. 8 Evolutionary stage illustrating a crust/ lithosphere of essentially Modern or Phanerozoic Earth characteristics, i.e. a geotherm intersecting the solidus at -90 km due to amphibole stability and a more oxidized upper zone of the asthenosphere (i.e. with OH" + CO3" in solution in melt of olivine melilitite character)[cf. figs. 2 & 3]. The model is drawn with the Enderby Land/MacRobertson Land area of Antarctica in mind. This was a period of extensive mafic dykes but also includes more undersaturated, nepheline-normative dykes. Mobile belts (Rayner Metamorphism - Ri) with charnockitic instrusives were also developed across the Archaean shield, the cartoon suggests that magmatic underplating may be an important part of this process.

Geol. Soc. Aust. Spec. Publ. 17,1-22. 0.

b . y r

(prior

A u s t r a l i a / A n t a r c t i c a

kimberlite/olivine

lamproite

separation]

magma

SHIELD

'/T

Iherzolite/harzburgite

± minor

(fluid

absent

a mphi

b ol e ± p hlog

C02-rich)

LITHOSPHERE

op it e

(CONDUCTIVE

phlogopite

120

± ilmenite

± CH4-H2

fluid

absent

rich

H 2 0 - C H a

H 2 0 - C 0 2 locally 1 5 0 KM

very SIC

variable stability ,

f02 to

LAYER)

apatite

rich rich from

fO?~MW 2

T ~ 950'C BOUNDARY T

YER

1400-1500°C

ASTHENOSPHERE (CONVECTIVE

YER)

Fig. 9 Model of thick lithosphere underlain by a boundary layer matching the 'kinked geotherm' or isobaric temperature discontinuity shown in figs 2 & 3. In Antarctica this is illustrated by the recent Gaussberg olivine lamproite volcano, and more generally in Gondwanaland is illustrated by Mesozoic to Tertiary kimberlitic to olivine melilititic magmas associated with early stages of continental rifting and thermal erosion of thickened lithosphere beneath them. Figs 6-9 have been presented as an evolutionary sequence but the evidence (e.g. Richardson 1986) that some areas beneath Archaean shields had attained fig. 9 characteristics by 3.1 b.yrs whereas other areas were at fig. 6 stage 2.7 b.yrs ago emphasizes that there is little global uniformity. In contrast the interpretations suggest a general mantle cooling (c.f. 'boundary layers' changes) within and beneath the asthenosphere and consequent changes in the scale of lithospheric units ('plates') and in the nature of heat transfer by magmatism. pedogenesis, but equally for interpreting mantle geophysical properties. Volatiles, particularly C-HO, are degassing from the earth and play a most important role in determining the melting behaviour of the mantle. The information from experimental studies of peridotite melting has been integrated with studies of modern magmatic provinces and mantle xenolith suites to develop specific models for different tectonic settings on the modern earth. Because of relative movement between lithospheric plates and asthenosphere, tran-

sient boundary layers, in which temperature rises more rapidly with depth, are formed and will lead to thermal erosion of the lithosphere and production of magmas reflecting earlier complex melting and enrichment processes in the lithosphere. Brief examination of the geological record leads to inferences that in the early history of the earth higher temperatures in the asthenosphere resulted in more efficient convective heat loss by eruption at specific centres, of extremely high

The Earth's Lithosphere and Asthenosphere.

temperature peridotitic komatiite lavas. However other regions, proto-continents or Archaean shields, were already in existence (at ~35km thickness) and some were underlain by thick, cold lithosphere to depths of 150-170km, i.e. into the diamond stability field. This paradox appears to require the existence of small proto-continents (shield) with thick, refractory lithosphere of > 150km thickness and of the continuing location of such proto-continents over the down-welling regions of asthenospheric convection. In neighbouring areas, the greenstone belts require extremely high temperature magmas and a refractory magmatic conduit or feeder system which becomes more refractory (to F094 dunite) at shallow depths. The speculative models illustrated in figs.4-9 are based on petrological and geological constraints and are presented as a basis for comparison with models deduced from geophysics and geochemistry.

BALLHAUS

C.,

BERRY

GREEN

D.H.

1990.

Experimental calibration of spinel in mantle lherzolite as an oxygen activity sensor. Contributions to Mineralogy & Petrology, in press.

BOYD F.R. 1984. Siberian geotherm based on lherzolite xenoliths from the Udachnaya kimberlite USSR. Geology

12, 5 2 8 - 5 3 0 .

BOYD F.R. 1987. High and low-temperature garnet peridotite xenoliths and their possible relation to the lithosphere-asthenosphere boundary beneath Southern Africa. In Nixon, RH. ed. Mantle Xenoliths. John Wiley, New York, (in press).

BOYD

F.R.

NIXON

RH.

1973.

Origin

the

ilmenite-silicate nodules in kimberlites from Lesotho and South Africa. In Nixon, P.H. ed. Lesotho Kimberlites. pp.254-268. Lesotho National Development Corporation, Maseru, Lesotho.

BREY G . P . & GREEN D . H . 1 9 7 6 . S o l u b i l i t y o f C 0 2 i n t h e

ACKNOWLEDGEMENTS Much of the material for this paper was originally presented as the 'Mawson Lecture' of the Australian Academy of Science, presented at the 4th International Conference on Antarctic Earth Science in Adelaide in 1982. The theme was further developed in a paper presented to the 7th Australian Geological Convention in August, 1984. Dr Barry Drummond is thanked for his patience and encouragement in urging completion of the manuscript and he and Dr A.L. Jaques are thanked for comments on the draft manuscript. I acknowledge the help and interest of my colleagues, particularly post- doctoral fellows and past graduate students at the University of Tasmania. I give particular thanks to Mrs June Pongratz for her work on preparing the diagrams and to Dr M. Barsdell and Mrs Karyl Whelan in the final preparation of the manuscript.

Earth's upper mantle. Contributions Petrology 55, 217.

to Mineralogy

and

BREY G . P . , BRICE W . R . , ELLIS D . J . , GREEN D . H . , HARRIS

K . L . & RYABCHIKOV I.D. 1983.

Pyroxene-carbonate

reactions in the upper mantle. Earth and Science Letters 62, 63-74.

CLARK

S.P.

REMGWOOD,

A.E.

1964.

Planetary

Density

distribution and constitution of the mantle. Reviews of Geophysics

2, 35-88.

COMPSTON

W.,

WILLIAMS

I.S.,

CAMPBELL

I.H.

GRESHAM J.J. 1986. Zircon xenocrysts from the Kambalda Volcanics: Age constraints and direct evidence for older continental crust below the Kambalda-Norseman greenstones. Earth and Planetary Science Letters 76, 299-311.

CRAWFORD

A.J.

KEAYS

R.R.

1978.

Cambrian

greenstone belts in Victoria marginal sea-crust slices in the Lachlan Fold Belt of southeastern Australia. Earth and Planetary Science Letters 41, 197-208.

REFERENCES: DUNCAN R . A . & GREEN D . H .

ARNDT N.T. 1977. Ultrabasic magmas and high degree melting of the mantle. Contributions to Mineralogy and Petrology 64, 205-221.

1987. T h e genesis

refractory melts in the formation of oceanic crust. Contributions to Mineralogy and Petrology 96,326-342.

Geol. Soc. Aust. Spec. Publ. 17,1-22.

EGGLER D.H. 1978. The effect of CO2 upon partial melting of peridotite in the system NA2-Ca0-Ah03-Mg0-Si02-C02 235kb with an analysis of melting in a peridotite -H2O-CO2 system. American Journal of Science 78, 305-343.

FOLEY S.F., TAYLOR W . R . & GREEN D . H . 1 9 8 6 . T h e r o l e

of fluorine and oxygen fugacity in the genesis of the ultrapotassic rocks. Contributions to Mineralogy and Petrology 94, 183-192.

EGGLER D.H. & BAKER D.R. 1982. Reduced volatiles in

GILL J.B. 1981. Orogenic andesites and plate tectonics. Springer-Verlag, New York, NY, 305p.

the system C-O-H: implications to meltings, fluid formation, and diamond genesis. In Akimoto S. and Manghnani M. eds. High-Pressure Research in Geophysics, pp.237-250, Center for Academic Publications Japan, Tokyo.

GREEN D.H. 1971. Compositions of basaltic magmas as indicators of conditions of origin: Applications to oceanic volcanism. Philosophical Transactions of the Royal Society London Series A 268, 707-725.

ELLIS D.J. 1983. The Napier and Rayner complexes of Enderby Land Antarctic - contrasting styles of metamorphism and tectonism. In Oliver R.L. James RR. and Jago J.B. eds. Antarctic Earth Science, pp.20-24. Australian Academy of Science. ELTHON O. & SCARFE C.M. 1984. High-pressure phase

equilibria of a high-magnesia basalt and the genesis of primary oceanic basals. American Mineralogist 69,1 -15.

FALLOON T.J. & GREEN D.H. 1986. Glass inclusions in

magnesian olivine phenocrysts from Tonga: evidence for highly refractory parental magmas in the Tongan arc. Earth and Planetary Science Letters 81,95-103. FALLOON T.J. & GREEN D.H. 1987a. Anhydrous partial melting of MORB pyrolite and other peridotite compositions at lOkbar: implications for the origin of primitive MORB glasses. Mineralogy and Petrology 37, 181-219. FALLOON T.J. & GREEN D.H. 1988. Anhydrous partial

melting of peridotite from 8 to 35kbars and the pedogenesis of MORB. Journal of Petrology Special Lithosphere Issue, 379-414. FALLOON T. J. & GREEN D . H . 1989.

T h e s o l i d u s of

carbonated, fertile peridotite. Earth & Planetary Science Letters 94, 364-370. FALLOON T. J. & GREEN D . H .

1990. S o l i d u s

carbonated fertile peridotite under conditions. Geology 18, 195-199.

fluid-saturated

FOLEY S.F. 1989. The genesis of lamproitic magmas in a reduced, fluorine- rich mantle. In 4th International Kimberlite Conference Kimberlites and Related Rocks, Vol. 1. Geological Society of Australia Special Publication 14, 616-631.

GREEN D.H. 1973a. Experimental melting studies on a model upper mantle composition at high pressure under water-saturated and water undersaturated conditions. Earth and Planetary Science Letters 19, 37-53. GREEN D.H. 1973b. Conditions of melting of basanite magma from garnet peridotite. Earth and Planetary Science Letters 17,456-465. GREEN D.H. 1975. Genesis of Archean peridotitic magmas and constraints on Archean geothermal gradients and tectonics. Geology 3,15-18. GREEN D.H. 1981. Pedogenesis of Archean ultramafic magmas and implications for Archean tectonics. In Kroner, A. ed. Precambrian Plate Tectonics. Elsevier, 469-489. GREEN D. H. 1990. The role of oxidation reduction and C-H-0 fluids in determining melting conditions and magma compositions in the upper mantle. Proceedings of the Indian Academy of Sciences (Earth & Planetary Sciences) 99,153-165. GREEN D .

WALLACE

1988.

Mantle

metasomatism by ephemeral carbonatite melts. Nature 336,459-462. GREEN D . H . , FALLOON T.J. & TAYLOR W . R .

1987.

Mantle-derived magmas - roles of variable source peridotite and variable C-H-0 fluid compositions. In Mysen B.O. ed.Magmatic Processes: Physicochemical Principles. Geochemical Society, USA. Special Publication 1, 139-154. GREEN D . H . , HIBBERSON W . O . & JAQUES, A . L . 1 9 7 9 .

Pedogenesis of mid- ocean ridge basalts, In McElhinny, M.W. ed. The Earth: Its origin, structure and evolution. Academic Press, London, 265-290.

The Earth's Lithosphere and Asthenosphere.

GREEN D.H. & LIEBERMANN R.C. 1976. Phase equilibria and elastic properties of the pyrolite model for the oceanic upper mantle. Tectonophys 32, 61-92.

O'REILLY S.Y. & GRIFFIN W.L. 1985. A xenolith derived geotherm for southeastern Australia and its geophysical implications. Tectonophysics 111, 41-63.

GREEN D.H. & RINGWOOD A.E. 1963. Mineral assemblages in a model mantle composition. Journal of Geophysical Resesarch 68, 937-945.

PARKER R.L. & OLDENBURG D.W. 1973. Thermal model of ocean ridges. Nature 242, 137-139.

GREEN D.H. & RINGWOOD A.E. 1967. The genesis of

basaltic magmas. Contributions to Mineralogy and Petrology 15, 103-190. S.L. 1983. Regional geobarometrygeothermometry and metamorphic evolution of Enderby Land, Antarctica./n Oliver R.L., James RR. and Jago J.B. eds. Antarctic Earth Science, pp.25-30. Australian Academy of Science. HARLEY

JAQUES A.L. & GREEN D.H. 1980. Anhydrous melting of

peridotite at 0- 15kbar pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology 73, 705-720. JENNER G. 1982. Pedogenesis of high-Mg andesites: An

experimental and geochemical study with emphasis on HMA from Cape Vogel, PNG. PhD Thesis, University of Tasmania (Unpublished). KURAT G., PALME H.J., SPETTEL B., BADDENHAUSEN H., HOFMEISTER H., PALME C. & WANKE H. 1986.

Geochemistry of ultramafic xenoliths from Kapfenstein, Austria; evidence for a variety of upper mantle processes. Geochimica Cosmochimica Acta 44, 45-60. MENGEL K. & GREEN D.H. 1989. Stability of amphibole and phlogopite in metasomatized peridotite under water saturated and water undersaturated conditions. In 4th International Kimberlite Conference Kimberlites and Related Rocks. Geological Society of Australia Special Publication 14, 571-581. NICKEL K.G. & GREEN D.H. 1984. The nature of the

upper mantle beneath Victoria, Australia as deduced from ultramafic xenoliths. In Kornprobst, J. ed. Kimberlites II: The mantle and crust!mantle relationships, pp. 161 -178. Elsevier.

PELTIER W.R. & JARVIS G.T. 1982. Whole mantle convection and the thermal evolution of the Earth. Physics ofthe Earth and Planetary Interiors 29,281 -304. RICHARDSON S.H. 1986. Latter-day origin of diamonds

of eclogitic paragenesis. Nature 322, 623-626.

RINGWOOD A.E. 1966a. The chemical composition and origin of the Earth. In Hurley P.M. ed. Advances in Earth Science, pp.287-356. M.I.T. Press. Cambridge, Mass. RINGWOOD A.E. 1975. Composition and petrology of the Earth's mantle. McGraw-Hill. 618pp. RYABCHIKOV I.D. & GREEN, D.H. 1978. Role of CO2 in the petrogenesis of potash-rich magmas. In Problems in the Petrology of the Earth's Crust and Upper Mantle. Nauka, Novosibirsk, p.49. SHERATON J.W. & COLLERSON K.D. 1984. Geochemical evolution of Archaean granulite-facies gneiss in the Vestfold Block and comparisons with other Archaean gneiss complexes in the East Antarctic Shield. Contributions to Mineralogy and Petrology 87, 51-64. SOBOLEV N.V. 1977. Deep-seated inclusions in kimberlite and the problem of the composition of the upper mantle. American Geophysical Union 279 pp. STOLPER E. 1980. A phase diagram for mid-ocean ridge basalts: preliminary results and implications for petrogenesis. Contributions to Mineralogy and Petrology 74, 13-27. TAKAHASHI E. 1986. Melting of a dry peridotite kLB-1

up to 14 Gpa: Implications on the origin of peridotitic upper mantle. Journal of Geophysical Research 91, 9367-9382.

OLAFSSON M . & EGGLER D.H. 1983. Phase relations of

TAKAHASHI E. & SCARFE C.M. 1985. Melting of peridotite to 14 Gpa and the genesis of komatiite. Nature 315, 566-568.

305-315.

TATSUMI Y. 1981. Melting experiments on a high-magnesian andesite. Earth and Planetary Science Letters 54, 357-365.

amphibole, amphibole-carbonate, and phlogopitecarbonate peridotite- petrologic constraints on the asthenosphere. Earth and Planetary Science Letters 64,

Geol. Soc. Aust. Spec. Publ. 17,1-22. TAYLOR W.R. & GREEN, D.H. 1987. The petrogenetic

role of methane: Effect on liquidus phase relations and the solubility mechanism of reduced C-H volatiles. In Mysen B.O. ed. Magmatic Processes: Physicochemical Principles. Geochemical Society, USA Special Publication 1, 121-138. TAYLOR W.R. & GREEN D.H. 1988. Measurement of

reduced peridotite C-O-H solidus and implications for redox melting of the mantle. Nature 332, 349-352. WALLACE M. W. & GREEN D. H. 1988. An experimental

determination of primary carbonatite composition. Nature 335, 343-346.

magma

WELLS P.R.A. 1977. Pyroxene thermometry in simple

and complex systems. Contributions to Mineralogy and Petrology 62, 129-139.

WENDTLANDT R.R. & MYSEN B.O. 1980. Melting phase

relations of natural peridotite + CO2 as a function of degree of melting at 15 and 30kb. American Mineralogist 65, 37-44.

WYLLIE P.J. 1977. Crustal anatexis - an experimental

review. Tectonophysics 43,41-71.

WYLLIE P.J. 1978. Mantle fluid compositions buffered in peridotite-CC>2-H20 by carbonates, amphibole, and phlogopite. Journal of Geology 86, 687-713. WYLLIE P.J. 1979. Magmas and volatile constituents. American Mineralogist 65, 469-500. WYLLIE P.J. 1987. Discussion of recent papers on

carbonated peridotite, bearing on mantle metasomatism and magmatism. Earth and Planetary Science Letters 82, 391-397.

D.H. Green Geology Department University of Tasmania Hobart 7005 Tasmania AUSTRALIA.

Geol. Soc. Aust. Spec. Publ. 17,23-40.

The Base of the Lithosphere Under Australia *

K.J. Muirhead and B .J. Drummond Bureau of Mineral Resources, Geology <& Geophysics, Canberra, A.C.T., Australia Rheological, temperature, geochemical and petrological definitions of the lithosphere are model dependent and therefore imprecise. An alternative definition based on seismic models is that the lithosphere/asthenosphere boundary corresponds to the top of a seismic low velocity layer within the upper mantle. The low velocity layer presumably occurs because of a drop in the shear modulus. It has therefore always been regarded as a potential zone of decoupling of the lithosphere from the underlying mantle, but within the Australian continent this may not be the case. A new seismic velocity model has been developed for the upper mantle under eastern Australia. It satisfies the available body wave data as well as the surface wave data. When compared to models from the shield regions of central and western Australia it shows that significant differences exist at depths below 100 km between the upper mantle seismic velocity structure under Phanerozoic eastern Australia and the shield regions of central and western Australia. In eastern Australia, a low velocity layer probably starts near 120 km depth, but no low velocity layers occur above 200 km depth under the Australian shield. Rather, the shallowest low velocity layer under northern Australia lies below the Lehmann discontinuity near 200 km depth. Teleseismic arrival times at seismograph stations in the shield regions are early relative to eastern Australian stations. The difference cannot be entirely accounted for by structures above 200 km depth and imply differences in seismic velocities at depths even greater than 200 km. The low velocity zone in eastern Australia is interpreted as a region where the geotherm lies above the upper mantle solidus, resulting in partial melting and a drop in the shear modulus. The refractory nature of the upper mantle in eastern Australia implied by the chemical signature of mantle nodules and alkali basalts throughout the region suggests that the low velocity zone has been in its present position relative to the continent for millions of years and must be drifting with the continent. The low velocity zone under eastern Australia therefore does not represent the zone of decoupling of the continent from the underlying mantle. The upper mantle under the shield regions is refractory and cold with temperatures below the solidus, at least to 200 km depth. It is probably depleted to depths greater than 200 km depth and therefore will be less dense than the surrounding mantle, making it isolated and unable to remix. The continent under the shield regions is therefore translating at depths much greater than 200 km. Thus, although the seismic lithosphere defines the rigid outer shell of the Earth, it does not define the depth at which the plates decouple from the underlying mantle. That is much deeper.

INTRODUCTION The lithosphere has traditionally been regarded as the rigid, stony, outer part of the Earth overlying a region which, because of its temperature, pressure and perhaps mineralogy, is weaker, and probably behaves in a viscous or plastic fashion when stresses are applied. The lithosphere plays an important role in large-scale tectonic processes such as mountain building and sedimentary basin formation, and the nature and depth distribution of the lithosphere have become key factors in influencing the search for minerals, especially diamonds. In the plate tectonic model, oceanic lithosphere is created at ocean ridges and is subducted and

destroyed at trenches. Thus the oceanic lithosphere has a well-defined history. Continental lithosphere is more complex. It is more buoyant and is therefore difficult to subduct and destroy; rather, it usually remains at the surface of the Earth and becomes reworked and very complicated. The job of defining the continental lithosphere therefore becomes difficult. It is difficult also because of the different ways in which the lithosphere can be defined (US Geodynamics Committee, 1983). Four principal definitions exist. The lithosphere may have a specific chemical or mineralogical character. However, the poor spatial sampling of rocks from the deep lithosphere as

formerly at Research School of Earth Sciences, Australian National University Canberra.

The base of the lithosphere.

nodules in volcanic or kimberlitic magmas means that the definition of a chemical or mineralogical lithosphere on a continent-wide scale is impossible, although suitable nodule populations probably exist in some regions to allow such a definition to be applied locally. The lithosphere, being the rigid outer shell of the Earth, can be defined as that part of the Earth where heat transfer occurs only by conduction. A different though related definition is that the base of the lithosphere is defined by the isotherm at which rocks lose their rigidity and so will readily deform. Turcotte & Schubert (1982) suggest that this occurs at about 1300°C. These definitions are

difficult to apply because of the uncertainties in estimates of geothermal gradients. Geothermal gradients in eastern Australia have been estimated by extrapolating surface temperatures and heat flux to depth (Sass et al. 1976; Cull & Conley, 1983) and by measuring the pressures and temperatures of equilibrium of mineral suites in nodules recently brought to the surface by volcanic magmas (O' Reilly & Griffin, 1985). Estimates of the depths to the 1300°C isotherm vary between about 85 km and 150 km. Estimates in other geological provinces would have similar uncertainties, so that defining the base of the lithosphere using thermal properties is not very useful. The uncertainties in 9

SPK^^TAU

2 5 / A / 1 96

Fig. 1. Locations of blasts (asterisks; O - Ord river, B - Bass Strait; M - Maralinga; F - Mount Fitton, K - Kunnanulling) and permanent observatories and temporary seismograph stations (dots) used to study the Australian lithosphere. The station mnemonics correspond to the labels in Figs. 2, 3 and 6 and may differ slightly from those used in previous publications. The dashed line marks the approximate eastern limit of Precambrian shield.

Geol. Soc. Aust. Spec. Publ. 17,23-40.

estimating geothermal gradients are discussed elsewhere in this volume by Cull (1991). The other two principal definitions of the lithosphere are currently popular. In one, the lithosphere is the rigid outer part of the Earth able to support surface loads such as sedimentary basins and mountain ranges by bending and flexing, while the underlying asthenosphere, which is viscous or plastic, is able to flow to accommodate the flexing of the lithosphere. This definition therefore delimits the lithosphere in terms of the effects of stresses applied for long periods of time. However, with this definition estimates of lithosphere thickness vary, depending on the type of rheology assumed for the lithosphere (eg. elastic, visco-elastic, etc.). The rheology may be temperature dependent, in which case the lithosphere definition is again qualified by the uncertainties in geothermal gradients. The estimates are also dependent on the spatial distribution of the surface load. The fourth definition is based on seismic results, where the base of the lithosphere is defined as the depth to a low velocity layer. In this definition, the response of the Earth to a transient stress field is the property being used. The seismic low velocity zone results from a low shear modulus,

probably caused by partial melting. In a plate tectonic framework, the low velocity layer has traditionally been regarded as the place where the lithospheric plate decouples from the underlying mantle, although whether the Earth responds in a similar fashion to both short term (seismic) and long term (tectonic) stresses remains to be proved.

The seismic definition of the lithosphere has been popular for a long time, mainly because it is the one definition which can be tested by direct measurement. Seismic studies using both natural and controlled sources have been probing the base of the lithosphere for many years. This paper summarises the results of such studies in the Australian region. In Australia, the results have suggested that the structures in the upper mantle under the shield regions of central and western Australia are different from those under the Phanerozoic provinces of eastern Australia. In the following sections, the seismic models from the two areas are presented separately and then compared. The models are then tested using the existing teleseismic residual data. The results of the tests, when combined with our current knowledge of the chemistry of the deep mantle and the results of experimental petrology have implications for the likely depth at which the

Fig. 2. Record section and travel-time plot for the Ord River blasts. Open circles, eastern line through CTA (Fig. 1); large dots, stations to the west; small dots, stations to the north and south. The travel-time curves are for Simpson's ORD model (Simpson, 1973; see Fig. 5). Velocity of reduction 8.25 km/s.

The base of the lithosphere.

continent moves relative to the mantle. This may be at depths greater than the low velocity zone, so that in the future we may have to divorce our traditional concepts of the lithosphere, whether they are based on rheological, temperature, chemical or seismological models, from the notion of continents and their roots translating over the underlying mantle.

CENTRAL AND WESTERN AUSTRALIA The data from body wave (P and S) and surface wave experiments have been used to produce models of the lithosphere in the shield regions of central and western Australia. The locations of the experiments are shown and the data summarised in Figs. 1-4. Velocity/depth models resulting from the experiments are shown in Fig. 5. Body wave studies in central and western Australia Four seismic refraction experiments have probed the deep lithosphere in central and western Australia (Fig. 1). The first was the Ord River experiment in which large quarrying blasts in northern Australia were recorded on both temporary and permanent seismograph stations in central and northern Australia. A record section of the most comprehensive line which traversed southwards across the centre of Australia is shown in Fig. 2. Also shown are the travel times to the stations at other azimuths not represented by the record section. Denham et al. (1972) noted that the apparent Pn velocities were high (up to 8.27 km/s) with an increase to 8.85 km/s at a distance of 1400 km. Similarly, the shear wave velocity increased near 1400 km, from 4.59 km/s to 4.84 km/s. They also noted that arrivals at the stations along the eastern line (BAC to CTA, Fig. 1; and shown as open circles in Fig. 2) were consistently 1.5 to 2.0 seconds earlier than for the other lines. Simpson (1973) used both first and later arrivals to produce the ORD model of the upper mantle shown in Fig. 5. The travel-time curves for the ORD model are superimposed on the record section in Fig. 2. His P-wave model contains seismic discontinuities at depths near 85 km and 170 km. The

discontinuities were interpreted as sharp velocity increases because the data were too poor to discriminate between first order discontinuities and more complex discontinuities. His shear wave model has a low velocity layer between depths of 65 and 80 km. Simpson's P-wave models were later confirmed by the work of Hales & Rynn (1978), who used explosions fired offshore as seismic sources for a set of instruments deployed to the south towards central Australia. They observed velocities in the uppermost mantle of 8.2 km/s, rising to 8.5 km/s below a discontinuity at a depth near 75 km. However, shear wave studies from this experiment (Hales et al., 1980b) did not confirm the low velocity layer for shear waves between 65 and 80 km depth interpreted by Simpson. Rather, they interpreted an Sn velocity in the uppermost mantle of 4.6 km/s, rising to 4.72 km/s under the discontinuity at 75 km depth. The models for northern Australia are similar to those for the shield regions of southern Australia (Finlayson et al., 1974) at a different (east/west) azimuth. Energy from large blasts in Western Australia (K in Fig. 1) and South Australia (F) was recorded along an east/west profile across southern Australia. Some stations were also deployed along a line to the north of the South Australian blast in order to reverse the Ord profile. The data from this experiment were supplemented by the travel-times from the blasts at Maralinga (M) (Bolt etal., 1958). The seismic recording station positions are shown in Fig. 1 and the travel-times are plotted in Fig. 3. The data from the profile into central Australia confirm Simpson's model for northern Australia, with a seismic discontinuity at about 60 km depth (85 km in Simpson's model). This discontinuity does not seem to be present farther south along the east/west profile. Finlayson et al. (1974) found that although the data could not preclude a low velocity layer being present, they did not require it. Hales et al. (1980a) used recordings of earthquakes to the north of Australia at arrays of portable seismographs in central and northern Australia to study the structure of the upper mantle. They produced a model (CAP8), the upper part of which is shown in Fig. 5. It is similar to Simpson's in the upper mantle. It has discontinuities near 75

Geol. Soc. Aust. Spec. Publ. 17,23-40.

km (85 km in Simpson's model) and 200 km (170 km). The differences between the models probably results from the different source regions in the two

projects and small uncertainties in the origin times and hypocentre locations of the earthquakes used by Hales et al. (1980a). Hales et al. (1980a) were

Jr-(a)

-f-

Kunanalling

O Mt Fitton

• Ord Dam O Bass

Strait

200

400

600

800

1000

1200 Distance (km)

1400 . •UMB

Kunanalling

O Mt Fitton

• Ord Dam O Bass

Strait

1000

1200 Distance (km)

1400

MEK

Fig. 3. Reduced travel times as a function of distance from the TASS survey, (a) north/south line through central Australia between Mt. Fitton (F, Fig. 1) and Ord River (O); (b) east/west line across southern Australia between Mt. Fitton (F) and Kunnanulling (K). The travel-time curves are for the TASS2a and TASS la models shown in Fig. 5 (from Finlaystm etal., 1974).

The base of the lithosphere.

able to extend the CAP8 model to greater depths than any of the previous studies that had used explosive energy sources. The Lehmann discontinuity near 200 km depth is recognised world-wide (Anderson, 1979) and was expected. However, the Hales et al. experiment showed the unexpected result of a marked low velocity layer below the Lehmann discontinuity. The evidence for the low velocity layer comes from record sections of deep events which occurred in the West Irian, Banda Sea and Flores regions. Fig. 4 is a record section of an event at 160-170 km depth in the Banda Sea. It shows that the velocity of first arrivals increases sharply near 1000 km distance. The branch bb represents waves refracted below the Lehmann discontinuity. The amplitudes of the arrivals along branch bb decrease rapidly at distances beyond 1400 km. However, the amplitudes decay without losing their higher frequencies; this indicates that the decay is caused by the energy being refracted into a low velocity layer rather than being attenuated in a region of low Q. Additional support for the low velocity layer below the Lehmann discontinuity was provided by

arrivals recorded in the Pilbara region of Western Australia from shallow earthquakes to the south of Sumba Island (Drummond et al., 1982). As predicted by the CAP8 model with a surface focus source, record sections of these events showed later arrivals with observed velocities between 8.75 and 8.83 km/s whose amplitudes decreased at distances beyond about 1500 km. Thus none of the models derived from the analysis of body waves in northern, central or southwestern Australia contain low velocity layers above the Lehmann discontinuity which seems to occur at about 200 km depth. Rather, the shallowest low velocity layer appears to be at depths of about 225 km. Surface wave studies in central and western Australia A number of surface wave studies of the Australian continent have been made (eg. Bolt & Naizi, 1964; Thomas, 1969). The most comprehensive was by Goncz et al. (1975) and Goncz & Cleary (1976) who used the two station pair method

• Observation site X Epicentre

1500 Distance (km)

2000 25/A/199

Fig. 4. Record section of an earthquake which occurred on October 14, 1975 at a depth of 167 km under the Banda Sea north of Australia and recorded on a north/south line of portable seismographs in northern Australia. Amplitudes on branch 'bb' decrease with distance without losing their high frequency component, implying that the energy is refracted into a low velocity layer, in this case near 220 km depth (after Hales et al., 1980a).

Geol. Soc. Aust. Spec. Publ. 17,23-40. of Dziewonski & Hales (1972) to analyse the longer period fundamental mode surface wave energy which penetrates to greater depths. In this method, the events to be analysed are chosen to be on the same great circle as two long period three component stations. The cross correlated output of the signals received at the two stations is equivalent to surface waves received at the second station from a point source at the first station. The method not only allows source uncertainties to be removed but it also allows the outputs from a number of events to be stacked in order to improve the signal to noise ratio. Using data from the long period World Wide Standard Network (WWSN) stations at Charters Towers, Adelaide and Mundaring (CTA, ADE, MUN, respectively, Fig. 1), three separate shield paths were analysed (Goncz et al., 1975). These paths showed similar dispersion so that the group velocities were averaged and inverted to obtain It

'I!

representative shear wave models, two of which are illustrated in Fig. 5. They are very similar to the shear wave model of Hales et al. (1980b) derived from body wave energy from explosions at sea off northern Australia and recorded in central and northern Australia. Within the limits of resolution none of the models required significant low velocity layers above 200 km depth in order to satisfy the data.

EASTERN AUSTRALIA Models of the lithosphere in eastern Australia have been produced from both body wave (P) and surface wave studies. The data are summarised in Figs. 6 and 7 and the velocity/depth models are shown in Fig. 8.

:ir: \ Cap

CAP8 —

ORD-Simpson TASS

TASS2A N. W.

VV7 W2

Shelf

CAP-shear

Velocity (km/s)

Fig. 5. Velocity/depth models for the shield regions of Australia derived from seismic body wave refraction experiments and inversion of surface wave dispersion data. Sources are: CAP8 - Hales et al. (1980a); ORD - Simpson (1973); TASS la, TASS2a - Finlayson etal. (1974); NW Shelf - Hales & Rynn (1978); Wl, W2 - Goncz & Cleary (1976); and CAP Shear - Hales et al. (1980b).

The base of the lithosphere.

Fig. 6. Composite record section and travel-time plot of the recordings of the blasts in Bass Strait (B, Fig. 1) and at Mt. Fitton (F) along the line of seismographs deployed between them. The travel time curves are for the compressional wave model of Muirhead et al. (1977), and are identical to those for the new compressional wave model for eastern Australia proposed in this paper and shown in Fig. 8. Bold-faced mnemonics indicate travel times and traces from the blast at Mt. Fitton (F, Fig. 1); other traces and travel times are for the Bass Strait blast (B).

Body wave studies in eastern Australia Only one seismic refraction experiment in eastern Australia recorded energy over a distance range sufficiently large to ensure penetration to depths of 100 km or more. In 1972, explosions at Mt. Fitton in South Australia (F, Fig. 1) and at sea in Bass Strait (B) were recorded along a line of temporary and permanent seismograph stations at distances up to 1200 km from the blasts (Fig. 1). A composite record section for the two explosions and the traveltime data are shown in Fig. 6. Muirhead et al. (1977) developed two models from the data. One had velocity discontinuities at depths near 100 and 200 km and did not include a low velocity layer, but it did not adequately model a large later arrival (labelled D in Fig. 6) which is interpreted as the retrograde cusp from the Lehmann discontinuity near 200 km depth. In their alternative model shown in Fig. 8, the Lehmann discontinuity is slightly shallower and is overlain by a low velocity channel. This model both reproduces the traveltime of the energy at cusp D and predicts the distance at which it is recorded. It has a low velocity channel with a constant velocity of 8.1 km/s, which constrained the top of the low velocity layer to a depth of 160 km. However, Muirhead et al. noted that there was considerable uncertainty in both the structure and position of the low velocity layer and that the data could have been modelled equally well by a thicker channel with less velocity contrast, or

by a thinner, higher contrast channel. This is discussed later. Surface wave studies in eastern Australia Goncz & Cleary (1976) derived a suite of shear wave models for eastern Australia. All of their models fit the dispersion curves within the specified error bounds and so no distinction can be made between them. The two shown in Fig. 8 indicate the range of possible velocity distributions within the low velocity zones. Goncz & Cleary chose model El as being a representative surface wave model for eastern Australia. Another surface wave study of eastern Australia was undertaken by Mills & Fitch (1977). They produced models of the crust and upper mantle by a controlled Monte Carlo inversion of the group velocities of long period data recorded at CTAfrom the 1973 Picton earthquake (just NE of INV, Fig. 1). Their model CTA2S1 is shown in Fig. 8. Their models are considerably different from those derived by Goncz & Cleary (1976); they do not have a low velocity zone in the depth range from 100 to 200 km, and the subcrustal shear wave velocities are anomalously low. A measure of the low velocities in the upper mantle in the Mills & Fitch (1977) models is given by the calculated travel times for the Picton - CTA path which in all cases is much larger than the observed time.

Geol Soc. Aust. Spec. Publ 17,23-40. CTA2S1 produces the time closest to the observed travel time, but it is still 6 s too late.

(a)

A surface wave study of eastern Australia by Ellis & Denham (1985) who analysed surface waves for paths between Victoria and Alice Springs (ASP, Fig. 1) produced the Ellis-Denham model in Fig. 8. This model is similar to the Goncz-Cleary El model although the velocity- thickness product of their low velocity channel is not as large. The surface wave studies and the single seismic refraction profile available in eastern Australia all indicate the presence of a significant low velocity channel in eastern Australia. However, the models from the surface wave studies in eastern Australia show considerable diversity, and some are alsQ inconsistent with the P-wave model of Muirhead et

50 Period

Distance (km) Event Depth (km) Observed Time (s) Model times

100

Period ( s ) I

Mills

Et Fitch

(95%

confidence

25/A/202

— data set #

CTA-RIV

CTA-TAU New

model

CTA2S1

Fig. 7(a). Phase velocity dispersion curves for the fundamental mode Rayleigh waves for three paths entirely within the Phanerozoic Tasman Orogenic Belt, and the average curve EA1 for eastern Australia (all from Goncz et al., 1975). The dots show the values calculated for the new model shown in Fig. 8. Fig. 7(b). Group velocities for fundamental mode Rayleigh waves for the Picton-CTA path with 95% confidence limits (Data Set #2, Mills & Fitch, 1977). Also shown are the group velocities for the three paths in the Tasman Orogenic Belt whose phase velocities are shown in (a) and the calculated group velocities for model CTA2S1 of Mills & Fitch and the new model in Fig. 8.

J-B PEM El CTA2FS1 CTA2FS2 CTA2S1 New Model

Key Jeffreys & Bullen (1940) Dziewonski et al. (1975) Goncz & Cleary (1976)

RIV-TAU

•

370.7*

limits)

£4 /

1606 27 369.9±4 370.1 347.4 367.8 386.8 383.3 376.3

J-B PEM El CTA2FS1 CTA2FS2 Mills & Fitch (1977) CTAS1 Ray bottoms between 205 and 210 km, ie. below the low velocity layer. Ray bottoms at about 80 km depth, ie. above the low velocity layer. *

Table 1. Travel times in seconds of short period S for the Picton earthquake recorder at Charters Towers (CTA).

The base of the lithosphere.

Depth 0.0 20.0 20.0 35.0 35.0 100.0 100.0 120.0 120.0 190.0 190.0 225.0

Vp 6.15 6.15 6.75 6.75 7.98 7.98 8.36 8.36 8.23 8.23 8.72 8.72

Vs Vp/Vs 3.50 1.76 3.50 3.725 1.81 3.725 4.333 1.84 4.333 4.544 1.84 4.544 4.050 2.03 4.050 4.744 1.84 4.744

(Depth is in kilometres; velocities are given in km/s)

Table 2. New model for eastern Australia.

al. (1977). This has led us to propose a new model for eastern Australia which is largely consistent with all data sets.

A new model for eastern Australia The principal difference to be reconciled between the models of the mantle in eastern Australia is that between the surface wave and body wave models. If the discontinuities in the P-wave model have any petrological or chemical significance, the shear wave models should have discontinuities at or near the same depths. The P-wave model of Muirhead et al. (1977) has a velocity increase at about 100 km depth, near the depth at which the low velocity layer starts in most of the surface wave models. Muirhead et al. (1977) found that observations of both the Bass Strait and Mt. Fitton explosions at distances greater than 900 km, including those to all of the Australian

T 100

ftJ hi: I i l I

New

Tf 1-4—fj 200

Model

Muirhead

et.

£ 2

CTA2S1 Ellis

Denham

EAST

10 Velocity (km/s)

25/A/203

Fig. 8. Velocity depth models for eastern Australia. The sources are: the new model for eastern Australia derived in this study; Muirhead et al1977 - shear wave model derived from the P-wave model assuming Vp/Vs=1.73; El, E2 - Goncz & Cleary (1976); CTA2S1 - Mills & Fitch (1977); and Ellis & Denham (1985).

Geol. Soc. Aust. Spec. Publ. 17,23-40.

seismic observatories as far away as Charters Towers (CTA), have arrival times which, when corrected for realistic crustal variations, are too early to lie on the Pn travel-time branch. Thus, there appears to be no doubt that a P wave velocity increase at a depth of about 100 km depth is a feature of the eastern Australian upper mantle structure, although the exact depth to the velocity increase in the body wave model may be uncertain. The shear wave model should also have a high velocity layer which starts near 100 km depth and overlies a low velocity layer. A new model which fits the P-wave travel times equally as well as the models of Muirhead et al. (1977) was produced. It is listed in Table 2 and drawn in Fig. 8. The currently available P-wave data do not discriminate between first order discontinuities and more complex velocity structures; all the discontinuities are therefore modelled as first order. The model has a velocity increase at 100 km depth, but the high velocity layer is thin and overlies a low velocity layer at about 120 km depth. The velocity - thickness product of the low velocity zone is the same as in the Muirhead et al. (1977) model, but the low velocity zone is thicker and has less velocity contrast. Also shown in Fig. 8 is a shear wave model derived by keeping the depths to the discontinuities the same as those in the P-wave model and varying the Vp/Vs ratio until the calculated surface wave dispersion curve matched the observed curves of Goncz et al. (1975). The Goncz et al. dispersion curves were favoured over those from the other studies because they were derived using the two station technique which should produce more reliable estimates, especially at longer periods, than the single station method, and because they were in a form which allowed averaging of data from several paths. The curves for three paths which fell entirely within the Phanerozoic Tasman Orogenic Province are shown in Fig. 7a. Also shown is the Goncz et al. curve EA1, which they derived by averaging the dispersion curves for these paths and others that fell mostly in the Tasman Province. The dots indicate the calculated phase velocities for the new model shown in Fig. 8 and listed in Table 2. The calculated phase velocities fall within the goodness of fit

criterion used by Goncz & Cleary (1976) to accept or reject models from their Monte Carlo inversion of the EA1 curve. Mills & Fitch (1977) published group velocity curves for the Picton earthquake recordings at CTA. Their data set #2, with 95% confidence limits shown, is plotted in Fig. 7b. Also shown are the group velocity dispersion curves for the three Tasman Orogenic Province paths whose phase velocity curves are shown in Fig. 7a. The dispersion curves show considerable scatter. Note, for example the difference (0.1 km/s at 45 s) between the curves for Picton to CTA (Mills & Fitch, 1977) and CTA to RIV (Goncz et al., 1975), which are almost identical paths. The differences are probably an indication of the low resolving power, especially at longer periods, of surface wave techniques in general, and the single station method compared to the two station method in particular. The calculated group velocity curve for the new model proposed in this paper are also shown, along with that for model CTA2S1 of Mills & Fitch (1977). The new model does not appear to fit the data as well as CTA2S1, although its dispersion curve falls within the 95% confidence limits of the data and lies within the family of group velocity dispersion curves for eastern Australia. One very important test for any model is whether it reproduces the observed travel times. Table 1 compares the calculated travel time for the new model for the Picton to CTA path with the observed time and the times for the Jeffreys & Bullen (1940) tables, model El of Goncz & Cleary (1976), several of the models of Mills & Fitch (1977) and the parametrically simple Earth model for continental areas produced by Dziewonski et al. (1975) (PEM in Table 1). The new model reproduces the travel time better than any of the models of Mills & Fitch (1977). It should be emphasised that many models could have been produced to fit the surface wave dispersion curves equally as well as the model presented here. However, this model is considered an advance over all previously published models because it (1) fits the observed P-wave data from the Bass Strait to Mount Fitton refraction profile,

The base of the lithosphere.

(2) fits the Goncz et al (1975) dispersion curves within the goodness of fit criterion defined by Goncz & Cleary (1976), (3) is consistent with the Mills & Fitch (1977) group velocity data within their 95% confidence limits, (4) reproduces the observed travel time from Picton to CTA, and (5) by virtue of its method of derivation, has the discontinuities at the same depths in the P- and S-wave models.

The residuals are therefore consistent with the presence of lower velocities in the upper mantle in the Phanerozoic regions of eastern Australia compared to the shield regions farther west. The largest positive residuals in eastern Australia all lie in the southeast corner where the velocities must be the lowest. Similar trends in travel time residuals have been observed elsewhere. Residuals from Precambrian regions are generally smaller (negative) than those of Phanerozoic platforms, which in turn are less than those of Phanerozoic orogenic zones (Jordan, 1981b).

DISCUSSION TELESEISMIC TRAVEL-TIME RESIDUALS Drummond et al (1989) produced a map of residuals of teleseismic travel-times relative to the global travel-time curve of Dziewonski & Anderson (1983). Teleseismic travel-time residuals are a indication of the time that energy with near-vertical paths takes to travel from deep in the Earth to the surface. Differences in residuals from province to province therefore reflect differences in structure at depth. With the exception of local variations in the residual field such as those reported by Lambeck & Penny (1984), most of central and western Australia has negative residuals. The residuals become positive and progressively higher towards the east. The boundary between Phanerozoic eastern Australia and the older shield basement regions of central and western Australia corresponds approximately to the easternmost -0.4 s contour in Fig. 9. The change from shield to non-shield basement correlates with an abrupt change of 0.44 s along a line through the stations between WRA and CTA in Fig. 1 (Wright et aly 1985). In the south along a line between F and B (Fig. 1), the change from west to east is not so abrupt, but the difference in the level of residuals from west to east is approximately 1.0 s (Cleary et al, 1972). These relative residuals cannot be explained in terms of crustal structure. Most of the effect must come from the mantle with a large proportion probably resulting from differences in upper mantle velocities below 200 km depth (cf. Finlayson, 1982).

Most attempts to describe the seismological characterictics of the upper mantle in terms of a penological model seem to be based on the creation of oceanic lithosphere at mid-ocean ridges, the phase changes it undergoes as it moves away from the ridges and becomes older and colder, and its eventual subduction at ocean trenches (eg. Wyllie, 1973; Green & Liebermann, 1976). The evolution of the deep lithosphere under continents must be more complex because the greater age of the continents and their roots allows more time for deep continental lithosphere to be modified by repeated

25/A/204

Fig. 9. Contour map of teleseismic travel time residuals for Australia relative to the Dziewonski & Anderson (1983) travel time curves. Contour interval 0.2 s. Contours are dashed in areas of sparse data (Drummond et al., 1989).

Geol Soc. Aust. Spec. Publ. 17,23-40.

magma extraction. Nevertheless attempts such as those of Green & Liebermann (1976) to account for the observed seismological models of the oceanic lithoshpere in terms of a peridotitic upper mantle undergoing phase changes and depletion are a useful starting point from which to study the upper mantle of the continents. Ringwood (1962) proposed that the primitive mantle is composed predominantly of olivine and pyroxene - his 'pyrolite' model. The precise mineralogy present at any depth is dependent on the pressure and temperature as well as any chemical heterogeneities (Green & Liebermann, 1976). At about 50 km under old oceans, spinel will transform to garnet, causing a small increase in the P-wave velocity (0.13 km/s) and a smaller increase in the S-wave velocity (0.03 km/s). At greater depths, probably near 85-90 km, the geotherms intersect the pyrolite solidus and partial melting will occur. The degree of partial melting is critically dependent on the amount of water present and in turn determines the composition of the magma (Green, 1973). Magma flux in the zone of partial melting could give rise to a vertical chemical zonation in the melt, and this would cause a zonation in the degree of partial melting (Green & Liebermann, 1976). The zone of partial melting would have low seismic velocities because the melt fraction would cause a severe drop in the shear modulus. The effect on the bulk modulus would be smaller so that the low velocity zone would be more pronounced for shear waves than for P- waves (Matsushima, 1986; Green & Liebermann, 1976). If the low velocity zone is chemically zoned, it would be more pronounced at the top (lower velocities) than at the bottom. A comparison of the Green & Liebermann (1976) penological model with the seismic models from both P- and S-wave observations of the upper mantle under Australia sheds some light on the likely nature of the upper mantle. The lithosphere under Australia is much older, especially in the shield regions, more depleted and refractory, and in places much colder than oceanic lithosphere. It is probably also much more chemically heterogeneous (eg. O'Reilly & Griffin, 1984). The models for the upper mantle in shield regions in Australia have several features in com-

mon, although they differ in detail. They have subcrustal velocities between 8.05 and 8.23 km/s (Fig. 5). At depths between 60 and 80 km, all models except TASS1 contain velocity increases to between 8.25 and 8.4 km/s. In the profiles which sampled the mantle below 150 km, there is an additional velocity increase to between 8.53 and 8.62 km/s at depths between 160 and 200 km. The shear wave models have equivalent discontinuities.

Finlayson (1982) suggested that the differences between the various models from the shield may represent real differences in the upper mantle beneath the Precambrian region. This is probably at least partly true. The differences in the arrival times from the Ord River blasts at stations to the south compared to those to the east and west (Fig. 2) possibly results from different depths to the discontinuity between 60 and 80 km. However, two additional factors probably also contribute to the differences and must be considered before any real significance is placed on them. Firstly, the refraction profiles were undertaken at what must be considered a reconnaissance scale, and all depths are not sampled with reversed data. Models based on earthquake data must have an added uncertainty, especially in the depths to the discontinuities, due to likely errors in the earthquake source parameters. Secondly, all of the models were based on the assumption that the Earth is horizontally stratified. To a first approximation this may be true for the major discontinuities in the upper mantle but it is not so for the crust/mantle boundary, and large variations in crustal structure and the depth to the crust/mantle boundary can have a serious effect on the interpreted depths and apparent velocities of discontinuities in the upper mantle. Thus the variations in the upper mantle models in Fig. 5 may reflect inadequte sampling of the structures of the upper mantle, coupled with the effects of lateral variations in the structures in and around the crust/mantle boundary. Nevertheless, the shield region does seem to be characterised by seismic discontinuities near 60-80 km and 160-200 km. The data can be satisfied without the need for a low velocity layer between these discontinuities, although this does not mean that a low velocity layer does not exist. However, if it is present, the velocity within it must be so close to the 'normal' velocities

The base of the lithosphere.

above and below it that further detailed experiments will be necessary to define it. In eastern Australia, the previous seismic models of the crust and upper mantle have now been superseded by the new model shown in Fig. 8 and listed in Table 2. The shear wave model is based on surface wave data averaged over all of eastern Australia and the velocities in each of the layers of the model are therefore very much vertical and lateral averages. The shear wave velocities above the low velocity zone appear to be low compared to those farther west in the shield, and to an apparent Sn velocity of about 4.6 km/s observed for earthquakes recorded at seismograph stations in southeast Australia. The amplitudes of Sn arrivals in southeast Australia are generally small and emergent (Cleary & Doyle, 1962), suggesting that shear waves are either attenuated in a region of low Q in the uppermost mantle, or travel through zones where the ratio Vs/r does not increase with depth but rather remains constant or decreases ('r'is the distance from the centre of the Earth) - ie., the energy encounters a low velocity zone. Thus the effects of the high geothermal gradient in eastern Australia on the shear modulus may extend over a broader depth range and to shallower depths than the effects on the bulk modulus. The results of Matsushima (1986) predict this. Hales et al. (1975) correlated the seismic discontinuity observed at 65-70 km depth under central and western Australia with the spinel/garnet transition. The transition may also explain the boundary at 100 km depth under eastern Australia. However, the seismic velocity under the boundary in central and western Australia is 8.28 km/s (an increase of 0.24 km/s across the boundary) and under eastern Australia it is 8.36 km/s (an increase of 0.38 km/s). Hales et al. (1975) estimated that up to 20% garnet would be required to produce a velocity of 8.4 km/s at 80-100 km depth, but the amount of garnet available from the pyrolite model is likely to be less - Green & Liebermann (1976) estimate 14% garnet giving an increase in P-wave velocity of only 0.12 km/s. The pattern of teleseismic travel-time residuals broadly correlates with changes in surface heat flow (Cull & Conley, 1983). Residuals are lowest in areas of low surface heat flow and greatest where

the heat flow is high. Surface heat flow in the shield regions is generally below world average but in the Phanerozoic provinces of eastern Australia it is mostly higher. This implies that the geothermal gradient is lower in the shield. Temperature affects seismic velocities in several ways. At sub-solidus temperatures, increasing temperature causes a decrease in seismic velocity. At above-solidus temperatures, the presence of even a small portion of partial melt can cause a large drop in the shear modulus, resulting in marked drops in seismic velocities, with shear wave velocities affected more than compressional wave velocities (Jackson, 1991). In central and western Australia, predicted geotherms (Cull & Conley, 1983) are unlikely to intersect the pyrolite solidus (Green & Liebermann, 1976), especially if the mantle is already depleted in volatiles, and no partial melting will occur. Thus no low velocity layer is expected. In eastern Australia, the geothermal gradients are higher. Estimated geothermal gradients (Sass et al., 1976; Cull & Conley, 1983) projected to upper mantle depths are likely to intersect the wet pyrolite solidus (Cull, 1991) causing partial melting and a low velocity zone. The base of the low velocity layer is less easily explained and Green & Liebermann (1976) summarise three possibilities: (i) the geotherm recrosses the watersaturated solidus, (ii) the water content of the mantle decreases below about 150 km depth, and (iii) water at depths below the low velocity zone is held in as yet undiscovered hydrous phases. These models assume a primitive mantle of constant composition. They do not explain the high velocities below the Lehmann discontinuity which seems ubiquitous under both continents and oceans (Drummond et al., 1982). In trying to explain the discontinuity, Leven et al. (1981) canvassed several models including changes of mineralogy (phase changes) and changes of chemistry. They preferred an explanation in which the discontinuity represents the zone of decoupling of the Australian plate from the mantle below, with the high velocities resulting from alignment of olivine and pyroxene crystals to form an anisotropic layer with high velocities observed along a north/south azimuth. This model is at variance with the observation of the Lehmann discontinuity along an

Geol. Soc. Aust. Spec. Publ. 17,23-40.

east/west azimuth across southern Australia (TASSla model, Fig. 5). Seismically, therefore, the upper mantle under Australia is both vertically and laterally heterogeneous. This reflects the different thermal regimes, from the colder shield regions to the hotter younger orogenic regions. It also reflects the different compositional maturity of the upper mantle between old and young regions. The upper mantle will evolve chemically from a primitive composition near that of pyrolite to a refractory lherzolite or harzburgite as partial melts form and are removed as basaltic magmas. The process will stop because (i) the removal of partial melts leaves behind a refractory residue which is harder to melt, and (ii) the removal of heat through conduction and as latent heat of fusion in magmas will cool the mantle and push the geotherm below the solidus. The development of this compositional and thermal maturity throughout Australia can be traced using the distribution in space and time of alkaline igneous rocks brought to the surface either as alkali basalts or as kimberlites, lamproites or carbonatites. The alkaline magmas are late stage partial melts of a refractory source and in Australia their emplacement is broadly contemporaneous with the cratonisation of the regions where they are found (Jaques et al, 1985). (Some exceptions occur. In the Kimberley region of northwest Australia, some regions were cratonised as early as the Proterozoic but are intruded by Tertiary lamproites, suggesting that although large areas of central and western Australia are apparently stabilised, processes are still at work deep in the mantle moving incompatible elements, water and perhaps CO2 into pockets where the solidus temperatures locally become lowered and partial melts can develop.) The cold, depleted upper mantle forms a thermal and chemical boundary layer (Jordan, 1975; 1981a). The partial melting and magma removal cause the upper mantle to become depleted in the key elements Al, Fe and Ca, rendering its density less than the surrounding mantle. Because of its lower density, the depleted upper mantle cannot mix easily with the rest of the mantle and becomes isolated. The base of the chemical boundary layer will move downwards with time as more of the

upper mantle is depleted. Jordan estimates that it can reach depths greater that 150 km. Clearly this is so in Australia. Depletion of the upper mantle in the east must be continuing today in the low velocity zone, so that in eastern Australia the chemically depleted upper mantle must extend at least to the base of the low velocity layer at depths as great as 200 km. In central and western Australia, the region between 100 and 200 km depth now lies below the solidus but must once have been sufficiently hot to have produced partial melts and been depleted. The differences in teleseismic travel time residuals between the shield regions and eastern Australia cannot be explained by the low velocities above 200 km depth in eastern Australia. Rather, they imply that the mantle below 200 km depth in central and western Australia has higher velocities than in the east (Finalyson, 1982). This is consistent with the removal of partial melts rich in iron relative to magnesium (Jordan, 1979) from depths greater than 200 km so that the zone of depletion under the shield probably extends to depths beyond 200 km. The chemically depleted boundary layer in Australia is thicker than the mechanically strong outer layer or lithosphere. The depth to the seismic low velocity layer suggests that the lithosphere is 100-120 km thick in eastern Australia, but in central and western Australia it is over 200 km thick. The continent is unlikely to translate across the underlying mantle at the base of the rigid outer part of the Earth. It probably translates at or below the bottom of the zone of chemical depletion. Two reasons can be given why this should be so. Firstly, Jordan's (1981) calculations show that the mantle above the zone of chemical depletion will not mix with the underlying mantle. Secondly, the upper mantle under continents is progressively depleted as the age of the geological provinces increase. The depletion occurs in the low velocity zone. If the low velocity zone was not fixed to the base of the continents, continents would be constantly drifting across new mantle, the source regions of magmas reaching the surface would not become progressively more refractory as the provinces get older, and volcanism would not stop as the provinces cratonised.

The base of the lithosphere.

ACKNOWLEDGEMENTS We thank our colleagues at the Research School of Earth Sciences, Australian National University, and the Bureau of Mineral Resources for their many helpful discussions and comments. Especially, we thank Prof. Anton Hales and Dr. Brian Kennett at the ANU and Dr. A.L. Jaques at the BMR. Barry Drummond publishes with the permission of the Director, Bureau of Mineral Resources. REFERENCES ANDERSON D.L. 1979. The deep structure of continents. Journal of Geophysical Research 84,7555-7560. BOLT B.A., DOYLE H.A., & SUTTON D.J. 1958. Seismic observations from the 1956 atomic explosions in Australia. Geophysical Journal of the Royal astronomical Society 1, 135-145. BOLT B.A. & NAIZI M. 1964. Dispersion of Rayliegh waves across Australia. Geophysical Journal ofthe Royal astronomical Society 9, 21-35. CLEARY J.R. & DOYLE H.A., 1962. Application of a seismograph network and electronic computer in near earthquake studies. Seismological Society of America, Bulletin 52, 673-682. CULL J.R 1991. Geothermal gradients in Australia. In Drummond B.J. ed. The Australian Lithosphere. Geol. Soc. Aust. Spec. Publ. 17, 147-156. CULL J.P. & CONLEY D. 1983. Geothermal gradients and heatflowin Australian sedimentary basins. BMR Journal ofAustralian Geology & Geophysics 8, 329-337. DENHAM D., SIMPSON D.W., GREGSON P.J. & SUTTON

D.J. 1972. Travel times and amplitudes from explosions in northern Australia. Geophysical Journal of the Royal astronomical Society 28, 225-235. DRUMMOND B.J., MUIRHEAD K.J. & HALES A.L. 1982.

Evidence for a seismic discontinuity near 200 km depth under a continental margin. Geophysical Journal of the Royal astronomical Society 70, 67-77.

DRUMMOND B.J., MUIRHEAD K.J., WELLMAN P. &

WRIGHT, C. 1989. A teleseismic travel-time residual map of the Australian continent. BMR Journal of Australian Geology & Geophysics, 11,101-105. DZIEWONSKI A.M. & ANDERSON D.L. 1983. Travel times and station corrections for P waves at teleseismic distances. Journal of Geophysical Research 88 3295-3314. y

DZIEWONSKI A.M. & HALES A.L. 1972. Numerical analysis of dispersed seismic waves. Methods of Computational Physics 11,39-85, Academic Press, New York & London. DZIEWONSKI A.M., HALES A.L. & LAPWOOD E.R. 1975.

Parametrically simple Earth models consistent with geophysical data. Physics of the Earth and Planetary Interiors 10, 14-48.

ELLIS R.M. & DENHAM D. 1985. Structure of the crust and upper mantle beneath Australia from Rayleigh- and Love-wave observations. Physics of the Earth and Planetary Interiors 38,224-234. RNLAYSON D.M. 1982. Geophysical differences in the

lithosphere between Phanerozoic and Precambrian Australia. Tectonophysics 84,287-312. FINLAYSON D.M., CULL J.P. & DRUMMOND B.J. 1974.

Upper mantle structure from the trans- Australia seismic refraction data. Geological Society ofAustralia Journal

21, 447-458.

GONCZ J.H. & CLEARY J.R. 1976. Variations in the structure of the upper mantle beneath Australia, from Rayleigh wave observations. Geophysical Journal of the Royal astronomical Society 44, 507-516. GONCZ J.H., HALES A.L. & MUIRHEAD K.J. 1975.

Analysis to extended periods of Rayleigh and Love wave dispersion across Australia,Geophysical Journal of the Royal astronomical Society 41, 81-105. GREEN D.H. 1973. Contrasted melting relations in a pyrolite upper mantle under mid-ocean ridge, stable crust and island arc environments. Tectonophysics 17, 285-297.

GREEN D.H. & LIEBERMANN R.C. 1976. Phase equilibria

and elastic properties of a pyrolite model for the oceanic upper mantle. Tectonophysics 32, 61-92.

Geol Soc. Aust. Spec. Publ 17,23-40. HALES A.L., MUIRHEAD K.J., RYNN J.M.W. and GETTRUST J.F. 1975. Upper mantle travel times in

Australia - a preliminary report. Physics of the Earth and Planetary Interiors 11, 109-118. HALES A.L. & RYNN J.M.W. 1978. A long range

controlled source seismic profile in northern Australia. Geophysical Journal of the Royal astronomical Society 55, 633-644. HALES A.L., MUIRHEAD K.J. & RYNN J.M.W. 1980a. A compressional velocity distribution for the upper mantle. Tectonophysics 63, 309-348.

1980b. Crust and upper mantle shear velocities from controlled sources. Geophysical Journal of the Royal astronomical Society 63, 659-670. HALES A.L., MUIRHEAD K.J. & RYNN J . M . W

JACKSON, Ian

1991. Petrophysical interpretation of seismological models for the continental lithosphere. In Dfummond B.J. ed. The Australian Lithosphere. Geol. Soc. Aust. Spec. Publ. 17, 81-114. JAQUES A.L., CREASER R . A . , FERGUSON J. & SMITH C . B .

1985. A review of the alkaline rocks of Australia. In J. Bristow ed. Alkaline and alkaline-ultrabasic rocks and their xenoliths. Geological Society of South Africa, Transactions 88,311-334.

JEFFREYS H. & BULLEN K.E. 1940. Seismological tables.

British Association for the Advancement of Science, Gray Milne Trust, London. 50p. JORDAN T.H. 1975. The continental tectosphere. Reviews

of Geophysics and Space Physics 13,1-12.

JORDAN T.H. 1979. Mineralogies, densities and seismic

velocities of garnet lherzolites and their geophysical implications. In Boyd F.R. & Meyer H.O.A. eds The mantle sample: inclusions in kimberlites and other volcanics. Proceedings of the Second International Kimberlite Conference, II, American Geophysical Union, 1-14. JORDAN T.H. 1981a. Continents as a chemical boundary

layer. Philosophical Transactions of the Royal Society of London A3Q1, 359-373. JORDAN T.H. 1981b. Global tectonic regionalisation of

seismological data analysis. Seismological Society of America, Bulletin 74, 1131-1141.

LAMBECK K. & PENNEY C. 1984. Teleseismic travel time anomalies and crustal structure in central Australia. Physics of the Earth and Planetary Interiors 34,46-56. LEVEN J.H., JACKSON I. & RINGWOOD A.E. 1981. Upper

mantle anisotropy and lithospheric decoupling. Nature 289,234-239.

MATSUSHIMA S. 1986. The effects of frequency on the elastic wave velocity in rocks at high temperatures under pressure. Tectonophysics 124, 239-259. MILLS J.M. & FITCH T.J. 1977. Thrust faulting and crust-upper mantle structure in east Australia. Geophysical Journal of the Royal Astronomical Society 48,351-384. MUIRHEAD K.J., CLEARY J.R. & FINLAYSON D . M . 1977.

A long-range seismic profile in southeastern Australia. Geophysical Journal of the Royal astronomical Society 48, 509-520. S.Y. & GRIFFIN W.L. 1984. Sr isotopic heterogeneity in primitive basaltic rocks, southeastern Australia: correlation with mantle metasomatism. Contributions tc Mineralogy and Petrology 87, 220-230.

O'REILLY

O'REILLY S.Y. & GRIFFIN W.L. 1985. A xenolith-derived

geotherm for southeastern Australia and its geophysical implications. Tectonophysics 111, 41-63.

RINGWOOD A.E. 1962. A model for the upper mantle. Journal of Geophysical Research 67, 857-867. SASS J.H., JAEGER J.C. & MUNROE R.J. 1976. Heat flow

and near-surface radioactivity in the Australian continental crust. United States Deptartment of the Interior Geological Survey Open-File Report 7 6 - 2 5 0 SIMPSON D.W. 1973. P wave velocity structure in the upper mantle in the Australian region. Ph.D. thesis, Australian National University, 212p, (unpublished). THOMAS L. 1969. Rayleigh wave dispersion. Bulletin of

the Seismological Society of America 59, 167-182.

TURCOTTE D.L. & SCHUBERT G., 1982.

Geodynamics, Wiley, New York, 450p. U.S. Geodynamics Committee, 1983, The lithosphere. Report of a workshop. National Academy Press, Washington.

The base of the lithosphere.

WRIGHT C., MUIRHEAD K.J. & DIXON A.E., 1985. T h e P

wave velocity near the base of the mantle. Journal of Geophysical Research 90,623-634.

WYLLIE P.J., 1973. Experimental petrology and global

tectonics - a preview. Tectonophysics 17, 189-209.

Muirhead K.J. & Drummond B J. Bureau of Mineral Resources, Geology and Geophisics, GPO Box 378 Canberra ACT 2601 AUSTRALIA

Geol. Soc. Aust. Spec. Publ. 17,41-58.

Velocity Variations and Isostatic Compensation in the Australian Region J. C. Dooley Bureau of Mineral Resources, Geology and Geophisics, Canberra, A.C.T., Australia. Densities of crustal layers inferred from seismic refraction velocities in Australia are used to calculate the mass of a crustal column at the site of each refraction survey. The variation of crustal mass shows that isostatic compensation is not complete at the base of the crust, and that there should be compensating mass variations in the upper mantle. Subcrustal densities inferred from Pn velocities do not in general give the required compensation; however intramantle seismological studies suggest density differences which could lead to compensation at a depth of about 110 km in eastern Australia; they also indicate a low S-velocity layer at about this depth, which could represent the asthenosphere for compensation. However the shield areas of Western Australia are anomalous, with a heavy crust underlain by lithospheric material with high velocities, which suggest a relatively high-density upper mantle instead of the low density needed for compensation of the heavy crust. There is moreover little or no evidence for a low S-velocity layer in the west. The present paper is an update of a previous investigation, including new data from the Lachlan Fold Belt, the Pilbara-Hamersley region, the McArthur Basin, and the Central Eromanga Basin, and revised models for many of the areas used previously. It is probable that the rocks under Archaean shield areas have been depleted of the iron-rich component with low melting point, leaving a refractory residue of dunite or peridotite with lower mean atomic weight; Birch has shown that such rocks may have a lower density for a given seismic velocity. However, even allowing for this, it seems that isostatic compensation cannot be attained completely until a depth of 150 km or more, implying significant strength of the rocks to this depth. The estimated depth of compensation tends to increase with age of the crust, consistent with a cooling and thickening lithosphere.

INTRODUCTION In this paper, an attempt is made to gain some information for the Australian continent on the thickness of the lithosphere, or equivalently, on the depth to the top of the asthenosphere, in the sense that the latter is unable to sustain differential vertical stresses over geological periods of time, so that isostatic compensation must occur at the top of (or above) it. This means that the total mass per unit area of a lithospheric column should be the same above a surface of constant depth in the asthenosphere. The mass of a column of lithosphere can be calculated if the density is known as a function of depth; however there is no direct method of obtaining the density at depths beyond those reached by drilling. Hence densities are commonly estimated from seismic velocities, which can be measured by refraction and wideangle reflection methods, and an empirical relation for converting velocity to density. Dooley (1976) used such an approach to estimate the variation of the mass per unit area of

the crust over Australia and surrounding marine areas; the difference between this and a corresponding column of mantle material was defined as the "crustal mass deficiency" (CMD). Although it appears that there is a substantial degree of isostatic compensation at the base of the crust, the variation in CMD showed that for complete compensation density variations to a greater depth must be taken into account. Dooley (1977) then used Pn velocities - and deeper seismic velocity information where available - to study how the CMD variation might be compensated in the uppermost mantle. It was found that in general, subcrustal lithospheric densities inferred from Pn did not provide the required lithospheric mass distribution; however in some parts of the continent, increases in seismic velocity observed at depths of 60 to 100 km suggest density changes which would lead to approximate compensation at about 130 km depth, corresponding to the top of a low velocity layer suggested by surface wave studies under the eastern part of the continent. The Western Australian shield was found to have the lowest CMD - i.e. the heaviest crust - and

Velocity variations, isostatic compensation.

also the highest P velocity, which implied a further increase in mass with depth relative to other parts of the continent. If the subshield lithosphere were to consist of refractory peridotite, depleted in the heavier elements by removal of iron-rich magma through igneous activity, its mean atomic weight would be low. Birch (1961) found in laboratory experiments that the density of such material would be lower than that obtained from the standard velocity-density relation. Using the lower density, the mass discrepancy with other areas was reduced slightly with increasing depth, but was still not eliminated at depths less than about 160 km. n

Since the 1976 calculation of CMD values, many new data have become available, and these have been used by Dooley and Moss (1988) to revise the calculation. In the present paper, the revised CMD values have been used in a similar fashion to that of Dooley (1977) to take another look at the properties of the lithosphere, and to compare these with the available lithospheric models obtained by seismic studies.

equal. For calculating the CMD for the refraction crustal models used here, allowance was made for variations in free-air gravity gf ; these were assumed to comprise a very broad global field as mapped by satellites gs, and regional anomalies gf - gs due to small departures from isostasy. The mass of material above sea level and the contribution from the sediments were also taken into account. Rock densities for CMD calculations were estimated from an empirical velocity-density relation used by Dooley (1976). In the more recent study (Dooley and Moss, 1988), a spline representation was used to facilitate computer calculation as follows: Velocity range Density, t/m km/s 2.0<V <3.6 2.38 + .106(V-3.6) 3.6<V <5.0 2.38 + .106(1/-3.6) + .036(1/-3.6) -.0125(1/-3.6) 5.0<V<6.4 2.82 + .312(1/-6.4) + .144(\/_6.4) + .038(1/ -6.4) 6A<V <9.0 2.82+ .312(1/-6.4) (2) 3

METHOD OF ANALYSIS Crustal Mass Deficiency A brief review is given here of the concept of crustal mass deficiency (CMD), as defined by Dooley (1976) to characterise standard models similar to those used by many authors in isostatic studies. The CMD is the difference between the mass of a column of crustal material of unit area, and the mass of the same column of mantle material of standard density between the Moho and sea level. The CMD for such models is given by: ms = Hs(pm

- ps)

A graph of this relation is shown in Fig. 1.

(1)

where Hs is the standard crustal thickness, p is the standard mantle density, and P-? is the standard crustal density. If isostasy were to prevail at the base of the crust, the CMD for all crustal columns should be m

Fig. 1 Velocity-density relation. The formula for the standard CMD is then: m

Zi Ti

Ap/ +

(gf - gs )/ 2nk - hpm

(3)

Geol. Soc. Aust. Spec. Publ. 17,41-58. where m is the CMD, Ti, p/ are the thickness and density of the i th layer Ap/= pm-p; k is the gravitational constant, h and gf are the elevation and free-air gravity anomaly averaged over a 4° x 4° lat./long. area around the site, and gs is the global gravity field deduced from satellite orbits.

complicated model, with velocity gradients and reversals, in the Tasman Geosyncline.

The standard CMD applies for zero elevation and zero gravity anomaly. Drummond and Collins (1986) have drawn attention to the fact that most measurements of seismic velocity in the laboratory at high pressures have been made at or near room temperature, and they have calculated corrections for geothermal gradient to reduce deep crustal velocities to surface temperature and a pressure of 1 GPa. These corrections would affect the relation between velocity and density. They have not been taken into account in 2 Typical crustal models, illustrating CMD the present study; neglect of them is expected to Fig. derivation - (a) Site 61, Yilgarn Block; (b) Site 51, affect all measurements to approximately the same Tasman (see Dooley & Moss, 1988). extent so that relative values are not likely to be Triangles Geosyncline represent velocities, and circles represent CMD affected much. values. CMD CALCULATIONS FOR CRUSTAL MODELS Since the 1976 study, many new deep seismic projects have been carried out. The present reassessment of the CMD includes data from the Lachlan Fold Belt, the Pilbara-Hamersley region, the McArthur Basin, and the Central Eromanga Basin; also, revised models are available for many of the areas used previously. The recent data used are discussed in more detail by Dooley and Moss (1988), and are summarised in their Table 2. Two examples of CMD derivations from crustal velocitydepth models are shown in Fig. 2 - one for a fairly simple model with constant velocity in each layer in the Yilgarn Block, and the other for a more

The locations and results of the calculations of CMD for the refraction models are shown in Fig. 3. The average value of all CMD for the continent is 18.5 kt/m ; values range from 14.8 to 22.4 kt/m . These values are somewhat higher than those found in 1976, i.e. the crust is lighter, but the range of values is about the same. Fig. 4 is a plot of CMD against depth to the Moho as measured by refraction surveys; the individual entries in Table 2 of Dooley and Moss (1988) provided 43 CMD/depth values. There is a correlation of CMD with depth - least-squares analysis gives a relation CMD = 10.8 + .17Hkt/m\ which is significant at the 1% level. A possible 2

Velocity variations, isostatic compensation.

To study the variation further, analysis of variance was undertaken for the CMD values. For this analysis, they were grouped in two ways firstly in the groups of Table 1, but with some modifications (GAL and CERO, SAUS and MLS, and MLW and YIL were combined) giving 11 groups; and secondly in the 6 groups of Fig. 4. In each case an estimate of variance from CMD values within the groups was compared with an estimate based on the mean values for the groups. The latter estimate in both cases was very significantly higher than the former. The analysis was repeated using departures from the regression line of Fig. 4 instead of the actual CMD values, with a similar result.

Fig. 3 Locations of refraction models, with CMD values. (1) McArthur Basin west (MCAW); (2) McArthur Basin east (MCAE); (3) North Australia Block (NAB); (4) Central Eromanga Basin (CERO); (5) Galilee Basin (GAL); (6) Bowen Basin (BOW); (7) Tasman Geosyncline north, including parts of the Lachlan Fold Belt and Sydney Basin (TGN); (8) Tasman Geosyncline south (TGS); (9) southern Australia, including Victoria, Bass Strait, Tasmania, and southeastern South Australia (SAUS); (10) from Maralinga to the southeast - Gawler Block (MLS); (11) from Maralinga to the west - Eucla Basin (MLW); (12) Yilgarn Block (YIL); (13) from southern Pilbara Block to northern Yilgarn Block Bangemall Basin (SPNY); (14) Pilbara Block (PIL).

Thus we conclude that there is a real variation in CMD between regions over the continent, and hence that isostatic compensation cannot be complete at the base of the crust. We now proceed to examine the available lithospheric models to see whether they indicate how compensation might be attained at a greater depth. The analysis is based on the whole of the variations in CMD, taking the apparent relation of CMD with depth as fortuitous.

LITHOSPHERIC MODELS Simple models from crustal survey data

explanation would be that the crustal densities were incorrect - a higher average density would reduce the rate of accumulation of CMD with depth. However the density contrast between crust and mantle would have to be less by about 0.17 t/m3 between about 30 and 55 km depth, and it seems unlikely that the estimates (based on laboratory measurements) would be wrong by that amount. In Fig. 4 the points are grouped in six main regions, and are marked by distinguishing symbols explained in the caption. The slopes of the regression lines for individual regions are mostly positive, but are not significantly different from zero. It could also be said that they do not differ significantly from the slope for all regions, except for Queensland, which has a negative slope. The points for Queensland and Vic/SAlie mostly above the regression line of Fig. 4, and those for the Pilbara lie mostly below the line.

First we examine the models corresponding to the crustal models used for calculating the CMD.

Fig. 4 CMD as a function of depth to Moho. Open circles - northern craton (MCAW, MCAE, NAB); open squares - Queensland (CERO, GAL, BOW); filled squares Tasman Geosyncline (TGN, TGS); filled triangles southern Australia (SAUS, MLS); filled circles southwestern craton (MLW, YIL); open triangles northwestern craton (SPNY, PIL).

Geol. Soc. Aust. Spec. Publ. 17,41-58.

Depth km

CMD kt/m 2

Velocity km/s

Density t/m 3

MCAW MCAE

53 40 60

18.2 18.0

8.47 7.80 7.91

3.47 3.26 3.29

Collins (1983)

NAB

52 61 70

20.6

8.20 8.26 8.30

3.38 3.40 3.41

Finlayson (1982a)

CERO

38 48-56 56

19.2

8.17 7.80 8.35

3.37 3.26 3.43

Finlayson etal (1984)

GAL BOW

36 45

19.1 18.1

8.00 8.20

3.32 3.38

Cull and Riesz (1972) Collins (1978); Leven (1980)

TGN

16.8

8.08

3.35

TGS

18.8

8.02

3.33

Finlayson etal (1979,1980); Finlayson and McCracken (1981)

SAUS MLS MLW

38 40 36

20.6 21.8 18.9

7.94 8.00 8.16

3.30 3.32 3.37

YIL

16.1

8.25

3.40

SPNY

21.4

8.17

3.37

PIL

36 42 50 57

14.6

8.12 8.29 8.34 8.50

3.36 3.41 3.43 3.4

Region

Reference

Dooley (1976)

Mathur (1974), Mathur et al. (1977) Drummond (1979) Drummond (1983)

Table 1 Velocities and densities used in lithospheric CMD calculations for crustal survey regions. Regions corresponding to code names are given in Fig. 3 legend.

Some of these surveys give no more information on the mantle than the velocity immediately below the crust - for these it is assumed for the present that the P n velocities remain constant with increasing depth - but some give evidence of velocity structure below the crust (or below the surface taken as the Moho by Dooley and Moss (1988) - as explained in that paper, it was not always clear what should

be regarded as the Moho). The velocities are then converted to densities using Equation 2, and the accumulations of mass deficiency with depth below the Moho are studied. For convenience, the abbreviation CMD has been generalised to represent "cumulative mass deficiency" in discussing the lithospheric structure.

Velocity variations, isostatic compensation.

—20 5o 15

Fig. 5 Mantle CMD variation below selected crustal models, based on crustal CMD and P from Table 1, together with subcrustal velocities where available from the refraction surveys. Triangles represent seismic "YJQ velocities; circles represent CMD values. The terminal points of the graphs are arbitrary; the velocity structures are not necessarily known to these depths. n

Depth (km)

120

Geol. Soc. Aust. Spec. Publ. 17,41-58.

The P n velocities used are the averages of the velocities for each region given by Dooley and Moss (1988) in their Table 2. These velocities and the corresponding densities are listed in Table 1; where there are subcrustal data available from the refraction surveys, these are also included. The lithospheric velocities for selected models from Table 1 are shown as a function of depth in the graphs of Fig. 5. The accumulation of CMD with depth for these models is also graphed in Fig. 5. The starting point for each graph is the average CMD for the region, as in Fig. 3. The assumption of constancy with depth leads to a straight horizontal line for the velocity graphs. Where the velocity is higher than 8.0 km/s, the mantle density from Equation 2 is higher than the standard mantle density of 3.32 t/m , and so the CMD decreases with depth; it increases with depth in the few cases where the velocity is below 8.0 km/s. If isostatic equilibrium were attained at some uniform depth in the lithosphere, then all of the CMD graphs should converge to a common value at or near that depth, and follow coincident paths for greater depths. Fig. 6 shows CMD graphs, including those shown in Fig. 5, for all the sites listed in Table 1,

superimposed on the same plot. It is clear that there is no common depth zone of convergence for all graphs. There are however three zones in which some degree of convergence occurs for a few graphs - at 70 to 85 km depth, with CMD between 18 and 20 kt/m2; at 80 to 95 km depth, with CMD between 15 and 17 kt/m2; and at 100 to 115 km depth, with CMD between 14 and 16 kt/m2. These three zones are shown as "boxes" on Fig. 6. The graphs converging in the first box are those for MCAE, NAB, GAL, TGS, and SPNY. In the second box we have CERO, BOW, TGN, and MLW; the graph for NAB also just clips one corner. The graphs for NAB, TGN, and MLW intersect in the third box, and those for CERO and BOW pass through a corner. Thus the same five graphs actually pass through the second and third boxes, though some of them only just do so. This leaves five graphs which do not pass through any of the boxes - those for SAUS and MLS are too high, whereas those for MCAW, YIL, and PIL are too low. There does not seem to be any consistent tectonic or geographic relation between the sites with graphs through any one box. However the sites with the highest CMD graphs are in sedimentary areas, whereas those with low CMD graphs are in cratonic areas.

The graphs for PIL and YIL start with low crustal CMD values, and also plunge more steeply than those of the first group. In physical terms this implies that not only is the crust heavier than average in these areas, but it is underlain by material with higher seismic velocity, and hence presumably higher density. The MCAW graph starts with a CMD value close to the average, but plunges steeply because of the high Pn velocity; it intersects the YIL graph at about 110 km depth, with a CMD value of about 15 to 16 kt/m2. However the PIL graph remains about 5 kt/m below the others. A substantial decrease in the density difference between the PIL lithosphere and the others Fig. 6 Mantle CMD variation below all crustal models would be needed to get convergence even within from Table 1 superimposed. "Boxes" indicate zones of the group; this would need to take the form of either partial convergence. Symbols for the various crustal models are as in Fig. 9, where the separate lines are easier a decrease in density for PIL, or an increase for the other two areas. to identify.

Velocity variations, isostatic compensation. Depth km

Velocity km/s

Density t/in

Reference

ORD

36-85 85-175 175

8.12 8.25-8.34 8.62

3.36 3.41 3.51

Denham et al. (1972)

DTC

45-76 76

8.18 8.38

3.35 3.44

Hales and Rynn (1978)

PCT

40-98 98

8.14 8.20

3.36 3.38

Mills and Fitch (1977)

TASS Model 2a

38-60 60-160 160 38-60 60-130 130-150 150

7.95 8.23 8.65 7.95 8.23 7.70 8.65

3.30 3.39 3.52 3.30 3.39 3.23 3.52

Finlayson etal. (1974); Muirhead et al. (1977)

43-50 52-60 63

7.96 7.61 8.00

3.31 3.20 3.32

Finlayson and McCracken (1981)

Model

Model 2b

MARU

Table 2 Mantle models used in CMD calculations. Lithospheric models based on upper mantle require unrealistically high velocities and densities studies to divert the first-box graphs so as to pass through We now investigate whether some of the graphs the second box. can be "bent" so as to converge in one of the three We examine some lithospheric models for boxes by making reasonable modifications to the velocity structure, while adhering to the velocity- various parts of Australia which extend to a greater density relation of Equation 2. Later the possibility depth than those obtained from the refraction surof modifying this relation will be examined. veys. These have been interpreted from recordings Many of the refraction surveys give no real of seismic waves generated by earthquakes or very information about the velocities deeper than 60 km large explosions. In general they do not correspond or so. The two highest curves (SAUS and MLS), exactly to the locations of the surveys used for and three of those passing through the first box, are determining CMD values, but an attempt has been reasonably parallel with P velocities close to 8.0 made to associate the lithospheric and crustal surkm/s; an increase in velocity at a moderate depth would bring them closer to the third box. It seems veys on the basis of reasonable geographical reasonable to expect that the relatively low P proximity and similarity of tectonic and geological velocity should not persist to a great depth. It would setting. n

Geol. Soc. Aust. Spec. Publ. 17,41-58. The locations of the lithospheric models are shown in Fig. 7, and the velocity structures are listed in Table 2 with the associated CMD values, and graphed in Fig. 8. The Ord model (ORD) is from Denham et al (1972); Darwin to Tennant Creek (DTC) is from Hales and Rynn (1978); TASS (Trans-Australia seismic survey) is from Finlayson

Fig. 7 Locations of published upper mantle models of Table 2. et al (1974) for the western part, and Muirhead et al (1977); Picton to Charters Towers (PCT) results from an earthquake study by Mills and Fitch (1977); and the profile westwards from Marulan (MARU) is from Finlayson and McCracken (1981). Several other published lithospheric models give the S-velocity structure, but little or no new information on P velocities, and thus do not enable estimates of density. However they may have some relevance to the present problem, as a low S-velocity zone may be an indication of partial melting, and hence of lower strength. This aspect will be considered later. Two models are shown in Fig. 8 for TASS models 2a and 2b of Finlayson et al (1974), one without and one with a low velocity zone at a depth of 160 km. The data were inadequate to distinguish definitively between these. The possible lowvelocity layer is deeper than the compensation zones discussed here; it could act as a deeper com-

pensation zone in some areas, but in view of its doubtful existence, little importance can be attached to it. As these profiles are not directly related to the crustal models, the crustal CMD values used for the initial points are to some extent arbitrary. The values used have not been corrected for gravity or elevation, so that they may differ somewhat from those for adjacent crustal models in Fig. 3. In taking a lithospheric model as representative of structure under a crustal model, the initial point should be set to coincide with an appropriate point on the CMD graph for that crustal model, i. e. the mantle CMD curves should be moved bodily up or down to the appropriate value. For want of better knowledge, we assume that the intra-lithospheric boundaries deeper than the branching point are at the same depth wherever applied. In addition to the mantle models discussed above, it is possible to use mantle data obtained by some crustal surveys, e. g. NAB, CERO, and PIL, to apply beneath other crustal survey regions. Table 3 sets out the crustal models to which each mantle CMD curve might be applicable, with the depth and CMD value for the starting point of the mantle model graph, and the constant values to be added to the curve - or vertical shifts of the graph - to make it pass through the appropriate points for the relevant crustal models. We now discuss the mantle CMD curves in groups geographically adjacent and with some tectonic similarity. The first, or northern craton, group (Fig. 9a) comprises MCAW, MCAE, and NAB. MCAW and NAB recorded some subcrustal structure, and high mantle velocities were attained; therefore there was no point in applying other mantle models. However MCAE shows increasing CMD with depth, and diverges widely from the others. Improvement is obtained by branching to either ORD or DTC models of Fig. 8, at a depth of 76 km; the latter is preferred as it passes through the third box. The graph for NAB, which crosses MCAE at a depth of 70 km, could also provide a satisfactory mantle model below that depth.

Velocity variations, isostatic compensation.

The low-CMD MCAW, having an already high velocity and density, cannot be improved by any of the nearby mantle models; a layer with substantially lower density would be needed to divert it towards the other curves. The second, or Queensland, group (Fig. 9b) comprises CERO, GAL, and BOW, all Phanerozoic basins. CERO has a low-velocity layer near the top of the mantle, underlain by high-velocity material, and provides its own mantle model. The other mantle model which might be applicable to this group is PCT; Fig. 9b shows this applied to both

GAL and BOW, being adjusted to pass through their crustal CMD points. The deeper part of CERO could also provide a model for GAL from 56 km, where the two models coincide. Thus suitable models for all sites in this group can be found giving convergence in the second box, and less so in the the third box. The third, or southeastern, group (Fig. 9c) comprises TGN, TGS, SAUS, and MLS; these are also Phanerozoic features. The MARU profile lies between TGN and TGS areas; the model has a low-

Figure 8 Velocity and CMD variation for the mantle models of Table 2 and Fig. 7. Triangles represent velocities, and circles represent CMD values.

Geol Soc. Aust. Spec. Publ. 17,41-58.

Initial point

Adjustment to mantle CMD kt/m2

Depth km

CMD kt/m2

Crustal model

ORD* DTC*

76 76

19.0 15.7

MCAE MCAE

0.4 3.7

PCT

18.8

GAL BOW

0.1 -0.5

MARU TASS

43 38

19.7 21.0

TGN TGS SAUS MLS

-2.9 -2.3 -0.4 0.8

PIL

14.6

YIL SPNY

1.4 7.4

Mantle model

Table 3 Depth - CMD starting points for mantle models and adjustments required for related crustal models. * signifies modified model.

velocity layer from 52 to 63 km depth. This model has been applied to TGN only, as the graph has the lowest CMD values; if applied to any of the other three high-CMD graphs, it would decrease the chance of convergence. Applying the MARU mantle to TGN puts the graph through the first box, but if this were done a higher density at a greater depth would have to be introduced to divert the graph back to the third box. The other mantle model which might be relevant here, PCT, would lower the TGN graph, and so it is not shown. The TASS model has been applied to the other three sites, and branches significantly from the simple models at 60 km depth. SAUS and MLS remain above the second and third boxes, though MLS passes through the first box. The fourth, or western group comprises MLW, YIL, PIL, and SPNY, all within the craton. The only one which shows any structure in the mantle is PIL, and this has been applied to the YIL and SPNY graphs in Fig. 9d. There is no other mantle model near enough to be considered reasonable for use at these sites. The TASS profile overlaps MLW; the

TASS models have a lower P n velocity than MLW, but a somewhat higher velocity below 60 km. There is little point in applying the lower velocity in contradiction to the measured one, but a combined model could be devised from 60 km; the graphs for the simple and combined models both pass through the second and third boxes. All of these models have high subcrustal velocities; PIL and YIL have low crustal CMD initial points, while SPNY has a veiy high one, and the graph passes through the first box. The apparent anomaly between SPNY and the neighbouring PIL and YIL models was noted and discussed by Drummond and Shelley (1981), who postulate crustal layering to explain the difference. Applying the PIL mantle model to SPNY diverts it through the second box, and leads to an intersection with YIL at about 130 km depth, but of course brings it no nearer to the PIL graph itself. The PIL and YIL graphs are well below all three boxes. Applying the PIL mantle model to YIL keeps it close to the PIL graph, but takes it further from everything else.

Velocity variations, isostatic compensation.

Thus there are feasible mantle structures which could give agreement for MLW and SPNY with the bulk of the graphs for other areas. However to bring PIL and YIL to converge with the others would need a layer of much lower density beneath them than is suggested by the measured velocities. Departures from velocity-density models

Where a suitable model shown in Fig. 9 already passes through the third box, it has been reproduced in Fig. 10. This applies to the original model for NAB, and to the combined model for MCAE with DTC model below 76 km in Fig. 10a; to GAL and BOW with the PCT mantle model in Fig. 10b; to the original TGN, and TGS with the TASS mantle model, in Fig. 10c; and to the original MLW in Fig. lOd.

Figs. 1 Oa, b, c, and d show graphs for the models of the four regions of Fig. 9, with modifications where appropriate as described below to improve convergence, particularly in the third box.

There is a possibility that the very high velocity under MCAW is underlain by a lower velocity layer at not too great a depth, in view of the subcrustal velocities recorded in neighbouring areas; a density of about 3.35 t/m3, corresponding to a velocity of

^ (a)

_ (b)

120

Depth (km)

Depth ( k m )

120

Figure 9 Combined CMD models, with mantle models of Fig. 8 applied to models of Fig. 5 at appropriate depths. (a) Northern craton - MCAW (filled stars); MCAE (open triangles); NAB (filled circles). (b) Queensland - CERO (open circles); GAL (inverted filled triangles); BOW (filled squares). (c) Southeastern Australia - TGN (inverted open triangles); TGS (filled triangles); SAUS (open stars); MLS (open squares). (d) Western cratons - MLW (filled diamonds); YIL (X); SPNY (+); PIL (open diamonds).

Geol. Soc. Aust. Spec. Publ. 17,41-58. 8.1 km/s, if introduced at 65 km depth, would divert the graph through the second and third boxes; such a model is shown in Fig. 10a. In Fig. 9b, the CERO graph passes through the second box but is low relative to the third box. Similarly to MCAW, an arbitrary but not unreasonable change to 8.20 km/s has been introduced at a depth of 80 km in Fig. 10b, diverting the graph through the third box. S AUS and MLS models result from earlier surveys, which may not have detected deep crustal increases in velocity; if so, their starting points may be too high. Their graphs in Fig. 9c, with the TASS mantle model, have been lowered by 2.5 kt/m in Fig. 10c, enabling convergence in the third box with other models. 2

The models in Fig. 10c for TGS, SAUS, and MLS, together with the TGN/MARU model of Fig.

9c, all have a CMD of about 18 kt/m at about 70 to 80 km depth. Thus it seems possible that isostasy could apply in the southeastern area at a depth of 80 km or so. 2

For YIL and PIL, a density of somewhat less than the standard mantle density (3.32 kt/m ), corresponding to a velocity of less than 8.0 km/s, would be needed to divert the graphs through the third box; this is much lower than the postulated velocity reversals for MCAW and CERO. Some evidence for P-velocity reversals in the mantle has been found under the Phanerozoic basins of eastern Australia, but not under the cratons of Western Australia. This is consistent with evidence from heat-flow measurements, which show lower values in the west; lower velocities might be expected to be associated with higher temperatures. It is also reasonable in relation to the different tectonic his3

Depth

(km) 25/A/179

Figure 10 CMD models of Fig. 9, modified as explained i: the text. Symbols are the same as in Fig. 9.

Velocity variations, isostatic compensation.

tones of the two regions. The cratons are very old and show no signs of major tectonic activity throughout the Phanerozoic, whereas the eastern part of the continent has undergone uplift and subsidence, basin formation, granitic intrusions, and volcanism, which imply higher temperatures at depth. Another possibility for explaining the discrepancy of the low CMD graphs is by departures from the velocity-density relation used. Birch (1961) derived a series of empirically based velocity-density curves for different mean atomic weights (MAW) of the rocks. A high MAW is generally associated with high iron content, and a relatively high density for a given seismic velocity. Parts of Birch's curves were shown by Dooley (1977, Fig. 5), together with part of the curve used here; the latter corresponded to a MAW of 21.5 in the velocity range concerned (8.2 to 8.5 km/s). From Birch's curves, the variation of density at a fixed velocity is 0.27 t/m per unit MAW. 3

According to the pyrolite model for the upper mantle (Ringwood, 1975) the mantle under shield areas consists of a refractory residue of dunite or peridotite whose MAW may be as low as 21.0. The iron content of the original material has been reduced by eons of plutonic activity, and the MAW is about the lowest that can reasonably be postulated for uppermost mantle rocks. A decrease in MAW of 0.5 would correspond to a density decrease of 0.14 t/m . Under YIL, the density could then be as low as 3.26 t/m , whereas 3.33 t/m , as shown in Fig. lOd, is low enough to divert the graph through the third box. For PIL, the density below 57 km depth could be reduced to 3.34 km/s; this model is shown in Fig. lOd, and still leaves the graph about 4kt/m below the third box. Application of the same density differential to the layers between 36 and 57 km depth would raise the graph another 2.8 kt/m , which would bring it close to the bottom of the third box, but it is not clear whether these layers should be regarded as pyrolite with low MAW. 3

Possibility of anisotropic velocities Drummond (1985) has investigated the possibility of anisotropic velocities beneath the Pilbara

crust as an explanation of the high observed velocities. The results could be explained by a southerly dip of 1° or 2° of the crust-mantle boundary, but his preferred model has a smaller dip with 2% anisotropy, the maximum velocity being at 30° east of north. There is also a suggestion of anisotropy in a different direction - about 40° west of north - about 15 km deeper than the Moho. Anisotropy of 2% could be caused by preferential alignment of olivine crystals during periods of tectonic stress; the density should presumably be related to the average seismic velocity, or about 1 % lower than the maximum. From Equation 2, this could lead to a density of about 0.025t/m lower than that obtained from the maximum velocity. If this differential were applied to the whole of the material below the base of the crust, the resulting increase in CMD at 110 km depth would be about 1.7 kt/m . The PIL graph would remain about 1 kt/m below the bottom of the third box in Fig. lOd. 3

As the evidence for anisotropy is inconclusive, the graph in Fig. lOd has not been modified. However anisotropy should be borne in mind as a possible explanation of the high observed velocities, not only in PIL, but also in MCAW and CERO. Possible zones of weakness So far we have been considering low-velocity layers, or reversals, as indicating possible low-density layers within the lithospheric column. They may also have significance as indicating a zone of weakness where compensation could take place; in this case they would be at the base of the lithosphere. Such zones would be more likely to be revealed by surface wave than by P-wave studies, because surface-wave velocities are more dependent on S than P velocities, and S velocities are more sensitive to temperatures approaching melting point. Also surface waves are better indicators than refraction studies of low velocities in the relevant depth range. Goncz and Cleary (1976), from surface-wave studies, found a layer with low S velocity under eastern Australia, centred at a depth of 130 km; the thickness of the layer could range from about 25 km with a velocity contrast of about 0.6 km/s, to about 80 km with a contrast of about 1.5 km/s. They

Geol. Soc. Aust. Spec. Publ. 17,41-58. found no evidence for a low-velocity layer under the western shield at depths less than 200 km. Mills and Fitch (1977), using surface waves from a shallow earthquake in southeastern New South Wales, found no pronounced low-velocity zone present generally under eastern Australia at a depth of less than 180 km, though the possibility of a lowvelocity zone deeper than 140 km under the PCT path could not be excluded. Hales et al. (1980), studying shear waves from explosions in the Banda Sea, found that there is no significant low-velocity zone under northern Australia to a depth of 150 km, and only a slight possibility of one above 200 km depth. Ellis and Denham (1985), by inverting surface-wave data over Australia including those of Mills and Fitch, found that a low-velocity zone was necessary to fit the data in eastern Australia, at about the same depth as the Goncz and Cleary model, but with a much smaller velocity-depth contrast. Under the western shield there was a possibility of a small low-velocity zone under one profile, but not under another nearby reversed one. The data were considered inadequate to determine whether a low-velocity zone exists near 100 km depth. Thus the evidence of S-wave velocity studies generally supports the existence of a low-velocity zone at about 110 km depth under eastern Australia, but no major such zone under the western shield. This is consistent with the results of the present study, in that the eastern low-velocity zone could be a zone of weakness enabling isostatic compensation of the lithospheric masses above that depth, corresponding to the third box where the graphs for eastern models converge, whereas the absence of any such prominent zone under the west is reflected by the difficulty in finding models with compensation at this depth for some of the western models. CONCLUSIONS The spread of crustal CMD values implies that isostasy is incomplete at the base of the crust, although the relatively small free-air anomalies indicate that departures from isostasy are small. Further, the variation of P velocities implies that, if crustal isostasy were complete, the mass balance would depart from isostasy with increasing depth. n

In some regions, density variations derived from subcrustal velocities are such as to compensate for the departures from crustal isostasy at depths of about 80 or 110 km; however in other regions departures from isostasy increase with depth on these models. Introduction of mantle structure shown by published models can be used to improve convergence to a common CMD value, and hence towards isostasy, in some areas, mostly by introducing higher velocities, and hence higher densities, than in the simple models. It is possible to construct models for the lithosphere under most of Australia using plausible seismic velocity-depth profiles and the velocity-density conversion of Equation 2, which attain complete isostatic equilibrium at a depth of about 110 km. "Plausible" generally means that layers of higher velocity have been introduced at depths too great for them to have been recorded during the field surveys, but nevertheless having velocities consistent with those observed at corresponding depths elsewhere. It is possible to construct such models attaining isostatic compensation within some regions or groups of surveys at lesser depths, say at 70 or 80 km, but not between all groups. Models leading to compensation at 110 km depth have been derived for all survey regions in eastern Australia. This does not mean that these models are correct, nor that the velocity-density relation used is accurate, nor that compensation is necessarily attained at that depth. The fit is significant because it shows that isostatic compensation at this depth is possible with a lithosphere consisting of generally accepted materials, but compensation at a lesser depth would require some change in the assumptions. However, it is reasonable to suppose that the low S-velocity layer at this depth shown by surfacewave evidence coincides with weak material, enabling compensation to occur. For some of the western shield areas, compensation at 110 km depth is not possible using similar models - in particular in the Pilbara and Yilgarn areas. By changing the velocity-density relation it is possible to attain compensation under the Yil-

Velocity variations, isostatic compensation.

garn, but for the Pilbara it is necessary to postulate densities at or beyond the limits of those appropriate on the depleted pyrolite model. Moreover, the high melting point of the pyrolite residue, and the low heat flow observed in the shield area, combine to make it unlikely that partial melting temperatures would be approached at this depth (Ringwood, 1975), so that a zone of weakness may not develop; this is consistent with the absence of evidence for a marked seismic lowvelocity zone in the west. Other investigations have suggested differences between the mantle structure and composition under eastern and western Australia, extending to even greater depths. Cleary (1967) and Everingham (1969) showed differences in travel-times from distant earthquakes and explosions. Finlayson (1982b) applied corrections for crustal structure to Cleary's data, and concluded that differences in seismic velocity structure persist to depths greater than 200 km. Wellman (1982) also concluded from a study of gravity and seismic refraction that density variations must exist in the lower lithosphere. The evidence suggests a substantial degree of isostatic compensation at the base of the crust, with more complete compensation occurring at greater depths which increase with age of the province. Very broadly, from Fig. 2 of Stewart (1985), the age of crustal formation for the Queensland and southeastern groups are within the range 1.8 to 0.8 Ga; they have possible compensation at about 80 km depth, though with different CMD for each group. The northern-group crust was formed between 2.6 and 1.8 Ga ago; compensation is possible at about 110 km depth. The western crust was formed between 3.5 and 2.6 Ga ago; models for compensation at 110 km depth are difficult to define, and a greater depth seems likely. This would be consistent with an initial crust overlying a weak layer, so that at the time of formation isostatic compensation was more or less complete at the base of the crust. As time went on, the material beneath the crust solidified, by cooling forming a thicker sub-crustal lithosphere, and/or by segregation of more mobile material from the mantle, i.e. underplating as proposed by Drummond and Collins (1986) for much of the continental crust, forming a thicker crust. Meanwhile the

crustal mass balance was disturbed by uplift and subsidence, erosion and deposition, thrusting and rifting, and other tectonic processes. As the thickness of the lithosphere increased, these disturbances were compensated at successively greater depths, while the earlier compensating surfaces were frozen in. ACKNOWLEDGMENTS Helpful discussions with B. J. Drummond and C. D. N. Collins are acknowledged. I. R. Qureshi and P. Wellman reviewed the manuscript and made constructive suggestions for improvements. REFERENCES BIRCH F. 1961. The velocity of compressional waves in rocks to 10 kilobars. Journal of Geophysical Research 66, 2199-2224. CLEARY J. R. 1967. P times to Australian stations from nuclear explosions. Seismological Society of America, Bulletin 57, 773-781. COLLINS C. D. N . 1978. Crustal structure of the central

Bowen Basin, Queensland. BMR Journal of Australian Geology & Geophysics 3, 2 0 2 - 2 0 9 .

COLLINS C. D. N. 1983. Crustal structure of the southern McArthur Basin, Northern Australia, from deep seismic sounding. BMR Journal of Australian Geology & Geophysics 8,19-34. CULL J. P. & RIESZ E. J. 1972. Deep crustal seismic reflection/ refraction survey between Clermont and Charters Towers, Queensland, 1971. Bureau of Mineral Resources, Australia, Record 1972/97. DENHAM D., SIMPSON D. W., GREGSON P. J. & SUTTON

D. J. 1972. Travel times and amplitudes from explosions in northern Australia. Geophysical Journal of the Royal Astronomical Society 28, 225-235. DOOLEY J. C. 1976. Variation of crustal mass over the Australian region. BMR Journal of Australian Geology & Geophysics 1, 291-296.

Geol. Soc. Aust. Spec. Publ. 17,41-58. DOOLEY J. C. 1977. Implications of Australian seismic and gravity measurements for the structure and composition of the upper mantle. BMR Journal of Australian Geology & Geophysics 2,1-5. DOOLEY J. C. & MOSS F. J. 1988. Deep crustal

FINLAYSON D. M., COLLINS C. D. N. & LOCK J. 1984.

P-wave velocity features of the lithosphere under the Eromanga Basin, Eastern Australia, including a prominent mid-crustal (Conrad?) discontinuity. Tectonophysics 101,267-291.

reflections in Australia 1957-1973 - II. Crustal models. Geophysical Journal 93,239-249.

FINLAYSON D. M., CULL J. P. & DRUMMOND B. J. 1974.

DRUMMOND B. J. 1979. A crustal profile across the Archaean Pilbara and northern Yilgarn Cratons, northwest Australia. BMR Journal of Australian Geology & Geophysics 4, 171-180.

21,447-458.

DRUMMOND B. J. 1983. Detailed seismic velocity/depth models of the upper lithosphere of the Pilbara Craton, northwest Australia. BMR Journal ofAustralian Geology & Geophysics 8, 35-51. DRUMMOND B. J. 1985. Seismic P-wave anisotropy in the

subcrustal lithosphere of north-west Australia. Geophysical Journal of the Royal astronomical Society

81, 497-519.

DRUMMOND B. J. & COLLINS C. D. N. 1986. Seismic evidence for underplating of the lower continental crust of Australia. Earth and Planetary Science Letters 79, 361-372.

DRUMMOND B. J. & SHELLEY H. M. 1981. Isostasy and

structure of the lower crust and upper mantle in the Precambrian terrains of northwest Australia. BMR Journal of Australian Geology & Geophysics 6, 137-143. ELLIS R. M. & DENHAM D. 1985. Structure of the crust and upper mantle beneath Australia from Rayleigh- and Love-wave observations. Physics of the Earth and Planetary Interiors 38, 224-234. EVERINGHAM I. B. 1969. P wave residuals at Australian

seismograph stations. Bureau of Mineral Resources, Australia, Record 1969/22. FINLAYSON D. M. 1982a. Seismic crustal structure of the

Proterozoic North Australian Craton between Tennant Creek and Mount Isa. Journal of Geophysical Research 87, 10569-10578. FINLAYSON D. M. 1982b. Geophysical differences in the

lithosphere between Phanerozoic and Precambrian Australia. Tectonophysics 84, 287-312.

Upper mantle structure from the Trans-Australia seismic refraction data. Geological Society of Australia Journal FINLAYSON D. M. & MCCRACKEN H. M. 1981. Crustal

structure under the Sydney Basin and Lachlan Fold Belt, determined from explosion seismic studies. Geological Society ofAustralia Journal 28, 177-190. FINLAYSON D. M., PRODEHL C. & COLLINS C. D. N. 1979.

Explosion seismic profiles and implications for crustal evolution in southeastern Australia. BMR Journal of Australian Geology & Geophysics 4, 243-252.

GONCZ J. H. & CLEARY J. R. 1976. Variations in the

structure of the upper mantle beneath Australia, from Rayleigh wave observations. Geophysical Journal of the Royal Astronomical Society 44, 507-516. HALES A. L. & RYNN J. M. W. 1978. A long range controlled source seismic profile in northern Aust. Geophysical Journal of the Royal Astronomical Society 55, 633-644.

HALES A. L., MUIRHEAD K. J. & RYNN J. M. W. 1980.

Crust and upper mantle shear velocities from controlled sources. Geophysical Journal of the Royal Astronomical Society 63, 659-670. 1980. The application of synthetic seismograms. Ph. D. Thesis, Australian National University, Canberra.

LEVEN J.

MATHUR S. P. 1974. Crustal structure in southwestern

Australia from seismic and gravity data. Tectonophysics

24, 151-182.

MATHUR S. P., Moss F. J. & BRANSON J. C. 1977. Seismic and gravity investigations along the Geotraverse, Western Australia, 1969. Bureau of Mineral Resources, Australia, Bulletin 191. MILLS J. M. & FITCH T. J. 1977. Thrust faulting and

crust-upper mantle structure in east Australia. Geophysical Journal of the Royal Astronomical Society

48, 351-38.

Velocity variations, isostatic compensation.

MUIRHEAD K. J., CLEARY J. R. & FINLAYSON D. M. 1977.

A long-range seismic profile in southeastern Australia. Geophysical Journal of the Royal Astronomical Society

48, 509-519.

RlNGWOOD E. R. 1975. Composition and petrology of the Earth's mantle. McGraw Hill, New York. STEWART A. J. 1985. Metamorphism. In BMR Earth

Science Atlas of Australia. Australian Bureau of Mineral Resources, Geology and Geophysics, Canberra.

WELLMAN P. 1982. Australian seismic refraction results, isostasy and altitude anomalies. Nature 298, 838-841. J. C. Dooley 66 Hawker Street Torrens ACT 2607 AUSTRALIA

Geol. Soc. Aust. Spec. Publ. 17,59-66.

Shear Wave Crustal Models for the Australian Continent David Denham

Bureau of Mineral Resources, Geology and Geophysics, Canberra, A.C.T., Australia S-wave crustal models for the Australian continent, obtained from body wave and surface wave studies, indicate shear velocities in the range 3.4 - 3.6 km s"1 for the upper crust, 3.8 - 4.2 km s"1 for the lower crust and 4.6 km s"1 in the topmost mantle. The lower crust, where the S-wave velocity gradients are higher than in the upper crust, corresponds to the region identified by Mathur (1983) which generates strong seismic reflections. The depth to the Moho usually ranges from 38-40 km continent-wide and appears to be independent of the age of cratonization. However, because these results are obtained by averaging large regions of the crust, it was not possible to detect local undulations in Moho depth. There is no evidence of any low velocity layer in the crust, although, in the upper crust, the S-wave velocity gradients are very small and the existence of minor low velocity regions are not precluded. Introduction Efforts to determine the structure of the crust and upper mantle have usually involved P-wave observations. This is because P-waves are easier to record than S-waves, and because P-wave first arrivals can be identified unambiguously from any seismic source, provided the signal level is above the microseisms. However, if the physical properties of the earth are to be determined completely, it is necessary to estimate the shear wave velocities. These velocities can be obtained directly by analysing body waves from explosions or earthquakes, or indirectly by analysing the phase and/or group velocities of surface waves. In this paper the results of both body and surface wave studies are reviewed and average S-wave crustal models for the Australian continent are presented.

Shackleford and Sutton (1981) for the Adelaide Geosyncline. They used explosions from two large open-cut mines (Leigh Creek, 30.4°S, 138.4°E and Iron Baron, 33.0°S, 137.1°E) and obtained Sg and Sn values of 3.43 ± 0.05 km/s and 4.44 ± 0.04 km s"1 respectively. However, their crustal thickness of 38 km agrees with that given by Doyle and Everingham (1964). The values obtained for the Adelaide Geosyncline are also lower than values obtained in other parts of the continent. For example Denham et al. (1972) gave either 4.59 ± 0.02 km s"1 or 4.75 ± 0.07 km s"1 for Sn, depending on how the data are grouped, from the Ord River explosions (16.13°S, 128.74°E), and Finlayson etal. (1980) using quarry blasts in southeastern Australia obtained values of 3,65 ± 0.02 km s"1 for the upper crust and 3.87 ± 0.03 km s~l for the lower crust. They did not obtain a reliable value for Sn.

Body Wave Studies Bolt et al. (1958) were among the first to study S-waves in the crust and upper mantle in Australia. They obtained values of 3.55 ±0.04 km s"1 and 4.75 ± 0.01 km s"1 for Sg and Sn, respectively, to the west of the 1956 Maralinga nuclear explosion (30.15°S, 131.59°E), and a crustal thickness of 39 km. To the southeast of the Emu (28.63°S, 132.20°E) and Maralinga test sites, Doyle and Everingham (1964) obtained values of 3.59 ± 0.01 km s"1 and 4.61 ±0.04 km s"1 for Sg and Sn and an equivalent crustal thickness (38 km). These velocities are higher than those determined by

One of the most detailed S-wave models produced from body waves was that given by Hales etal. (1981). They used explosions from a two ship refraction program in the region centred on 11°S, 130°E to record S-waves in the distance range 90 1150 km south of the source region. Their models are reproduced in Table 1. These indicate velocities of 3.5 8 and 3.70 km s~1 for the upper and lower crust and mantle volocities of 4.60 km s"1 and 4.72 km s"1 for the depth ranges 45-76 km and below 76 km respectively. The mantle results are consistent with these obtained by Denham et al. (1972).

Shear wave crustal models for Australia.

These and other body-wave studies of S-wave velocities tend to identify the significant high velocity layers in the crust or upper mantle and do

not necessarily give an overall model for the crustal structure; this often can be estimated from studies of surface waves, which essentially sample the

OROGENIC DOMAINS Palaeozoic Proterozoic Archaean

H H H

PLATFORM COVERS Mesozoic to Cainozoic Late Proterozoic to Palaeozoic Early to Mid Proterozoic Fault

25/A/177

Fig. 1. Propagation paths (1-14) for S-wave crustal models shown in Fig. 2. BR, CD, PT, SD, SR and WG represent the epicentres of the 1979 Broome, 1979 Cadoux, 1973 Picton, 1972 Simpson Desert, 1979 Scott Reef and the 1982 Wonnangatta earthquakes respectively. The seismograph stations are represented by the standard three letter codes. Paths 1-7 are from Ellis and Denham (1985), 8 and 9 from Mills and Fitch (1977) and 10-14 from Denham and Ellis (1986).

Geol. Soc. Aust. Spec. Publ. 17, 59-66.

1 1

S - W a v e Velocity ( k m / s )

2 L_

20 •

40 -

11 2

•

. i

l.l

(

—,—I

b-r-

4 L

40 -

25/A/1 76

Fig. 2. S-wave crustal models for the paths indicated in Fig. 1.

Shear wave crustal models for Australia.

whole crust. To obtain detailed crustal structure it is necessary to analyse body wave data using synthetic seismograms, and special modelling techniques. However, these have not been tried yet for S-waves in Australia.

S-WAVE STRUCTURE FROM SURFACE WAVES Bolt and Niazi (1964) used phase and group velocity observations from Rayleigh waves to estimate average crustal thickness and crust and upper mantle S-wave velocities. They used paths from Brisbane to Perth and Charters Towers to Perth and data in the period range 15 to 40 s. The average crustal thickness (for a one layer crust) was found to be in the range 30-35 km and the Sg and Sn velocities were estimated to be 3.70 km s" and 4.80 km s"1 respectively. Thomas (1969) examined in more detail the S-wave structure using Rayleigh wave phase velocities over seven paths between Australian World Wide Standard Seismograph Stations. He found a lack of uniqueness in the models in so far that the data seemed to fit both two layer and three layer crusts equally well, within the limits of the observations. Table 2 shows his adopted models for the different paths. Goncz et al. (1975) and Goncz and Cleary (1976) used Rayleigh wave observations for nine paths across Australia in the period range 10-200 s. They found the presence of a low shear velocity layer in the upper mantle beneath eastern Australia, centred at a depth of

about 130 km but found no evidence for a similar layer beneath central and/or western Australia. The shear velocity profiles in the upper 50 km for successful models found by Monte Carlo inversion are given in Table 3. Mills and Fitch (1977) examined multimode group velocities from the 1973 Picton earthquake and used a Monte Carlo technique to obtain Earth models for paths to Charters Towers and Adelaide (see Fig. 1). A suite of models was obtained that fitted the observations and two 'average' models are shown in Fig. 2. Their models have subcrustal shear velocities of 4.2 - 4.3 km s"1, which are some 0.4 km s"1 slower than the data obtained from body wave studies and also slower than the Denham and Ellis (1986) and Ellis and Denham (1985) models. Ellis and Denham (1985) examined fundamental and first higher modes of Rayleigh and Love wave group velocities along 7 paths across the Australian continent (Fig. 1). The data were jointly inverted by a controlled Monte Carlo procedure to obtain regional shear wave velocity models for the crust and upper mantle (Fig. 2). The crustal models obtained from this work show that in eastern Australia the upper crusts are both thicker and have lower velocities than those in western Australia. The crust-upper mantle boundary does not appear to be sharply defined whenever there is a well dispersed wave train available. In most models the lower crust/uppermantle region appears to be characterised by a region of higher velocity gradient. There does not appear to be a dependence of Sn velocity on the age of cratonization or crustal thickness and because of the presence of a gradient rather than a sharp disconLand Area

Shelf Area S-wave velocity (kms"1)

Layer thickness (km)

S-wave velocity (kms -1 )

1.38

1.8

3.58

12.8

1.58

0.3

3.70

31.1

3.58

9.4

4.48

1.1

3.70

19.9

4.48

13.6

4.60

31.0

4.60

31.0

4.72

124.0

4.72

124.0

Table 1. Crustal models obtained by Hales etal. (1981) for northern Australia.

Geol. Soc. Aust. Spec. Publ. 17,59-66.

Path Model

ADE •CTA

ADE -RIV

3.54

3.74

3.63

3.40

3.50

3.58

3.24

4.70

3.63

4.54

3.91

4.72

4.61

4.07

34.1

28.9

36.6

19.5

30.6

35.2

20.1

4.79

hi h2

ADE • TAU

ADE- MUN

4.55

11.5

4.55

21.2

19.5

Path Model

CTA -RIV

CTA - MUN

RIV - TAU

3.45

3.58

3.63

3.59

3.41

4.82

4.65

3.70

4.62

4.16

P3 hi

34.0

34.6

18.9

30.2

19.7

4.75

21.3

4.71

20.7

pi and p2 are S-wave velocities in kms"1. hi and h2 are thicknesses in km. Table 2. Crustal models obtained by Thomas (1969).

Western Australia Model

3.61

3.57

3.58

3.62

3.67

3.68

4.65

4.57

4.53

4.57

Eastern Australia Model

3.50

3.56

3.51

3.61

3.57

3.66

3.64

4.64

4.63

4.67

TABLE 3. Crustal models obtained by Goncz and Cleary (1976)

tinuity it is difficult to define a Moho in terms of S-wave velocities. Rather we can say that for most of the continent Sn falls within the range 4.55 - 4.65 km s"1. A more detailed study, based on synthetic seismograms and group velocities of Rayleigh waves from the 1982 Wonnangatta earthquake, was carried out by Denham and Ellis (1986) for eastern Australia. For the five paths examined from this earthquake, a two-layer crust, overlying a Moho at about 40 km, is required to fit the observations. In the upper layer down to about 25 km, the velocity gradients are very small and shear velocities are in the range 3 . 4 - 3 . 6 km s"1. In the lower crust the S-wave velocities increase gradually to about 4.6 k m s ' 1 at the Moho, where the velocity gradients are again very small. The region where the S-wave velocities increase significantly with depth corresponds to the region of the lower crust which generates strong seismic reflections in deep seismic sounding experiments (Mathur, 1983). The thin crust postulated by Johnson (1973) beneath Bass Strait and Tas-

Shear wave crustal models for Australia.

mania is not substantiated by the surface wave data and a crustal thickness of abowt 40 km is also required for this path. For all the longer profiles (~1800 km) where the surface waves are dispersed more widely the crust/mantle boundary is identified with a more gradual change in S-wave velocity than indicated on the shorter profiles ( - 7 0 0 km).

DENHAM, D . & ELLIS, R . M . 1 9 8 6 . D e t e r m i n a t i o n of

earthquake focal mechanisms and crustal structure in eastern Australia using surface waves. Tectonophysics 121,109-123.

DENHAM, D . , SIMPSON, D . W . , GREGSON, P.J., & SUTTON,

D.J. 1972. Travel times and amplitudes from explosions in northern Australia. Geophysical Journal of the Royal astronomical Society, 28, 225-235. DOYLE,

CONCLUSIONS Apart from the low Sn velocities obtained by Mills and Fitch (1977), the S-wave crust and upper mantle structure appears not to vary greatly over the continent. This may reflect the lack of resolution available in the analyses of S-waves and surface waves, or it may mean that the paths being analysed are so long that local undulations are smoothed out. On average, the upper crustal Swave velocities vary from 3.4 - 3.6 km s"1, the lower crustal velocities from 3.8 - 4.2 km s" 1 , and the topmost mantle from 4.55 - 4.60 km s" 1 .

H.A.

EVERINGHAM,

I.B.

1964.

Seismic

velocities and crustal structure in southern Australia. Geological Society of Australia, Journal 11, 141-150. ELLIS, R . M . & DENHAM, D . 1 9 8 5 . S t r u c t u r e o f t h e c r u s t

and upper mantle beneath Australia from Rayleigh- and Love-wave observations. Physics of the Earth and Planetary Interiors 38, 224-234. FINLAYSON, D . M . , COLLINS, C . D . N . & DENHAM, D .

1980. Crustal structure under the Lachlan Fold Belt, southeastern Australia. Physics of the Earth and Planetary Interiors 21, 321-342. GONCZ, J . H . , HALES, A . L . & MUIRHEAD, K . J . 1 9 7 5 .

Analysis to extended periods of Rayleigh and Love wave disperson across Australia. Geophysical Journal of the Royal astronomical Society 41, 81-105. ACKNOWLEDGMENTS GONCZ, J . H . & CLEARY, J . R . 1 9 7 6 . V a r i a t i o n s in t h e

I thank Chris. H. Fitzgerald for preparing the diagrams, Helen Tozer and Paulette Lhotka for typing the text, Clive Collins and Anton Hales for reviewing the manuscript and making several suggestions to improve its contents, and the Director, Bureau of Mineral Resources, Geology and Geophysics for approval to publish this paper.

structure of the upper mantle beneath Australia, from Rayleigh wave observations. Geophysical Journal of the Royal astronomical Society 44, 507-516. HALES, A . L . , MUIRHEAD, K . J . & RYNN, J . M . W . 1 9 8 0 .

Crust and upper mantle shear velocities from controlled sources. Geophysical Journal of the Royal astronomical Society

REFERENCES

63, 659-670.

JOHNSON, B.D. 1973. Time term analysis ofthe data obtained during the Bass Strait upper mantle project (Operation BUMP). Australian Society of Exploration Geophysicists, Bulletin 4, 15-20.

BOLT, B . A . , DOYLE, H . A . , & SUTTON D . J . 1958. S e i s m i c

Observations from the 1956 Atomic Explosions in Australia. Geophysical Journal of the Royal astronomical Society 1, 135-145.

BOLT, B . A . , & NIAZI, M . 1 9 6 4 . D i s p e r s i o n of R a y l e i g h

Waves across Australia. Geophysical Journal of the Royal astronomical Society 9, 21-36.

MATHUR, S.P. 1983. Deep crustal reflection results from the Central Eromanga Basin, Australia. Tectonophysics 100, 1 6 3 - 1 7 3 .

MILLS, J.M. & FITCH, T.J. 1977. Thrust faulting and crustupper mantle structure in East Australia. Geophysical Journal ofthe Royal astronomical Society 48, 351-384.

Geol. Soc. Aust. Spec. Publ. 17,59-66.

David Denham Bureau of Mineral Resources, Geology and Geophysics GPO Box 378 THOMAS, L. 1969. Rayleigh wave dispersion in Australia. Canberra ACT 2601 Seismological Society ofAmerica, Bulletin 59,167- 182. AUSTRALIA.

SHACKLEFORD, P.R.J. & SUTTON, D.J. 1981. A first interpretation of crustal structure in the Adelaide Geosyncline in South Australia using quarry blasts. Geological Society of Australia, Journal 28,491-500.

Geol Soc. Aust. Spec. Publ. 17,67-80.

The Nature of the Crust-Mantle Boundary Under Australia from Seismic Evidence C.D.N. Collins

Bureau of Mineral Resources, Geology and Geophysics, Canberra, ACT, Australia The seismological definition of the crust-mantle boundary is based on the identification of the boundary which separates low seismic P-wave velocities (less than about 7.8 km/s) associated with predominantly crustal rock types, from higher velocities (greater than 7.8 km/s) usually associated with mantle rock types, ie. the Moho discontinuity. The Moho under Australia has been identified by both refraction and reflection surveys and exhibits a number of characteristic features. The depth to the Moho is greatest under the Proterozoic North Australian Craton and Central Australia, and under the Palaeozoic Lachlan Fold Belt, and is shallowest under the Archaean Pilbara Craton, and possibly Tasmania. The Moho is observed to be a transition zone in which the velocities increase to upper mantle values over several kilometres, and is most gradational where it is deepest. The Moho is often observed on near-vertical incidence reflection profiles as a zone of short coherent reflections below which few, if any, reflections are seen. The depth at which upper-mantle velocities are first observed coincides with the top of the non-reflective zone. Model studies have shown that the Moho most likely consists of a zone of alternating subhorizontal layers of relatively high and low velocity, with individual layers having a limited extent. Possible interpretations of this model include meta-sedimentary/volcanic layering, layered intrusions into the lower crust, underplating, ductilely deformed layering, partial melting, and multiple melts and cumulates.

Introduction The crust-mantle boundary is usually identified with the seismic velocity discontinuity discovered by A. Mohorovicic in 1909 from near-earthquake recordings, and since then established as a near universal feature under all continents and oceans. This discontinuity, the "Moho", occurs at depths commonly between 30 and 50 km in continental areas. Early interpretations of earthquake data in Australia (eg. De Jersey, 1946) and refraction experiments using nuclear explosions at Maralinga (Bolt and others, 1958) and quarry blasts in the Snowy Mountains (Doyle and others, 1959), confirmed the presence of the Moho under the Australian continent as well. Early seismic refraction surveys were interpreted using only seismic travel times and, as a result of the interpretation methods used, velocity changes with depth were commonly modelled as abrupt velocity discontinuities. More recently, improvements in data quality and developments in interpretation methods have made it possible to model tht amplitudes of the various seismic phases. They show that the changes in velocity at most boundaries within the deep crust, and at the Moho, occur over transition zones rather than abruptly. As

a consequence, it is difficult to precisely define at any particular location a single depth at which the velocity reaches or exceeds those expected in the upper mantle. In all areas of Australia where the data quality allows amplitude modelling, the Moho is defined by a velocity gradient, or transition zone, from compressional velocities typical of lowercrustal rocks (less than about 7.8 km/s) to velocities typical of mantle rocks (greater than 7.8 km/s).

Seismic Evidence for the Crust-Mantle Boundary The seismic definition of the crust-mantle boundary comes largely from arrivals generated from earthquakes and explosive sources, which travel within the upper mantle below the Moho at velocities of around 8 km/s and are refracted back to the surface, and from seismic energy reflected from the Moho. Reflected energy recorded in seismic profiling may be classified into two types: near-vertical reflections, when the angle of incidence of the energy is close to normal to the boundary, and wide-angle reflections for larger angles of incidence (typically 45-55 degrees). Some examples of seismic data which show these arrivals from the Moho are now discussed.

The nature of the crust-mantle boundary.

Figure 1 shows the location of some seismic surveys designed to record refracted and wideangle reflection energy. Figure 2 shows record sections from the Pilbara Craton, the Bowen Basin, and the North Australian Craton. To facilitate comparison between these three examples the same plotting parameters have been used in each case. Each trace has been distance normalised such that the true recorded amplitude is plotted at 100 km distance, while at larger distances the amplitudes have been increased by a factor proportional to distance. The travel-time branches of arrivals which have travelled as refracted energy through the upper mantle (traditionally termed Pn), and the wideangle reflections from the Moho transition zone (termed PmP) are shown. For simplicity, other

branches of the travel-time curves which represent seismic phases from crustal boundaries have not been labelled. Some variations in amplitude between traces are probably due to local recordingsite conditions. However, these do not affect the relative amplitudes of the various seismic phases within each trace. These three sections illustrate the seismic arrivals on which the depth of the Moho and the velocity structure above and below it are based. They also exhibit the different character of these arrivals in each case: the Pilbara section (a) contains large wide-angle PmP reflections and relatively small Pn refracted arrivals. In the Bowen Basin (b), the amplitudes of Pn arrivals, relative to the PmP reflections, are larger than in the Pilbara. The North Australian section (c) contains large Pn

NORTH AUSTRALIAN CRATON 'NORTHEAST OROGENS

( B O W E N BASIN RLLBARA/C RATON CENTRAL

AUSTRALIAN

MOBILE

• A- -M- 4- 4- -I- 4- 4A- 4- 4- 4- 4- 4- 4- 4- + 4- -H 4- -b + -b + -b + + 4-4-4y

,4- 4 - 4 - YILGARN 4 - 4* Y

+ +

BLOCK

EROMANGA BASIN

+ (

+ + +]•

V Y + + + + + + 4-/ V W 4 + 4- + + / ' AtA + + V .

• TASMAN—GEOSYNCLINE

BELTS

IULLARBORV BLOCK

GAWLER

L/\CRATON

CURNA MONA7 CRATON •:/NEW ENGLAND FOLD BELT

/• /SYDNEY BASIN

LACHLAN FOLD BELT Tasman

Geosync/ine

Stabilised fold belts within the Tasman Geosync/ine

26/A/23

Fig. 1. Location of selected seismic profiles and major geological province boundaries. The eastern limit of Precambrian basement is marked by the boundary of the Tasman Geosyncline.

100

120

140

160

180

200

220

240

280

260

300

180

200

220

240

260

280

300

Distance ( k m ) 14 r

i i I

; r

12 - ; oo 2

l 10 |

] i j

8 i

l ? i % -»

S I

1 DC

}

: i

! 1 |

100

120

140

160

Distance ( k m )

Upper Mantle refracted arrival 0

, Upper Mantle reflect,on

. , Crustal arrivals

Start of Moho transition zone

Fig. 2. Seismic record-sections from (a) Pilbara Block, (b) Bowen Basin, (c) North Australian Craton. Representative velocity/depth models for these three areas are shown below. Trace amplitudes are multiplied by Distance/100, ie. at 100 km, amplitude is multiplied by 1, at 200 km, amplitude is multiplied by 2, etc. Arrivals refracted through the upper mantle (Pn) and wide-angle reflections from the Moho transition zone (PmP) are shown. For simplicity, travel-time branches of other phases have not been included.

The nature of the crust-mantle boundary. refracted arrivals, and no clear PmP reflections. The sharpness of the boundary has a profound These differences in the seismic character indicate effect on the amplitude and frequency characthat the change in velocity with depth across the teristics of the reflected and refracted seismic enerMoho is different in all three provinces. gy (Sezawa and Kanai, 1935). The effect on the

Moho models

Moho

models

Abrupt

velocity

10 km

gradient

Amplitude-distance

Pmp

Abrupt •--

Moho transition zone

10 km

velocity

discontinuity

curves

discontinuity

gradient

Pn Abrupt

velocity

10 km

gradient

Vp (km/s)

Fig. 3. Amplitude-distance curves for Pn and PmP phases for two Moho models.

discontinuity

71 Geol. Soc. Aust. Spec. Publ. 17,67-80. relative amplitudes can be seen for the two simple Moho marks the base of the crust, this gives an cases shown in Fig. 3 adapted from Braile and indication of the range of depths to the crust-mantle Chiang (1986). When the discontinuity is abrupt boundary within each province, and also within the (solid model), the amplitudes of the wide-angle Australian continent. reflections are much larger than the refracted arThe individual velocity/depth curves within rivals. As the transition zone increases in thickness (across a 10 km depth range in the dotted model), each region are interpreted from different traverses the wide-angle reflections become less prominent in the region, and illustrate the heterogeneity of the while the refracted arrivals increase relatively in crust. The different crustal character of these six amplitude. The distance from the source at which regions, and in particular, the Moho transition the maximum amplitude reflections occur also in- zones, can be gauged from Fig. 4. The increase in creases as the thickness of the boundary increases. velocity from crustal to upper-mantle values is By comparing the observed data with synthetic most transitional, and the Moho occurs at the seismograms generated from various types of greatest depths, in the Proterozoic North Australian boundary, it is possible to define the form of the Craton. The Moho is sharpest and shallowest in the transition zone at the boundary. In the three ex- Archaean Pilbara Block and the Mesozoic Bowen amples illustrated in Fig. 2, the Moho beneath the Basin. Pilbara is relatively sharp and well defined comMost of the available Moho depths and Pn pared with the Moho beneath the North Australian velocities for the Australian continent (Collins, Craton which is transitional in nature. 1988) are plotted in Figs. 5a and 5b, respectively. Figure 4 shows velocity/depth curves which The values are averages measured along traverses have been interpreted from six regions within the which are usually 200-300 km long. Some old Australian continent. The ages of these six regions, results have not been included if they conflict with defined here by the last major tectonic events to those derived from later, better quality data. The have occurred within them, range from Archaean depths of the Moho in Fig. 5a include many interto Mesozoic. These six regions are chosen because pretations based on travel-time data alone, in which good-quality data are available, and the interpreta- the Moho is interpreted as a step-like boundary tion of the data has included modelling of the between crustal and mantle velocities. These are amplitudes of the various seismic phases. The 8.0 shown as light figures. Depths which are interkm/s velocity is marked, as are the estimated min- preted using amplitude data as well as travel-times imum and maximum depths to the top of the Moho to interpret the velocity distribution are shown in transition zone within each region. Assuming the bold type, and are the depths to the bottom of the Velocity ( k m / s )

Velocity ( k m / s )

Pilbara Craton (Archaean)

N. Australian Craton (Proterozoic)

Lachlan Fold Belt (Lower Palaeozoic)

Sydney Basin (Permian & M e s o z o i c )

Eromanga Basin (Palaeozoic & M e s o z o i c )

Bowen Basin (Permian & M e s o z o i c ) 26/A/26

Fig. 4. Velocity/depth curves. Each curve represents a different traverse in each province. The velocity of 8 km/s is marked. The horizontal lines are minimum and maximum depths of the start of the transition to upper mantle velocities.

The nature of the crust-mantle boundary.

Moho transition zone; ie. the depths plotted are those at which upper mantle material is first encountered. The regions in which the Moho transition is most gradational coincide with the areas of deepest Moho shown in Fig. 5a, ie. the North Australian Craton and Lachlan Fold Belt.

identified by Cull and Denham (1979) as areas in which anomalously high heat flow determinations may have biased the heat flow averages. In the

The variations in Pn within each geological province must be treated with caution. While they may be due to real lateral variations in some cases, they may be due to inadequate sampling in others, eg. where the traverses are not properly reversed, some of the differences may be due to the effects of dip of the Moho. The variations may also be due to seismic anisotropy in the mantle which causes the velocities to be azimuth-dependent, as is the case in the Pilbara (Drummond, 1985). The trends which are seen in Fig. 4 are also seen in the larger data set of Fig. 5a; the Pilbara and Bowen Basin have a shallower Moho than the Lachlan Fold Belt and the North Australian Craton. Figure 5b also shows another trend which has long been recognised, namely that the Pn velocities are generally higher in the shield regions of Australia compared to the Tasman Geosyncline (eg. Dooley,1971). As seismic velocities are temperature dependent, Dooley (1971) suggested this trend may be due to the higher geothermal gradients observed in eastern Australia. Figure 6a shows Pn velocity plotted against heat flow. The Pn velocities for each region are the averages of the values shown in Fig. 5b, and the heat flow values are averages from the contours given by Cull and Denham (1979) for averages over a three degree grid. Excluding the values for the North Australian Craton (3) and the Eromanga Basin (8), there is evidently a correlation of Pn velocity and heat flow. This correlation was noted by Cull and Denham (1979) who showed that it was consistent with the effects of the temperature derivative of P-wave velocity (approximately -5.0 X 10~4 kms' 1 °C' 1 , eg. Jackson, 1991). Thus it is not necessary to invoke gross changes in sub-crustal lithology to explain the overall trend in Pn velocities, providing the geotherms derived from surface heat flow are accepted. The two regions which do not fit the general trend in velocity with heat flow, the North Australian Craton and the Eromanga Basin, are

Fig. 5. a) Moho depths, b) Pn velocities. Determinations from individual traverses are shown, and include data from early surveys where no later surveys exist. Moho depths in bold type are depths at which Pn is first observed below the velocity transition zone; depths in light type are interpreted as step- boundaries from travel-time data only. Major geological province boundaries are shown as in Fig. 1. Sources of data are summarised in Collins (1988).

Geol Soc. Aust. Spec. Publ. 17,67-80. North Australian Craton heat flow measurements are high near Tennant Creek due to concentrations of radioactive isotopes in the crust. In the Eromanga basin, the average heat flow values are high due to large scale vertical movement of heated ground water at the southwestern margins. There does not seem to be a correlation between Moho depth and heat flow (Fig. 6b), nor is there a correlation between Moho depth and Pn velocity (Fig. 6c). These conclusions differ from the results presented by Meissner and others (1987) for Europe in which a negative correlation between Moho depth and heat flow was found. However, if the Archaean shield data (1) and (2) are ignored in the Australian data, a weak negative correlation may be seen between Moho depth and heat flow, and a weak positive correlation may be seen between Moho depth and Pn velocity. Meissner and others (1987) also report a positive correlation between Moho depth and age, partly derived from the heat flow result. The depth of the Moho in Australia also appears to increase with age (see Fig. 4) due to a thickening of the lower crust, probably by underplating (Drummond & Collins, 1986). However, in marked contrast to the European data, which includes Archaean data, some of the shallowest Moho depths occur under the oldest regions, the Archaean Pilbara Craton (Fig. 4) and Yilgarn Block.

Moho, and by definition the Moho is a velocity discontinuity. Near-vertical reflection data map the occurrence of velocity contrasts at depth in the form of reflections, but at lower crustal depths the velocities cannot be derived from the reflection data alone. To identify Moho reflections it is necessary to know the velocities of the rocks above and below the reflecting boundaries. This is best done where there is coincident or nearby refraction data. In some cases, the change in reflection character of

Reflection Character of the Crust-Mantle Boundary Near-vertical reflections from the deep crust, and probably the Moho, have been recorded in Australia since the 1950s on single records and short traverses (Mathur, 1984; Dooley and Moss, 1988; Moss and Dooley, 1988). Since the mid 1970s improvements in data quality through digital recording and processing have enabled the recording of deep events on continuous profiles. These profiles allow the groups of reflection events seen on individual records to be correlated over large distances, supporting their interpretation as reflections within the deep crust, and not artifacts related to individual recording sites. Interpretation of the nature of these events remains problematical, and in particular the identification of the crust-mantle boundary. The crustmantle boundary is seismically defined by the

Fig. 6. a) Pn velocity vs Heat Flow, b) Moho depth vs Heat Row, c) Pn velocity vs Moho depth. Pn velocities and Moho depths are averages for each region; heat flow values are averages for each region derived from three-degree grid contours of Cull and Denham (1979). Data shown from (1) Pilbara, (2) Yilgarn, (3) North Australian Craton, (4) Gawler Craton, (5) South Australia (East Gawler Craton), (6) Cape York (Northeast Orogen), (7) Bowen Basin, (8) Eromanga Basin, (9) Lachlan Fold Belt, (10) Victoria (Southern Lachlan Fold Belt), and (11) Tasmania.

The nature of the crust-mantle boundary.

the section at the appropriate depth may be construed as the transition from crust to mantle, but without any petrologically dependent data, such as velocity, the interpretation must remain speculative. In seeking examples showing the reflection character of the Moho, it is therefore necessary to choose recordings from areas where the approximate depth to the Moho is known from coincident, or nearby, refraction data (Fig. 7). The velocity/depth curves derived from the refraction data have been converted to velocity/two-way travel time curves so that direct comparison can be made with the adjacent reflection sections. The first example in Fig. 7, from the Archaean Yilgarn Block (Site G, Fig. 1), shows a reflection section recorded at Hines Hill (Mathur, 1974), which is 69 km north of the centre of the refraction traverses shown. The velocity/two-way time curve displayed is from the nearest refraction traverse (this preliminary interpretation is published in Drummond and Collins, 1986). The Moho transition zone defined by the refraction data coincides with the band of reflections betweeen 9 and 12 seconds. Below 12 seconds, within the upper mantle, the number of reflections decreases abruptly. There are no reflection sections available for the Archaean Pilbara Block, but the velocity/depth curves from that region (Drummond, 1983) would suggest a similar picture. (km/s)

Yilgarn Block

The Proterozoic North Australian platform area is represented by a short reflection profile in the McArthur Basin at site B (Fig 1); the velocity/twoway-time curve is derived from the western traverse shown (Collins, 1983). The reflection section is characterized by reflectors throughout the section, with no obvious change in character towards the bottom of the record. Similar sections have been recorded in the Georgina and Drummond Basins (Mathur, 1983b). The latter area, while in Phanerozoic Eastern Australia, shows the typical character of Precambrian reflection sections, and Mathur postulates this site is an isolated part of the Precambrian Craton. Because of the thickness of the crust, longer recording times may be required to show any changes in reflection character at the crust-mantle boundary in these areas. Below the Proterozoic Arunta Block, where recording times are believed to be sufficiently long, there is also no distinct change in reflection character at the crust-mantle boundary (Goleby and others, 1989). However, this region has had a long and complicated tectonic history which may have modified the original character.

(km/s)

McArthur

Basin

At site C in the Lachlan Fold Belt (Fig. 1), the reflection bands seen within the crust (Pinchin, 1980) correlate with changes in the velocity gradient (Finlayson and others, 1979) (Fig. 7). At the top of the Moho a weaker band of reflections occurs, below which no reflections can be seen. The record is too short (16 seconds) for d e f i n e con-

L a c h l a n Fold Belt

(km/s)

Eromanga

Basin

(km/s)

Bowen Basin

Fig. 7. Reflection profiles and corresponding velocity/ two-way-time curves. The arrow marks the start of the transition to upper mantle velocities.

Geol. Soc. Aust. Spec. Publ. 17,67-80. elusions to be drawn about the upper mantle character; from the refraction data, true mantle velocities are not reached until about 16 seconds. The reflection section from the Eromanga Basin (Site D, Fig 1) is part of 1400 km of profiling to 20 seconds two-way time within the basin (Mathur, 1984). The top of the reflection band at about 8 seconds coincides with a prominent midcrustal discontinuity defined by coincident refraction data (Finlayson and others, 1984). The bottom of the band at about 13 seconds coincides with the Moho as defined by the refraction data, below which few reflections are seen. The few sub-Moho reflections may be side-reflections or diffractions from the crust, and the upper most mantle may not exhibit any reflections at all. The reflection section from the Bowen Basin (Site F, Fig 1; Mathur, 1983) was recorded within the Denison Trough, south of the refraction line from which the velocity/depth curve was obtained (Collins, 1980). The refraction profile was recorded along the main axis of the basin. If allowance is made for the different locations of the data, a correlation can be made between the depth to the lower crustal velocity discontinuity interpreted from the refraction data and the top of the reflection bands, seen at about 8 seconds in the reflection section. The base of the lower band of reflectors, at about 12 seconds, coincides with the Moho defined by the refraction model. The thickness of the transition zone is not well defined by the refraction data, but a transition, rather than a sharp boundary, is interpreted from the relative amplitudes of the wideangle reflections observed from the Moho. None of the examples presented here show the Moho as a discrete reflection which can be traced for any large distance. Rather, the transition to upper mantle velocities is marked by a broad band of discontinuous reflections. Below the transition where the velocity reaches 8 km/s or more, few, if any, reflections are recorded. So far, the boundary has not been seen as a single continuous reflection event extending for more than a few kilometres anywhere under the Australian continent. If the Moho does correspond to the petrological crustmantle boundary, the seismic results indicate that there is no abrupt change in chemistry or mineral-

ogy at the base of the crust, though the base of this petrological transition zone may be well defined. The Nature of the Crust-Mantle Boundary The depth to the Moho transition zone is well determined from the seismic refraction travel-time data, and detailed analysis of the amplitudes of the refracted arrivals and wide-angle reflections has shown a velocity transition zone occurs at the Moho which may extend over a number of kilometres in depth. Vertical reflection profiles show that this zone corresponds to a band of short discontinuous reflections, and hence the transition zone has a fine structure. These major features of the Moho under Australia are common to observations from many parts of the world, and a considerable effort has (a)

Oistance (km)

(c)

(b) 25 26 27

J" £ a

29 30 31 32

6 0

6 5

7.0 7 0

7.5 7.5

Velocity (km/s) (km/s)

8.0 80

8 5

6.0

6.5

7.0

7.5

Velocity ( k m / s )

8.0

8 5 26/A/35

Fig. 8. a) Hypothetical velocity model of a Moho transition zone 3 km thick which satisfies many of the observed features of reflected and refracted seismic arrivals from the Moho (from Braile and Chiang, 1986). b) Representative velocity-depth curve for this model at a position mid- way (6 km) along it. c) Velocity-depth curve derived from exposed section of the Ivrea-Verbano Zone (Hale and Thompson, 1982). Depth values are arbitrary.

The nature of the crust-mantle boundary.

been made to interpret them in terms of the lithology at depth. Model studies of the amplitude and frequency characteristics of near-vertical and wide-angle reflections and refracted arrivals from the deep crust and upper mantle have been undertaken by a number of workers (eg. Fuchs, 1969,1970; Clowes and Kanasewich, 1970; Davidova and others, 1972; Meissner, 1973; Pajchel, 1976; Deichmann and Ansorge 1983; Braile and Chiang, 1986). They all agree that the Moho is a velocity transition zone. Moreover, detailed analysis shows that rather than a smooth increase in velocity with depth, the form of the transition zone which best fits the amplitude, frequency and spatial characteristics of the observed data is a series of alternating high and low velocity layers, or laminae, with individual thicknesses of 50 to 200 m, and which on average have an increase in velocity with depth. Data from widely varying regions are consistent with modelling results, which seems to indicate that the gross structure of the Moho is laterally continuous. However, the fine structure is probably laterally variable over short distances, resulting in the discontinuous character of near-vertical reflections. Braile and Chiang (1986) illustrated a hypothetical Moho transition zone incorporating the above features which produces a synthetic seismic response similar to observed data (Fig. 8a,b). Direct evidence of the nature of the crustmantle boundary may come from xenoliths brought to the surface in volcanic eruptions, and from exposures of what are believed to be slices of lower crust brought to the surface by tectonic processes. The Ivrea-Verbano zone in northern Italy is an example of the latter. Within Australia the Musgrave Block, Fraser Range and Central Arunta Province contain rocks from deep crustal levels which have been exhumed (Doepel, 1973; Forman and Shaw, 1973; Fountain and Salisbury, 1981). The seismic response of the Central Arunta Province has been modeled (Goleby and others, 1989) but none of these Australian examples are from deep enough levels to represent the crustmantle boundary. Hale and Thompson (1982) produced synthetic seismograms for a geological section through the Ivrea-Verbano Zone, which is interpreted as a tec-

tonically exposed section through the lower crust and upper mantle (Berckhemer, 1969). This body has a laminated structure comprising granulite facies gneisses of alternating arenitic and volcanic origin, with ultramafic lenses occurring with increasing frequency towards the bottom of the sequence. The total thickness of this laminated zone is about 6 km (Fig. 8c), and is manifest in the synthetic seismic section as a zone of reflections two seconds long. It would correspond to the thickness of the velocity transition at the Moho. In this case, the transition zone represents meta-sedimentary/volcanic layering and intrusions; other possible petrological models which have been proposed include layered intrusions into the lower crust, underplating, ductily deformed layering, lenses of partial melting and solidified layers of multiple melts and cumulates. Interpretations of xenolith data from eastern Australia imply that the Moho may not always correspond to a chemical crust-mantle boundary. O'Reilly and Griffin (1985) derived pressuretemperature curves for southeastern Australia from xenoliths found in Tertiary-Recent volcanics. They concluded that the lower crust was composed of dominantly mafic rocks and that the geotherm was exceptionally high, probably due to volcanism and ponding of magma at the crust-mantle boundary. The xenolith suites include ultramafics in the form of spinel lherzolites, and temperature estimates for these place them at depths as shallow as 25 km on the basis of the xenolith derived geotherm. Griffin and O'Reilly (1987) therefore contend that the top of the crust-mantle boundary transition zone is at a depth of only about 25 km under southeastern Australia. The observed velocities at this depth (6.2-7.1 km/s) are less than expected for upper mantle rocks, and this is attributed to intermixed mafic rocks, pyroxenes and amphiboles in the lherzolites, and the high temperature. Below their crust-mantle boundary, between 25 and 55 km depth, the xenolith lithology consists of spinel lherzolite, interleaved with spinel- and garnetpyroxenites and basaltic cumulates, with an increasing proportion of ultramafic rocks with depth. From the xenolith-derived geotherm they postulate a phase transition from spinel lherzolite to garnet lherzolite at about 55 km which is identified as the seismically defined Moho.

Geol. Soc. Aust. Spec. Publ. 17,67-80.

A similar geotherm to southeast Australia has been derived for east-central Queensland by Griffin and others (1987) from xenolith data. Again, the authors contend that geothermometry on spinel lherzolites in the xenolith suite show that they become dominant at depths below 30 km, which is the depth at which they place the crust-mantle boundary. This is at least 6 km above the seismically defined Moho in central Queensland (Cull & Riesz, 1972; Collins, 1980; Leven, 1980; Finlayson and others, 1984). The shallow, relatively sharp Moho under the Pilbara Block is interpreted by Griffin and O'Reilly (1987) as a phase change of lower crustal mafic rocks to eclogite facies, due to the cooler geotherm acting on essentially the same petrographic column as southeast Australia. However, the long term gravitational instability of eclogites resting on a less dense upper mantle was pointed out by Ringwood and Green (1966). The validity of the xenolith-derived crust-upper mantle column has not been universally accepted because of the uncertainties that exist with the method. For example, McDonough and others (1991) question whether the garnet pyroxenites used in the thermobarometric calculations are fully equilibrated to the geotherm, and point out the uncertainties in the geothermometers applied to the spinel lherzolite xenoliths. The depth assigned to the xenoliths is very sensitive to the temperature estimates because of the steepness of the xenolithderived geotherm. They also consider it unlikely that the spinel-garnet transition would be observed as a velocity discontinuity, since they believe the evidence shows it occurs over a large depth range. This questions the interpretation that the seismically defined Moho in eastern Australia is merely a phase transition within the upper mantle. In central-western Victoria, where the geothermal gradient is high, Gibson and others (1981) interpret the Moho at a depth of about 36 km, where the velocity exceeds 7.8 km/s. In northern Tasmania, which also has a high geotherm, Richardson (1980) interprets the Moho at a depth of only 25 km, below which the velocity exceeds 7.9 km/s. These high velocities occur at substantially shallower depths than expected if the xenolith-derived petrography and geothermal gradient are assumed. In the North Australian Craton, the depth to the Moho is greater than in the Lachlan Fold Belt, and

yet it is a geothermally cooler region. These examples are at odds with the concept that the depth of the Moho varies mainly as a consequence of different geothermal gradients acting on essentially similar petrographic columns. Reflection data suggest that the Moho coincides with a zone of discrete bodies such as intrusions, and not merely with a metamorphic front or phase change. These observations would tend to suggest that generally the seismically defined Moho transition zone does indeed mark the petrological boundary between the crust and upper mantle.

ACKNOWLEDGEMENTS I am grateful for the many discussions I have had with my colleagues at BMR, particularly within the Explosion Seismology Group, and I thank Barry Drummond, Cedric Wright and Jim Leven for many useful comments and suggestions on this paper. The figures were drafted by John Rayner-Sharpe. This paper is published with the permission of the Director of the Bureau of Mineral Resources.

REFERENCES BERCKHEMER H. 1969. Direct evidence for the composition of the lower crust and the Moho. Tectonophysics

8, 97-105.

BOLT B . A . , DOYLE H . A . & SUTTON D . J . 1 9 5 8 . S e i s m i c

observations from the 1956 atomic explosions in Australia. Geophysical Journal of the Royal astronomical Society 1, 135-145. BRAILE L.W. & CHIANG C . S . 1 9 8 6 . T h e continental

Mohorovicic discontinuity: Results from near-vertical and wide-angle seismic reflection studies. In: Barazangi M. & Brown L. (Eds), Reflection Seismology: A Global Perspective, Geodynamics Series 13, American Geophysical Union, Washington, 257-272.

CLOWES R . M . & KANASEWICH E . R .

1970.

Seismic

attenuation and the nature of reflecting horizons within the crust. Journal of Geophysical Research 75, 6693-6705.

The nature of the crust-mantle boundary.

COLLINS C.D.N. 1980. The crustal structure of the central

Bowen Basin from deep seismic sounding. MSc thesis, University of Queensland (unpubl).

COLLINS C.D.N. 1983. Crustal structure of the southern

McArthur Basin from deep seismic sounding. BMR Journal of Australian Geology & Geophysics 8, 19-34. COLLINS C.D.N. 1988. Seismic velocities in the crust and upper mantle of Australia. Bureau of Mineral Resources Australia, Report 277. CULL J.P. & DENHAM D. 1979. Regional variations in

Australian heat flow. BMR Journal of Australian Geology & Geophysics, 4, 1-13. CULL J.P. & RIESZ E.J. 1972. Deep crustal seismic reflection/refraction survey between Clermont and Charters Towers, Queensland, 1971. Bureau of Mineral Resources Australia, Record 1972/97 DAVIDOVA N.I, KOSMINSKAYA I.P., KAPUSTIAN N.K. & MICHOTA G.G. 1972. Models of the Earth's crust and

M-Boundary. Zeitchschriftfur Geophysik 38, 369-393.

DE JERSEY N.J. 1946. Seismological evidence bearing on crustal thickness in the south-west Pacific. Papers of the Department of Geology, University of Queensland, 3. DEICHMANN N. & ANSORGE J. 1983. Evidence for

DRUMMOND B.J. 1982. Seismic constraints on the chemical composition of the crust of the Pilbara craton, northwest Australia. Revista Brasileira de Geociencias 12, 113-120. DRUMMOND B.J. 1983, Detailed seismic velocity/depth

models of the upper lithosphere of the Pilbara Craton, northwest Australia. BMR Journal ofAustralian Geology & Geophysics 8, 35-51. DRUMMOND B.J. 1985. Seismic P-wave anisotropy in the

subcrustal lithosphere of north-west Australia. Geophysical Journal of the Royal astronomical Society 81,497-519. DRUMMOND B.J., & COLLINS, C.D.N. 1986. Seismic

evidence for underplating of the lower continental crust of Australia. Earth and Planetary Science Letters 79, 361-372.

FINLAYSON D.M., COLLINS C.D.N. & LOCK J. 1984.

P-wave velocity features under the Eromanga Basin, eastern Australia, including a prominent mid-crustal (Conrad?) discontinuity. Tectonophysics 101, 267-291.

FINLAYSON D.M., PRODEHL C. & COLLINS C.D.N. 1979.

Explosion seismic profiles and their implications for evolution in southeastern Australia. BMR Journal of Australian Geology & Geophysics 4, 243-252.

lamination in the lower continental crust beneath the Black Forest (southwestern Germany). Journal of Geophysics 52, 109-118.

FORMAN D.J. & SHAW R.D. 1973. Deformation of the crust and mantle in Central Australia. Bureau of Mineral Resources Australia Bulletin 144, 20 pp.

DOEPEL J.J.G. 1973. Norseman, W.A., Western Australian Geological Survey 1:250,000 Geological Series Explanatory Notes, 40 pp.

FOUNTAIN D.M. & SALISBURY M.H. 1981. Exposed cross

DOOLEY J.C. 1971. Seismological studies of the upper mantle in the Australian region. Proceedings of the 2nd Symposium ofthe Upper Mantle Project, December 1970 Hyderabad, 113-146. DOOLEY J.C. & Moss F.J. 1988. Deep crustal reflections

sections through the continental crust: implications for crustal structure, petrology and evolution. Earth and Planetary Science Letters 56, 263-277.

FUCHS K. 1969. On the properties of the deep crustal

reflections. Zeitschrift fur Geophysik 35, 133-149.

FUCHS K. 1970. On the determination of velocity depth

in Australia 1957-1973 - II. Crustal models. Geophysical Journal of the Royal astronomical Society 93,239-249.

distributions of elastic waves from the dynamic characteristics of the reflected wave field. Zeitschrift fur Geophysik 36, 531-548.

DOYLE H.A., EVERINGHAM I.B., & HOGAN T.K. 1959.

GIBSON G., WESSON V. & CUTHBERTSON R. 1981.

Seismic recordings of large explosions in southeastern Australia. Australian Journal of Physics 12, 222-230.

Seismicity of Victoria to 1980. Geological Society of Australia Journal 28, 341-356.

Geol Soc. Aust. Spec. Publ. 1 7 , 6 7 - 8 0 .

GOLEBY B.R., SHAW R.D., WRIGHT C., KENNETT B.L.N. & LAMBECK K. 1989. Geophysical evidence for

MECHEE J., PRODEHL C. & FUCHS K. 1983. T h e

"thick-skinned" crustal deformation in Central Australia. Nature, 337: 325-330.

long-range seismic refraction experiment in the Rhenish Massif. In Fuchs K. et. al. eds, Plateau Uplift, Springer-Verlag, Berlin.

GRIFFIN W.L. & O'REILLY S.Y. 1987. Is the continental Moho the crust-mantle boundary? Geology 1 5 , 2 4 1 - 2 4 4 .

MEISSNER R. 1973. The Moho as a transition zone. Geophysical Surveys 1, 195-216.

GRIFFIN W.L., SUTHERLAND F.L. & HOLLIS J.D. 1987.

MEISSNER R. & DOHR G. 1973. Observations of deep crustal reflections in Central Europe. Geophysics 38, 1213.

Geothermal profile and crust-mantle transition beneath east-central Queensland: volcanology, xenolith petrology and seismic data. Journal of Volcanological and Geothermal Research 31, 177-203. HALE L.D. & THOMSON G.A. 1982. The seismic

reflection character of the continental Mohorovicic discontinuity. Journal of Geophysical Research 87,

4625-4635.

HERZBERG C.T., FYFE W.S. & CARR M.J. 1983. Density

constraints on the formation of the continental Moho and crust. Contributions to Mineralogy & Petrology 84, 1-5. JACKSON I.N.S. 1991. The penological basis for the

MEISSNER R., WEVER T. & FLUH E.R. 1987. The Moho in Europe - implications for crustal development. Annales Geophysicae 5B, 357-364.

Moss F.J. & DOOLEY J.C. 1988. Deep crustal reflection recordings in Australia 1957-1973 - I. Crustal models. Data acquisition and presentation. Geophysical Journal of the Royal astronomical Society 93, 229-237. OLIVER J. 1982. Changes at the crust-mantle boundary. Nature 299, 398-399.

interpretation of seismological models for the continental lithosphere. In Drummond B.J. ed. The Australian Lithosphere. Geol. Soc. Aust. Spec. Publ.Yl, 81-114.

O'REILLY S.Y. & GRIFFIN W.L. 1985. A xenolith-derived

The application of synthetic seismograms to the interpretation of crustal and upper mantle structure. Ph.D. Thesis, Aust. Nat. Uni., Canberra, (unpublished).

PAJCHEL J. 1976. Nature of deep seismic boundaries in

LEVEN J.H. 1980.

MCDONOUGH W.F., RUDNICK R.L. & MCCULLOCH M.T.

1991. The chemical and isotopic composition of the lower eastern Australian lithosphere. In Drummond B.J. ed. The Australian Lithosphere. Geol. Soc. Aust. Spec. Publ.11, 163-188. MATHUR S.R 1974. Crustal. structure in southwestern Australia from seismic and gravity data. Tectonophysics 24, 151-182. MATHUR S.R 1983(a). Deep reflection experiments in

northeastern Australia. Geophysics 48, 1588-1597.

MATHUR S.P. 1983(b). Deep reflection probes in eastern Australia reaveal differences in the nature of the crust. First Break I, 9-16. MATHUR S.P. 1984. Improvements in seismic reflection

techniques for studying the lithosphere in Australia. Tectonophysics 105, 373-381.

geotherm for southeastern Australia and its geophysical implications. Tectonophysics, 111, 41-63.

the earth's crust in the light of dynamic interpretation of seismic wave field. Publications of the Institute of Geophysics of the Polish Academy of Science 101, 103-115.

PINCHIN J. 1980. Intracrustal seismic reflections from the Lachlan Fold Belt near Canberra. BMR Journal of Australian Geology & Geophysics 5, 305-309.

1980. Crustal Seismology. Ph.D. Thesis, University of Tasmania, Hobart, (unpublished).

RICHARDSON R.G.

RINGWOOD A.E. & GREEN D.H. 1966. An experimental

investigation of the gabbro-eclogite transformation and some geophysical implications. Tectonophysics 3, 383-427.

SEZAWA K. & KANAI K. 1935. The effect of sharpness of discontinuities on the transmission and reflection of elastic waves. Bulletin of the Earthquake Research Institute, Tokyo University 13, 750-755. SMITHSON S.B. & BROWN S.K. 1977. A model for the lower continental crust. Earth and Planetery Science Letters 35,134-144.

80 C. D. N. Collins Bureau of Mineral Resources, Geology and Geophysics GPO Box 378 Canberra ACT 2601 AUSTRALIA

The nature of the crust-mantle boundary.

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

The Petrophysical Basis for the Interpretation of Seismological Models for the Continental Lithosphere Ian Jackson Research School of Earth Sciences, Australian National University, Canberra. A.C.T. Australia. The petrophysical basis for the interpretation of seismological models for the lithosphere is outlined within a framework provided by theoretical descriptions of the elasticity/anelasticity of rock. Low-porosity cracked rocks are modelled by an elastic matrix containing a dilute-to-moderate concentration of spheroidal inclusions of low aspect ratio. Applications of the theory to both dry and fluid-saturated rock are reviewed. The theory predicts a remarkable sensitivity of elastic moduli and wave velocities to the presence of porosity associated with cracks of low aspect ratio. Compressional and shear wave velocities are affected differentially such that Vp/Vs is reduced by the presence of dry cracks. The measurement of wave velocities representative of dry rocks in situ necessitates the conduct of experiments at sufficiently high pressures to guarantee closure of most of the crack porosity. The influence of fluid saturation upon the wave velocities for cracked rocks depends upon the frequency at which the rock is interrogated. Four regimes - glued, saturated isolated, saturated isobaric, and drained - defined according to the spatial scale for fluid flow, are anticipated at progressively lower frequencies. Vp/Vs should always be increased by the presence of fluid-saturated cracks - in contrast to the predictions for dry cracked rocks. Pronounced dispersion (frequency dependence of moduli and wave velocities) and attenuation are expected for frequencies comparable with those for fluid flow, within a crack, between adjacent interconnected cracks, or between the pore space of a laboratory specimen and an external reservoir. Laboratory measurements of wave velocities and attenuation are then reviewed with emphasis on the most widely applied ultrasonic methods and the most general results. The observed influences of pressure, temperature and fluids upon wave velocities are generally consistent with the predictions of the theory outlined previously. The pronounced frequency dependence of wave velocities and attenuation predicted by the theory, and observed in the few laboratory studies to date, poses special problems for the close simulation of the conditions of seismic wave propagation. It is stressed that laboratory measurements (particularly those at ultrasonic frequencies) will not necessarily be directly applicable under seismological conditions because of the commonly different fluid flow regimes which prevail. The connection between rock fabric and elastic anistropy is briefly explored. It is shown that velocity-density systematics serve to systematise much of the available data but do not eliminate the need for further laboratory wave-velocity measurements especially in conjunction with seismological and/or petrological/geochemical studies. A few selected applications of petrophysical data to the interpretation of seismological models of the continental lithosphere are presented. Some simple general observations concerning the structure and metamorphic grade of the continental crust are followed by a discussion of wave velocities for granulite facies rocks. It is concluded that mafic garnet granulites and eclogites may be an important constituent of the lower part (z>30 km, Vp>7 km s" ) of many thick crustal sections. Anisotropy and regional variability of Vp in the uppermost mantle are related, respectively and via laboratory measurements, to preferred orientation of olivine and variation of Moho temperature. Reference is made to the emerging use of petrophysical data in the interpretation of fine structure in the crust (e.g. low-velocity zones and subhorizontal reflectors). Finally, petrophysical evidence bearing on the nature of the asthenosphere is briefly discussed. 1

INTRODUCTION Insight into the structure of the lithosphere is primarily of seismological derivation. Refraction studies provide information concerning the average structure in the form of models describing the spatial variation of wave velocities in one, two or three dimensions. Complementary resolution of fine structure is available from traveltimes of near-ver-

tical reflections from sub-horizontal discontinuities in acoustic impedance (density x velocity). In some cases it is possible to establish an unambiguous correlation between surface geology and the structure at depth inferred from such seismological studies. More commonly, however, the latter serve to constrain only the geometrical distribution of contrasting lithological units. In order to draw secure conclusions concerning the

Petrophysical basis for seismological models.

lithological constitution of the lithosphere, and thereby some insight into processes that are operative, it is essential that laboratory investigations of the relevant rock properties (elastic wave velocities and attenuation) be conducted under controlled conditions which simulate as closely as possible those of seismic wave propagation in the lithosphere. The purpose of this paper is to outline some of the principles which form the basis for this petrophysical calibration of the seismological probe. At the outset it is appropriate to stress the complexity of geological materials. Rocks are multiphase aggregates of elastically and thermally anisotropic crystallites. There is often a significant macroscopic anisotropy associated with a preferred orientation of minerals (fabric) acquired during deformation. Moreover, rocks sampled at the Earth's surface are almost invariably pervasively microcracked as a result of intergranular stresses encountered during cooling and decompression. The mechanical properties of rocks tested at ambient conditions are therefore generally unrepresentative of the corresponding properties in situ. Thus elastic wave velocities measured on rocks at ambient conditions are profoundly influenced by extrinsic factors (such as crack porosity and crack aspect ratio) and are not closely representative of the intrinsic mineralogical makeup of the rock. It is therefore desirable that measurements on dry rocks be conducted under conditions of high confining pressure in an attempt to close most of the crack porosity in order to retrieve 'intrinsic' velocities. The extent to which this strategy succeeds may be judged by comparison of measured velocities with those calculated for ideal aggregates (zero porosity) of the constituent minerals from single-crystal elasticity data. Under other circumstances it may be the grain boundary cracks and their interconnectedness which are of primary interest. In rocks saturated by fluid (water, gas, oil, silicate melt) the periodic stresses associated with wave propagation will cause periodic fluid flow over spatial scales determined by the period of the applied stress, the geometiy of the interconnected network of cracks and pores, and the viscosity of the fluid. Since such fluid flow will result in a degree of stress relaxation, it follows that the elastic

moduli M (stress/strain) and the associated wave velocities V = (M/p) (p being the density) may be strongly frequency dependent for fluid-saturated rocks. 1/2

These and other considerations complicate the laboratory simulation of the conditions of seismic wave propagation. It is therefore appropriate to begin this review with an outline of the theory of the elasticity/anelasticity of cracked solids in order to provide the essential framework for the review of laboratory results and their application in the interpretation of seismological models, which follows. It should be stressed that attention will be focussed throughout upon low-porosity 'crystalline' rocks of igneous and metamorphic origin. The properties of high-porosity sedimentary rocks, which are somewhat different by virtue of the much shorter timescales for fluid flow, are of great importance in the context of exploration for hydrocarbons, have been studied in some detail, and are reviewed elsewhere (e.g. Bourbie et al., 1987). ROCKS AS ELASTIC/ANELASTIC MEDIA: THEORY Cracks in rocks It has long been recognized that the elastic properties of most low-porosity rocks at and near ambient pressure are strongly influenced by the presence of cracks (e.g. Birch, 1961). The pervasive microcracking of most igneous and metamorphic rocks is attributed to the intergranular stresses which arise during cooling and decompression as a consequence of the anisotropy and intergranular variability of thermal expansion and compressibility (e.g. Birch, 1961; Nur and Simmons, 1970). By analogy with the morphology of cracks induced by thermal cycling in the laboratory (Batzle et al., 1980), it is expected that cracks produced by cooling and decompression in nature would be characterized by very low aspect-ratios (minimum/maximum dimension) a < 10" . Vesicles containing fluids trapped during crystallization, partial healing of cracks formed during the early stages of cooling and decompression, and subsequent etching and alteration due to the per3

Geol. Soc. Aust. Spec. Publ. 17, 81-114. eolation of corrosive metamorphic fluids and groundwater, all contribute to the development of a broad distribution p(a) of generally higher aspect-ratios (Sprunt and Brace, 1974; Batzle et al., 1980). Distributions of crack aspect-ratio ranging from 1 for equant voids down to 10" at the limit of detection have been measured by scanning electron microscopy of ion-bombarded surfaces of granitic rocks by Sprunt and Brace (1974) and Hadley (1976). 4

Pressure-induced crack closure The intuitive expectation that cracks of low aspect-ratio a might be relatively easily closed by the application of uniaxial or hydrostatic pressure P derives support from the calculation by Walsh (1965) which indicates that c

Pc - E.oc

(1)

Since the Young's modulus E is of the order of 100 GPa for most rocks, cracks of aspect-ratio 10" in dry jacketted rock specimens should be closed by the application of-100 MPa (1 kbar) hydrostatic pressure. The applicability of equation (1) for order-of-magnitude estimates of crack closure pressures is supported by the direct SEM observation of the closure of cracks of very low aspect-ratio (such as those induced by thermal cycling) by uniaxial stresses less than 30 MPa (Batzle et al., 1980). Both static determinations of incompressibility (e.g. Brace, 1965) and ultrasonically measured elastic wave velocities (e.g. Birch, 1960, 1961; Simmons, 1964a,b; Christensen, 1974) for dry jacketted rocks increase markedly with increasing pressure below -300 MPa but approach values characteristic of the uncracked matrix at higher pressures. Only cracks with relatively low aspectratios (< 10" ) will be closed under these conditions; it follows that the elastic properties of cracked solids must be particularly sensitive to the presence of (low aspect-ratio) crack porosity. 3

Elasticity of cracked solids - perturbation theory Many theoretical descriptions of the elasticity of cracked solids are based upon calculation of the influence of low aspect-ratio ellipsoidal inclusions (dimensions a>b»c) on the potential energy of the

composite (elastically isotropic matrix plus inclusions) under prescribed conditions of applied stress. In general, both the volume fraction cp of the inclusions (i.e. the 'crack porosity') and their aspect-ratio a influence the elastic moduli of the composite. In many situations, however, these quantities appear in the expressions for the effective elastic moduli only as the ratio cp/a which is closely related to the crack density parameter 71

introduced by O'Connell and Budiansky (1974). In this expression, N is the number of ellipsoidal inclusions per unit volume and A and P are respectively the area (7tab) and perimeter of the principal cross-section of the low aspect-ratio ellipsoidal inclusion; the brackets denote expectation value. With this parameterization (equation 2), ellipsoidal and spheroidal (a=b»c) inclusions have essentially the same influence upon the elasticity of the cracked solid (O'Connell and Budiansky, 1974). It is therefore sufficient to consider the case of spheroidal inclusions of aspect ratio a=c/a for which 8 = Na = (3/4tc). (cp/a)

(3)

For a composite comprised of a matrix (with bulk and shear moduli K and G) which contains a small volume fraction 9 of isolated low aspect-ratio spheroidal inclusions ( a « l , moduli K and G), Walsh (1969) derived the following expressions for the effective bulk (K) and shear(G) moduli: ( K / K ) - = l 9 ( l - K / K ) 3K+4G _ 1

(G/G)~ l 1=

3 g + 4 g + 3 m G ( 3 K + G ) / ( 3K

f 4 G )

1 -G/G) [ 1 4Q 3 Q^3^^2G)/(3K+4G) +

7ta

2(3K+2G+2G) 3K+4G+37iaG(3K+G)/(3K44G)

( 4 ) J

With the substitution G=icor|, this model yields the frequency (co)-dependent complex moduli for the case in which isolated cracks are saturated with a fluid of viscosity T|. Thus, the effective moduli and wave velocities may be calculated for models

Petrophysical basis for seismological models.

of both dry (K,r| negligibly small for air) and fluidsaturated rock. Elasticity of cracked solids: the self-consistent approximation Owing to the neglect of interactions between cracks, such 'perturbation' theory (see also Kuster and Toksoz, 1974) is strictly applicable only in the limit of low crack density. One approach to the incorporation of these interactions (O'Connell and Budiansky, 1974) is based on a 'self-consistent approximation' in which each inclusion is considered as if it were embedded in a medium with the effective elastic moduli of the composite. The static moduli for 'soft-fluid' saturation evaluated in the limit oc->0 but with the parameter £2=(K/K)/a held constant (equations 22-24 of O'Connell and Budiansky, 1974) are as follows:

K/K=l-

16 1—v D8 l-2v v /

Elasticity of dry rocks The bulk modulus (K~10~4 GPa) and viscosity (r|~10~5 Pa.s) of air at ambient conditions are so small that both K and G = icorj are negligible. Since K-100 GPa for most rocks, Q. = (K/K)/a is also negligibly small unless a < 10~5. The effective moduli deriving from the perturbation model (equations 4 with K=G=0 - c.f. Nur and Simmons, 1969a) are expressible in terms of the crack density e (equation 3) as follows: K/K=[l+m(v).8]

G/G=1-||(1-V)[D+^]8

m(v)=-

(v-v)(2-v)

(l-v 2 )[D( l+3v)(2-v)-2( l -2v)]

-1

with

with 8=

low crack densities for which crack interactions are unimportant. At the moderate (but generally poorly known) crack densities characteristic of igneous and metamorphic rocks, however, the various theories predict significantly different effective moduli - leaving substantial ambiguity in the quantitative interpretation of the elasticity of cracked solids.

16 l - v 2 =10/3 for v = 1/4 9 l-2v

G/G = [l+n(v).8]

and

-l

with

r -o\ 4Q l - v 2

D=[l+-

371 l - 2 v IKr v /

n(v) = (5)

32(l-v)(5-v) =152/105 for v = 1/4 45(2-v)

(6) v and v being Poisson's ratio for the matrix and the composite respectively. These expressions are applicable under both dry (Q=0) and 'saturatedisolated' conditions (£2^0, see later discussion for explanation of these conditions). Bruner (1976) and Henyey and Pomphrey (1982) have argued that the theory of O'Connell and Budiansky over-accounts for crack interaction, and have proposed an alternative self-consistent approximation. Results deriving from the alternative self-consistent theories will be compared below with each other and with the predictions of the simpler perturbation theory. It will be shown that all of these theories converge, as expected, at

Formally identical expressions (to first order in 8) have been derived by Hudson (1981, equations 24). The corresponding results for the self-consistent theory of O'Connell and Budiansky (1974) derive from equations (5) with Q = 0 (and hence D=l): K/K=l-m(v).8 G/G=l-n(v).e _ _ _ 16(1 - v )(10v-3vv-v) V _ V 45 (2-v)

(7)

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

pressures <300 MPa at which such low aspect-ratio cracks are expected to close. The same amount of porosity associated with cracks of aspect-ratio 10~2 (8=0.024) would survive to much higher pressures (-1 GPa) but would reduce the wave velocities by only 2-3%. Secondly, the relative reduction in compressional wave velocity Vp is greater than that for the shear wave velocity Vs - resulting in a reduction of the ratio Vp/Vs .

The last of these equations must be solved for v before K/K and G/G can be calculated. An approximate solution for the effective elastic moduli based on the alternative self-consistent approximation (Bruner, 1976, equations 11) yields K/K«( 1 -2v)e _16e/9 /( l-2ve~ 8e/5 ) and G/G«( l + v ) e _ 1 1 l - f v e " 8 e / 5 )

The previously-mentioned convergence of the various models at low crack densities may be verified by examination of equations (6), (7) and (8); the effective moduli predicted by the various models for 8 = 0.024 are indistinguishable. However, at moderate crack densities (8=0.24, Table 1) comparable with those measured for competent granitic rock (Hadley, 1976), the different assumptions concerning crack interaction lead to substantial differences in calculated moduli and wave velocities.

(8)

These alternative theoretical descriptions of the elasticity of dry cracked solids are compared for the geophysically interesting case v = 1/4, 8 = 3/471 = 0.24 (corresponding, for example, to (p = a =10~3) in Table 1. The three models are in agreement on two fundamental points. Firstly, very low crack porosities can produce substantial reductions (-15-30%) in elastic wave velocities a provided that the aspect-ratio is also low (-10" ). This result is consistent with the pronounced pressure dependence of wave velocities observed for

Anelasticity of fluid-saturated rocks Since stresses may be relaxed by fluid flow within or between cracks on appropriate timescales DRY

WATER SATURATED Regime

flat

low a

'soft fluid'

Walsh

K/K

0.96

G/G

0.78

Vp/Vp

0.94

Vs/Vs

0.88

Vp/Vs Vp/Vs

1.06

Theory*

drained (IV)

isobaric (HI)

isolated (II)

glued (I)

low a

flat

flat OB

Bruner

Walsh

1.00

0.41

0.50

0.56

0.79

0.62

0.70

0.74

0.95

0.91

0.71

0.77

0.80

0.89

0.79

0.84

0.86

1.06

1.07

1.16

0.90

0.92

0.93

Table 1. Effective elastic moduli and wave velocities - for a cracked solid both dry and water-saturated with matrix Poisson's ratio a=l/4, and crack density 8=3/471 (corresponding, for example, to crack porosity cp= K)'3 and aspect ratio cx=10"3). The 'low a ' and 'soft fluid' calculations for the saturated isolated regime are based on K=2.3 GPa, K=100 GPa and a= 10"3 (i.e. £2=K/Ka=23). The perturbation theory presented by Walsh (1969) and the self-consistent theories of O'Connell and Budiansky (1974, 1977 OB) and Bruner (1976) all yield results which are applicable when a « l (low a). The effective moduli deriving from the self-consistent theories are evaluated either in the limit a->0 (flat) or as qc^O with_£2=K/Koc constant ('soft fluid' saturation). The effect of porosity on density is neglected in the calculation of Vp and Vs.

Petrophysical basis for seismological models.

it is obvious that the effective elastic moduli and hence wave velocities will depend upon the frequency at which they are measured. The following four regimes, illustrated in Fig. 1 and arranged below in order of decreasing frequency, may be encountered (O'Connell and Budiansky, 1977):

I. Glued regime

Saturated Isolated

Saturated feobanc

Glued

At sufficiently high frequencies (co»cov defined below) the effective rigidity G=icoT] of the fluid approaches that (G) of the matrix so that there will be no opportunity for significant relaxation of Drained

z;u O to <d

J3 b

* £

rata // ^ //

-C<D o w<0

/ / / / / / / / / , . \ \ \ N \ V V \ X / / / / / / / / / < . X • • • V • \ » \

P1 < P2

4 i

< i

t t

{..

•p(a)-

«-

Increasing Frequency u>

Fig. 1. Fluid flow regimes, effective bulk and shear moduli (K,G) and shear mode dissipation QG" for fluid-saturated rock (after O'Connell and Budiansky, 1977; see text). The sketches are intended to illustrate the response of an idealized 'rock' containing spheroidal cracks of a single low aspect-ratio a, to the application of either hydrostatic pressure (upper row) or pure shear stress (lower row). The principal modification needed for a rock with a distribution p(a) of aspect ratios is relaxation of the bulk modulus in the saturated isobaric regjime due to fluid flow in response to pressure (P) gradients between cracks of different aspect ratio as indicated on the K(co) plot. 1

Geol. Soc. Aust. Spec. Publ. 17, 81-114. shear stress by fluid flow even within individual cracks. Under these circumstances relative displacement parallel to the plane of a crack of its opposite faces is inhibited and the shear modulus is therefore unaffected by the presence of cracks (Fig. 1).

The self-consistent theory derived by O' Connell and Budiansky (1977) provides a complete description of the anelastic behaviour of fluidsaturated solids (Regimes I, II and HI and the transitions between them) in the limit of flat cracks (a-»0):

II. Saturated isolated regime

K/K = 1 32 1—v [(2-v')D+3C]e G/G=l- 45 2-v'

At somewhat lower frequencies ( C O V » C G » C O F - defined below), shear stresses within the fluid will be completely relaxed, but fluid flow in response to pressure gradients between adjacent cracks of different orientation will not occur (Fig. 1). The effective moduli for this regime are described by the theories of Walsh (1969) and O'Connell and Budiansky (1974) outlined above (equations (4) and (5), respectively). ID. Saturated isobaric regime At yet lower frequencies ( C O F » C O » C O D defined below), fluid flow between adjacent cracks of different orientation, in response to an applied shear stress, will maintain constant and uniform pore pressure (Mavko and Nur, 1975). The effective shear modulus is therefore unaffected by the presence of fluid and is given by the results presented earlier for the dry cracked solid (Fig. 1). Only for media with a spectrum p(a) of crack aspect ratios will hydrostatic compression give rise to gradients in pore fluid pressure driving flow, between cracks and voids (pores) of more equant shape, which results in partial relaxation of the bulk modulus (Budiansky and O'Connell, 1980). Otherwise the bulk modulus is unchanged from regime II. IV. Drained regime At very low frequencies ( C O « C O D - defined below) fluid flow between the pore volume of the cracked solid and an external reservoir maintained at constant pressure may occur on the timescale of the deformation. Under these conditions (termed "drained') neither bulk nor shear modulus is affected by the presence of fluid - the effective moduli being those presented above for the dry cracked solid.

where v' (which is not necessarily equal to v) is the root of the following equation: 16 1—v>2 [2(1-2V)C~(2-V )(1+3V)D]8. v'=v+-45 2-v' ,

D and C are given by -idCOF/CO D= 1-idcoF/co with j 9 fl-2v 16 1-v and -iecoy/co C- 1-iecov/co with K 2-v 1-v'

(9)

Approximate expressions have been derived for the characteristic frequencies for viscous relaxation (cov~aG/r| - Walsh, 1969) and for fluid flow between cracks (C0F~Ka /r| - O'Connell and Budiansky, 1977). The characteristicfrequencyCOD for fluid flow into and out of the stressed solid is a function of its characteristic dimension R and the hydraulic diffusivity K and is given by COD ~ 27CK/R , where K = kK/rjcp and k is the permeability (e.g., Gordon, 1974; Jones and Nur, 1984). 3

Equations (9) yield an effective shear modulus which, in general, will be both complex and frequency-dependent leading to dispersion and at-

K/K *

G/G*

v or v' being the root of:

1 0 1

,1 32 1-ve 15 2-v

32 (1-V )(1-2V)

V=VH— -

(0p»t0»0)D 1 1

(2-v)

, 32 (l-v')(5-v') le 45 (2-v') 16(l-v )(10v-3vv'-v') v =v - — — -e 45 (2-v') 2

IV drained (dry) A>D»(0 -

CI)V»CO»COF

III saturated isobaric

C0F~Ka /n

Ruid flow between adjacent cracks

D**

Viscous relaxation within fluid cov-Ga/rj

G)»COv 0

C**

II saturated isolated

Ruid flow in to and out of cracked solid cor>~27ikK/n6R

I glued

,1 - 16 1 - v e 9 l-2v ,l - 32 (l-v)(5-v) e 45 (2-v) 2

16 (1-V )(10V-3VV-V) v=v - — -e 45 (2-v) 2

* Effective moduli evaluated in the limit (a->0) of flat circular cracks. The case of spheroidal cracks of low but non-vanishing aspect ratio ('soft-fluid' saturation - O'Connell and Budiansky, 1974) is characterized by a different effective bulk modulus well approximated by K/K = 1 - K<))/K but similar shear modulus provided that Q = K / K a » l . ** See text. Table 2: The regimes of viscoelastic behavior of fluid-saturated cracked solids (after O'Connell and Budiansky, 1977)

Geol. Soc. Aust. Spec. Publ. 17, 81-114. tenuation of both compressional and shear waves (Fig. 1). However, there are four special parts of the frequency range, corresponding to the regimes described above, for which the imaginary part of the shear modulus (and hence both dispersion and attenuation) will be negligible. The 'elastic'moduli for all four regimes are compared in Table 2.

saturation at ambient conditions of low aspect-ratio cracks.

(ii) The results deriving from the self-consistent theory of O'Connell and Budiansky (1974) and from the perturbation theory of Walsh (1969) are indistinguishable at moderate crack densities (8=0.24) - in marked contrast to the The result K/K = 1 (equations (9), Table 2) is a situation for dry cracked solids. Examination consequence of the derivation of the effective of the self-consistent theory shows that the moduli in the limit oc-^0, in which case the inexpressions for K/K and G/G are indeed only fluence of fluid with K^O is to prevent relative mildly non-linear in 8 provided that Q » l . displacement of the opposite faces of a crack normal to the plane of the crack (O'Connell and Budiansky, 1974). The previously mentioned Modulus dispersion and dissipation results (equations (5) ) for 'soft-fluid saturation' The anelasticity implied by the general form of (c.f. Hudson (1981), equations 36 and 37) provide a more physically realistic description of the bulk equations (9) is associated with the transitions bemodulus for regimes I-III. For £2»1, which is the tween regimes I, II and III. By virtue of observation case for saturation of low aspect-ratio cracks (ii) above, useful insight into the nature of this (oc<10~ ) with water at ambient conditions (K=2.3 anelasticity may be gleaned from a simplified form GPa) or with basaltic melt (K~15 GPa), these equa- of these equations in which the self-consistent description is abandoned leading to the following tions reduce, with the help of equation (3) to expressions for the real and imaginary parts of G = G + iGi (O'Connell and Budiansky, 1977): K/K~l-||=l-K(p/K (10) 5Gf G = G- 5Gv 1+(cotv) 1+(C0TF) and to expressions for G/G and v which approach those given for the saturated isolated regime (II) in and Table 2. Gi = SGvcoxv2+ SGfcotf2 l+(C0Ty) 1+(CGTf) The predictions of the various theoretical treatments for a water-saturated cracked rock (with with v=l/4 and 8=3/471=0.24) are compared with previously discussed predictions for the same rock under dry conditions in Table 1. The behaviour of Ty = 1/coye water-saturated rock (regimes II and ID) is seen to differ from that of dry rock in two important 5Gy/G = 32 "l-v" 8 = 32 for v = 1/4 15 2-v 3 ? respects: 3

(i) The relative reduction of bulk modulus under saturated conditionsjs much less than that for the shear modulus; Vp/ Vs is therefore greater for the fluid-saturated solid (regimes II and EI) than for the matrix - the effect being most pronounced for the saturated isobaric regime (III). Note also that the wave velocities obtained for 'soft-fluid' saturation are well-approximated by the results for 'flat' cracks - a consequence of the fact that Q » l for water

for viscous relaxation and tf = 1/COFd 09

5Gf/G=~(1-v)£=~ for v = 1/4

forfluidflowbetween adjacent cracks

(11)

Petrophysical basis for seismological models.

The expressions forJ>Gv and 5GF are readily derived from those for G in Table 2. As noted by O'Connell and Budiansky (see also Walsh, 1969), these expressions describe the behaviour of a standard anelastic solid (e.g. Nowick and Berry, 1972) with two distinct relaxation mechanisms - each with its own characteristic relaxation time Ti and associated shear modulus relaxation 8Gi. It is evident from equations (11) that, for small crack densities 8, the relaxation of shear modulus due to viscous relaxation (8Gv) is greater (by a factor 12/7 for v=l/4) than that associated with fluid flow between adjacent cracks ( 8GF). Inspection of the shear moduli for regimes II and HI in Table 1 indicates that, for 8 = 0.24, 8Gv(0.21 G)> 8GF(0.17 G). However, the reverse is the case for high crack densities (>0.4 - see O'Connell and Budiansky, 1977, Fig.6). The phase velocity V and attenuation yfor plane travelling waves u = A exp(-yx) exp [ico(t-x/V)] are readily derived from the real and imaginary parts of the complex moduli via the expression V * = V r H V i = V M * / p ( M * is G for shear and K + 4 G / 3 for compressional waves): V=(V VVi /Vr r

y = coVi/(Vr +Vi ) 2

(12)

The quality factor Q, an alternative useful measure of dissipation, is given by QM" = Im(M*)/Re(M*) 1

(13)

attenuation is confined to a frequency interval C0o/10<C0<10(0o (e.g. Jackson, 1986). Generalization of the above theory to accommodate a continuous distribution of crack aspect-ratios (and hence of relaxation times) results in dispersion and attenuation over a much wider range of frequencies (Fig.7, O'Connell and Budiansky, 1977). Water-saturated rocks It remains to assess the application of this theory to laboratory and seismological studies of rocks saturated with water or silicate melt. For water saturation, r|~10~ Pa.s (Kaye and Laby, 1973) so that 10 Hz<cov ~ Ga/ri<10 Hz and 10Hz< COF~ Ka /ri<10 Hz for 10" < a< 10" . The characteristic frequency COD ~ 27iKK/R|(pR for the draining of a water-saturated cylindrical laboratory rock specimen of radius ~10" m and low crack porosity (10~ -10~ ) and permeability typical of a competent granite (10" -10 m - Brace et al., 1968) lies in the range 10" -10 Hz. Thus, the characteristic frequency coy for viscous relaxation is much higher than even ultrasonic frequencies (~10 HZ) SO that the dispersion and attenuation associated with the transition from the glued regime to the saturated-isolated regime are not expected to be experimentally observable. Fluid flow between adjacent cracks, on the other hand, is expected to contribute to dispersion and attenuation over a broad range of frequencies corresponding to those of resonance and ultrasonic methods (O'Connell and Budiansky, 1977). At the relatively low frequencies of resonant systems with external inertia and sub-resonant methods, the drained regime may be encountered with elastic moduli given by those of the dry cracked solid. Only in very special environments, such as permeable near-surface fault zones and partially saturated aquifers, is the drained regime likely to be encountered by seismic waves. 9

The dissipation QG \ obtained by substitution of G and Gi given by equations (11) into equation It follows that laboratory measurements on (13), attains maxima of amplitude approximately water-saturated rocks, whether conducted at SGv/2G and 8GF/2G for small crack densities. The ultrasonic or much lower frequencies, will not above discussion of the relative magnitudes of 8Gv necessarily sample the regime of seismic wave and SGF also applies directly, therefore, to the propagation. As an illustration of the consequenrelative magnitudes of the maximum dissipation ces, consider the case (cp=a=10~ ) 2of Table 121with2 associated with the two relaxation mechanisms. For the further assumptions that R = 10" m and 10 m the standard anelastic solid with a single charac- <k< 10"17 m 2. The characteristic frequencies are teristic frequency C0o, most of the dispersion and coy = 10 Hz and COF = 10 Hz with COD lying in the r

Geol. Soc. Aust. Spec. Publ. 17,81-114. range 10"1-101 Hz. Ultrasonic measurements at 10f\ Hz would thus sample the_saturated-isolated regime with_V /Vp~ 0.95 and V /V - 0.89 compared with Vp/Vp - 0.91 and V /Vs - 0.79 under normal seismological conditions within the saturated isobaric regime. Low-frequency subresonant studies with co <10" Hz on the other hand, will sample the drained regime with Vp/Vp - 0.71 and Vs/Vs - 0.79. These predictions of the theory are tested below for consistency with experimental observations of effective moduli and dissipation for fluid-saturated rocks. p

Partially molten rocks For basaltic melt, the viscosity r| varies between - 1 and 1 0 Pa.s within ~ 3 0 0 ° C of its liquidus to 3 GPa (Kushiro et al,, 1 9 7 6 ) ; thus 10 Hz<cov<10 Hz and 1 0 Hz< CQFCIO^HZ for 10" < a < 1 0 " . Somewhat lower characteristic frequencies 10" Hz< CQF< 1 0 Hz have been estimated by Mavko ( 1 9 8 0 ) for fluid flow ('melt squirt') between adjacent parts of the interconnected network of channels (tubules) along grain edges which represents the equilibrium distribution of partial melt under hydrostatic conditions (Waff 9and Bulau, 1 9 7 9 ; Bulau et al., 1 9 7 9 ) . Since COD^R , characteristic frequencies of order 10" HZ<COD<10HZ, lower than those for melt squirt, are anticipated for the draining of a cm-sized laboratory specimen of a partially molten ultramafic rock of 1 mm grain size. These estimates of the various characteristic frequencies suggest that viscous relaxation might be detected at ultrasonic frequencies - particularly for andesitic (r\ -10 -10 Pa.s - Kushiro et al 1976) or even more siliceous melts. Flow of melt between cracks or tubules, on the other hand, might account for substantial attenuation and dispersion across a very broadfrequency range encompassing most of the seismic band. Experimental investigation of effective bulk moduli within the saturated isobaric regime may, however, be compromised by the draining of cm-sized specimens on 0.1-1000 s timescales; this complication does not arise in the measurement of shear moduli which are unaltered by the transition from saturated isobaric to drained conditions (Fig. 1). Furthermore, it should be noted that a given melt fraction is much more effective in lowering the elastic wave velocities when dis6

tributed as an intergranular film of low aspect-ratio than it is when concentrated in intergranular tubules (Mavko, 1980). Generalization of the theory Finally, each of the theories discussed above, when generalized by incorporation of a pressuredependent distribution of crack aspect-ratios, provides a model which links the aspect-ratio distribution observed at ambient pressure with the pressure-dependence of the elastic wave velocities. In this way, Hadley (1976) has calculated Vp(P) and Vs(P) for Westerley granite from the measured distribution p(a) of crack aspect-ratios.. The selfconsistent theory of O'Connell and Budiansky (1974-) was found to provide a better match to Vp(P) and Vs(P) measured by Nur and Simmons (1969a) than the perturbation theory of Kuster and Toksoz (1974). Conversely, the measured pressure dependence of elastic wave velocities may be inverted via any of the alternative models for the distribution at ambient pressure of crack aspect-ratios (Toksoz et al., 1976; Cheng and Toksoz, 1979; Zimmerman, 1985). This review has been restricted to the description of the anelasticity of elastically isotropic media consisting of an isotropic matrix containing randomly oriented spheroidal inclusions. The theory of Hudson (1981) referred to above, is in fact more general and provides a description of the anelasticity of anisotropic media containing a population of aligned dry or fluid-saturated cracks (see also Crampin, 1984).

LABORATORY MEASUREMENTS OF WAVE VELOCITIES AND ATTENUATION The elastic/anelastic properties of rocks are amenable to laboratory investigation over a wide range of frequencies by application of a variety of well-established techniques. The overwhelming majority of available information derives from wave propagation measurements at ultrasonic frequencies (~10 HZ). The techniques used and the information deriving from such experiments will therefore be discussed at length. 6

Petrophysical basis for seismological models.

However, interest in mechanisms of seismicwave attenuation, and the possibility of substantial anelastic dispersion between ultrasonic and the much lower seismic frequencies (<10Hz) especially for fluid-saturated rocks, make it imperative that more experimental work be performed at substantially lower frequencies. The methods available have been reviewed elsewhere (Nowick and Berry, 1972; Jackson, 1986; Peselnick and Liu, 1987) and will therefore be described only very briefly here. Measurement of the frequencies and the rate of decay of free oscillations of rock specimens of various geometries provides data typically at frequencies of order l O ^ z . The addition of external inertia results in systems, often referred to as 'pendulums', with very much lower resonant frequencies topically l-10 2 Hz. At still lower frequencies (~10~ -1 Hz) corresponding to those of the teleseismic frequency band, rock anelasticity may be investigated by the observation of forced (subresonant) oscillations of rock specimens. Finally, quasistatic methods may be applied in which the strain resulting from the application of step-function stress is measured. Ultrasonic methods The potential of ultrasonic pulse-transmission methods for the characterization of rocks was first demonstrated by Hughes and his collaborators (e.g. Hughes and Jones, 1950; Hughes and Cross, 1951; Hughes and Maurette, 1957). In particular, it was shown that (i)

the wavelengths (typically a few mm) of elastic waves generated by piezoelectric transducers with resonance frequencies of order 1 MHz (10 6 Hz) are short enough relative to the dimensions (few cm) of readily available rock specimens and yet long enough relative to grainsize (-mm) to permit wave propagation experiments without excessive scattering;

(ii) both longitudinal and shear wave velocities may be measured on specimens of suitable geometry (length/diameter <5) by means of mode-conversion methods. However, most subsequent measurements have been performed with mode-specific transducers (oriented discs cut from single-crystal quartz,

BaTi03 or LiNb03, or polycrystalline ferroelectrics such as lead-zirconate-titanate PZT); (iii) the time interval between application of an electrical pulse to the transmitting transducer and the generation of a corresponding signal associated with the incidence of the elastic wave upon the receiving transducer at the far end of a specimen (a few cm in length) is readily measured within ±(l-2)% - adequate accuracy for seismological application. Consistency in the measurement of elastic-wave travel-times may be enhanced by the use of a calibrated mercury delay-line (Birch, 1960);

(iv) such experiments (unlike flexural or extensional mode resonance experiments) are very readily performed within a pressure vessel, which may be externally or internally heated. It has already been demonstrated in the previous section that porosity in the form of relatively low aspect-ratio cracks has a profound influence upon the elastic properties of rocks and that such porosity can be closed by the application of a few 100 MPa pressure. Accordingly, it is only by the conduct of highpressure experiments (on jacketted specimens) that wave velocities can be measured which are representative of the intrinsic elastic properties of the rock in situ;

(v) dispersion of shear waves in dry rocks at ambient temperature is negligible (i.e. <1%) for frequencies between 104and 10 Hz.

General results The following is an outline of some of the principal features of rock elasticity/anelasticity which have emerged from numerous published studies based on the methods described above. It is not intended to be a comprehensive review of all that has been done and is certainly not a compilation of numerical data. For such information, the reader is referred to a thorough review by Gebrande and Kem (1982) and a compilation of velocity data by Christensen (1982).

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

Pressure dependence of elastic wave velocities The ultrasonic methods pioneered by Hughes et al. have been systematically applied in numerous studies conducted under high-pressure conditions (typically to 1 GPa=1000 MPa=10 kbar) at ambient temperature (e.g. Birch, 1960; Simmons, 1964a,b; Kanamori and Mizutani, 1965; Christensen, 1966, 1974; Babuska, 1972; Manghnani etai, 1974). As expected from the discussion above, the measured wave velocities are low and very pressure-sensitive for most low-porosity igneous and metamorphic rocks for pressures <300 MPa. At higher pressures, the mean elastic wave velocities (averaged over three mutually orthogonal propagation directions) gradually approach those calculated, for the corresponding uncracked elastically isotropic composite, from the single- crystal elastic moduli for the constituent minerals. The most definitive comparison of this kind derives from the work of Babuska(1972) who measured P- and Swave velocities for 13 representative propagation directions relative to the rock fabric (see later discussion of anisotropy) for carefully chosen blocks of Twin Sisters' dunite and Stillwater bronzitite. His mean velocities for these essentially monominerallic rocks compare favourably with Voigt-Reuss-Hill (VRH) averages of the singlecrystal elastic moduli for olivine and orthopyroxene of comparable composition (Table 3, see also Jackson et a/.(1990) and O'Reilly et al. (1990)). Furthermore, it has been shown that the pressure dependence of wave velocities in the pres-

sure range 1-3 GPa is generally consistent with VRH averages of the pressure derivatives of the corresponding single-crystal elastic moduli. The pressure derivatives of compressional wave velocity (3Vp/5P) for mafic and ultramafic rocks of mantle origin are of order +0.10 km.s^GPa" 1 with values up to 50% larger for pyroxene-rich rocks (Christensen, 1974). Velocity-density

systematics

Velocity measurements at ambient pressure on drill cores ranging in density p from 1.3 to 2.8 Mg.m show that the compressional wave velocity is well approximated by a function only of density known as the Nafe-Drake relationship (see Talwani et al, 1959, Fig.2; Grant and West, 1965, p.200). This function is essentially linear for the range 2.2<p<2.8 Mg.m"3 with a very steep slope 3V p /3p of about 6 km.s"1 /Mg.m" 3 which presumably represents the influence of porosity on density and velocity. For low-porosity rocks of higher density (>2.8 Mg.m ) Vp is less sensitive to variation of density with av p /3p~ 3 km.s _1 /Mg.m" 3 (Woollard, 1959). Birch (1961) demonstrated that compressional wave velocities measured at 1 GPa on a wide variety of low porosity igneous and metamorphic rocks, with mean atomic weight m between 20 and 24, may be adequately described by an expression of the form

Density Mg.m

vP km.8 1

Vs km.s"1

Reference

Twin Sisters dunite (94% F090)

3.31

8.38

4.84

Babuska (1972)

Olivine (F093)

3.31

8.42

4.89

Kumazawa and Anderson (1969)

Stillwater bronzitite (93% En83)

3.29

7.89

4.59

Babuska (1972)

Orthopyroxene (En84)

3.34

7.85

4.76

Kumazawa (1969)

Rock/mineral

Table 3. Mean elastic wave velocities at 1 GPa for dunites and bronzitites compared with Voigt-Reuss-Hill averages of the corresponding single-crystal elastic moduli (at zero pressure).

Petrophysical basis for seismological models.

= a(m) + bp 1

(14)

with b ~ 3 km.s" /Mg.m and a standard deviation of about 0.3 km.s' 1 . The tendency for rocks rich in CaO to have higher velocities than those predicted by equation (14) led to a refinement (Simmons, 1964a) of the velocity-density model in which a(in) is replaced by A-B(m-21) + C[CaO]. Both the general applicability of such velocity-density systematics and the modification for CaO content have been confirmed by a host of subsequent studies (see Table 4 and Gebrande and Kem, 1982). Alternative elasticity-density systematics expressed in terms of the seismic parameter K/p = Vp2 9 - 4V S 73 (Anderson, 1967) or the bulk sound velocity V<p = (K/p) 1/2 (e.g. Wang, 1969) also provide adequate descriptions of the velocity-density data (e.g. Manghnani et al., 1974; Gebrande and Kern, 1982). Velocity-density systematics have thus been useful in systematizing a large body of experimental data and indeed in certain geophysical applications. Nevertheless, the RMS deviation of data from any of the alternative systematics is sufficiently large (~0.2 km.s"1) to complicate the interpretation of high-quality seismological models. Accordingly, there remains strong motivaNumber of data*

tion for laboratory wave velocity measurements performed in conjunction with regional seismological and petrological/geochemical investigations particularly of the continental lithosphere. Elastic anisotropy and rock fabric Measurements of elastic wave velocities have traditionally been performed for three mutually orthogonal propagation directions - often chosen to correspond with the principal axes of any obvious fabric. Pronounced velocity anistropy (Vmax ~ Vmin)/Vmean is observed in rocks at low confining pressures and is attributed in part to the preferred orientation of low aspect ratio porosity such as cleavage cracks in mica and amphibole grains (Babuska and Pros, 1984). Velocity anisotropy which persists to pressures (-1 GPa) sufficient to suppress such low aspect-ratio porosity, has been attributed to preferred orientation of the constituent mineral grains (Birch, 1961; Christensen, 1966). Compressional wave velocity anisotropy measured at high pressure is greatest for olivine-dominated ultramafics (7-12%), amphibolites and schists (both -10%), in which the highly elastically anisotropic olivine, amphibole and mica minerals display a strong preferred orientation (Babuska,

a km.8 1

b km.s"1/Mg.m"3

-0.98

2.76

Reference

Compressional wave velocity 65 (U) it

20-24

45 (U)

~21

-1.34

45 (C)

~21

26 (C)

21.6-22.5

22 (U) 20 (C)

a(m) = -0.98-0.7(m-21)+4.6[CaC)]

Birch (1961, solution 5) Simmons (1964a)

2.92

Birch (1961, solution 1)

-1.87

3.05

Birch (1961, solution 2)

-1.85

2.87

Manghnani et al, (1974)

20.6-21.3

-0.88

1.63

Christensen (1968)

21.6-22.5

-0.33

1.40

Manghnani et al. (1974)

Shear wave velocity

* C denotes velocities (1 GPa) and densities corrected for pressure-induced dimensional changes; U denotes data which have not been thus corrected. The study of Manghnani et al. (1974) was restricted to granulite facies rocks and eclogites. Table 4. Velocity-density systematics V = a(m) + bp

Geol Soc. Aust. Spec. Publ. 17, 81-114.

1984). Crustal igneous rocks are typically much less anisotropic (1-3%), and intermediate behaviour is observed for gneisses (3-7%), granulites (<3%), pyroxenites and eclogites (both -4%). Convincing seismological measurements of compressional-wave velocity anisotropy and associated shear-wave birefringence in the uppermost suboceanic mantle (e.g. Hess, 1964; Raitt et al., 1969; Forsyth, 1975) have focussed attention on the elastic anisotropy of olivine-rich rocks. Measurements on single-crystal olivine (F093, Kumazawa and Anderson, 1969) yield compressional wave velocities of9.89,7.73 and 8.43 km.s"1 respectively for propagation parallel to [100], [010] and [001]. The combination of this extreme velocity anisotropy (25%) with mechanisms capable of producing strong preferred orientation during deformation both in nature and in the laboratory (e.g. Nicolas and Poirier, 1976) explains the very pronounced elastic anisotropy found in deformed olivine-rich rocks. Detailed investigations in which velocities were measured in as many as 13 different propagation directions (Christensen and Ramananantoandro, 1971; Babuska, 1972) have revealed compressional and shear wave velocity anisotropics of 10-15% and 4-9% respectively in samples of Twin Sisters dunite. The relationship between the rock fabric and its elastic anisotropy has long been qualitatively understood; high compressional wave velocities are observed parallel to the preferred orientation of the fast [100] direction (Space group setting: Pbnm) and the lowest Vp is measured parallel to a concentration of the slowest [010] axes (Birch, 1961; Christensen, 1966; Christensen and Ramananantoandro, 1971). More recently, procedures due to Voigt and Reuss have been used to calculate the elastic anisotropy from the observed distribution of grain orientations (Crosson and Lin, 1971; Babuska, 1972; Peselnick and Nicolas, 1978; Seront etal. 1989; Panned evetal., 1989). Quantitative consistency (within 2-3% in velocity) has been demonstrated in these studies between measured and calculated elastic anisotropy.

Temperature dependence of wave velocities Intergranular stresses, associated with anisotropic thermal expansion of individual grains and differential thermal expansion between grains of different minerals, cause pervasive microcracking in rocks heated at or near ambient pressure. As a consequence, temperature derivatives of elastic wave velocities representative of conditions at depth in the Earth can be measured only under confining pressure sufficient to prevent thermal microcracking. It is evident from the pioneering resonance experiments of Birch (1943) and from the more recent ultrasonic work of Christensen (1979) and Kern (1982) that the required confining pressure for the measurement of 'intrinsic' velocities is of the order of 1 MPa K"1. Gas-medium apparatus for the ultrasonic measurement of elastic wave velocities under conditions of both high-pressure and high temperature (up to 1 GPa and 1300°C) has been described by Hughes and Cross (1951), Peselnick and Stewart (1975), Ramananantoandro and Manghnani (1978), Christensen (1979), and Murase and Kushiro (1979). An alternative approach, which has been widely applied to ultrasonic measurements on rocks, is the use of solid-medium apparatus capable of quasihydrostatic pressure to 2-5 GPa and temperatures to 500-900°C (Fielitz, 1971; Matsushima and Akeni, 1977; Volarovich et al., 1977; Ito et al., 1977; Levykin and Vavakin, 1978; Kem, 1982). Temperature derivatives of compressional wave velocity 0Vp/3T)p are typically - (4-6) x 10"4 km.s"1 K"1 for a wide variety of rock types (Christensen, 1979; Kern, 1982). Temperature derivatives measured for essentially monominerallic rocks are either comparable with or slightly greater than VRH derivatives calculated from the temperature dependence of the corresponding single-crystal elastic moduli. For example, for dunites, measured temperature derivatives of -6.1 x 10"4 and -5.6 x 10" km.s^K"1 (Ramananantoandro and Manghnani, 1978; Christensen, 1979; respectively) are slightly larger in magnitude than the VRH average of -4.8 x 10"4 km.s^K"1 (Kumazawa and Anderson, 1969) for olivine of comparable composition.

Petrophysical basis for seismological models.

For shear wave velocities, temperature derivatives measured at relatively low temperatures typically <500°C, whether by resonance (-10 kHz) or ultrasonic techniques (~1 MHz), are of order -2 x 10"4 km.s^K" 1 for a wide variety of rocks - with consistently higher values reported for dunite and marble (Birch, 1943; Kern, 1982). Since the variation of velocity with depth for material of constant chemical composition and mineralogy is given by

dV (dV) dP (dv) dz " dP \ )

and dP/dz = pg, it follows that the critical thermal gradient for which dV/dz = 0 is (dT/dz)cr = -pg@V/3P) T / 0V/dT) P

(16)

The typical values for (3Vp/3P)x - 0 . 1 km.s"1 GPa' 1 and (av P /3T) P - -5 x lO'^m.s^K" 1 presented above result in a critical thermal gradient of only 6K.km~1 which will usually be exceeded by the geothermal gradient under crustal conditions giving rise to the possibility of crustal low-velocity zones (Christensen, 1979 and later discussion). The role of fluids Systematic investigation of the influence of water saturation upon the elastic wave velocities of low-porosity rocks appears to have begun with the ultrasonic study of Nur and Simmons (1969a). It was found that compressional wave velocities were dramatically increased (8Vp=27-61 % for three low porosity granites) by water saturation at ambient pressure, whereas shear wave velocities were only slightly affected (SVs = 0-7%). It was also demonstrated that 5Vp is positively correlated with crack porosity (p, and that 8Vp decreases progressively with increasing effective pressure P e to - 0 at 300 MPa. (The effective pressure is given by P e = P-nPf where P and Pf are respectively the confining pressure and pore pressure and n~l - Todd and Simmons, 1972). Takeuchi and Simmons (1973) showed that the velocities of water-saturated rocks of low porosity are increased (8Vp~5%, 8Vs

-10%) essentially to those of the uncracked matrix by freezing of the pore fluid. The important role of the bulk modulus K of the fluid phase is evident from the work of Spencer and Nur (1976) in which velocities were measured at 100 MPa confining pressure as functions of increasing temperature, with pore pressure either held constant at 10 MPa (P e =90 MPa) or allowed to rise with increasing temperature to 100 MPa (P e =0). In the former case, the crack density is maintained at a relatively low level by the effective pressure, but K and Vp decrease at an accelerating rate with_increasing temperature between 150 and 300°C. Vs is essentiallyjndependent of temperature in this range, so that Vp/Vs decreases substantially with increasing temperature. Inthe latter case, increasing pore pressure maintains K essentially constant to ~300°C (by counteracting the influence of increasing temperature) and also reduces the effective pressure thereby increasing the crack density. Vs decreases steadily with increasing temperature with Vp essentially constant to 300°C; Vp/Vs therefore increases substantially with increasing temperature. Most of these ultrasonic observations are qualitatively consistent with the theory reviewed above for the saturated-isolated regime (Nur and Simmons, 1969a; Takeuchi and Simmons, 1973; O'Connell and Budiansky, 1974). In particular, the observed high and low Vp/Vs ratios for saturated anddryconditions, respectively^and the sensitivity of Vp/Vs to the bulk modulus K of the saturating fluid are all predicted by the theory. Moreover, inversion of the velocity-pressure data (via the theory) yields plausible pressure-dependent crack parameters cp and a , or e (Takeuchi and Simmons, 1973; O'Connell and Budiansky, 1974). Minor discrepancies between observations and theory are suggestive of incomplete saturation/dehydration and/or slightly increased crack densities in saturated rocks (O'Connell and Budiansky, 1974). Strong evidence for pronounced frequency dependence of the effective moduli of water-saturated low-porosity rocks is provided by the work of Gordon and his collaborators at the lower frequencies of resonance (~10 4 Hz) and sub-resonant forced oscillation (~10~2 Hz) experiments. Comparison of their own measurements of both extensional (E) and shear (G) moduli with ultrasonic data (Nur and

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

Simmons, 1969b) for Rhode Island granite shows that the modulus increment 8M associated with water saturation decreases dramatically with decreasing frequency and ultimately becomes slightly negative at the mHz frequencies of the sub-resonant experiments (Gordon and Rader, 1971; Gordon, 1974). Since 8M = Ofor all moduli only under drained conditions (Table 2), it would appear that the mHz experiments belong to this regime. The fact that SM is actually slightly negative rather than zero suggests that another factor is operative - either lubrication (Gordon, 1974), or increased crack density resulting from water saturation (O'Connell and Budiansky, 1977). Gordon's (1974) interpretation of the observed modulus relaxation under water-saturated conditions was based in part on the results of a parallel experiment on the same rock with glycerol saturant. A peak in dissipation and substantial associated modulus dispersion observed at 50 kHz and r|~4 Pa.s were attributed to viscous relaxation. Since the same mechanism would be operative at much higher frequencies -200 MHz (cocT| = constant) under water-saturated conditions, it was concluded that the observed modulus relaxation between MHz and mHz frequencies must be interpreted in terms of fluid flow. (The alternative attribution of the peak for glycerol saturation to local flow (O' Connell and Budiansky, 1977) carries the alternative implication that COF ~ 200 MHz for water saturation, and therefore that all of the relaxation between MHz and mHz frequencies would arise from fluid flow into and out of the specimen (i.e. the transition between saturated isobaric and drained regimes). Since this transition involves no shear modulus relaxation (Table 2), this interpretation is at variance with the observations for Rhode Island granite). The interpretation favoured above draws further support from the measurements by Murphy (1984,1985) at both ultrasonic (100-200 kHz) and sonic (1-7 kHz) frequencies on water-saturated Sierra White granite. Not only are the ultrasonic velocities consistent with the theory of O'Connell and Budiansky (1977) for the saturated isolated regime, but there is also evidence in the form of pronounced dissipation and modulus dispersion for the transition to saturated-isobaric conditions at sonic frequencies. The inference that saturated-iso-

lated conditions are commonly encountered at ultrasonic frequencies is entirely reasonable in view of the characteristic frequencies 10 Hz< COF <107 Hz estimated above for fluid flow between adjacent cracks of aspect ratio 10"4<a<10"2. Note, however, that saturated isobaric behaviour might be observed at ultrasonic frequencies for rocks in which cracks of aspect ratio ~10"2 are predominant (COF ~107 HZ). In this regard, it should be noted that the insensitivity of Vs to water saturation (5Vs<l%) for several of the rocks studied by Nur and Simmons (1969a) is a characteristic of the saturated isobaric regime. It is concluded that laboratory studies of the elasticIanelastic properties of water-saturated lowporosity rocks at mHz-MHz frequencies indicate that both wave velocities and attenuation are frequency dependent - in general accord with the theory outlined above. Ultrasonic measurements ( c o M H z ) commonly sample the saturated isolated regime, although saturated isobaric conditions may also be encountered in rocks where wider cracks (CL~10~2) are predominant. Subresonant methods at frequencies of ~1 mHz probe the drained regime. Pronounced frequency dependence of wave velocities and attenuation at intermediatefrequencies is associated with fluidflow on spatial scales ranging from that of interconnected adjacent cracks to the characteristic dimension (~10~2m) of laboratory rock specimens. It follows that great care must be exercised in the application of laboratory measurements, in the interpretation of seismological data. It should also be noted in passing that considerable modulus dispersion is evident even for nominally diy rock - much of it probably related to the presence of adsorbed water which is responsible for much of the dissipation in rocks at ambient conditions (Tittmann, 1977). The relatively low Q (typically -100) measured, particularly at kHz frequencies, for a wide variety of rocks, has been attributed to the deformation of thin capilliary films of adsorbed water at points of contact between grains (Murphy, 1982; Bulau et a/., 1984). The reorganization of water molecules hydrogenbonded to silanol groups (Si-OH) on the opposing surfaces provides a plausible molecular mechanism (Bulau et al., 1984) for the macroscopic viscoelas-

Petrophysical basis for seismological models.

ticity of the capilliary film. It has also been demonstrated, however, that Q may be increased to values >1000 either by outgassing of adsorbed water under conditions of ultrahigh vacuum (Tittmann, 1977) or by the application of an effective pressure of order 100 MPa (Katahara et al., 1982; Jackson et al., 1984; Jackson & Paterson, 1987).

APPLICATIONS TO THE SUB-CONTINENTAL LITHOSPHERE The nature of the continental crust - some tentative generalizations Laboratory measurements of elastic wave velocities and attenuation have been widely used in the interpretation of seismological models for the continental lithosphere. At the outset of this very selective discussion, it is appropriate to stress the very considerable local heterogeneity and regional variability of the continental crust. Field mapping of exposed crustal sections, the diverse chemistry and mineralogy of crustal xenoliths and, particularly, the results of recent seismic reflection profiling all attest to the considerable complexity of the continental crust (e.g. Smithson and Brown, 1977; Schilt et al, 1979; Kay and Kay, 1981; Barazangi and Brown, 1986a,b; Matthews and Smith, 1987; Leven et aU 1990). Nevertheless, some broad generalizations may be attempted. The total thickness of the continental crust is generally defined by the depth at which the compressional wave velocity Vp, determined by seismic refraction studies, attains values (7.8-8.2 km.s"1) characteristic of the ultramafic material of the uppermost mantle. A correlation has been proposed between total crustal thickness and the nature of the boundary between the crust and upper mantle (Prodehl, 1977; Bamford and Prodehl, 1977; Drummond and Collins, 1986) Relatively thin continental crust of-30-40 km total thickness is generally associated with a relatively sharp Moho - represented by a depth interval of less than 5 km probably with a laterally variable finely laminated structure, across which Vp increases from lower crustal values (typically 6.7-7.0 km.s"1 at -30 km) to those characteristic of the upper mantle (Braile and Chiang, 1986). Thick continen-

tal crust of ~40-60 km total thickness typically displays broadly similar velocity-depth structure in the depth interval to 30 km. Below 30 km depth, however; Vp is consistently greater than 7.0 km.s'1, often increasing steadily via generally steep velocity gradients, rather than abrupt discontinuities, to values approaching those of the upper mantle. In the Australian context, the respective seismological models for the Pilbara Craton (Drummond, 1983) and the Lachlan Fold Belt (Finlayson et al., 1979) provide examples of these thin and thick crustal stereotypes. Ringwood (1975) associated broadly similar thin and thick crustal models with stable continental and young orogenic regions, respectively. On the other hand, it has been suggested in a survey of recent seismological studies of the Australasian crust that crustal thickness may increase with increasing 'cratonic' age at least for post-Archaean provinces (Drummond and Collins, 1986). In fact, however, the similarities among the seismological models for the early Proterozoic North Australian craton and various provinces within the Palaeozoic Tasman Geosyncline are as striking as the differences which were used to support the correlation of crustal structure with age. In particular, most of these models have total crustal thicknesses of ~4055 km with generally high lower crustal velocities and velocity gradients and 'transitional' rather than sharp Moho's. Prodehl (1977) associates thick crust and the absence of a sharp Moho with both young geosynclinal provinces and stable continental regions. More recently, Nelson et al. (1987) have associated relatively thin crust (z<35 km) underlain by a sharp, reflective Moho with late extensional episodes, and relatively thick crust (z~45 km) terminated at depth by an unreflective Moho with the cratonic interior of North America. Thus there appears as yet to be no generally accepted tectonic interpretation of the considerable regional variability of the continental crust. Nevertheless, the stereotypes described above provide a useful basis for subsequent discussion. A further generalization of considerable value concerns the metamorphic grade of crustal rocks. Increasing pressure and temperature associated with increasing depth in the crust will generally result in increasing metamorphic grade - although exceptions may arise, particularly by preservation

Geol Soc. Aust. Spec. Publ. 17,81-114.

of non-equilibrium assemblages associated with underthrusting or with the cooling below ~800°C of mafic intrusions (Furlong and Fountain, 1986; Sobolev and Babeiko, 1987) A general progression is therefore expected from unmetamorphosed sediments and igneous rocks in the uppermost crust through greenschist and amphibolite facies in the upper-middle crust to granulite and even eclogite facies in the lower crust (e.g. Fountain and Salisbury, 1981) Thus, in the best-studied cross-section of continental crust exposed in the Strona-Ceneri and Ivrea-Verbano zones of the Southern Alps (Berckhemer, 1969; Fountain, 1976) a thick series of metamorphic rocks is encountered with properties which correlate well with those of the northern Italian crust as determined by regional geophysical surveys. Greenschist facies rocks of the upper crust, which are overlain by sediments, grade downwards into rocks of the amphibolite facies - gneisses and schists, and migmatites associated with occasional granite intrusives - with a mean density of 2.7 Mg.m*3 and an average compressional wave velocity at 600-1000 MPa of 6.5 km.s"1 (Fountain, 1976). Vp increases abruptly to -7.1 km.s"1 at the amphibolite/granulite facies transition with the appearance in the section of amphibolites, and then more steadily to -7.5 km.s"1 associated with mafic and metapelitic rocks (mean density 3.06 Mg.m" ) of the granulite facies. Near the base of the crustal section these granulites are interleaved with lenses of ultramafic material. A broadly similar crustal structure has been inferred for the Colorado Plateau of North America by comparison of velocities measured on crustal xenoliths with seismological models (Padovani et al., 1982).

Wave velocities in lower crustal granulites Clear evidence for the importance of granulite facies rocks in the lower crust, deriving from studies of relevant phase equilibria, of exposed deep crustal sections and of crustal xenoliths, has provided the motivation for a number of laboratory studies of their elastic wave velocities. Measurements on a suite of granulites of diverse origin by Christensen and Fountain (1975) indicated that both density and compressional wave velocity in-

crease systematically with increasing proportion of ferromagnesian minerals (amphibole, pyroxene and garnet) and also with increasing metamorphic grade - measured by the pyroxene/hornblende ratio. Felsic representatives of this suite of granulites contain more than 80% (feldspar+quartz) and have densities of 2.7-2.85 Mg.m"3 and Vp(l GPa) of 6.5-6.8 km.s"1. The intermediate and mafic members, which contain -40-60% (feldspar ± quartz) and -40-60% (pyroxene ± hornblende ± garnet), range in density from 2.9 to 3.1 Mg.m"3 and in compressional wave velocity (at 1 GPa) from 7.0 to 7.5 km.s"1. Generally similar results have been reported in several studies of regional suites of granulite-facies rocks (Manghnani et al., 1974; Evans, 1980; Chroston and Evans, 1983). For example, compressional wave velocities measured by Manghnani et al (1974) on granulites from the Adirondack Mountains range between 6.5 and 7.1 km.s"1 (1 GPa) for rocks with >50% SiC>2 and between 7.0 and 7.3 km.s'1 for the more mafic representatives. Mafic (and generally very magnesian) rocks of higher metamorphic grade dominate the lower crustal xenolith suite studied by Jackson and Arculus (1984). The unaltered mafic representatives are garnet granulites and eclogites with >70% (clinopyroxene (30-75%) + garnet (1040%)), densities of 3.27-3.4 Mg.m"3 and compressional wave velocities (at 400 MPa) of 7.5-8.0 km.s"1. These velocities may be compared with the range of 7.7-8.5 km.s"1 (at 1 GPa) for a worldwide suite of eclogites (Manghnani et al, 1974). The observed steady increase of Vp for mafic rocks from -7.0 to 7.5 km.s'1 in the pyroxene granulite facies, through 7.5-8.0 km.s"1 for garnet granulites to 7.7-8.5 km.s"1 for eclogites, is in excellent accord with the predictions of Green (1967; see also Ringwood, 1975) from velocity-density systematics. Inferred pressures (0.8-1.2 GPa) and temperatures (800-900°C) of equilibration for the mafic garnet granulite and eclogite xenoliths studied by Jackson and Arculus suggest an origin at depths of 30-40 km in the lower crust beneath the Adelaide Geosyncline of southern Australia. Correction of the measured velocities to in situ conditions via pressure and temperature derivatives discussed earlier in this paper 0Vp/3P= 0.1 km.s" GPa"1, aVp/aT = -5xl0"4 km.s^.K"1) yields values of 6.9-7.7 km.s"1. These velocities are comparable

100

Petrophysical basis for seismological models.

with those calculated for depths of 25-40 km along a 'tectonic' geotherm (heat flow 90 mW.rrf ) for underplated basaltic material in equilibrium assemblages of quartz tholeiite (7.2-7.4 km.s ) or olivine gabbro (7.5-7.7 km.s* ) compositions (Furlong and Fountain, 1986).

and Braile, 1982). Explanation of the observed variation of Vp in terms of temperature is strongly reinforced by the observation by these authors that the inferred value of 3Vpn/9T is comparable with temperature derivatives (~-5xl0" km.s^.K" ) measured in the laboratory for olivine-rich rocks. Furthermore, it may be noted that regional comIt follows that all of the characteristic features variability (e.g. in the degree of depleof the lower part (z>30 km) of thick continental positional of a lherzolitic upper mantle is not expected crust - high velocities of 7.0-7.7 km.s' , tion) contribute to significant variability in comprespronounced velocity gradients and the absence of tosional wave velocity (Jordan, 1979). The tendancy a sharp Moho - can be explained by a high proportion of mafic material in the garnet granulite facies. for Vp to decrease upon depletion, by removal of high-velocity phases (spinel or garnet) containing It must be remembered, however, that there exists a variety of alternative lithologies capable of much of the low melting-point fraction, is offset by producing comparably high lower crustal the increased velocities associated with the more velocities. These include pyroxenites, metapelitic magnesian character of the residual olivine and rocks of the garnet granulite facies, and physical pyroxene. mixtures of ultramafics with granulites of intermediate composition (Ringwood, 1975; Fountain, Fine structure in seismological models for the 1976). On the other hand, velocities measured on crust intermediate and mafic pyroxene granulites, when above discussion has focussed entirely on corrected for crustal temperatures, compare the The interpretation of the gross features of lithosfavourably with typical velocities for the continen- pheric velocity-density models. However, the comtal crust at depths of 15-30 km. Roughly similar velocities would also be expected for amphibolitic bination of increasingly detailed travel-time and assemblages (Christensen and Fountain, 1975; amplitude information from seismic refraction studies and the input from near vertical-incidence Fountain, 1976). reflection profiling provides for resolution of a great deal of fine structure as well (e.g. Mueller, P velocity anisotropy and regional variability 1977; Mooney and Brocher, 1987). Important feaThe continental lithosphere is also charac- tures needing explanation include upper crustal terized by both regional variability and elastic shear wave birefringence (Crampin, 1984; Cramanisotropy of the upper mantle. The considerable pin et al, 1986), upper- to mid-crustal low-velocity lateral heterogeneity of the continental lithosphere zones commonly associated with high attenuation severely complicates the study of any azimuthal (e.g. Landisman et al, 1971; Mueller, 1977) and variation of wave velocities. Nevertheless, strong reflections off the Moho and other sub-horizontal cases have been made for upper mantle anisotropy structure (particularly in the lower crust - e.g. Matin parts of the continental lithosphere (e.g. Bamford thews and Cheadle, 1986) and off dipping thrust et al, 1979; Fuchs, 1983). By analogy with the faults (e.g. Dohr and Meissner, 1975; Schilt et al, interpretation of P anisotropy in the oceanic litho- 1979). sphere mentioned earlier, the velocity anisotropy of the continental lithosphere is also attributed Observations in a growing number of regions primarily to a preferred orientation of olivine crys- of shear wave birefringence ('splitting') in the tals developed during deformation in the regional upper continental crust have been interpreted in stress field. terms of arrays of vertical water-filled microcracks aligned normal to the (usually horizontal) axis of The regional variability of P velocities (i.e. least compressive stress (Crampin, 1984; Crampin those of the uppermost mantle) is well correlated et al, 1986). The marginal resolution of the aswith surface heat-flow and thus also with inferred sociated compressional wave velocity anisotropy is Moho temperature (Cull and Denham, 1979; Black consistent with the fact (Table 1) that Vp is reduced 2

Geol. Soc. Aust. Spec. Publ. 17, 81-114. much less than Vs by the presence of liquid-filled cracks of low aspect ratio (Crampin et al, 1986). Reference was made earlier in this paper to the work of Christensen (1979) in which it was demonstrated that critical gradients for compressional wave velocity inversion in zones of constant composition and mineralogy are of order 6K.km . However, substantially higher gradients, which may be realized only in regions of relatively high heat-flow (e.g. Smith et al. 1975), will generally be required to offset the tendancy for Vp to increase with increasing depth in the crust as a consequence of the general trend towards higher metamorphic grade and more mafic composition. Alternatively, zones of relatively low velocity and high attenuation in both the uppermost crust and at greater depth in the upper crust may be caused by the presence of fluids under high pore pressure Pf (Gordon and Davis, 1968; Nur and Simmons, 1969a; Smithson and Shive, 1975; Mueller, 1977; Jones, 1985). Elastic wave velocities in low-porosity rocks increase with increasing effective pressure (P =P-nPf with n~l) below ~200 MPa as a consequence of the closure of relatively low aspect-ratio cracks aclO" , Todd and Simmons, 1972). Above -200 MPa the velocities approach those of the uncracked matrix. Even in the saturated-isolated regime, 0.1 volume % pore water in cracks of aspect ratio 10~ is sufficient to reduce the compressional and shear wave velocities by 5% and 10% respectively (Table 1). For the saturated-isobaric conditions, which are likely to prevail throughout much of the upper crust, the corresponding deficits in velocity are approximately 10% and 20%. Very locally, in special environments such as permeable near-surface fault zones, drained conditions may be encountered with even lower compressional wave velocities. It is clear that very substantial velocity contrast is to be expected between regions of high and low effective pressure in the crust. The observed general decrease of attenuation with increasing depth in the upper crust (Thouvenot, 1983) may be attributable to increasing effective pressure (c.f. Jackson and Paterson, 1987 and above). Exceptions in the form of high attentuation (low Q) may be anticipated in regions of low effective pressure (where pore fluid pressure approaches lithostatic) or high subsolidus temperatures where thermally activated relaxation mechanisms may be operative. _1

101

Laboratory measurements of elastic wave velocities are also being applied in the interpretation of deep reflectors within the crust and upper mantle. There is a general consensus that the amplitudes, frequency content and commonly multicyclic character of such reflections cannot be explained by a simple discontinuity in seismic impedance (i.e. the product of density and elastic wave velocity). Rather, the observed reflections are thought to result from constructive interference associated with laminated horizons composed of thin (-100 m) layers of alternately high and low impedance (e.g. Fuchs, 1969; Blundell and Raynaud, 1986; Braile and Chiang, 1986; Sandmeier et al., 1987). Such laminated horizons appear to be laterally variable with a characteristic scale of order 1 km (Braile and Chiang, 1986; Sandmeier et al., 1987) and are variously attributed to compositional layering, velocity anisotropy associated with preferred orientation especially of layer silicate minerals in mylonite zones, or with variations in fluid pressure (see Matthews and Cheadle, 1986, for a brief review). Synthetic seismograms have been calculated by Hale and Thompson (1982) and Fountain (1986) for the crustal cross-section exposed in the IvreaVerbano zone from the velocities measured by Fountain (1976). Intercalated layers of mafic granulite and lherzolite at the base of the crustal section are responsible for the appearance in the synthetics of considerable lower crustal/Moho reflectivity. This approach has been extended by Fountain et al (1987) and McDonough andFountain (1988) who have calculated synthetic reflection profiles from one-dimensional impedance models based upon the observed geology and measured wave velocities for exhumed metamorphic terranes. It was concluded that the reflectivity commonly associated with mylonite zones and the lower crust may be primarily attributable to lithologic heterogeneity (see also Hurich and Smithson, 1987). Pronounced compressional wave velocity anisotropy associated with preferred orientation of layer silicate minerals in highly deformed rocks from mylonite zones (Jones and Nur, 1984; Christensen and Szymanski, 1988) may also contribute to the observed reflectivity. Indeed, it was demonstrated (see also Fountain et al., 1984) that a laminated zone, comprising 100-

102

Petrophysical basis for seismological models.

150m thick layers alternately of relatively undeformed (and hence elastically isotropic) and highly deformed rock (with slow velocity normal to foliation), is capable of producing reflections such as those associated with the Wind River thrust in Wyoming (Brewer et al., 1982; Smithson et al, 1979). High pore fluid pressure associated with active faults may also contribute to the observed reflections by virtue of the large impedance contrast between essentially dry and fluid-saturated rocks (Jones and Nur, 1984; Matthews and Cheadle, 1986; and above). The observation that exposed granulite terranes, which are well known to be essentially anhydrous, are apparently less reflective than underlying lower crust in situ has been used in support of a role for fluids in the reflectivity of the lower crust (Klemperer et al., 1987). There is thus a range of factors which may be responsible for crustal reflections. An improved understanding of their relative importance is an important goal of ongoing research into deep seismic profiling (Matthews et al., 1987; Warner, 1990). The nature of the sub-continental asthenosphere It would be inappropriate to conclude even this highly selective discussion of the application of petrophysical data in the interpretation of seismological models for the subcontinental lithosphere without some reference to the underlying asthenosphere. In the context of modern geodynamics, the term 'lithosphere' describes the outer shell of the Earth which remains relatively rigid over geologically long periods of time. The lithosphere is thought to vary substantially in thickness between oceanic (-100 km) and continental (-200 km) regions, and is laterally discontinuous forming a set of plates in essentially rigid relative motion at the surface of the convecting planet. The lithospheric plates constitute the upper thermal boundary layer across which heat is lost by conduction from the convecting interior. Strongly temperature-dependent rheology is responsible for the contrast between the rigidity of the plates and the much lower viscosity of the underlying mantle. The depth of closest approach of the mantle geotherm and the appropriate solidus corresponds

to an inferred weak viscosity minimum known as the asthenosphere. For a review of these fundamentals of the plate tectonics hypothesis, see Turcotte and Schubert (1982). The temptation to associate anomalies in geophysical observables, such as seismic wave velocities and attenuation and electrical conductivity, with this inferred viscosity minimum of the asthenosphere has proved irresistible. Moreover, the requirement that basalts be produced, at midocean ridges and other widespread locations, by partial melting of an ultramafic upper mantle has led to the plausible (but facile) association of all of these phenomena with a laterally extensive zone of partial melt in the upper mantle. The possibility that partial melting is indeed responsible for low velocities (by -10%), high attenuation (Q" - 10~ ) and high electrical conductivity (-0.1 S.m" ) in the upper mantle beneath oceanic lithosphere and continental rift zones has been explored by Shankland et al. (1981). They concluded that flow of melt between adjacent cracks, described by the theory of O'Connell and Budiansky (1977) and reviewed above, might account for the reduced wave velocities and pronounced dissipation observed in the seismic frequency band. A crack density 8 of 0.2-0.3 associated with a (logarithmic) distribution of crack aspect ratios over the range (10" , 10" ) would suffice; the corresponding crack porosity is only 0.3-0.5%. 1

If high electrical conductivity is also to be explained by the presence of melt, approximately 3% of additional porosity is required in the form of an interconnected arrangement of more equant pores. A network of grain edge tubules, believed to be the equilibrium configuration for melt under hydrostatic conditions (see above), could provide the necessary combination of low and more equant aspect ratios (Shankland et al., 1981; Mavko, 1980). The volumetrically dominant more equant component of the porosity exerts less influence upon the elastic wave velocities than the much smaller fraction of crack porosity (above). However, the presence of a more equant component of the porosity would serve to extend to higher frequencies the dissipation associated with local fluid flow, and also to provide for significant relaxation of the bulk modulus K by fluid flow between adjacent parts of the network having very different

Geol. Soc. Aust. Spec. Publ. 17, 81-114. aspect ratios (Shankland et al., 1981; see also Fig. 1).

Despite the general plausibility of the above scenario, especially for regions of very high heat flow and active volcanism such as mid-ocean ridges and continental rift environments, it must be stressed that the minimum requirement for an explanation of the observed upper mantle geophysical anomalies is high temperature (Shankland et al., 1981). Attention has been drawn to the wide variety of thermally activated relaxation mechanisms capable of yielding reduced wave velocities and pronounced attenuation at high subsolidus temperatures (e.g. Jackson and Anderson, 1970; Nowick and Berry, 1972; Goetze, 1977; Karato and Spetzler, 1990). Such mechanisms typically involve the migration, either of point defects such as substitutional or interstitial impurities or of dislocations or grain boundaries, in response to stresses applied on appropriate timescales. The potential importance of solid state relaxation mechanisms is underscored by recent studies of ultramafic rocks in which the dissipation Q" is substantial under high subsolidus temperatures (Jackson et al, 1991) and increases relatively smoothly with increasing temperature without a dramatic discontinuity at the solidus (Berckhemer etal., 1982; Sato et al, 1989). 1

Finally, it is appropriate to emphasize the fact that tectonic stresses and those associated with seismic wave propagation interrogate the Earth's upper mantle on vastly different timescales. Given the inevitable frequency dependence of the anelastic response of material as complicated as the upper mantle, it is dangerous to equate low seismic velocities and high attenuation with low creep strength at geological timescales. Different physical mechanisms are almost certainly involved. Thus, although the presence of a few per cent of appropriately distributed melt might be expected to result in reduced wave velocities and increased attenuation via local fluid flow, there need not be a dramatic reduction in the creep strength above the solidus. In the presence of a few per cent of melt, the applied load continues to be borne by the crystalline matrix, but the presence of melt may influence the rheology in a way which will depend upon the

103

dominant deformation mechanism (Van der Molen and Paterson, 1979). For example, if the deformation of a partially molten ultramafic rock involves the crystal plasticity of olivine, the presence of melt may increase the strain rate by provision of a fast path for diffusion (Bussod and Christie, 1987) or reduce it by withholding water which is an important weakening agent (Karato, 1986). Certainly, no simple equivalence exists between the asthenosphere and the seismologically defined upper mantle zones of low velocities and high attenuation. CONCLUSIONS A review has been presented of the theory of the elasticity/anelasticity of cracked solids which provides a firm basis for the understanding of laboratory measurements of wave velocities and attenuation on both dry and fluid-saturated rocks of low porosity, and also for the seismological application of such measurements. While there is some divergence among the various theoretical descriptions for large crack densities, the greater practical difficulty is associated with the determination of the distribution p(a) of crack aspect ratios. In the following summary of important predictions of the theory, the numerical results quoted are based upon the self-consistent theory of O'Connell and Budiansky (1977). 1. The influence of porosity (p upon the elastic moduli or wave velocities of cracked solids is described in terms of a crack density parameter e given for spheroidal pores of aspect ratio a by the expression e = (3/4rc). (cp/a). 2. Since the closure pressure P for pores of aspect ratio a is given approximately by Ea (where E - 100 GPa is the Young's modulus), it follows that cracks of aspect ratio 10~ are closed by pressure of order 100 MPa (1 kbar). c

3. The form of the expression for the crack density parameter 8 indicates that low aspect ratio (i.e. crack) porosity will be much more effective in lowering elastic moduli and wave velocities than more equant porosity. Indeed, a mere 0.1 volume % of dry cracks with aspect

104

Petrophysical basis for seismological models. ratio 10- 3 results in compressional and shear wave velocities respectively 29% and 21% lower than those for the uncracked matrix. For the same volume fraction of wider cracks of aspect ratio 10 , the reductions in velocity are an order of magnitude smaller. The presence of dry cracks is reflected in a lower ratio Vp/Vs than for the matrix.

The spatial range of fluid flow in stressed saturated rocks will be a strong function of the timescale of the deformation. Elastic moduli and wave velocities are therefore predicted to be functions of the frequency at which the rock is interrogated. Four regimes defined by the scale of fluid flow - glued, saturated isolated, saturated isobaric and drained - are anticipated (at progressively lower frequencies) in the model of O'Connell and Budiansky (1977). Pronounced modulus/velocity dispersion (frequency dependence) and attenuation are associated with the transition between adjacent regimes; such transitions are expected when the period of the imposed stress is comparable with the characteristic time for viscous relaxation (by fluid flow within a crack), fluid flow between adjacent interconnected cracks, or the draining of the pore space of a laboratory rock specimen to an external reservoir maintained at constant pressure. Water saturation at ambient conditions of a cracked solid with 9 = a = 10"3 (as above) results in compressional and shear wave velocity reductions relative to those of the matrix of 5% and 11% respectively for the saturated isolated regime and 9% and 21 % for the saturated isobaric regime. Vp/Vs is thus always greater for the saturated solid than for the matrix (irrespective of regime) in contrast with the behaviour of dry cracked solids.

Owing to the expectation of pronounced frequency dependence of both wave velocities and attenuation for fluid-saturated media, care must be exercised in the application of laboratory results to the interpretation of seismological data (see later conclusions). General results emerging from laboratory studies (predominantly at ultrasonic frequencies)

of low-porosity 'crystalline' rocks have been assessed within this theoretical framework, as follows: 5.

The marked pressure dependence of wave velocities below ~ 300 MPa is attributed to closure of low aspect ratio cracks; at pressures of order 1 GPa both wave velocities and their pressure derivatives approach the values calculated for ideal (zero porosity) aggregates from single-crystal elasticity data. 5Vp/dP is typically of order 0.10 km s GPa"1 for a wide variety of crustal and upper mantle lithologies.

For rocks with a wide range of chemical composition and metamorphic grade there exist striking correlations between density and elastic wave velocities. For low-porosity crystalline rocks, velocities measured at 1 GPa confining pressure yield 3Vp/3p ~ 3 km.s"1 /Mg.m"3. The predictive power of these correlations is limited by the considerable scatter (RMS deviation ~ 0.2 km.s"1) leaving scope for further laboratory measurements especially in conjunction with detailed geochemical and/or seismological studies.

Elastic wave velocity anisotropy measured at high pressure is most pronounced in peridotites, amphibolites and schists where it is caused by the preferred orientation of olivine, amphibole and mica minerals respectively.

Temperature derivatives of wave velocities must be measured under conditions of high confining pressure (~ 1 MPa/K) in order to suppress microcracking induced by intergranular stresses of thermal origin. Ultrasonic measurements to ~ 800 K (~ 500°C) at 500 MPa yield typical values for (3Vp/3T)p of -(4-6) x 10" km.s^.K"1. Substantially more negative derivatives might be expected at higher subsolidus temperatures and (the lower) seismic frequencies as a consequence of anelastic relaxation.

Measured pressure and temperature derivatives may be combined to estimate critical thermal gradients of order 6 K.km'1 for which dVp/dz = 0 in the crust and upper mantle. The

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

fact that this gradient will usually be exceeded under crustal conditions may provide locally for the existence of low-velocity zones in regions of the Earth's crust where increasing metamorphic grade and/or progressively more mafic compositions do not guarantee positive avp/az. 10. Laboratory studies of the elasticity!(inelasticity of water-saturated rocks are not yet definitive, but there are indications that ultrasonic measurements commonly sample the saturated isolated regime; only for rocks dominated by relatively wide cracks a - 10~2 is the saturated isobaric regime encountered at ultrasonic frequencies MHz). At the very low frequencies mHz) of sub-resonant methods, the drained regime is encountered. Pronounced dispersion of wave velocities and marked attenuation observed at frequencies intermediate between mHz and MHz are attributed both to fluid flow between adjacent cracks and also to the draining of cm-sized laboratory specimens. It follows that conditions of seismic wave propagation are not easily simulated in the laboratory. For example, for the case of water saturation at ambient conditions of a rock with (p=oc= 10"3' the characteristic frequency CDF for fluid flow between adjacent cracks is ~ 10 4 Hz so that MHz ultrasonic measurements would_ sample the saturated isolated regime with Vp/Vp - 0.95 and Vs/Vs ~ 0.89 (relative to matrix). If a typical competent granite permeability k (10~2 ^kclO" 1 7 m ) is also assumed, cm-sized specimens will be drained with characteristic frequencies in the range 10"2 - 102 Hz. Subresonant methods at frequencies below 102 Hz wouldtherefore encounter the drained regime with V P / V P ~ 0.71 and Vs/Vs ~ 0.79. For this particular set of conditions, only resonant techniques at frequencies of ~10 3 Hz would duplicate the saturated isobaric conditions with Vp/Vp - 0.91 and Vs/Vs - 0.79 expected to apply under seismological conditions. At the higher viscosities of silicate melts, viscous relaxation (by fluid flow within each crack) might be detected at ultrasonic frequencies for andesitic or more siliceous melts.

105

Flow of melt between adjacent cracks or tubules might account for substantial dispersion and attenuation across a very broad frequency range encompassing most of the seismic band. Experimental investigation of effective bulk moduli within the saturated isobaric regime may, however, be compromised by the draining of laboratory specimens on 1 - 1000 s timescales. Selected applications were presented of the foregoing to the interpretation of seismological models for the sub-continental lithosphere: 11. Two tentative generalizations concerning the nature of the continental crust were presented. Firstly, it was noted that there appears to be a general correlation between crustal thickness and the character of the Moho such that thin crust (z<35 km) is underlain by a sharp Moho while thicker crust is characterized by high velocities and velocity gradients below 30 km depth and a more gradational transition to typical upper mantle velocities. Secondly metamorphic grade will generally increase systematically with depth - although nonequilibrium assemblages associated with underthrusting and the cooling of mafic intrusions may be preserved at depth. 12. Wave velocity measurements for lower crustal granulites were briefly reviewed. Mafic garnet granulites and ecologites have in situ lower crustal compressional wave velocities of 6.9 -7.7 km s"1. An increase with increasing depth of the proportion of mafic material arising perhaps from basaltic underplating, might therefore explain high lower crustal velocities and velocity gradients in thick continental crust and also a diffuse rather than sharp Moho. 13. Regional variability of Pn velocities characteristic of the uppermost mantle correlates with heat flow and hence with inferred Moho temperatures in a way that is consistent with measured temperature derivatives of elastic wave velocities. Evidence for upper mantle elastic anisotropy within the continental lithosphere is interpreted in terms of the preferred orientation of olivine.

106

Petrophysical basis for seismological models.

14. Quantitative interpretation of crustal fine

structure such as upper crustal shear wave

splitting, upper-mid crustal low-velocityzones and sub-horizontal reflectors in the lower crust in terms of petrophysical data is in its infancy. It is likely that microcracking, lithological layering, velocity anisotropy associated with preferred orientation of layer silicate minerals in highly deformed (mylonite) zones and variations in fluid pressure all contribute to the complexity of the seismic signature of the continental crust. 15. Upper mantle zones of low seismic wave velocities and high attenuation are at-

tributable to thermally activated anelastic relaxation mechanisms. In zones of particularly high heat flow and active volcanism, a few percent of partial melt may be plausibly invoked to explain both the anelasticity and anomalous electrical conductivity. Elsewhere, solid state relaxation mechanisms may predominate. The connection commonly made between low seismic wave velocities (and high attenuation) and the low creep strength of an inferred asthenosphere is without a secure foundation.

ACKNOWLEDGEMENTS This manuscript has been significantly improved as a consequence of thoughtful reviews by B.J. Drummond, B.L.N. Kennett, R.D. Shaw, S.F. Cox and R.J. O'Connell. D. Devir, J. Keller, R. Colombo and M. Davern are thanked for invaluable assistance in the preparation of the manuscript.

BABUSKA V. 1984. P wave velocity anisotropy in crystalline rocks. Geophysical Journal of the Royal Astronomical Society 7 6 , 1 1 3 - 1 1 9 . BABUSKA V. & PROS Z. 1 9 8 4 . Velocity anistropy in granodiorite and quartzite due to the distribution of microcracks. Geophysical Journal of the Royal Astronomical Society 7 6 , 1 2 1 - 1 2 7 . BAMFORD D . , JENTSCH M . & PRODEHL C. 1979. P

anisotropy studies in northern Britain and the eastern and Western United States. Geophysical Journal of the Royal Astronomical Society 5 7 , 3 9 7 - 4 2 9 . BAMFORD D . & PRODEHL C. 1 9 7 7 . Explosion seismology and the continental crust-mantle boundary. Geological Society of London, Journal 1 3 4 , 1 3 9 - 1 5 1 . BARAZANGI M . & BROWN L. eds. 1986a. Reflection seismology: a global perspective, American Geophysical Union, Washington, Geodynamics Series, 13, 311 p. BARAZANGI M . & BROWN L. eds. 1986b. Reflection seismology: the continental crust, American Geophysical Union, Washington, Geodynamics Series, 14, 339 p. BATZLE M.L., SIMMONS G, & SIEGFRIED R.W. 1 9 8 0 .

Microcrack closure in rocks under stress: direct observation. Journal of Geophysical Research 85,

7072-7090.

BERCKHEMER H. 1969. Direct evidence for the composition of the lower crust and the Moho. Tectonophysics 8, 97-105. BERCKHEMER H., KAMPFMANN W., AULBACH E. & SCHMELING H. 1982. Shear modulus and Q of forsterite

and dunite near partial melting from forced oscillation experiments. Physics of the Earth and Planetary Interiors 29, 30-41.

REFERENCES

BIRCH F. 1943. Elasticity of igneous rocks at high temperatures and pressures. Geological Society of America, Bulletin 54,263-286.

ANDERSON D . L .

1967. A seismic equation of state. Geophysical Journal of the Royal Astronomical Society 13, 9-30.

BIRCH F. 1960. The velocity of compressional waves in rocks to 10 ki lobars, I. Journal of Geophysical Research 65, 1083-1102.

BABUSKA V. 1972. Elasticity and anisotropy of dunite and bronzitite. Journal of Geophysical Research 77, 6955-6965.

BIRCH F. 1961. The velocity of compressional waves in rocks to 10 kilobars, II. Journal of Geophysical Research 66, 2199-2224.

Geol. Soc. Aust. Spec. Publ. 17, 81-114. BLACK P.R. & BRAILE L.W. 1982. P velocity and cooling of the continental lithosphere. Journal of Geophysical Research 87, 10557-10568. N

BLUNDELL D.J. & RAYNAUD B. 1986. Modelling lower

crust reflections observed on BIRPS profiles. In Barazangi M. and Brown L. eds, Reflection seismology: a global perspective, pp. 287-295, American Geophysical Union, Washington, Geodynamics Series, 13. BOURBI£ T., COUSSY O. & ZINSZNER B. 1987. Acoustics of porous media, Gulf Publishing Company, Houston, 334 p. BRACE W.F. 1965. Some new measurements of linear

compressibility of rocks. Journal of Geophysical Research 70, 391-398.

BRACE W.F., WALSH J.B. & FRANCOS W.T. 1968.

Permeability of granite under high pressure. Journal of Geophysical Research 73, 2 2 2 5 - 2 2 3 6 .

BRAILE L.W. & CHIANG C.S. 1986. The continental Mohorovicic discontinuity: results from near-vertical and wide-angle seismic reflection studies. In Barazangi M. and Brown L. eds. Reflection seismology: a global perspective, pp. 2 5 7 - 2 7 2 . American Geophysical Union, Washington, Geodynamics Series, 13. BREWER J.A., ALLMENDINGER R.W., BROWN L.D., OLIVER J.E. & KAUFMAN S. 1982. COCORP profiling

across the Rocky Mountain front in Southern Wyoming, I. Laramide structure. Geological Society of America, Bulletin 93, 1242-1252. BRUNER W,M. 1976. Comment on 'Seismic velocities in dry and saturated cracked solids' by R.J. O'Connell and B. Budiansky. Journal of Geophysical Research 81, 2573-2576.

BUDIANSKY B. & O'CONNELL R.J. 1980. Bulk dissipation in a heterogeneous material. In Nemat-Nasser S. ed. Solid Earth Geophysics and Geotechnology, pp. 1-10. American Society of Mechanical Engineers, New York. BULAU J.R., WAFF H.S. & TYBURCZY J.A. 1979.

Mechanical and thermodynamic constraints on fluid distribution in partial melts. Journal of Geophysical Research 84, 6 1 0 2 - 6 1 0 8

107

BULAU J.R., TITTMANN B.R., ABDEL-GAWAD M . & SALVADO C. 1984. The role of aqueous fluids in the

internal friction of rock. Journal of Geophysical

Research 89,4207-4212.

BUSSOD G.V. & CHRISTIE J.M. 1987. Experimentally deformed lherzolite at hypersolidus conditions: an example of diffusion-controlled power-law creep. IUGG XIX General Assembly, Abstracts 1, 196. CHENG C.H. & TOKSOZ M.N. 1979. Inversion of seismic velocities for the pore aspect-ratio spectrum of a rock. Journal of Geophysical Research 84, 7 5 3 3 - 7 5 4 4 . CHRISTENSEN N.I. 1966. Elasticity of ultrabasic rocks. Journal of Geophysical Research 71, 5 9 2 1 - 5 9 4 1 . CHRISTENSEN N.I. 1968. Chemical changes associated with upper mantle structure. Tectonophysics 6, 3 3 1 - 3 4 2 . CHRISTENSEN N.I. 1974. Compressional wave velocities

in possible mantle rocks to pressures of 30 kilobars. Journal of Geophysical Research 79, 4 0 7 - 4 1 2 . CHRISTENSEN N.I. 1979. Compressional wave velocities

in rocks at high temperatures and pressures, critical thermal gradients and crustal low-velocity zones. Journal of Geophysical Research 84, 6 8 4 9 - 6 8 5 7 . CHRISTENSEN N.I. 1982. Seismic velocities. In Carmichael R.S. ed, Handbook ofphysical properties of rocks. Vol. II. pp. 1-228. CRC Press, Boca Raton. CHRISTENSEN N.I. & RAMANANANTOANDRO R . 1971.

Elastic moduli and anisotropy of dunite to 10 kilobars. Journal of Geophysical Research 76, 4 0 0 3 - 4 0 1 0 . CHRISTENSEN N.I. & FOUNTAIN D . M . 1975. Constitution

of the lower continental crust based on experimental studies of seismic velocities in granulite. Geological Society of America, Bulletin 86, 227-236. CHRISTENSEN N.I. & SZYMANSKI D.L. 1988. Origin of

Reflections from the Brevard Fault Zone. Journal of Geophysical Research 9 3 , 1 0 8 7 - 1 1 0 2 .

CHROSTON P.N. & EVANS C.J. 1983. Seismic velocities of granulites from the Seiland Petrographic Province, N. Norway: Implications for Scandinavian Lower Continental crust. Journal of Geophysical Research 52, 14-21.

108

Petrophysical basis for seismological models.

CRAMPIN S. 1984. Effective anisotropic elastic constants 76, 135-145.

D.M. 1976. The Ivrea-Verbano and Strona-Ceneri zones, Northern Italy : a cross-section of the continental crust - new evidence from seismic velocities of rock samples. Tectonophysics 33, 145-165.

CRAMPIN S., MCGONIGLE R. & ANDO M. 1986.

FOUNTAIN D.M. 1986. Implications of deep crustal

for wave propagation through cracked solids. Geophysical Journal of the Royal Astronomical Society

Extensive-dilatancy anisotropy beneath Mount Hood, Oregon and the effect of aspect ratio on seismic velocities through aligned cracks. Journal ofGeophysical Research 91,12703-12710.

CROSSON R.S. & LIN J.W. 1971. Voigt and Reuss prediction of anisotropic elasticity of dunite. Journal of Geophysical Research 76, 570-578. CULL J.R & DENHAM D. 1979. Regional variations in

FOUNTAIN

evolution for seismic reflection interpretation. In Barazangi M. and Brown L. eds. Reflection seismology: the continental crust, pp. 1-7. American Geophysical Union, Washington, Geodynamics Series, 14. FOUNTAIN D.M., HURICH C.A. & SMITHSON S.B. 1984.

Seismic reflectivity of mylonite zones in the crust.

Geology 12,195-198.

FOUNTAIN D.M., MCDONOUGH D.T. & GORHAM J.M.

Australian heat flow. BMR Journal of Australian Geology & Geophysics 4,1-13.

1987. Seismic reflection models of continental crust based on metamorphic terrains. Geophysical Journal of the Royal Astronomical Society 89, 61-66.

DOHR G.R & MEISSNER R. 1975. Deep crustal reflections

FOUNTAIN D.M., & SALISBURY M.H. 1981. Exposed

in Europe. Geophysics 40, 25-39.

DRUMMOND B.J. 1983. Detailed seismic velocity/depth models of the upper lithosphere of the Pilbara craton, N.W. Australia. BMR Journal of Australian Geology & Geophysics 8,35-51. DRUMMOND B.J. & COLLINS C.D.N. 1986. Seismic evidence for underplating of the lower continental crust of Australia. Earth and Planetary Science Letters 79, 361-372. EVANS C.J. 1980. The seismic velocities of the Ox

Mountain granulites of Ireland and the implications for the interpretation of the crustal structure of north Britain. Geophysical Journal of the Royal Astronomical Society 63,417-426.

FIELITZK. 1971. Elastische Wellengeschwindigkeiten in

verschiedenen Gesteinen unter hohem Druck und bei Temperaturen bis 750°C. Zeitschrift fur Geophysik 37, 943-956.

cross-sections through the continental crust: implications for crustal structure, petrology and evolution. Earth and Planetary Science Letters 56,263-277. FUCHS K. 1969. On the properties of deep crustal

reflectors. Zeitschrift fur Geophysics 35, 133-149.

FUCHS K. 1983. Recently formed elastic anisotropy and petrological models for the continental subcrustal lithosphere in Southern Germany. Physics of the Earth and Planetary Interiors 31, 93-118. FURLONG K.P. & FOUNTAIN D.M. 1986. Continental

crustal underplating: thermal considerations and seismic-petrologic consequences. Journal of Geophysical Research 91, 8285-8294.

GEBRANDE H. & KERN H. 1982. Elastic wave velocities and constants of elasticity of rocks and rock-forming minerals. In Angenheister G. ed. Physical Properties of Rocks, Vol. lb. pp. 1-140. Landolt-Bornstein, Group V, Springer, Berlin.

Explosion seismic profiles, and implications for crustal evolution, in southeastern Australia. BMR Journal of Australian Geology & Geophysics 4,243-252.

GOETZE C. 1977. A brief summary of our present day understanding of the effect of volatiles and partial melt on the mechanical properties of the upper mantle. In Manghnani M.H. and Akimoto S. eds. High-pressure research: Applications in Geophysics, pp. 3-23. Academic Press, New York.

FORSYTH D.W. 1975. The early structural evolution and

GORDON R.B. 1974. Mechanical relaxation spectrum of

FINLAYSON D.M., PRODEHL C. & COLLINS C.D.N. 1979.

anisotropy ofthe oceanic upper mantle. Geophysical Journal ofthe Royal Astronomical Society 43, 103-162.

crystalline rock containing water.Journal ofGeophysical Research 79, 2129-2131.

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

109

GORDON R.B. & DAVIS L.A. 1968.

Velocity and attenuation of seismic waves in imperfectly elastic rock. Journal of Geophysical Research 73, 3917-3935.

HURICH C.A. & SMITHSON S.B. 1987. Compositional variation and the origin of deep crustal reflections, Earth and Planetary Science Letters 85, 416-426.

GORDON R.B. & RADER D. 1971. Imperfect elasticity of rock: its influence on the velocity of stress waves. In Heacock J.G. ed Structure and physical properties of the Earth's crust, pp. 235-242. American Geophysical Union, Washington.

ITO H., MIZUTANI H., ICHINOSE K. & AKIMOTO S. 1977.

GRANT F.S. & WEST G.F. 1965. Interpretation theory in

applied geophysics. McGraw-Hill, New York.

GREEN T.H. 1967. An experimental investigation of sub-solidus assemblages formed at high pressure in high-alumina basalt, kyanite eclogite and grosspydite compositions. Contributions to Mineralogy and Petrology 16, 84-114.

Ultrasonic wave velocity measurements in solids under high pressure using solid pressure media. In Manghnani M.H. and Akimoto, S. eds High-pressure research: applications in Geophysics, pp. 603-622. Academic Press, New York. JACKSON D.D. & ANDERSON D.L. 1970.

Physical mechanisms of seismic wave attenuation, Reviews of Geophysics and Space Physics 8, 1-63.

JACKSON I. 1986. The laboratory study of seismic wave attenuation. In Hobbs B.E. and Heard H.C. eds Mineral and rock deformation: laboratory studies, pp. 11-23. American Geophysical Union, Washington.

HADLEY K. 1976. Comparison of calculated and observed crack densities and seismic velocities in westerly granite. Journal of Geophysical Research 81,

JACKSON I. & ARCULUS R.J. 1984. Laboratory wave

HALE L.D. & THOMPSON G.A. 1982. The seismic

JACKSON I. & PATERSON M.S. 1987. Shear modulus and

3484- 3494.

velocity measurements on lower crustal xenoliths from Calcutteroo, South Australia. Tectonophysics 101, 185-197.

reflection character of the continental Mohorovicic discontinuity. Journal of Geophysical Research 87,

internal friction of calcite rocks at seismic frequencies: pressure, frequency and grainsize dependence. Physics of the Earth and Planetary Interiors 45, 349-367.

HENYEY F.S. & POMPHREY N. 1982. Self-consistent moduli of a cracked solid. Geophysical Research Letters 9, 903-906.

JACKSON I., PATERSON M.S. & FITZGERALD J. D. 1991.

4625-4635.

HESS H. 1964. Seismic anisotropy of the uppermost mantle under oceans. Nature 203, 629-631. HUDSON J.A. 1981. Wave speeds and attenuation of

elastic waves in material containing cracks. Geophysical Journal of the Royal Astronomical Society 64, 133-150. HUGHES D.S. & JONES H.J. 1950. Variation of elastic

moduli of igneous rocks with pressure and temperature. Geological Society of America, Bulletin 61, 843-856. HUGHES D.S. & CROSS J.H. 1951. Elastic wave velocities

in rocks at high pressures and temperatures. Geophysics 16, 577-593.

HUGHES D.S. & MAURETTE C. 1957. Variation of elastic

wave velocities in basic igneous rocks with pressure and temperature. Geophysics 22, 23-31.

Seismic wave dispersion and attenuation in Aheim dunite: an experimental study.Geophysical Journal International, (submitted).

JACKSON I., RUDNICK R. L., O'REILLY S. Y. & BEZANT

C. 1990. Measured and calculated elastic wave velocities for xenoliths from the lower crust andupper mantle. Tectonophysics 173, 207-210.

JACKSON I., PATERSON M.S., NIESLER H. & WATERFORD

R.M. 1984. A high pressure study of rock anelasticity at low strain and seismic frequency. Geophysical Research Letters 11, 1235-1238. JONES T.D. 1985. Nature of seismic reflections from the

crystalline basement: COCORP Wind River Line, Wyoming. Journal of Geophysical Research 90, 67836791.

JONES T.D. & NUR A. 1984. The nature of seismic

reflections from deep crustal fault zones. Journal of Geophysical Research 89, 3153-3171.

110

Petrophysical basis for seismological models.

JORDAN T.H. 1979. Mineralogies, densities and seismic

velocities of garnet lherzolites and their geophysical implications. In Boyd F.R. and Meyer H.O.A. eds The mantle sample : inclusions in kimberlites and other volcanics. Proceedings of the Second International Kimberlite Conference, II, pp. 1-14. American Geophysical Union, Washington.

KANAMORI H. & MIZUTANI H. 1965. Ultrasonic measurement of elastic constants of rock under high pressures. Earthquake Research Institute, Bulletin 43, 173-194.

KARATO S. 1986. Does partial melting reduce the creep strength of the upper mantle? Nature 319, 309-310. KARATO S. & SPETZLER H. 1990. Defect microdynamics

in minerals and solid -state mechanisms of seismic wave attenuation and velocity dispersion in the mantle. Reviews of Geophysics. 28, 399-421. KATAHARA K.W., MANGHNANI M.H., DEVANI N. & TITTMANN B.R. 1982. Pressure dependence of Q in

selected rocks. In Akimoto S. and Manghnani M.H. eds. High-pressure research in geophysics, pp. 147-158. Centre for Academic Publications, Tokyo. KAY R.W. & KAY S.M. 1981. The nature of the lower continental crust : inferences from geophysics, surface geology and crustal xenoliths. Reviews of Geophysics and Space Physics 19, 271-297. KAYE G.W.C. & LABY T.H. 1973. Tables of physical and chemical constants and some mathematical functions. 14th Edition. Longman, London. KERN H. 1982. P- and S-wave velocities in crustal and

mantle rocks under the simultaneous action of high confining pressure and high temperature and the effect of rock microstructure. In Schreyer W. ed. High- pressure researches in Geoscience, pp. 15-45. E. Schweizerbart'sche Verlagsbuchhandlung, Stuttgart. KLEMPERER S.L. & the BIRPS Group, 1987. Reflectivity of the crystalline crust: hypotheses and tests. Geophysical Journal of the Royal Astronomical Society 89, 217-222.

KUMAZAWA M. 1969. The elastic constants of single-crystal orthopyroxene. Journal of Geophysical Research 74, 5973-5980.

KUMAZAWA M. & ANDERSON O.L. 1969. Elastic moduli, pressure derivatives and temperature derivatives of single-crystal olivine and single- crystal forsterite. Journal of Geophysical Research 74, 5961-5972. KUSHIRO I., YODER H.S. & MYSEN B . 0 . 1 9 7 6 . Viscosities of basalt and andesite melts at high pressures. Journal of Geophysical Research 81, 6351-6356. KUSTER G.T. & TOKSOZ M.N. 1974. Velocity and

attenuation of seismic waves in two-phase media : Part 1. Theoretical formulations. Geophysics 39, 587-606.

LANDISMAN M., MUELLER S. & MITCHELL B.J. 1971.

Review of evidence for velocity inversions in the continental crust. In Heacock J.G. ed. The structure and physical properties of the Earth's crust, pp. 1134. American Geophysical Union, Washington. LEVEN J. H., FINLAYSON D. M., WRIGHT C., DOOLEY J. C. & KENNETT B.L.N, eds. 1990. Seismic probing of the

continents and their margins. Tectonophysics 173,1-642

A.I. & VAVAKIN V.V. 1978. Study of elastic-wave velocities and density of rocks and minerals at pressures to 20 kbar and temperatures to 500 C. Izvestiya, Earth Physics 14,346-352. LEVYKIN

MANGHNANI M.H., RAMANANANTOANDRO R. & CLARK

S.P. 1974. Compressional and shear wave velocities in granulite facies rocks and eclogites to 10 kbar. Journal of Geophysical Research 79, 5427-5446. MATSUSHIMA S. & AKENI K. 1977. Elastic wave velocities in the Ichinomegata ultramafic nodules : composition of the upper most mantle. In Manghnani M.H. and Akimoto S. eds. High-pressure Research : Applications in Geophysics, pp. 65-76. Academic Press, New York. MATTHEWS D.H. & CHEADLE M.J. 1986. Deep reflections from the Caledonides and Variscides west of Britain and comparison with the Himalayas. In Barazangi M. and Brown L. eds. Reflection seismology : a global perspective. pp. 5-19. American Geophysical Union, Washington, Geodynamics Series, 13. MATTHEWS* D.H. & SMITH C. (Editors) 1987. Proceedings Second International Symposium on Deep Seismic Reflection Profiling of the Continental Lithosphere. Geophysical Journal of the Royal Astronomical Society 89,1-498.

Geol. Soc. Aust. Spec. Publ. 17, 81-114.

MATTHEWS

D.H.,

HOBBS

R.,

KLEMPERER

S.L.,

MCGEARY S., PEDDY C.P., WARNER M . R . & SMITH C . A . ,

1987. Some unsolved BIRPS problems. Geophysical Journal of the Royal Astronomical Society 89, 209-216. MAVKO G.M. 1980. Velocity and attenuation in partially molten rocks. Journal of Geophysical Research 85, 5173-5189. MAVKO G.M. & NUR A. 1975. Melt squirt in the asthenosphere. Journal of Geophysical Research 80, 1444-1448. MCDONOUGH D . T . & FOUNTAIN D . M . 1 9 8 8 . R e f l e c t i o n

characteristics of a mylonite zone based on compressional wave velocities of rock samples. Geophysical Journal 93, 547-558. MOONEY W . D . & BROCHER T . M . , 1 9 8 7 .

Ill

NOWICK A.S. & BERRY B.S. 1972. Anelastic

relaxation

in crystalline solids. Academic Press, New York, 677p. NUR A. & SIMMONS G. 1969a. The effect of saturation on velocity in low porosity rocks. Earth and Planetary Science Letters 7,183-193. NUR A. & SIMMONS G. 1969b. The effect of viscosity of a fluid phase on velocity in low-porosity rocks. Earth and Planetary Science Letters 7, 99-108. NUR A. & SIMMONS G. 1970. The origin of small cracks in igneous rocks. Int. J. Rock Mech. Min. Sci. 7,307-314. O'CONNELL

R.J.

BUDIANSKY B .

1974.

Seismic

velocities in dry and saturated cracked solids. Journal of Geophysical Research 79, 5412-5426.

Coincident

seismic reflection/refraction studies of the continental lithosphere: a global review. Geophysical Journal of the Royal Astronomical Society 89,1-6. MUELLER S. 1977. Anew model for the continental crust. In Heacock J.G. ed. The Earth's Crust, pp. 289-317American Geophysical Union, Geophysical Monograph Series 20,. Washington. MURASE T. & KUSHIRO I. 1979. Compressional wave

velocity in partially molten peridotite at high pressures. Carnegie Institution Washington Yearbook 78, 559-562. MURPHY W.F. 1982. Effects of partial water saturation on attenuation in Massilon sandstone and Vycor porous glass. Acoustic Society of America, Journal 71, 1458-1468. MURPHY W.F. 1984. Seismic to ultrasonic velocity drift: Intrinsic absorption and dispersion in crystalline rock. Geophysical Research Letters 11, 1239-1242. MURPHY W.F. 1985. Sonic and ultrasonic velocities: theory versus experiments. Geophysical Research Letters 12, 85-88.

O'CONNELL R . J . & BUDIANSKY B . 1 9 7 7 . V i s c o e l a s t i c

properties of fluid- saturated cracked solids. Journal of Geophysical Research 82, 5719-5735. O'REILLY

S.Y.

JACKSON

BEZANT

1990.

Equilibration temperatures and elastic wave velocities for upper mantle rocks from eastern Australia: implications for the interpretation of seismological models. Tectonophysics 185, 67-82. PADOVANIE.R. HALL J . & SIMMONS G . 1 9 8 2 . C o n s t r a i n t s

on crustal hydration below the Colorado Plateau from V p measurements on crustal xenoliths. Tectonophysics 84, 313-328.

PANNETIER J. WENK H . R . & KERN H . 1 9 8 9 . F a b r i c

changes and associated elastic anisotropy during progressive deformation of granite rocks EOS. Trans American Geophysical Union 70,457 (abstract). PESELNICK L. & STEWART R.M. 1975. A sample assembly

for velocity measurements of rocks at elevated temperatures and pressures. Journal of Geophysical Research 86, 3765-3769.

NELSON K . D . ALLMENDINGER R . POTTER C . HAUSER E . LATHAM T. BROWN L . OUVER J . & KAUFMAN S. 1 9 8 7 .

PESELNICK L. & NICOLAS A. 1978. Seismic anisotropy in

Regional variations in the reflection character of the continental Moho. IUGG XIX General Assembly Abstracts 1, 59.

an ophiolite peridotite : application to oceanic upper mantle. Journal of Geophysical Research 83,1227-1235. PESELNICK

NICOLAS A. & POIRER J.P. 1976. Crystalline

plasticity

and solid-state flow in metamorphic Wiley-Interscience, London, 444p.

rocks.

Liu

H.-P.

1987.

Laboratory

measurement of internal friction in rocks and minerals at seismic frequencies. In Methods of Experimental Physics, 24A, Academic Press, New York, 349-369.

112

Petrophysical basis for seismological models.

PRODEHL C. 1977. The structure of the crust-mantle

SIMMONS B. 1964b. Velocity of shear waves in rocks to

RAITT R.W., SHOR G.G., FRANCIS T.J.G. & MORRIS G . B .

Upper-crustal low velocity layers: possible effect of high temperatures over a mantle upwarp at the Basin Range-Colorado Plateau transition. Earth and Planetary Science Letters 2 8 , 1 9 7 - 2 0 4 .

boundary beneath North America and Europe as derived from explosion seismology. In Heacock J.G. ed The Earth's Crust: ItsNature and Physical Properties, pp. 349-369. American Geophysical Union, Washington Geophysical Monograph Series, 20. 1969. Anisotropy of the Pacific upper mantle. Journal of Geophysical Research 74, 3095-3109.

RAMANANANTOANDRO R . & MANGHNANI M . H . 1978.

Temperature dependence of the compressional wave velocity in an anisotropic dunite : measurements to 500°C at 10 kbar. Tectonophysics 47, 73-84.

10 kilobars. Journal of Geophysical Research 69, 1123-1130.

SMITH R.B., BRAILE L.W. & KELLER G . R . 1975.

SMITHSON S.B. & SHIVE P.N. 1975. Field measurements of compressional w^ve velocities in common crystalline rocks. Earth and Planetary Science Letters 27,170-176.

RLNGWOOD A . E . 1975. Composition and Petrology of the Earth's Mantle. McGraw-Hill, New York, 618.

SMITHSON S.B. & BROWN S.K. 1977. A model for the lower continental crust. Earth and Planetary Science Letters 35,134-144.

S ANDMEIER K.J., WALDE W . & WENZEL F. 1987. Physical properties and structure of the lower crust revealed by one- and two-dimensional modelling. Geophysical Journal of the Royal Astronomical Society 89, 3 3 9 - 3 4 4 .

SMITHSON S.B., BREWER J.A., KAUFMAN S „ OLIVER J.E. & HURICH C . A . 1979. Structure of the Laramide Wind

SATO H . SACKS I.S. MURASE T. MUNCELL G . &

FUKUYAMA H . 1989. Qp- Melting temperature relation in peridotite at high pressure and temperature: attenuation mechanism and implications for the mechanical properties of the upper mantle. Journal of Geophysical Research 94, 10647-10661. SCHILT S., OLIVER J., BROWN L., KAUFMANN S., ALBAUGH D., BREWER J., COOK F., JENSEN L., KRUMHANSL P., LONG G . & STEINER D . 1979. T h e

River uplift, Wyoming, from COCORP deep reflection data and from gravity data. Journal of Geophysical Research 84, 5 9 5 5 - 5 9 7 2 .

SOBOLEV S.V. & BABEIKO A.Y. 1987. Phase transformations in the lower crust and its seismic structure. IUGG XIX General Assembly Abstracts, 1, 65. SPENCER J.W. & NUR A . M . 1976. The effects of pressure, temperature and pore water on velocities in Westerly Granite. Journal of Geophysical Research 81, 899-904.

heterogeneity of the continental crust: results from deep crustal seismic reflection profiling using the vibroseis program. Reviews of Geophysics and Space Physics 17, 354-368.

SPRUNT E.S. & BRACE W.F. 1974. Direct observation of microcavities in crystalline rocks. Int. J. RockMech. Min Sci. and Geomech. Abstr. 11,139-150.

SERONT B., MAINPRICE D . & CHRISTENSEN N . I. 1989.

TAKEUCHI S. & SIMMONS G .

The complete seismic properties of an anorthositecomparison between LPO and laboratory measurements EOS. Transactions American Geophysical Union 70, 460-461.

SHANKLAND T.J., O'CONNELL R.J. & WAFF H.S., 1981.

Geophysical constraints on partial melt in the upper mantle. Reviews of Geophysics and Space Physics 19, 394-406.

SIMMONS G . 1964a. Velocity of compressional waves in various minerals at pressures to 10 kilobars. Journal of Geophysical Research 69, 1117-1121.

1973. Elasticity of water-saturated rocks as a function of temperature and pressure. Journal of Geophysical Research 78, 3310-3320. TALWANI M . , SUTTON G . H . & WORZEL J.L. 1959. A

crustal section across the Puerto Rico trench. Journal of Geophysical Research 6 4 , 1 5 4 5 - 1 5 5 5 .

THOUVENOT F. 1983. Frequency dependence of the

quality factor in the upper crust: a deep seismic sounding approach. Geophysical Journal of the Royal Astronomical Society 73,427-447.

Geol. Soc. Aust. Spec. Publ. 17, 81-114. TTTTMAN B.R. 1977. Internal friction measurements and

WALSH J.B. 1965. The effect of cracks on the

their implications in seismic Q structure models of the crust. In Heacock J.G. ed The Earth's Crust: its nature and physical properties, pp. 197-213. American Geophysical Union .Geophysical Monograph Series, 20.

Research 70,381-389.

TODD T. & SIMMONS G. 1972. Effect of pore pressure on

74,4333-4337.

the velocity of compressional waves in low porosity rocks. Journal of Geophysical Research 77,3731-3743.

TOKSOZ M.N., CHENG C.H. & TIMUR A. 1976. Velocities

of seismic waves in porous rocks. Geophysics 41, 621-645.

TURCOTTE D.L. & SCHUBERT G. 1982. Geodynamics:

Applications of continuum physics to geological problems. Wiley, New York, 450 p.

VAN DER MOLEN I. & PATERSON M.S. 1979. Experimental

113

compressibility of rock. Journal of Geophysical

WALSH J.B. 1969. New analysis of attenuation in partially melted rock. Journal of Geophysical Research WANG C.Y. 1969. Equation of state of periclase and some of its geophysical implications. Journal of Geophysical Research 74,1451-1457. WARNER M.R. 1990. Basalts, water or shear zones in the lower continental crust? Tectonophysics 173,163-175. WOOLLARD G.P. 1959. Crustal structure from gravity and

seismic measurements. Journal ofGeophysical Research

64, 1521-1544.

deformation of partially melted granite. Contributions to Mineralogy and Petrology 70, 299-318.

ZIMMERMAN R.W. 1985. Inversion of seismic velocity

VOLAROVICH M.P., BAYUK Y.I. & SHAGINYAN G.S. 1977.

950.

A study of longitudinal wave velocity and absorption in rocks and minerals at pressures up to 15 kbar and temperatures to 600°C. Izvestiya, Earth Physics 13, 503-509.

WAFF H.S. & BULAU J.R. 1979. Equilibrium fluid

distribution in an ultramafic partial melt under hydrostatic stress conditions. Journal of Geophysical Research 84,6109-6114.

data for determining pore aspect-ratio distributions. EOS Transactions of the American Geophysical Union, 66,

Ian Jackson Research School of Earth Sciences Australian National University GPO Box 4 Canberra ACT 2601 AUSTRALIA.

Geol. Soc. Aust. Spec. Publ. 17,115-116.

115

Extrapolation of Laboratory Rheological Behaviour in Tectonophysics M.S. Paterson Research School of Earth Sciences, Australian National University, Canberra, ACT., Australia At relatively low temperatures and pressures rocks fail under non-hydrostatic stress by brittle fracture. However, they tend to become ductile as temperature and pressure are raised, although the presence of high pore fluid pressures can greatly extend the brittle field. At lower crustal and upper mantle conditions, ductile behaviour is usually assumed, except where the occurrence of earthquakes points to brittle failure. This note deals only with creep of rocks in the ductile field. Such creep is invoked in considering large scale tectonic processes in the lower crust and upper mantle and therefore the predictions of creep behaviour under natural conditions from laboratory observations would be very useful in constraining tectonophysical models. However, this prediction is hampered by many uncertainties and lack of knowledge, even concerning what are the potentially important variables. In relating laboratory observations to tectonophysical processes, three types of consideration enter:

(shortening positive) at a given temperature T and confining pressure 02 = 03. Not much attention has been given to the analysis of transient creep. The test results are usually fitted to a steady state "flow law" of the form: £ = Ac n exp( - ^ j r )

where A, n and Q are empirical parameters, R is the gas constant and T the absolute temperature. In tectonophysics other types of stress state are involved and so a generalisation of this flow law is needed. In engineering creep calculations for general stress states it is often assumed, following Odqvist, that the flow law can be expressed in terms of an "effective stress" o and an "effective strain rate" 8* in the form (neglecting elastic strains): a

{(G1-G2) 2

+ (C2-03) 2 + ( a 3 - 0 i ) 2 } 1/2

(a) The form of the flow criterion for general stress states £•* =

(b) The environmental and internal variables that need be considered (c) The circumstance that different mechanisms may dominate behaviour under different conditions, giving rise to a variety of flow laws for a given rock. These topics are now discussed in turn. Almost all laboratory rheological measurements have been made in axial compression under superposed confining pressure, that is, with Oi>O2 = 03 where Oi, 02, G3 are the principal stresses (compression positive). The quantity a = 01 - G3 is reported as the "flow stress", determined as a function of the axial strain rate 8i = 8

(1)

{ (ei-e2)2 /

+ (£2-£3) 2 + ( e 3 - e i ) 2 } 1 / 2 In this case it is seen that a ( = 01 - 03) and 8 ( = 81) in the experiments can be viewed as effective stress and effective strain rate and that it is appropriate* to extend * * (1) to general stress states in terms of a and 8 , remembering, however, that the form (1) is essentially empirical at this point and that other forms might have been used. Even accepting (1) as the basic form of flow law, f [ e*, g*,T )= 0, with primary variable 8* , o and T, the flow behaviour in practice can be influenced by the following additional variables:

116

Extrapolation of laboratory rheology.

(a) Pressure or mean stress. The influence of pressure is relatively small in the laboratory range and no precise measurements are available. However, it is thought to be significant in extrapolating to upper mantle conditions, when it can be allowed for by increasing the value of Q. (b) Grain size. Under some conditions the creep rate can be strongly dependent on grain size. In this case the pre-exponential factor in (1) is rewritten as A/d where d is the grain size and p is another empirical constant, usually in the range 0 to 3. p

(c) Water activity. Intragranular hydrolytic weakening is an effect of major importance, well known in quartz and olivine and probably occurring in other silicates, although the exact manner in which the dependence on water activity should be incorporated in the flow law has yet to be established. Other thermodynamic variables such as oxygen activity may also prove to be important in particular cases. (d) Presence of fluid phases. The pore pressure effects that are well known in the brittle field are probably unimportant in the ductile field. However, the presence of fluids in the grain boundary regions can have important rheological effects if conditions are such that diffusive transfer of material through the fluid phase contributes significantly to the strain or if it assists intergranular accommodation. These effects can be expected to introduce some grain size dependence in A and possibly affect the other parameters too. (e) Preferred crystallographic orientation. This would tend to introduce anistropy of flow stress, requiring different choice of parameters for different orientations. Little is known about flow laws for anisotropic rock. The third aspect of flow behaviour to be considered is the variety of deformation mechanisms or rheological regimes. Many mechanisms are known or are conceivable, by means of which rock

can flow. The flow law can therefore be expected to vary, in values of parameters and in the variables that need to be incorporated, according to the mechanisms predominating. These mechanisms may include dislocation movement controlled by a variety of factors, displacement of material by diffusion via various paths, and cataclastic processes involving microcracking and granular flow. Correspondingly, laboratory measurements have established different rheological regimes within which different values of the parameters are found when flow law (1) is fitted and different dependences on variables such as grain size appear. Before making any justifiable extrapolation of a laboratory-determined flow law, valid for certain ranges of strain rate, stress, temperature and other variables, to tectonophysical conditions in the Earth it must be established that the same flow mechanism or rheological regime is involved. Microstructural evidence on flow mechanism in

laboratory and field is therefore of primary importance in guiding the application of laboratory findings in geology. In the absence of this evidence prediction by extrapolation must be seen as speculative unless the general form of the rheological behaviour within the Earth can be independently established. When the latter is possible, comparison with experiment may permit conclusions about conditions within the Earth, such as temperature and other thermodynamic variables. Otherwise, on present knowledge, only very broad limits can be placed on the rheology of the lower crust and upper mantle by extrapolation from experiments. Footnote: The subject of this note is treated in more detail in PATERSON M.S. 1987. Problems in the extrapolation of laboratoiy rheological data. Techonophysics 133, 33-43.

M.S. Paterson Research School of Earth Sciences Australian National University GPO Box 4 Canberra ACT 2601 AUSTRALIA.

Geol. Soc. Aust. Spec. Publ. 17,117-120.

117

Australian Long-Wavelength Magnetic Anomalies P. Wellman, A.S. Murray, and M.W. McMullan Bureau of Mineral Resourses, Geology and Geophysics, Canberra, ACT., Australia A long-wavelength total-field magnetic anomaly map has been prepared for Australia using near surface observations (Wellman et al, 1985). The field can be upward continued and compared with satellite observations to check its reliability, and it can be compared with heat flow, seismicallymeasured crustal thickness, gravity anomalies and sediment thickness, to investigate the sources of the anomalies. The near surface magnetic observations used are of two types (Dooley & McGregor, 1982): 1. A third-order survey of spot readings on the ground. These have an even geographic distribution, but adjacent readings are not consistent because of near-surface high magnetization. 2. Aeromagnetic observations at 3 km altitude. These are along widely-spaced long profiles, so the data do not have an even geographic distribution. Adjacent readings are consistent because of the filtering of short-wavelength anomalies by the 3 km altitude.

processes tend to suppress short-wavelength anomalies, but preserve long-wavelength anomalies. The maps for 400 km altitude and 200 km grid spacing from near-surface and satellite data have a similar distribution of major highs and lows, which indicates that both maps reflect real longwavelength magnetic anomalies. However, the anomalies on the map derived from near surface observations have a minimum wavelength of about one half, and an amplitude of about twice that of the MAGSAT anomalies. We attribute the differences to smoothing of the MAGSAT data, the smoothing being caused by differences in the altitude of the satellite passes, and by differences in the regional field removed from the observations of each pass. The long-wavelength magnetic anomalies mapped from near surface and satellite observations could be due to magnetisation in the upper

Maps of the two data sets using a 200 km grid have the same major features, so the data sets are consistent with each other. A near surface composite map was prepared using the combined third-order survey and longprofile data. The total 18 000 observations were corrected for secular variation, and for the regional field model MGST 4/81 (Langel et al., 1981), but were not corrected for diurnal variation. Fig. 1 shows the anomalies when the grid size is 100 km. This grid size is the minimum desirable smoothing. The MAGSAT satellite anomalies (Langel et al, 1982) are for 400 km altitude, and 200 km grid spacing. In order to compare near-surface with satellite observations, the near-surface observations on a 100 km grid have been upward continued to 400 km and regridded to a 200 km grid. These

Fig. 1. Long-wavelength magnetic total intensity anomalies over Australia. 100 km grid, 1.5 km mean altitude, 100 nT contour interval, negative anomalies stippled (from Wellman et al., 1985).

118

Long-wavelength magnetic anomalies.

crust, the lower crust, or the deeper lithosphere. Several possibilities are discussed below. There is no apparent correlation Australia-wide of the longwavelength magnetic anomalies with variations in heat flow given by Cull (1982) and Cull & Conley (1983). This makes it unlikely that the crust has a uniform magnetisation with the Curie point in the crust, because this would result in hotter crust having a shallower Curie point, resulting in a negative correlation between magnetic anomaly and heat flow. There is also no apparent correlation of long-wavelength magnetic anomalies with crustal thickness determined by seismic refraction results as listed by Wellman (1982) and Finlayson et al. (1984). It is unlikely therefore that the crust has uniform magnetisation, with a Curie point below the crust, because this should give a positive correlation between crustal thickness and magnetic anomaly, given that pristine mantle material in-situ is non-magnetic (Wasilewski, 1983). There is a rough correlation between long-wavelength magnetic lows and sedimentary basins. The effect of low magnetisation of sedimentary basin rocks relative to average crustal values has been modelled. The calculated anomalies due to this cause are much smaller than the observed anomalies, so this cause is an insignificant component of longwavelength magnetic anomalies. However, long-wavelength magnetic anomalies (Fig. 1) do show a good spacial correlation with upper crustal geological structure. Magnetic highs in eastern Australia correlate with exposed basement, and in central and western Australia they correlate with outcropping and subcropping craton boundaries as determined by the gravity trend pattern, and major gravity 'dipole' anomalies (Wellman, 1978). These geological structures are likely to be underlain by related structures in the sub-crustal lithosphere, however these mantle rocks are likely to be non-magnetic. Because of these spatial correlations, the longwavelength magnetic highs are likely to be due mainly to rock of high apparent susceptibility in the upper or lower crust. In conclusion, the long-wavelength magnetic anomalies have a spacial correlation with upper crustal geological features, but not with heat flow, or crustal thickness, so they are likely to be due to variations in the apparent magnetisation in the

upper or lower crust. Therefore it seems unlikely that rocks in the subcrustal lithosphere contribute significantly to the long-wavelength magnetic anomalies.

Acknowledgements This paper is published with the permission of the Director, Bureau of Mineral Resources, Geology and Geophysics.

References CULL, J.P. 1982. An appraisal of Australian heat flow

data. BMR Journal ofAustralian Geology & Geophysics

7, 11-21.

CULL, J.P. & CONLEY, D. 1983. Estimates of temperature and heat flow in the sedimentary basins of Australia. BMR Journal of Australian Geology & Geophysics 8, 329-337.

DOOLEY, J.C. & MCGREGOR, P.M. 1982. Correlative

geophysical data in the Australian region for use in the MAGSAT project. Australian Society of Exploration Geophysicists, Bulletin 13, 63-67. FINLAYSON, D.M., COLLINS, C.D.N., & LOCK, J. 1984.

P-wave velocity features of the lithosphere under the Eromanga Basin, eastern Australia, including a prominent mid-crustal (Conrad?) disconformity. Tectonophysics 101*267-291.

LANGEL, R.A., BERBERT, J., JENNINGS, T., & HORNER, R.

1981. MAGSAT data processing: an interim report for investigators. NASA TM 82160, Nov. 1981.

LANGEL, R.A., PHILLIPS, J.D., & HORNER, R.J. 1982.

Initial scaler magnetic anomaly map from MAGSAT. Geophysical Research Letters 9, 269-272.

WASILEWSKI, P.J. 1983. Magnetic mineralogy of the crust and upper mantle. EOS, 64, 213-214. WELLMAN, P. 1978. Gravity evidence for abrupt changes in mean crustal density at the junction of Australian crustal blocks. BMR Journal of Australian Geology & Geophysics 3 , 1 5 3 - 1 6 2 .

Geol. Soc. Aust. Spec. Publ. 17,117-120.

WELLMAN, P. 1982. Australian seismic refraction results, isostasy, and altitude anomalies. Nature 298, 838-841. WELLMAN, P., MURRAY, A.S., & MCMULLAN, M.W.

1985. Australian long-wavelength magnetic anomalies. BMR Journal of Australian Geology & Geophysics 10.

P. Wellman A.S Murray M.W. McMullan Bureau of Mineral Resources Geology and Geophysics GPO Box 4 Canberra ACT 2601 AUSTRALIA.

Geol. Soc. Aust. Spec. FwM. 17,121-140.

121

Electrical Studies of the Australian Lithosphere S.C. Constable Scripps Institution of Oceanography, A-030, La Jolla, CA 92093, USA. About 20 studies of the Australian deep lithosphere using electrical techniques have been conducted, mostly employing magnetometer array and magnetotelluric methods. Many of these studies detected an increase in conductivity at a depth of 90-200 km, presumably representing the base of the lithosphere. The Australian sedimentary basins are invariably shown to be conductive, resistivities of 1-10 Qm often being measured. Such resistivities reflect the high salinity of the groundwater and high porosities in the sediments. Apart from locally anomalous regions, little structure has been distinguished within the sub-basement lithosphere. Resistivities of 1,000-10,000 Om are usually cited, with no evidence for an electrical Moho. Most studies use simple layered structures to model the data, but hypothesis testing provides a means of extracting definitive answers in the face of the non-uniqueness of electromagnetic interpretation. Inversion techniques developed within the last decade provide tools for making such tests. Finally, electrical experiments in Australia seem vulnerable to current channelling, and interpretations often include unrealistically low resistivities in the upper crust. INTRODUCTION The goal of a geophysicist is to understand the internal structure of the Earth using measurements which must be made at the Earth's surface. His tools include the gravity, magnetic and electric fields, seismic waves, heat flow and rheological properties. Within the Earth's uppermost kilometre or so geophysical investigations provide a cheaper and more versatile alternative to drilling. At greater depths geophysicists provide important constraints on models of subsurface structure and composition, many of which cannot be obtained by other means. This paper is a review of electrical conductivity investigations carried out in Australia, with emphasis on the study of the lithosphere. Electromagnetic (E.M.) methods, like seismic techniques, use both natural and man-made sources. The ability to control the energy source endows these two techniques with a flexibility not found in other geophysical methods. Fewer deep electrical studies have been made in Australia than in western Europe, North America, the U.S.S.R. and possibly even Africa. The geographical distribution of studies within Australia is also biased, with nearly all sites lying in a belt from north central Australia, through eastern South Australia and into southwest New South Wales, Victoria and northern Tasmania, with most of Queensland and all of Western Australia essentially untouched (Fig. 1). Consequently, a coherent picture of the electrical structure for the Australian

lithosphere has not yet emerged; despite this an attempt is made here to draw existing work together wherever possible. Field studies have been grouped together in broad categories dealing with the coast effect, sedimentary basins, conductivity anomalies and shield areas. However, before these summaries, the first parts of this paper provide a brief introduction to the conductivity of rocks and the use of electrical methods. Attention has been paid to the various inversion techniques used to obtain models of conductivity from field data, as an understanding of their limitations is essential to the assessment of the published material. Also, in spite of the limited amount of field work carried out in Australia, this country has made significant contributions to the theory (eg. Parkinson, 1962), interpretation (eg. Jupp and Vozoff, 1975) and instrumentation (eg. Chamalaun and Walker, 1982) of geomagnetism. Recent improvements in instrumentation and computers have greatly assisted the collection and interpretation of electrical sounding data, but these have placed the field into an adolescent rather than mature stage. Three-dimensional (3D) forward modelling (ie., predicting experimental results given an earth structure) remains very difficult, and the understanding of the one-dimensional (ID) inverse problem (ie., solving for the earth structure given experimental results) is still subject to progress. Much of the world's magnetometer array data are still being collected using the Gough-Reitzel magnetometer developed in the mid-sixties, which requires data to be hand digitized from

122

Electrical studies, Australian lithosphere.

Fig. 1. Positions of the various electrical experiments reported in the literature and discussed in the text. Table 1 gives the key reference for each study. photographic film. It follows that this is an easy time in which to be critical of earlier work and of current work still using earlier techniques, but a critical assessment is necessary if researchers outside the field are to understand the value of the E.M. method without expecting too much or too little from it. ELECTRICAL CONDUCTIVITY OF ROCKS

moving through a crystal lattice. Defects may be intrinsic, caused by impurities inherent in the crystal which do not easily fit into the lattice structure, or extrinsic, a term applied to defects which are created by an environmental influence, such as temperature. Because the number of vacancies and other defects, as well as the mobility of ions, increases with temperature, the resistivity of silicate minerals decreases with increasing temperature, to roughly 10 Qm at 1000 °C. When a rock melts the mobility of ions is further increased, and so the conductivity of silicate melt is on the order of 1 Clm. It is quite probable that this transition in mobility is not as sharp as is usually imagined for a melting process. Silicate melts retain relatively long range ordering, and Piwinskii and Duba (1974) showed that the conductivity of albite crystals increased with time as they were held at high temperature but 3

Electrical conductivity has units of siemens per metre (S/m); resistivity, in ohm-metres (Qm), is just the reciprocal of conductivity. The resistivity of dry silicate minerals is very high, as much as 10 £2m at surface temperatures. Electric charge is conducted in most silicates by the movement of ions and ion vacancies in the form of crystal defects 14

Geol. Soc. Aust. Spec. Publ. 17,121-140.

below their solidus, presumably as disordering within the solid crystal structure took place. Mafic rocks are more conductive than felsic rocks. However, this effect is small compared with that of temperature, and amounts to only half an order of magnitude between granites and gabbros (Kariya and Shankland, 1983). Because of the high resistivity of mineral grains, the conductivity of rocks at normal crustal temperatures ( <600 °C) is determined by the amount of fluid or other conductive phases (usually water) in cracks, fissures, joints and pores. An empirical relation (Archie's law) between porosity and resistivity is given by Array Studies 1. 2. 3. 4. 5. 6. 7.

Bennett & Lilley (1973) Everett & Hyndman (1967a) Gough etal. (1974) Lilley (1976) Chamalaun (1985) White &Milligan (1984) White & Polatayko (1978,1985), White & Hopgood (1979), White & Milligan (1985) 8. Woods & Lilley (1979) 9. Woods & Lilley (1980) 10. Parkinson &Hermanto (1986) Magnetotelluric Studies 2. Everett & Hyndman (1967b) 3. Tammemagi & Lilley (1973) 11. Cull (1982b) 12. Cull &Spence (1985) 13. Jupp et al. (1979) 14. Moore et al. (1977) 15. Spence & Finlayson (1983) 16. Tammemagi & Lilley (1971) 17. Vozoff etal. (1975) 18. Whiteley& Pollard (1971) Resistivity Soundings 19. Constable (1985) 20. Constable et al. (1984)

Table 1. This table provides the key to Fig. 1, which shows positions and approximate areal extents of Australian electrical studies.

123

rrock —

where O is the porosity expressed as a fraction and rw is the water resistivity. Groundwater resistivity, which is determined by the amount of dissolved salts and to a lesser degree temperature, is usually between 0.05 and 100 £lm for bulk samples, but is probably limited to about 10 Qm when in situ because of the effect of ionic exchange capacity in mineral grains (Keller and Frischknecht, 1966). In the expression, m is a constant which depends on how well the water is interconnected, and although it may range between 1.2 and 2.6, is often assumed to be 2.0. It is important to realize that it is the conductivity of free water that enhances the conductivity of a wet rock. Water of hydration, which is part of the crystal structure, will not lower rock resistivities below that of dry silicate minerals (01hoeft, 1981). Pressure has very much a second order effect on rock conductivities compared with temperature. There is a small initial increase in conductivity for a dry rock as the mineral grains are pressed into better contact with each other. However, if the rock is wet, then there will be a decrease in conductivity as porosity is reduced (assuming that hydrostatic pressure is less than lithostatic pressure). However, deep drilling suggests that some degree of porosity is maintained to depths of many kilometres in the crust. It is likely that this residual porosity will be comparatively more efficient at lowering resistivity than bulk porosity because surface effects and polarization probably result in the conductivity of grain surfaces being considerably higher than that of the bulk groundwater. Pressure has little effect on the conductivity of minerals at crustal and upper mantle depths, but is important for its capacity to force phase changes deeper in the mantle. Olivine is thought to be a major constituent of the upper mantle, and so has been the subject of many laboratory studies, but Duba & Nicholls (1973) showed that reliable results for olivine conductivity may only be obtained if the oxidation state of the olivine is controlled during the experiment. The best published results to date are probably those of Duba etal. (1974), who give the resistivity for a single crystal of composition Mgi .sFeo. i9Si04 as a function of temperature, between 600 °C (=106 Qm) and 1700 °C («20 tan). However, little is

124

Electrical studies, Australian lithosphere.

understood about the role of grain boundaries on rock conductivity. It is possible that surface defects and impurities lower resistivity appreciably, in which case the single crystal values would represent an upper bound for whole rock resistivities (Shankland and Waff, 1977).

subject, while Kaufman and Keller (1981) and Vozoff (1986) give more recent treatments. The early applications of M.T. studies were directed at making ID soundings of resistivity structure, but by measuring over a wide bandwidth and occupying many stations the M.T. method has the potential for resolving 2D or even 3D structure moderately well.

COLLECTION AND INTERPRETATION OF EARTH CONDUCTIVITY DATA

In a G.D.S. study an array of three component magnetic variometers is deployed with spacings ranging from a few kilometres to several hundred kilometres between instruments. The book by Rokityansky (1982) deals with several aspects of G.D.S. experiments. Such arrays are extremely good at delineating lateral variations in electric current, caused either by local induction over a lateral change in conductivity, or by regional induction in a large feature such as an ocean or sedimentary basin which in turn drives currents through the area of the magnetometer array ('current channelling'). It is a characteristic of magnetometer arrays that the depth to the observed currents is difficult to estimate, but under favourable conditions, associated with a lack of lateral changes in conductivity and a source field having a horizontal gradient, the array data may be transformed into M.T. impedances, allowing an estimate of vertical conductivity structure to be made (the 'spatial gradient method').

Estimates of the Earth's conductivity are obtained by measuring the response to an applied electromagnetic field. The natural, external magnetic field may be used as the applied field, or a man-made, 'controlled', source may be employed. For both methods, the lower the frequency of the applied field the deeper the energy penetration in the Earth. In Australia, the only controlled source method that has been used for deep studies is resistivity sounding, in which the source field is a direct current passed through the ground by means of two electrodes; the receiver consists of two more electrodes and a voltmeter. Because the frequency is fixed (at D.C.), the depth of investigation is controlled by the source-receiver separation, and it is difficult to obtain the separations needed to study the deeper lithosphere. However, separations of 20-200 km are possible, allowing crustal structure to be examined. Transient electromagnetic (T.E.M.) methods have mostly been used to explore for ore bodies, where penetration depths of only a few hundred metres are adequate, but it is possible to use T.E.M. methods for investigating deeper structure (e.g. Vozoff et al., 1985). Transient methods have been successfully employed on other continents to study deep crustal structure, but they have yet to be used for that purpose in Australia. The two main natural source methods are magnetometer array studies (geomagnetic depth sounding or G.D.S.) and magnetotelluric (M.T.) studies. In M.T. experiments, the horizontal magnetic and electric fields are observed simultaneously; essentially the magnetic field is assumed to be the source field and the orthogonal electric field the response. M.T. sounding has been popular in exploration, so the literature contains many papers on the subject. Cagniard (1953) is credited with the original concept. Vozoff (1972) presented an early review of the

After the natural or controlled source measurements have been made, the first step in interpreting the results in terms of geological or physical structure is to model them by conductivity variations in the Earth. This is not so hard to do if the object of the experiment is to delineate qualitatively a lateral change in conductivity, as is often the case with G.D.S. studies, but if a quantitative assessment of conductivity is desired then the problem rapidly becomes very difficult. Typically a ID distribution of conductivity with depth is desired from M.T. or controlled source data. The relative ease with which the forward problem (computation of the response of an assumed model) may be solved has led many to underestimate the difficulty of the inverse problem (estimation of some property of the real Earth from data collected over it). This difficulty arises both from the nonlinear nature of the forward solution, which limits the application of established linear inverse theory, and from the

Geol. Soc. Aust. Spec. Publ. 17,121-140.

non-uniqueness of the E.M. inverse problem. Although it is possible to show that an infinite quantity of perfect data will uniquely determine the Earth's conductivity at all depths, this result is lost when a finite number of data with errors is considered. Any set of real data may be satisfied by an infinite number of different conductivity models if it can be modelled at all. It is generally considered that inverse problems are composed of two parts; the construction of a model which fits the data and the assessment of the reliability of this model. Most workers restrict themselves to model construction, partly because model assessment is much more difficult and partly because the non-uniqueness is not generally appreciated, producing false confidence in whichever of the infinite number of models has been found. One of the first serious attempts at model assessment in geophysics was made by Backus and Gilbert (1967, 1968), who presented the notion of resolving kernels. These are weighted averages of the model over depth which are uniquely determined by the data. Unfortunately this theoiy is only valid for the linear problem and not strictly appropriate for the interpretation of conductivity data, although it has sometimes been applied to the linearized E.M. problem (eg. Parker, 1970). Oldenburg's (1983) 'funnel' functions provide a similar approach to M.T. studies, using an iterative method to achieve bounds on box-car averages of conductivity with depth. Another approach to the problem is to force the construction method to produce a model having a particular characteristic. One useful property a model may have is that of globally minimizing the misfit between the model response and the observed data. This has been accomplished for both M.T. and resistivity sounding in one dimension by Parker (1980, 1982), using the usual c2 measure of misfit (the sum of the squared deviations between the data and the response of the theoretical model). The models produced using this criterion are totally non-physical, being delta functions of conductivity for M.T. sounding (called a D + model by Parker) and vanishingly thin layers for resistivity sounding (bilayer models), so this approach is best used to test for the existence of ID models satisfying data to within the desired misfit. It should be noted that

125

the most commonly used least squares model construction methods aspire to these non-physical models by seeking to minimize misfit, although the restrictive model spaces employed (such as those of a fixed number of layers or Hilbert spaces) prevent their attainment. Another useful property of a model is that of being maximally smooth. First suggested for M.T. sounding by Parker & Whaler (1981), a practical algorithm for obtaining such models for any sounding technique was presented by Constable et al. (1987). The advantage of this method is that the models are physically reasonable and presumably contain no complexities of structure not featured in the real Earth. The most common way in which data are interpreted is by finding a model consisting of a few (about 2 to 5) layers, either by trial and error modelling or by repeated modification of a linearized starting model. The modeller usually seeks to minimize %2, and so if too many layers are included the model tends towards the extreme solutions described above. Thus the choice of the number of layers (ie. the parameterisation) heavily biases the outcome of the modelling. The iterative schemes are also biased by the structure of the starting model. Monte Carlo modelling (the automatic generation and testing of a huge number of random models) suffers in a similar way; close examination will usually reveal that the 'random' selection of models is in fact restricted to only 2 or 3 layer structures, which again has a great influence on the outcome. Care must also be taken to ensure that the bounds necessarily used to restrict the range of models considered do not influence the results.

THE COAST EFFECT The term 'coast effect' refers to the distinctive behaviour of the geomagnetic field near a continental coastline caused by the large lateral variation in conductivity between the continental crust/mantle and the ocean and oceanic crust/mantle. A review of the coast effect is given by Parkinson and Jones (1979). In the absence of lateral conductivity and source-field variations, telluric currents induced by time variations in the external magnetic field

126

Electrical studies, Australian lithosphere.

Australian Coast Effect •

Everett & Hyndmon (1967) Shield

x Everett & Hyndmon (1967) Fold Belt o White & Hopgood (1979)

\rxi X

+ White & Polotoyko (1978) a

c o oc D LL.

White ic Polotoyko (1985) Geosyncline

X White & Polotoyko (1985) Croton O Bennett J

W c o

200 300 Kilometres from Coast

Fig. 2. A compilation of magnetometer data illustrating the coast effect over the Australian continent at a period of 1 hour. The solid lines show the response for an infinitely conducting ocean in a half-space having a finite resistivity of 100 ftm (lower curve) and 500 Qm (upper curve) (Ravel et al., 1981). The broken lines are models from Cox et al. (1971) in which the ocean has a realistic conductance but the lithosphere is approximated by material of infinite resistance and an infinitely conductive mantle placed at depths of 240 km (lower curve) and 420 km (upper curve). produce a secondary field having no vertical component. Thus, the time-varying magnetic field is strongest in the horizontal plane. As one approaches a strong lateral variation in conductivity, such as a coastline, induced currents concentrated to the side of an observer produce a vertical magnetic field component, causing the time-varying magnetic field to be strongest in a plane which tilts up towards the conductor. Parkinson (1962) defined a vector which is the downward normal to this preferred plane, projected onto the horizontal surface. Although other induction vectors which represent the transfer function between the vertical and horizontal fields may be defined, the real parts of these vectors are usually constructed so as to point towards regions of increased current flow, whether these be oceans or other regions of

anomalous conductivity. Induction vectors are much favored by those who deploy magnetometer arrays, as maps may be drawn for various frequencies in which these arrows simply point towards regions of interest. It is important to realize that a measurement of the coast effect is only a measure of conductivity contrast, and not absolute conductivity. Thus an increased coast effect may be achieved not only by decreasing the conductivity of the continental side but also by increasing the conductivity of the oceanic side. There is reason to believe that the ocean water alone does not account for the magnitude of the observed coast effect, and that a contrast between the oceanic and continental mantles is also required (Cox et al., 1971).

Geol. Soc. Aust. Spec. Publ. 17,121-140.

Figure 2 presents a compilation of coast effect measurements taken in Australia, all at a period of 1 hour. One of the first studies was a pioneering experiment by Everett and Hyndman (1967a). Ten sites on the Precambrian shield of southwestern Australia and 2 sites on the fold belt rocks of southeastern Australia were occupied with a set of 3 component, digitally recording, proton precession magnetometers. Four of the western sites were occupied by recording voltmeters as well, allowing M.T. analyses to be made (Everett and Hyndman, 1967b). The authors noted that the coast effect was considerably larger in the west than the east, from which they concluded that the shield was more resistive than the fold belt. This is easily understood in terms of the lower porosity of igneous shield rocks and the higher porosity and heat flow in the fold belt region. In 1969 and 1970 Bennett and Lilley (1971) deployed the same magnetometers at five stations in southeastern Australia. The data collected are in basic agreement with the two southeastern stations of Everett and Hyndman. White and Polatayko, (1978, 1985), White and Hopgood (1979) and White and Milligan (1985) have deployed various magnetometer arrays in central south Australia, between Spencer Gulf and the Victorian border, occupying on the order of 100 sites in all. Two lines were digitized for Fig. 2 from an induction arrow map of White and Polatayko (1985, Fig. 4), one to the west of Spencer Gulf on the Gawler Craton and one to the east of the gulf on the Adelaide Geosyncline. The data are quite scattered, although they are consistent with the results for the east coast and generally below those for the W. A. shield. Attempts are usually made to group coast effect data into clear categories of 'shield' and 'fold belt', which from inspection of Fig. 2 may have suppressed what appears to be either a continuum of continental conductivities or a veiy noisy estimation process, or, most probably, both. There is also the question as to whether the ocean alone could produce the observed coast effect, or whether enhanced conductivity beneath the oceanic crust is required. We may go some way to answering both these questions by examining numerical models of the coast effect. The two broken lines on Fig. 2 belong to models from Cox et al. (1971) in which the ocean has a realistic conductivity and thickness

127

but the lithosphere is approximated by material of infinite resistance. The models must be terminated by conductive material, and the two lines show an infinitely conductive mantle placed at depths of 240 km (lower curve) and 420 km (upper curve). For comparison, the solid lines show the response from a complementary modelling scheme whereby the ocean is approximated by an infinite conductor embedded in a half-space having a finite resistivity (Fischer etal., 1978, Ravel etal., 1981), in this case 100 Qm (lower curve) and 500 Qm (upper curve).

Because the ocean is not many skin depths deep at a period of 1 hour, the models of Cox et al. (1971) are more applicable for the Australian coast effect data, and it is clear that these models generally do a better job of fitting that data. Considering the error suggested by scatter within data sets and Bennett & Lilley's error bars, these models provide an entirely adequate fit. Since neither of the modelling schemes requires any contrast other than the ocean water, it may be concluded that the data provide no constraint on an oceanic/continental lithospheric conductivity contrast. Cox et al. (1971) were able to show that in California an enhanced oceanic mantle conductivity was required. However, they collected offshore electric field data, and it is important to observe that these were vital to their interpretation. Their land data alone could be modelled adequately using a conductivity contrast provided only by the ocean, and so it is not surprising that the Australian data, all collected landward of the continental shelf, are not able to provide constraints on the conductivity of the oceanic mantle. The coast effect data do, however, seem to be sensitive to the continental resistivity structure. There is a tendency for data from shield regions to display a larger coast effect than data from elsewhere, but without good error estimates it is impossible to determine the significance of this. The parameters associated with the modelling schemes are reasonably well determined, the depth to the conductor in the Cox et al. model being 300+100 km and the half-space resistivity of the Ravel et al. model being 300±200£2 m. In 1971 an array of Gough-Reitzel magnetometers were deployed in southeastern

128

Electrical studies, Australian lithosphere.

Australia throughout the states of Victoria and New South Wales to study the coast effect associated with the Tasman Sea and Bass Strait (Bennett and Lilley, 1973, 1974 and Lilley and Bennett, 1972). The authors modelled the data from a coastal station (normalized by an inland station) covering the period range of 0.5 hour to 1 day and concluded that the coast effect for the Tasman Sea could not be explained by sea water alone. They postulated a zone of partial melting in the upper mantle at a depth of no more than 50 km having a resistivity no more than 20 Qm. Although we have seen that the Bennet and Lilley (1971) data at a period of 1 hour do not require a contrast in conductivity other than that provided by the ocean, by looking at the coast effect over a frequency range of more than 1.5 decades the second experiment has much more power to discriminate between conductivity structures. Unfortunately, Bennet and Lilley's (1974) modelling relied heavily on an assumed ID conductivity structure as a background to the ocean/mantle structure, and so we cannot be sure that the data could not be fit by a different 1D model without a lateral contrast in the mantle, or, more likely, that an oceanic/continental contrast exists but would not be required to be as large for a different background structure.

however, they have attracted a disproportionate number of M.T. studies, clearly reflecting their potential as petroleum prospects. Sediments are much more porous than igneous or metamorphic rocks, and so are much more conductive, because all but the very near-surface rocks are wet. As in most arid lands, salts accumulate in the near-surface groundwater of inland Australia, dropping the sediment resistivities to as low as 1Q m. Sediments lose porosity, and therefore conductivity, with both age and depth.

The analysis of the first M.T. site of the Tasman' project (Ferguson et al., 1985), in which seafloor M.T. stations and land G.D.S. stations were deployed in a line across the Tasman Sea from New Zealand to New South Wales, also suggests an unusually high conductance for the upper 80 km of oceanic mantle. The response function is very anisotropic, however, and analysis of further stations shows this behaviour to be typical (Lilley et al., 1986). In view of the very large transverse resistance for the oceanic upper mantle measured by Cox et al. (1985), it is very likely that current channelling in the oceanic region between Australia and New Zealand is responsible for the low resistivity estimates of Ferguson et al., as well as affecting the data of Bennett and Lilley.

Moore et al. (1977) report results from five M.T. sites in the southern Cooper Basin. Sediment resistivities of 1-5 Qm to a depth of about 3 km were modelled. The resistivity of the deeper lithosphere was given values of 100-2000 Qm in the ID modelling. The lower values are unusual, but the authors suggest a 2D structure in which the lower lithosphere has a value of 1000 Qm. The ID models for four of the sites, and the 2D model, suggest a lowering of resistivity at a depth of about 90 km.

SEDIMENTARY BASINS A good proportion of Australia is covered by sedimentary basins. In terms of electrical methods,

There have been two M.T. surveys in the Murray Basin; Vozoff et al. (1975) describe 5 M.T. sites in a N-S line near the South Australia/New South Wales border and more recently Cull and Spence (1985) and Cull (1985) report an E-W traverse across the northern Murray Basin and onto the Broken Hill Block. Both studies showed sediment resistivities to lie between 1 and 20 Qm over a thickness of about 5 km, although Cull and Spence (1985) model 500-600 Qm sediments at depth. Vozoff et al. (1975) noted that the difference in porosity implied by the resistivities between the sediments (10%) and basement (2.5 %) accounted for the 25 mGal gravity low observed over Pooncarie.

In the northeastern Officer Basin, combined M.T. and Schlumberger resistivity soundings were carried out at 12 sites by Jupp et al. (1979). One of the objects of this study was the detection of the overthrust contact of the Musgrave Block with the sediments, and the results showed that overthrusting may be less extensive than suggested by modelling of gravity data. The familiar pattern of sediment resistivities in the range 1-10 Qm to a depth of about 3km underlain by lithosphere of 500 -10 4 Qm was presented. However, the authors modelled the lowering of resistivity to 10 Qm at a

Geol. Soc. Aust. Spec. Publ. 17,121-140. depth of 200-250 km, in contrast to the 90-100 km values from M.T. studies in the Cooper and Murray basins to the east.

129

CONDUCTIVITY ANOMALIES

Following the deployment (by Bennett and Lilley) of the 1971 array in southeastern Australia a second array (Lilley, 1976) was deployed across Bass Strait, to provide better coverage of a pronounced anomaly off the southwest edge of the first array. It was well established that the centre of the anomaly was over the coast of Victoria near the Otway Ranges, and it is thought to be associated with a conductive body at depth, possibly elongated in a SW-NE direction. Bennett and Lilley (1974) present convincing arguments against the anomaly being associated with induction or current channelling in the Otway Basin. The anomaly is coincident with a region of seismic activity, and Lilley (1976) demonstrates the striking agreement beThere have been two other electrical studies in tween the anomaly position and the projected the Eromanga Basin, both using M.T. sounding. present position of a southward migrating central Whiteley and Pollard (1971) report an M.T. sound- volcano province discussed by Wellman and Mcing in the same region as the 1977 magnetometer Dougall (1974). Lilley (1976) goes on to conclude array study. Resistivity sounding was used to deter- that the anomaly is a response to a region of hot mine shallow structure, and their experiment rock and partial melt associated with the volcanic showed the basin sediments to have resistivities of province. As is characteristic of G.D.S. studies, 1.5-55 Qm over a thickness of 4 km. The data were while the lateral position of the conductor is well modelled by a 10 Qm lithosphere underlain by 20 determined, its depth is not. Lilley (1976) argues Qra mantle at a depth of 90 km. Spence and Fin- that depths of between 10 and 75 km are consistent layson (1983) present data from 12 M.T. sites with various aspects of the data. situated further south in the basin. Although the Parkinson and Hermanto (1986) report the results showed some variation across the 200 km profile, they may be characterized as follows: very results of deploying magnetometers at about 40 sites in northeastern Tasmania. They observe a conductive (3-5 Qm) Jurassic-Cretaceous sediments extend from the surface to a depth of about pronounced conductivity anomaly coincident with 2 km, underlain by Permo-Triassic sediments of the Tamar lineament. Although the anomaly correresistivity 50-500 Qm. The basement, thought to be lates well with a change in geological province, the the Thompson Fold Belt rocks at a depth of about authors found it difficult to ascribe a physical cause 3-4 km, has resistivities of 1000-2000 Qm to a to the feature, principally because high conducdepth of40-60 km, whereupon the resistivity starts tivities of about 2S/m (0.5 Qm) were modelled very to drop again. By 60-150 km the resistivity has near the surface. Their suggestion of a fracture zone dropped to 1-10 Qm, which appears to be as- would require more modest conductivities, but could be consistent with electric current being sociated with the base of the lithosphere. channelled across the north-eastern corner of TasCull (1982b) describes the interpretation of 34 mania between two parts of the ocean. An analogue M.T. soundings across the McArthur Basin in the modelling study by Dosso et al. (1985) examined Northern Territory. The primary concern of the the effect of induction and current chanelling in the project was to study sedimentary structure, and waters surrounding Tasmania. They found that the interpretations for structure below 10 km deep are deflection of induced currents by the Tasmanian not presented. Sediment resistivities were between coastline and chanelling through the Bass Strait 20 and 500 Qm to a depth of up to 10 km, underlain were strong when the source E-field was oriented E-W. by basement of 1000-18,000 Qm.

In 1977 a magnetometer array was deployed in the Eromanga Basin to examine more closely the anomalous south-eastern station of a larger array deployed in 1976 (Woods and Lilley, 1980). The authors considered the results to be indicative of D.C. current flow, or "current channelling", and modelled a sheet of 2100 A total current. By equating high and low density regions of the current sheet with basement lows and highs they were able to obtain a qualitative map of basement topography, identifying one known and one previously unknown basement high.

130

Electrical studies, Australian lithosphere.

In 1971 an array of 25 Gough-Reitzel magnetometers was deployed across South Australia and into western Victoria (Gough et al., 1972, 1974). Telluric recorders were operated at nine of the magnetometer sites to provide an E-W M.T. traverse (Tammemagi and Lilley, 1973). The aim of these experiments was to study the deep structure associated with the Adelaide geosyncline, of which the Flinders Ranges form a surface expression. The G.D.S. study delineated a pronounced N-S trending conductivity anomaly along the eastern flank of the Flinders Ranges. M.T. stations over this anomaly were characterized by very low apparent resistivities. Both Gough et al. (1974) and Tammemagi and Lilley (1973) considered the anomaly to be caused by a buried conductive body, but their estimates of depth differed; Gough et al. (1974) considered it to be in the lower crust or upper mantle, and Tammemagi and Lilley (1973) considered it to be no deeper than 10 km. The detection of this conductivity anomaly has prompted considerable recent activity in the region. Chamalaun (1985) deployed twenty digitally recording magnetometers across the central section of the 1971 array. The array spacing was one third of that of the 1971 experiment. The Flinders anomaly was clearly detected and shown to trend in a more SW-NE direction than N-S. On the basis of frequency dependence and line current modelling, Chamalaun (1985) considers the conductive region to lie less than 30 km deep. However, he notes that the coast effect dominates the data over the entire array at low frequencies, making it difficult to rule out the possibility of a deep conductor affecting the data. Many magnetometer sites have been occupied south of the original 1971 array, from the Eyre Peninsula to the Victorian border, by Anthony White and coworkers. The studies of White and Polatayko (1985) and White and Milligan (1985) show the southern extension of the Flinders anomaly to swing westward, probably into Spencer Gulf. Data from arrays to the southeast of this region (White and Polatayko, 1978,1985, White and Hopgood, 1979 and White and Milligan, 1985) display a normal coast effect with little evidence of anomalous current flow.

In 1981 and 1982 Constable (1985) conducted six 20 km electrode spacing resistivity soundings in the region, as part of a trial of new equipment. A buried conductor was not detected, suggesting it to be deeper than 5 km. However, the resistivities of the sediments east of the Flinders Ranges (the Frome embayment) were measured as 2-9 Hm, compared with 500-800 Qm for the metasediments forming the ranges and 3000 Hm for the Willyama Complex to the east. Since the sediments were up to 1 km thick in places, Constable (1985) suggested that current channelling through those sediments, between the Great Artesian Basin to the north and the Southern Ocean to the south, could have been at least partially responsible for the G.D.S. and M.T. results. This interpretation was supported by the results of Woods and Lilley (1980), in which 2100 A currents were detected flowing in surface sediments along the eastern edge of the Great Artesian Basin. There is growing evidence for a shallow flow of current in the Flinders anomaly, and that the anomaly cuts across several geological structures to pass from the Frome Embayment, across the geosyncline sediments, into Spencer Gulf (and possibly then even entering the Gawler Block). This strongly suggests that the Flinders anomaly is an expression of a regional flow of telluric current taking a path of least resistance, rather than local induction in a buried conductive body. Several workers (Woods and Lilley, 1980; Chamalaun, 1985; White and Polatayko, 1985; Constable, 1985) have suggested that the current flow continues north to link with the currents observed by Woods and Lilley (1980). An anomalous region on the Eyre Peninsula was investigated by White and Milligan (1984, 1986) using an array of 40 closely spaced (5 -km) magnetometers. A region, about 70 km long, of considerable N-S current flow was delineated, which the authors took to indicate a large fracture system less than 15 km deep in the Gawler Block. White and Polatayko (1985) note the possibility that this anomaly may be in electrical contact with the Flinders anomaly via the water and sediments of Spencer Gulf, but also that this cannot be confirmed using the present data. White and Milligan (1986) note that their conductor coincides with the western flank of a 30 mGal gravity high having similar strike to the conductivity anomaly.

Geol. Soc. Aust. Spec. Publ. 17,121-140.

DEEP CRUSTAL STRUCTURE In 1970, Tammemagi and Lilley (1971) occupied 4 M.T. stations across the Lachlan fold belt, between Moruya and Griffith in N.S.W. Most of the data were strongly anisotropic (which would normally reflect 2D or 3D structure), results were only obtained between 103 and 104 s periods and the phase data were very scattered, so the ID interpretations given by the authors should be treated with caution. However, the apparent resistivities suggest that the resistivity for the deep lithosphere increases progressively towards the west, by an order of magnitude over the 420 km traverse. Tammemagi and Lilley (1971) suggest that this reflects a temperature difference of about 200 °C at a depth of about 200 km (ie., colder to the west). This is consistent with the olivine resistivity versus temperature data of Duba et al. (1974) and the heat flow compilation of Cull (1982a) which shows a difference of up to 30 mW/m2 across the Lachlan fold belt. Jupp et al (1979) modelled a drop in resistivity at 200 - 250 km under the Officer Basin and Moore et al. modelled a drop at about 90 km under the Cooper Basin. Both studies estimated the lithospheric resistivity as 103-104 Qm. The Murray basin M.T. studies modelled the resistivity of the deep crust and lithosphere as 10 3 -5xl0 3 Qm. Vozoff et al. (1975) also modelled a drop in resistivity to 10-50 £lm at about 100 km deep for all stations, and associated this with the base of the lithosphere, but Cull and Spence (1985) (also Cull, 1985) did not detect such a feature. Although it is within the power of the M.T. method to resolve such differences, these soundings were designed to study crustal, not lithospheric, structure, and so only the very longest period data are sensitive to this conductive feature. Therefore the range of depths suggested by these interpretations may not be real. It should be noted here that the phase of the M.T. function responds to a drop in resistivity at shorter periods than the apparent resistivity (amplitude), yet in only half of the above studies is it clear that phase has been inverted. Of particular interest was Cull and Spence's interpretation that anisotropy in the data at periods greater than 10 s was caused by strong vertical foliations in the Willyama Complex (the substance of the Broken Hill Block) between 15 and 35 km deep. Persistent anisotropy in the eastern

131

soundings suggest an extension of an anisotropic Willyama Complex under the Murray Basin sediments at least as far as Menindee. Lilley et al. (1981 a, b) applied a spatial gradient analysis to the 1971 S.E. Australian array data. In this method the array data are assumed to be free from effects caused by lateral variations in conductivity, and horizontal gradients of the horizontal magnetic field are used to calculate complex impedances, which may then be interpreted just as M.T data are. In 1976 Woods and Lilley (1979) had deployed 21 magnetometers in central Australia, across the southern Northern Territory and extending into western Queensland. Apart from the reversal of the magnetic field at one station in the S.E. corner of the array, the results showed no anomalous structure and so the data were also considered suitable by Lilley et al. (1981a, b) for a ID analysis using the spatial gradient method. The Gough-Reitzel magnetometers, however, operate at lower frequencies than most M.T. systems, and in this study data were restricted to daily harmonics (periods between 2xl0 4 and 9xl0 4 s) in order to ensure a gradient in the source field. Trial and error fitting and Monte Carlo modelling were used to interpret both data sets, which led the authors to conclude that there exists an order of magnitude difference in conductivity between S.E. and central Australia at depths of 200-300 km. However, this hypothesis was never tested directly, and the limitations of the modelling methods they used have been noted earlier in this paper. Although the work of Lilley et al. (1981b) shows that models may have large differences in conductivity and fit the data, it can be demonstrated that these spatial gradient data do not require a difference in conductivity, that is, that a single model fits both data sets. The D + algorithm of Parker (1977), constructing the best possible ID model (in a %2 sense) for a given data set, was used on the data of Lilley et al. to produce a joint fit well within the error bars (to a root mean square, or r.m.s., misfit of 0.8 standard errors). One new technique which avoids the problems of fitting layered models is the algorithm of Constable et al. (1987) which constructs the smoothest (in the sense of the integrated first or second derivative) models fitting the data to a specified level of misfit. Figure 3 shows the models fitting the spatial gradient data

Electrical studies, Australian lithosphere.

132

(A) Spatial Gradient Models

2.5 E 2.0

c -4>s —' :> 1 . 5

'cn 0)

1.0

Central Aust., rms=0.80

S.E. Aust., rms=0.80

.5 -

Joint model, rms=0.85

I I I I I I 11

1o 4

I I I I I I 11

1 o5 Depth, m

iiit 10

(B) Responses of Joint Mode! Centrol Aust.

5.6 i - o -5.2 4.8 4.4 i i i J I L 30000 60000 9000 •XT

30000

60000

Period, s

90000

Period, s

Fig. 3.(a) One-dimensional models fitting the spatial gradient data of Lilley et al. (1981), illustrating the similarity in Earth structure allowed by the data sets. The solid and coarsely broken lines show the result of fitting smooth models to the two data sets independently, and even though we are probably overfitting the data (an r.m.s. misfit of 0.8 is close to the minimum possible) there is a remarkable similarity. However, the direct test is to fit the two data sets simultaneously, and to do so we have to relax the misfit a little, to r.m.s. = 0.85 (finely broken line). (b) The data and the response of the joint model (the sign has been reversed on the imaginary components of c, which are the lower curves).

Geol. Soc. Aust. Spec. Publ. 17,121-140.

sets to an r.m.s. error of 0.8 and having the smallest (slope)2 integrated over all depths. The idea of a large difference in conductivity at 200-300 km depth is not supported. The figure also shows a smooth model which fits both data sets simultaneously to an r.m.s. misfit of 0.85. If the data errors are correct, it is most probable that the real Earth's structure fits the data to an r.m.s. misfit of 1.0. There is evidence that the errors on the spatial gradient data are over-estimated, which allows fits significantly better than r.m.s. = 1.0 to be made. However, error estimation is a difficult process, and we must be glad that any errors at all were presented, for much research is published without uncertainty estimates. The acceptance criterion used by Lilley et al. ensured that the r.m.s. misfit was 1.0, by forcing the model to pass through all the error bars (a veiy unlikely event), so we have exceeded the r.m.s. restriction implied by the original modelling. We have not disproved the hypothesis of differing conductivity between the two regions, but rather have shown that it cannot be proved that a difference does exists. However, had our test failed (ie. even D + modelling could not fit both data sets acceptably well) we would have proved that a difference in conductivity must exist. Such a result would have been more conclusive than a great many Monte Carlo simulations. Departing from natural source field methods which have been the more commonly used in Australia, Constable et al. (1984) conducted three resistivity soundings on the central Australian shield, using electrode spacings of up to 200 km. The results showed the upper crust to have a resistivity of at least 6 x 103 Qm in the northern section (the Tennant Creek Block) and at least 4.5 x 104 Qm to the south (the Arunta Block). The difference could be explained by a lower porosity in the Arunta rocks, but also correlates with a higher heat flow over the Tennant Creek Block. The resistivity sounding method is preferentially sensitive to the resistivity-thickness product (or transverse resistance) of the resistive part of the crust, which was 8

well determined to be 2x10 Qm for the Tennant Creek Block and less well determined as 3x10 Qm for the Arunta. The data also suggested a lowering of resistivity to less than 103 Qm at a depth of 10-20 km. A problem with large-scale resistivity soundings is their sensitivity to lateral

133

variations in resistivity, which may be impossible to distinguish from structure with depth, but Constable et al. noted that even if the surrounding sedimentary basins were infinitely conductive they would not produce the observed effect. World-wide conductivity studies have suggested a vertical conductivity structure for continents which includes a conductive (10-100 Qm) layer in the mid-crust (10 - 15 km deep) which is usually, but not always, present, and a conductive (about 10 Qm) layer in the upper mantle at depths of 100 - 200 km. The only evidence for a mid-crustal conductor in Australia comes from the central Australian resistivity soundings of Constable et al. (1984). The G.D.S. data would not detect such a feature because they are too low frequency and not very sensitive to vertical structure. The magnetotelluric soundings have generally been carried out over deep sedimentary basins which make them less sensitive to thin conductive layers in the midcrust. Elsewhere in the world there is a correlation of crustal low resistivity zones with seismic low velocity zones or the top of a feature often termed the Conrad discontinuity. This is an indication that the feature may be present in Australia, as crustal low velocity zones have been detected there (eg. Finlayson et al. 1980). The reason for these zone's existence is in doubt; free water in the mid-crust is a favourite theory and would account for both the seismic and electrical signature. However, why the water should not be present above or below the zone is not clear. Carbon (as graphite) is common in sedimentary rocks and could lower the conductivity dramatically (Duba and Shankland, 1982). Magnetite is also very conductive, and a few percent of interconnected magnetite derived from serpentinisation of a mafic rock can lower the resistivity to about 10 Qm (Brace and Orange, 1968, Stesky and Brace, 1973). (Note that Jones (1987) draws the opposite conclusion from these papers. The interested reader is encouraged to study them himself.) Finally, although Karia and Shankland (1983) showed that dry mafic rocks are only slightly more conductive than dry felsic rocks, wet basic rocks measured by Lee et al. (1983) appeared to be 1-2 orders of magnitude more conductive than predicted by Archie's Law (felsic rocks of similar porosity obeyed Archie's Law).

134

Electrical studies, Australian lithosphere.

Thus an acid to basic transition in a wet crust could account for the observed crustal conductivities. That a crustal conductive zone has not been seen in most of the Australian experiments might be partially due to the fact that many of the studies were carried out before it was generally appreciated that this zone existed. Not so with the upper mantle low velocity zone/low resistivity zone. Everett and Hyndman (1967b), for example, make it clear their preferred depth to this layer (70 km in the southwestern Australian shield) is taken from other work. One suspects that this drop in resistivity is often included in models because it is expected to be there. This is not an incorrect thing to do, but makes it hard to know whether the layer is required by, or merely consistent with, the data, and what the variance associated with its depth is. Again, the reason for this zone's existence is still in doubt. Its depth associates it with the base of the seismic and rheological lithospheres and fluids again provide the most popular hypothesis, this time in the form of partial melting of mantle material. Another hypothesis is appealing in this context. Olivine conductivity is mostly controlled by crystal defects (there may be an additional, electronic component to conduction in olivine, Shankland, 1975). Defects created by deformation associated with plate motion should therefore enhance the conductivity at the base of the lithosphere. It is not known whether this effect is of sufficient magnitude to account for the observed conductivities, but it must be remembered that temperature is already providing a background of increasing conductivity with depth. Carbon is also a possibility for upper mantle conduction until shown otherwise (Duba and Shankland, 1982).

models for inversions are approximately correct. Because electrical methods lack the resolution to distinguish between a sharp change in conductivity and a gradual one, if a sharp transition is known to exist this may be included as additional information by an appropriate choice of layered model, thereby improving the resolution of the method. Thus, for example, if an electrical method is used to determine the depth to a sediment/basement interface and the thickness of sediment is roughly known, the results of a sensibly posed layered inversion will estimate the required depth quite accurately. However, if the depth to a rise in conductivity driven by a gradual increase in temperature with depth (as in parts of the upper mantle) is modelled by a layered structure, the results are as likely to depend on the layering used as the real Earth structure. It is too much to expect that even a well executed electromagnetic sounding will determine, by itself, the structure of the Earth; the inverse problem is too non-unique. Some additional information must be included in the interpretation. This may be that the Earth's conductivity is smoothly varying, that there are a certain number of sharp changes at approximately known depths, or that there is a relationship between conductivity and some other measurable parameter (say, density). The models found from the E.M. data will, of course, only be as good as these other assumptions. The current trend towards fitting E.M. data by 2 dimensional models will only make these statements more applicable, because it is very difficult to collect enough extra data to overconstrain the model to the level of the 1D layered interpretations. There are indications that regularization approaches similar to that of Constable et al. (1987) for ID inversion will prove invaluable for 2D inversion schemes.

DISCUSSION Many of the conductivity studies undertaken in Australia are difficult to interpret because of the inadequate analysis techniques used and the possibility of current channelling affecting the experiments. The layered models almost always presented to interpret data are only meaningful if the real earth is known, a priori, to be layered, and the starting

Hypothesis testing is a much more reliable approach than trying to determine too much about the Earth's structure with too little information. We saw above that the hypothesis that S.E. and central Australia had different conductivities may be tested directly. If no model could be found to fit both data sets simultaneously the hypothesis would be true. The actual nature of any model found to show that this hypothesis in not necessarily true is relatively unimportant, and could in fact be totally non-physi-

Geol. Soc. Aust. Spec. Pubi 17,121-140. cal, as are the models generated by Parker's D algorithm.

Current channelling presents a different problem; Jones (1983) gives an excellent review. It occurs when induction in a large conductive body such as an ocean or perhaps a big sedimentary basin drives a current flow through a different region, usually smaller and which may be geographically removed from the larger body. The danger is then that the results of a natural source field experiment such as M.T. or G.D.S., operated over the region where currents are being channelled rather than induced, will be interpreted on the assumption that the currents have been induced locally. This will result in a model in which the currents are replaced by conductive regions. If the currents are large, the conductivity will be large, perhaps unrealistically so. This is a familiar situation in Australia; the Flinders conductivity anomaly was modelled as being 0.1 dm; the upper mantle under the Tasman sea was initially modelled as being unusually conductive; the latest anomaly in Tasmania is being given a value of 0.5 Qm. It is true that the channelled current flow will preferentially follow conductive paths, but the interpretation of the depth and magnitude of conductive regions will be distorted by incorrect assumptions of local induction.

CONCLUSIONS It appears from the Australian studies, particularly those using M.T., that there is a welldefined electrical lithosphere for the continent whose lower boundary is characterized by a drop in resistivity from greater than 1000 Qm to 1-50 Hm. The measured depth to this boundary varies between 90 and 200 km, but whether this variation is real or due to experimental and interpretational uncertainty has never been tested. All the modern M.T. studies have been designed to examine sedimentary structure and porosity (which they have accomplished rather well), not deep lithospheric structure. Spatial gradient analyses from array studies have been restricted to less than a decade of frequency, offering limited resolution. The one deep controlled source study performed could not penetrate to such depths.

135

It seems reasonable to expect an enhanced conductivity at the base of the rheological lithosphere, as ductile flow, partial melting and an increase in temperature would all be expected to increase conductivity. An increase in pore fluid (water or melt) decreases both seismic velocity and electrical resistivity and probably accounts for the observed coincidence of seismic low velocity zones and electrical conductors. This would suggest that the base of the lithosphere is a region of partial melting. Whatever the reason for the increased conductivity, it appears that the M.T. method provides a useful tool for mapping lithospheric thickness, as it is sensitive to a lowering of resistivity at depth. Within the lithosphere electrical properties are not easily characterized. The sedimentary basins are very conductive, nearly always having resistivities less than 1000 Qm, and usually 1-50 Qm throughout their entire thicknesses, which may reach 10 km. This reflects their porous nature and the high salinity of the ground water. The bulk of the lithosphere has been given values between 10 and 10 Qm by the various studies. Few experiments have resolved structure within the region between basement and the asthenosphere, and more work needs to be done before much can be said on the basis of electrical studies. A few things seem clear, however. There does not appear to be a pronounced electrical Moho; this is consistent with the idea that the Moho is a compositional change which would produce only a small resistivity contrast, difficult to detect with present techniques. Mid-crustal resistivities are too small to be consistent with dry rock laboratory values, suggesting that crustal conductivity is determined by something other than the silicate rocks, probably 'free' (ie., not structurally bound) water in concentrations of 1-5%. 5

The peril of unrecognized current channelling is the estimation of too large a conductivity for a body when natural source soundings are used. It seems likely that this has happened several times in the history of Australian conductivity studies. Attention has also been drawn to the extent to which the choice of modelling method may determine the outcome of E.M. interpretation.

136

Electrical studies, Australian lithosphere.

ACKNOWLEDGEMENTS

CONSTABLE S.C. 1985. Resistivity studies over the

The author would like to thank Cathy Constable, Alan Jones, Ted Lilley and A1 Duba for their comments on the manuscript, and Bob Parker and Chip Cox for many useful discussions over the last five years.

BACKUS G. & GILBERT F. 1967. Numerical applications

of a formalism for geophysical inverse problems. Geophysical Journal of the Royal astronomical Society 13, 247-276

BACKUS G. & GILBERT F. 1968. The resolving power of gross Earth data. Geophysical Journal of the Royal astronomical Society 16, 169-205. BENNETT D.J. & LILLEY F.E.M. 1971. The effect of the southwest coast of Australia on transient magnetic variations. Earth and Planetary Science Letters 12, 392-398. BENNETT D.J. & LILLEY F.E.M. 1973. An array study of daily magnetic variations in southeast Australia. Journal of Geomagnetism and Geoelectricity 25, 39-62. BENNETT D.J. & LILLEY F.E.M. 1974. Electrical conductivity structure in the southeast Australian region. Geophysical Journal of the Royal astronomical Society 37, 191-206.

BRACE W.F. & ORANGE A.S. 1968. Further studies of the effect of pressure on electrical resistivity of rocks. Journal of Geophysical Research 73, 5407-5420. CAGNIARD L. 1953. Basic theory of the magneto-telluric

method of geophysical prospecting. Geophysics 18, 605-635. CHAMALAUN F.M. 1985. Geomagnetic deep sounding

experiment in the central Flinders Ranges of South Australia. Physics ofthe Earth and Plantary Interiors 37, 174-182.

F.M.

CONSTABLE S.C., MCELHINNY M.W. & MCFADDEN P.L.

1984. Deep Schlumberger sounding and the crustal resistivity structure of central Australia. Geophysical Journal of the Royal astronomical Society 79, 893-910. CONSTABLE S.C., PARKER R.L. & CONSTABLE C.G. 1987.

REFERENCES

CHAMALAUN

Flinders conductivity anomaly. Geophysical Journal of the Royal astronomical Society 83,775-786.

WALKER

1982.

microprocessor based digital fluxgate magnetometer for geomagnetic deep sounding studies. Journal of Geomagnetism and Geoelectricity 34,491-507.

Occam's inversion; a practical algorithm for the inversion of electromagnetic data. Geophysics 52, 289-300.

CONSTABLE S.C. & DUBA A. 1990. The electrical conductivity of olivine, a dunite, and the mantle. Journal Geophysical Research 95, 6967-6978. C o x C.S., CONSTABLE S.C., CHAVE A.D. & WEBB S.C.

1986. Controlled-source electromagnetic sounding of the oceanic lithosphere. Nature 320, 52-54.

C o x C.S., FILLOUX J.H. & LARSEN J.C. 1971.

Electromagnetic studies of ocean currents and electrical conductivity below the ocean floor. The Sea 4(1), Maxwell (ed.), Wiley. CULL J.R 1982a. An appraisal of Australian heat-flow data. BMR Journal ofAustralian Geology & Geophysics 7, 11-21. CULL J.R 1982b. Magnetotelluric profiles in the McArthur Basin of northern Australia. BMR Journal of Australian Geology & Geophysics 7, 275-286. CULL J.P. 1985. Magnetotelluric sounding over a Precambrian contact in Australia. Geophysical Journal of the Royal astronomical Society 80, 661-675. CULL J.R & SPENCE A.G. 1985. Magnetotelluric

soundings across the boundary of the Precambrian Willyama Complex and the Cainozoic Murray Basin, southeastern Australia. Bureau of Mineral Resources, Australia, Report 250. DeGROOT-HEDLIN C. & CONSTABLE S.C. 1990. Occam's

inversion to generate smooth, two-dimensional models from magnetotelluric data. Geophysics, 55 (12). Dosso H.W., NIENABER W.& PARKINSON W.D. 1985. An analogue model study of electromagnetic induction in the Tasmanian region. Physics of the Earth and Planetary Interiors 39, 118-133.

Geol. Soc. Aust. Spec. Publ. 17,121-140. DUBA A., HEARD H.C. & SCHOCK R.N. 1974. Electrical

conductivity of olivine at high pressure and under controlled oxygen fugacity. Journal of Geophysical Research 19,1667-1673. DUBA A. & NICHOLLS I.A. 1972. The influence of

oxidation state on the electrical conductivity of olivine. Earth and Planetary Science Letters 18,59-64. DUBA A. & SHANKLAND T.J. 1982. Free carbon and the

137

JONES A.G. 1983. The problem of current channelling: a critical review. Geophysical Surveys 6,79-122. JONES A.G. 1987. M T and reflection: an essential

combination. Geophysical Journal of the Royal astronomical Society 89,7-18.

JUPP D.L.B., KERR D.W., LEMAIRE H., MILTON B.E., MOORE R.F., NELSON R. & VOZOFF K. 1979. Joint

electrical conductivity of the Earth's mantle Geophysical Research Letters 9,1271-1274.

magnetotelluric-DC resistivity survey - eastern Officer Basin. Australian Society of Exploration Geophysicists, Bulletin 10, 209-212.

EVERETT J.E. & HYNDMAN R.D. 1976a. Geomagnetic

JUPPD.L.B.&VOZOFF K. 1975. Stable iterative methods

variations and electrical conductivity structure in southwestern Australia. Physics of the Earth and Planetary Interiors 1, 24-34.

for the inversion of geophysical data. Geophysical Journal of the Royal astronomical Society 42, 957-975.

EVERETT J.E. & HYNDMAN R.D. 1967b. Magneto-telluric

conductivity of dry lower crustal rocks. Geophysics 48,

investigations in southwestern Australia, Physics of the Earth and Planetary Interiors 1, 49-54. FERGURSON I.J., FILLOUX J.H., LILLEY F.E.M., BINDOFF N.L. &MULHEARN,RJ. 1985. Aseafloormagnetotelluric

sounding in the Tasman Sea. Geophysical Research Letters 12,545-548.

FINLAYSON D.M., COLLINS C.D.N. & DENHAM D. 1980.

Crustal structure under the Lachlan fold belt, southeastern Australia. Physics of the Earth and Planetary Interiors 21, 321-342. FISCHER G., SCHNEGG R-A. & USADEL K.D. 1978.

Electromagnetic response of an ocean-coast model to E-polarization induction. Geophysical Journal of the Royal Astronomical Society 53, 599-618. GOUGH D.I., LELLEY F.E.M. & MCELHINNY M.W. 1972.

A polarization-sensitive magnetic variation anomaly in South Australia. Nature 239, 48-91.

KARIYA K.A. & SHANKLAND T.J. 1983. Electrical 52-61.

KAUFMAN A. & KELLER G.V. 1981. The Magnetotelluric Sounding Method, Elsevier, Amsterdam. KELLER G.V. & FRISCHKNECHT F.C. 1966. Electrical Methods in Geophysical Prospecting, Pergamon Press, London. LEE C. D., VINE F.J. & Ross R.G. 1983. Electrical

Conductivity models for the continental crust based on laboratory measurements on high-grade metamorphic rocks. Journal of Geophysical Research 72,353-371. LILLEY F.E.M. 1976. A magnetometer array study across

southern Victoria and the Bass Strait area, Australia. Geophysical Journal of the Royal astronomical Society

46, 165-184.

LILLEY F.E.M. & BENNETT D.J. 1972. An array

GOUGH D.I., MCELHINNY M.W. & LILLEY F.E.M. 1974.

experiment with magnetic variometers near the coasts of southeast Australia. Geophysical Journal of the Royal astronomical Society 29,49-64.

36, 345-362.

LILLEY F.E.M., FILLOUX J.H., FERGURSON I J., BINDOFF N.L. & MULHEARN P.J. 1986. The Tasman project of

A magnetometer array study in southern Australia. Geophysical Journal of the Royal Astronomical Society

HIRSCH L.M. & WANG C.-Y. 1986. Electrical conductivity of olivine during high-temperature creep. Journal of Geophysical Research 91,10429-10441. HYNDMAN R.D. & SHEARER P.M. 1989. Water in the

lower continental crust: modelling magnetotelluric and seismic reflection results. Geophysical Journal International 98, 343-365.

seafloor magnetotelluric exploration, presented at the 8th workshop on electromagnetic induction in the Earth and Moon, Neuchatel, Switzerland.

LILLEY F.E.M., WOODS D.V. & SLOANE M.N. 1981a.

Electrical conductivity from Australian magnetometer arrays using spatial gradient data. Physics of the Earth and Planetary Interiors 25, 202-209.

138

Electrical studies, Australian lithosphere.

LILLEY F.E.M., WOODS D.V. & SLOANE M.N. 1981b.

Electrical conductivity profiles and implications for the absence or presence of partial melting beneath central and southeast Australia. Physics of the Earth and Planetary Interiors 25,419-428. MOORE R.F., KERR D.W., VOZOFF K. & JUPP D.L.B.

1977. Southern Cooper Basin magnetotelluric survey, South Australia, 1974. Bureau of Mineral Resources, Australia, Record 1977/41. Oldenburg D.W. 1983. Funnel functions in linear and nonlinear appraisal. Journal of Geophysal Research 88, 7387-7398. OLHOEFT G.R. 1981. Electrical properties of granite with implications for the lower crust. Journal of Geophysical Research 86, 931-936. PARKER R.L. 1970. The inverse problem of electrical conductivity in the mantle. Geophysical Journal of the Royal astronomical Society 22, 121-138. 1980. The inverse problem of electromagnetic induction: existence and construction of solutions based on incomplete data. Journal of Geophysical Research 85,4421-4428. PARKER R.L.

PARKER R.L. 1982. The inverse problem of resistivity sounding. Geophysics 49, 2143-2158.

PIWINSKH A.J.& DUBA A. 1974. High temperature electrical conductivity of albite. Geophysical Research Letters 1, 209-211. RAVEL U., WEAVER J.T. & DAWSON T.W. 1981. The

ocean-coast effect re-examined Geophysical Journal of the Royal astronomical Society 115-123. ROKITYANSKY I.I. 1982. Geomagnetic Investigations of

the Earth's Crust and Upper Mantle. Springer-Verlag, Heidelberg.

SHANKLAND T.J. 1975. Electrical conduction in rocks and minerals: parameters for interpretation. Physics of the Earth and Planetary Interiors 10, 209-219. SHANKLAND T.J. & WAFF H.S. 1977. Partial melting and the electrical conductivity anomalies in the upper mantle. Journal of Geophysical Research 82, 5409-5417. SPENCE A.G. & FINLAYSON D.M. 1983. The resistivity structure of the crust and upper mantle in the central Eromanga Basin, Queensland, using magnetotelluric techniques. Geological Society of Aust, Journal 30,

1-16.

STESKY R.M. & BRACE W.F. 1973. Electrical conductivity of serpentinized rocks to 6 kilobars. Journal of Geophysical Research 78, 7614-7621.

PARKER R.L. & WHALER K.A. 1981. Numerical methods

H.Y. & LILLEY F.E.M. 1971. Magnetotelluric studies across the Tasman geosyncline, Australia. Geophysical Journal of the Royal astronomical Society 22, 505-516.

PARKINSON W.D. 1962. The influence of continents and oceans on geomagnetic variations. Geophysical Journal of the Royal astronomical Society 6,441-449.

TAMMEMAGI

for establishing solutions to the inverse problem of electromagnetic induction. Journal of Geophysical Research 86, 9574-9584.

PARKINSON W.D. & HERMANTO R. 1986. The Tamar conductivity anomaly. Exploration Geophysics 17, 34-35.

PARKINSON W.D. & JONES F.W. 1979. The geomagnetic

TAMMEMAGI

H.Y. & LILLEY F.E.M. 1973. A magnetotelluric traverse in southern Australia. Geophysical Journal of the Royal astronomical Society 31, 433-445.

VOZOFF K. 1972. The magnetotelluric method in the exploration of sedimentary basins. Geophysics 37, 98-141.

coast effect. Reviews of Geophysics and Space Physics

VOZOFF K. ed. 1986. Magnetotelluric Methods, Society of Exploration Geophysicists Rep. Ser., 5.

PARSONS B.& SCLATER J.G. 1977. An analysis of the variation of ocean floor bathymetry and heat flow with age. Journal of Geophysical Research 82, 803-827.

VOZOFF K., KERR D.W., MOORE R.F., JUPP D.L.B. & LEWIS R.G. 1975. Murray Basin magnetotelluric study.

17, 1999-2015.

Geological Society of Australia, Journal 22,361-375.

Geol. Soc. Aust. Spec. Publ. 17,121-140.

139

WOODS D.V. & LILLEY F.E.M. 1980. Anomalous VOZOFFK., LEBROCQK., MOSS D., ZILE M. & PRIDMORE D. 1985. Deep transient electromagnetic soundings for geomagnetic variations and the concentration of telluric petroleum exploration. Centre for Geophysical currents in southwest Queensland, Australia. Exploration Research, Report, Macquarie Univ., Sydney, Geophysical Journal of the Royal astronomical Society 62, 675-689. Australia. WELLMAN P. & MCDOUGALL 1.1974 Cainozoic igneous activity in eastern Australia. Tectonophysics 23,49-65. WHITE A.W. & HOPGOOD D.N. 1979. The island and

Note added in proof:

The gestation of this volume has been long, and the reader is warned that only the literature up to about 1987 has been searched. Since then the Geomagnetism and Geoelectricity 30,479-484. author himself has addressed the issue of grain boundary conduction in olivine rocks (Constable WHITE A. & MILLIGAN P.R. 1985. Geomagnetic Duba, 1990), showing that it probably is not variations across the southern Adelaide Geosyncline, and important, and has extended smooth modelling to South Australia. Journal of Geomagnetism and 2D (deGroot-Hedlin and Constable, 1990). Geoelectricity 37,715-728. Hyndman and Shearer (1989) have presented a WHITE A. & MILLIGAN P.R. 1986. A crustal conductor on comprehensive discussion of mid-crustal conductors and Hirsch and Wang (1986) have shown that Eyre Peninsula, South Australia. Nature 310, 219-222. deformation of olivine is not likely to be responWHITE A.W. & POLATAYKO O.W. 1978. The coast effect sible for the mantle high conductivity zone, leaving in geomagnetic variation in South Australia. Journal of fluids and carbon as remaining contenders. Work has continued on the Tasman MT data set, extendGeomagnetism and Geoelectricity 30,109-120. ing the interpretation to 2D (Kellet, unpub. thesis, A.N.U.) and 3D (pers. comm. with Graham HeinWHITE A. & POLATAYKO O.W. 1985. Electrical conductivity anomalies and their relationship with the son and Ted Lilley). Most recently 57 digitally tectonics of South Australia. Geophysical Journal of the recording magnetometers have been deployed in an Australia-wide G.D.S. array (pers. comm. with Royal astronomical Society 80,757-771. Francois Chamalaun and Charlie Barton), the first WHITELEY R.J. & POLLARD P.C. 1971. A combined deep time such an array has encompassed an entire conresistivity and magneto-telluric sounding in the tinent. Eromanga Basin, Queensland. Search 2, 103-105. S.C. Constable Scripps Institution of Oceanography, A-030 WOODS D.V. & LILLEY F.E.M. 1979. Geomagnetic induction in central Australia. Journal of Geomagnetism LaJolla, California, 92093 United States of America. and Geoelectricity 31,449-458.

coast effect in geomagnetic variations around St. Vincent's Gulf, South Australia. Journal of

Geol. Soc. Aust. Spec. Publ. 17,141-146.

141

The State of Knowledge of the Asthenosphere from Electromagnetic Soundings: Australia and Surrounding Oceans F.E.M. Lilley Research School of Earth Sciences, Australian National University, Canberra, Australia Much strength is achieved in interpretation in earth science by bringing to bear on the same problem information obtained from independant means. In geophysics, this process may be possible by examining quite separate physical processes. Thus the asthenosphere of the earth, for which some of the best geophysical evidence has come from seismology, stands out as a particularly attractive target for geomagnetic induction physics. The electrical conductivity of the asthenosphere should be greatly increased if partial melt is present. Further, shearing can increase electrical conductivity, by aligning conducting domains. An international project on the Electrical Conductivity of the Asthenosphere (ELAS) has been mounted, and the present paper summarises Australian results relevant to the ELAS project, as at 1984.

Introduction The Australian continent has a number of features which are advantageous for carrying out geomagnetic induction studies. It is a large, flat continent, ideal for observations. Travel by light aircraft has proved well-suited to overcoming the long distances involved. The weather is generally good if the appropriate season is chosen. A sparse population density contributes to undisturbed field operating conditions, and the whole continent is within one national boundary. The adjoining oceans are generally wide, thus simplifying the analysis of geomagnetic coast effects. The continent lies in mid-latitudes, and so generally enjoys geomagnetic source fields which vary smoothly with distance. The focus of the source field for the magnetic quiet daily variation passes across the continent. The whole continent lies within the Indian-Australian plate, but within this plate Australia contains a great variety of geological structures: the western part of the continent is generally old, Precambrian Shield; the eastern part of the continent is generally younger, being Palaeozoic and post Palaeozoic in age. Geomagnetic induction physics in Australia Within the Australian setting as described, a variety of studies of natural geomagnetic induction physics have been carried out. These studies can be grouped into four main categories:

1. Magnetotelluric measurements on land. At high frequencies such measurements probe sedimentary basins (Vozoff etal., 1975; Jupp etal., 1979; Dekker& Hastie, 1981; Spence & Finlay son, 1983) and in some environments contribute as an electrical prospecting method to ore search. At lower frequencies such measurements are used to probe crustal and upper mantle structure. 2. Magnetometer array studies on land. Typically these exercises have involved the simultaneous operation of some twenty recording magnetometers, working as temporary geophysical observatories. The instruments are situated some one hundred kilometres distance from each other, a whole array thus covering an area some hundreds of kilometres in horizontal scale length. The first such array exercise in Australia was that reported by Gough et al. (1974). The capability for such work in Australia has increased recently with the development of the magnetometer reported by Chamalaun & Walker (1982). Magnetometer array studies have produced two main classes of information on earth conductivity structure: 2.1 Where strong departures from horizontal layering exist in the electrical conductivity of the earth, recording magnetometers operating in arrays (or in lines, or in many cases individually) have mapped places where natural telluric currents are greatly concentrated along some preferred path of current flow. A recent example of such work in

142

Electromagnetic soundings.

Australia is the crustal conductor reported by White & Milligan (1984). 2.2 Where such strong departures from horizontal layering are absent, so that one-dimensional structure is demonstrated to be present on a large scale by the magnetometer array exercise itself, then the nature of the array data (simultaneous over two horizontal dimensions) has been fully exploited to produce information on the electrical conductivity profile of the earth down to depths of hundreds of kilometres (Woods, 1979; Lilley etal., 1981a, b). 3. Coast effect studies. Australian magnetic observatories have generally been at or near coasts, and so have played an important part from an early stage of the subject in the elucidation of the geomagnetic coast effect (see for example, Parkinson & Jones, 1979). Until the development of the continental-shelf magnetometer of White (1979), all observations of the coast-effect had been on the land side of coastlines. 4. Magnetotelluric studies at sea. There is currently being pursued a major project, the Tasman Project of Seafloor Magnetotelluric Exploration (TPSME), which is a collaborative project between the Scripps Institution of Oceanography in California, the Royal Australian Navy Research Laboratory, and the Australian National University. The Hinders University of South Australia and the Bureau of Mineral Resources are also participants in the project. The most difficult and hazardous part of this project, the period of marine and field observations, has now been completed, from December 1983 to March 1984 (inclusive). During the observation period, some twentyone seafloor instruments of the Scripps Institution of Oceanography brought to Australia by J.H. Filloux operated at nine sites on the floor of the Tasman Sea, between Australia and the Lord Howe Rise, recording (not all parameters at all sites) fluctuations in three components of the natural magnetic field, three components in the natural electric field, the ambient hydrostatic pressure, and the ambient temperature. The hydrographic section of the Royal Australian Navy provided deployment and retrieval cruises on the recently-commissioned oceanographic vessel HMAS COOK.

Simultaneously, on the line of the seafloor traverse continued from the coast inland into continental eastern Australia, eight Gough-Reitzel variometers recorded fluctuations in the three components of the geomagnetic field, as did the Canberra Magnetic Observatory of the Bureau of Mineral Resources. Where the line crossed the offshore edge of the continental shelf, a site was occupied for some weeks by the continental shelf magnetometer of Hinders University. The observation sites of the TPSME experiment are shown in Fig. 1. The first results of the TPSME experiment were reported by Ferguson et al., 1985; Filloux etal., 1985; Lilley etal., 1986; andBindoff etal., 1986.

Results relevant to the ELAS project From a great deal of information about earth electrical conductivity obtained by the above methods, the following items stand out to the present reporter as being relevant to the ELAS project: 1. Long-period magnetotelluric studies. Magnetotelluric observations at long periods all indicate electrical conductivity to increase with depth into the earth (Everett & Hyndman, 1967; Tammemagi & Lilley, 1971, 1973). From the point of view of the ELAS project, the interesting results are whether such increases in conductivity are sufficiently sharp to indicate the presence of an asthenosphere, and at what depths do such increases occur. In these respects, Vozoff et al. (1975) interpreted an increase in conductivity to occur at a depth of some 100 km beneath southeastern Australia. Moore et al. (1977) obtained a similar result of 90 km for the Cooper Basin in South Australia. Spence & Finlayson (1983) for southwestern Queensland have an increase in conductivity of three orders of magnitude (from 10~3 to 10° S.m"1) over the depth range 60 to 150 km. There is thus some consistency amongst these results, which are generally for the eastern half of Australia, of an increase in conductivity to a value of order 10"1 S.m"1 at a depth of 100km. Below the Officer Basin, which is in the western half of the continent, Jupp

Geol. Soc. Aust. Spec. Publ. 17,141-146. et al. (1979) find this increase in conductivity to occur at a greater depth, of order 250 km. 2. Magnetometer array studies. Using the spatial gradient method, Woods (1979) and Lilley et al. (1981a, b) have analysed data from magnetometer arrays in central and southeast Australia to obtain conductivity distributions for the two

143

regions as shown in Fig. 1. The results depend particularly on the analysis of quiet magnetic daily variations, and are new in the sense that application of the spatial gradient method to such data is still being explored. The resolution of the method is not yet clear. Nevertheless, there appears to be a definite indication that the gross electrical conductivity structure is substantially different under

4Q_

ZEALA ND

100

200

300 400 500 DEPTH (KM)

600

700

100

200

300 400 500 DEPTH (KM)

600

700

Fig. 1. The solid circles on the map show the 1983-84 sites of instruments operated for the Tasman Project of Seafloor Magnetotelluric Exploration. The two electrical conductivity distributions below the map are results from magnetometer array studies in central Australia and southeast Australia, and are taken from Lilley et al. (1981b).

144

Electromagnetic soundings.

southeast Australia relative to central Australia. Lilley et al. (1981b) interpret the result for the southeast Australian profile as indicating the presence of a small degree of partial melting, and in these terms the presence of an asthenosphere, at a depth of order 200 to 300 km. 3. The Tasman Project of Seafloor Magnetotelluric Exploration. While the observations for this project are now secured, data analysis is as yet at too early a stage to quote any results. However, it is relevant in this context to refer to the result of Bennett & Lilley (1974) who found, modelling the geomagnetic coast effect using array data on the land side only, that the highly-conducting sea water was not by itself sufficient to account for the geomagnetic coast effect at the long periods of the geomagnetic daily variation, and that a deeper contrast in electrical conductivity was needed between oceanic and continental geology. Such a contrast could be provided by an asthenospheric step downwards, from beneath ocean to beneath continent. One aim of the TPSME experiment is to supplement these earlier results with observations on both sides of the coastline.

add much new information concerning the structure of the Tasman Sea. The continued development of geomagnetic induction physics in Australia is important because further magnetometer array studies should contribute more knowledge of the asthenosphere, and of the electrical conductivity structure of the continent. Information on electrical conductivity anomalies has practical application to such matters as accurate aeromagnetic surveying (Lilley, 1982), and in understanding telluric 'noise' problems in the use of electrical methods to search for ore bodies and sedimentary basin hydrocarbon structures (Vozoff, 1984). Acknowledgements The author wishes to acknowledge the contributions of all whose work is reported in this paper, and to thank his collaborators in those projects in which he has himself participated. References

Conclusion The investigation, by exploiting geomagnetic induction physics, of a possible asthenosphere beneath the Australian continent and the surrounding oceans has commenced, but it is far from concluded. The present evidence from magnetometer array studies is for a relatively well developed asthenosphere beneath the southeastern region of the continent, in contrast to a less well developed asthenosphere beneath the centre of the continent. The results of magnetotelluric studies generally agree with this basic description, but interpret the rises in conductivity to occur at lesser depths, possibly because the magnetotelluric observations detect the commencement of conductivity increases whereas the magnetometer array results are giving bulk, averaged (and so deeper) conductivity values. The Tasman Project of Seafloor Magnetotelluric Exploration, currently under analysis, should

BENNETT, D.J., & LILLEY, F.E.M. 1974. Electrical conductivity structure in the southeast Australian region. Geophysical Journal of the Royal astronomical Society

37, 191-206.

BINDOFF, N.L., FILLOUX, J.H., MULHEARN, P.J., LILLEY, F.E.M., & FERGUSON, I.J. 1986. Vertical electric field

fluctuations at the floor of the Tasman Abyssal Plain. Deep-Sea Research 33, 587-600. CHAMALAUN, F.H.,

WALKER, R.

1982.

microprocessor based digital fluxgate magnetometer for geomagnetic deep sounding studies. Journal of Geomagnetism and Geo electricity 34,491-507. DEKKER, D.L., & HASTIE, L.M. 1981. Sources of error

and bias in a magnetotelluric depth sounding of the Bowen Basin. Physics of the Earth and Planetary Interiors 25, 219-225. EVERETT, J.E., & HYNDMAN, R.D. 1967. Magnetotelluric investigations in southwestern Australia. Physics of the Earth and Planetary Interiors 1,49-54.

Geol. Soc. Aust. Spec. Publ. 17,141-146. FERGUSON, I.J., FILLOUX, J.H., LILLEY, F.E.M., BINDOFF, N.L., & MULHEARN, P.J. 1985. A seafloor

magnetotelluric sounding in the Tasman Sea. Geophysical Research Letters 12, 545-548.

145

PARKINSON, W.D., & JONES, F.W. 1979. The geomagnetic

coast effect. Reviews of Geophysics 17, 1999-2015.

FILLOUX, J.H., LILLEY, F.E.M., FERGUSON, I.J., BINDOFF, N.L., & MULHEARN, P.J. 1985. The Tasman Project of

SPENCE, A.G., & FINLAYSON, D.M. 1983. The resistivity structure of the crust and upper mantle in the central Eromanga Basin, Queensland, using magnetotelluric techniques. Geological Society ofAustralia, Journal 30, 1-16.

GOUGH, D.I., MCELHINNY, M.W., & LILLEY, F.E.M.

TAMMEMAGI,

JUPP, D.L.B., KERR, D., LEMAIRE, H., MILTON, B.E., MOORE, R.F., NELSON, R., & VOZOFF, K. 1979. Joint

TAMMEMAGI,

Seafloor Magnetotelluric Exploration. Exploration Geophysics 16, 221-224. 1974. A magnetometer array study in southern Australia. Geophysical Journal of the Royal astronomical Society 36, 345-362. magnetotelluric-DC resistivity survey, eastern Officer Basin (Australia). Australian Society of Exploration Geophysicists, Bulletin 10, 209-212. LILLEY, F.E.M. 1982. Geomagnetic field fluctuations

over Australia in relation to magnetic surveys. Australian Society of Exploration Geophysicists, Bulletin 13, 68-76.

LILLEY, F.E.M., WOODS, D.V., & SLOANE, M . N . 1981a.

Electrical conductivity from Australian magnetometer arrays using spatial gradient data. Physics of the Earth and Planetary Interiors 25, 202-209. LILLEY, F.E.M., WOODS, D.V., & SLOANE, M . N . 1981b.

Electrical conductivity profiles and implications for the absence or presence of partial melting beneath central and southeast Australia. Physics of the Earth and Planetary Interiors 25,419-428.

H.Y., & LILLEY, F.E.M. 1971. Magnetotelluric studies across the Tasman Geosyncline, Australia. Geophysical Journal of the Royal astronomical Society 22, 505-516. H.Y., & LILLEY, F.E.M. 1973. A magnetotelluric traverse in southern Australia. Geophysical Journal of the Royal astronomical Society 31,433-445. WHITE, A., 1979. A sea floor magnetometer for the

continental shelf. Marine Geophysical Researches 4, 105-114.

WHITE, A., & MILLIGAN, P.R. 1984. A crustal conductor

on the Eyre Peninsula, South Australia. Nature 310, 219-222.

WOODS, D.V. 1979. Geomagnetic depth sounding studies

in Central Australia. Unpublished Ph.D. thesis, Australian National University, Canberra, 238 pp.

VOZOFF, K., KERR, D., MOORE, R.F., JUPP, D.L.B., & LEWIS, R.J.G. 1975. Murray basin magnetotelluric study.

Geological Society of Australia, Journal 22, 361-375.

LILLEY, F.E.M., MULHEARN, P.J., FILLOUX, J.H., BINDOFF, N.L., & FERGUSON, I.J. 1986. Pressure

VOZOFF, K. 1984. Regions of anomalously large telluric

MOORE, R.F., KERR, D.W., VOZOFF, K., & JUPP, D.L.B.

F.E.M. Lilley Research Scool of Earth Sciences Australian National University GPO Box 4 Canberra ACT 2601 AUSTRALIA.

fluctuations on the open ocean floor: mid-Tasman Sea at 38°30'S, 162°38'E near the Lord Howe Rise. Australian Journal of Marine and Freshwater Research 37, 27-37. 1977. Southern Cooper Basin magnetotelluric survey, South Australia, 1974. Bureau of Mineral Resources, Australia, Record 1977/41.

interference in Induced Polarisation: the Cobar Syndrome. Exploration Geophysics 15,175-178.

Geol. Soc. Aust. Spec. Publ. 17,147-156.

147

Geothermal Gradients In Australia J.P. CULL Department of Earth Sciences, Monash University, Clayton, Vic., Australia. Models of regional tectonism are constrained by the physical and chemical properties of the lithosphere. Most of the critical rheological properties are dependent on temperature. Consequently representative geothermal gradients must be constructed for detailed interpretations in any geological province. Geothermal gradients are usually calculated using analytical expressions describing equilibration by thermal conduction. Simple exponential distributions have been assumed for the radiogenic heat source with vertical equilibration in a uniform crust of constant thermal conductivity. Errors can now be assessed using numerical models incorporating the effects of macroscopic heterogeneity. Random models have been generated consisting of discrete blocks representing a complex geological section. Uniform physical properties have been assigned to each segment conforming on average to conventional estimates. Temperatures were calculated for each segment assuming constant boundary conditions. A range of ±20% in heatflowfor any province can generate variations of ±160 °C at depths of 50km; lateral equilibration is essential to accommodate the corresponding inflections in the geothermal gradient. Geothermal gradients calculated for Australia are consistent with recent xenolith data but estimates of temperature based on electrical sections are poorly resolved. Precambrian cratons can be distinguished from more recent blocks in terms of the homologous temperature (actual temperature / melting temperature). Areas of high heatfloware characterised by plastic flow at depths of 45 ± 15 km consistent with concepts of crustal underplating/intrusion in eastern Australia. Introduction Geothermal gradients and seismic soundings provide major constraints for models of regional tectonism. Structural units can be identified using seismic velocity profiles and bulk density can be assessed using empirical correlations. However phase diagrams must also be calculated from observations of surface heat flow to describe a corresponding mineral assemblage. Gross surface features can then be explained in terms of a passive rheological response to lithosphere dynamics (eg Lambeck 1984). Primitive concepts involving uniform layers with horizontal boundaries are no longer tenable for any advanced interpretation of regional geophysical data. Seismic profiles in Australia are now characterised by vertical gradations in velocity (eg Finlayson et al. 1983) and geological complexity is emphasised by Smithson (1978). In these circumstances any notion of a single "geotherm" must be dispelled. Several geothermal models have been constructed for Australia by Sass and Lachenbruch (1979). Essentially they were able to distinguish three groups representing the Archaean, the

Proterozoic, and the Cenozoic. However there was no attempt to resolve minor variations in any particular craton. Gross physical properties were described by analytical expressions approximating the effects of macroscopic heterogeneity. Errors in calculating temperature must now be assessed for each province prior to formulating more detailed tectonic models. Complex geological sections can be analysed using numerical approximations to describe thermal equilibration between uniform segments. Physical constants from within the range of conventional estimates can be assigned to each unit. However a comprehensive theory is required for extrapolation between segments at greater depth.

Heat Flow and Thermal Conductivity Temperatures close to the surface of the Earth are determined primarily by climatic conditions including diurnal and seasonal fluctuations in solar radiation. Consequently geothermal gradients are usually measured at depths exceeding 100m. Some data have been obtained in tunnels and mines but access is generally provided by deep drilling.

148

Geothermal Gradients in Australia.

Borehole temperatures are monitored using probes which incorporate a thermistor sensor and measurements are quoted to ±0.01 °C. However all temperatures are disturbed during drilling and groundwater migration can cause a consistent bias. Consequently some data may contain a scatter of ±20 °C (Cull & Conley 1983). In these circumstances gross errors will result from any attempt to extrapolate the near-surface geothermal gradient. Instead all calculations must be based on models consistent with surface heat flow. Heat flow Q (W m 2 ) is calculated from the expression Q = P *

(1)

where (3 (°C km' 1 ) is the geothermal gradient and X (W m^K" 1 ) is the thermal conductivity of the rocks in which the gradient is established. A linear gradient implies a constant thermal conductivity and a uniform stratigraphy. Uniform crustal sections are rare and complex temperature profiles are normally observed in regions of constant heat flow. Thermal conductivities have been measured in Australia primarily on core samples related to mineralised horizons. Cull and Conley (1983) have demonstrated a normal distribution for thermal conductivity with a mean value of 3.3 + 0.1 W

Fig. 1. Heat flow and geothermal domains in Australia; Archaean in the west, Proterozoic in central Australia, and Cenozoic in the east (Cull 1982; Sass & Lachenbruch 1979).

m^K" 1 (standard error of the mean). However data for softer materials contained in sedimentary basins are comparatively rare. Estimates can be provided for coal measures at depths less than 1000 m (Cull & Denham 1979, Hyndman 1967) but core from deeper oil wells are poorly studied. Consequently representative values must be assigned using other techniques. Models of basin formation involving progressive compaction of sediments have been suggested in studies of oil maturation (Angevine & Turcotte 1981). Compaction can be expected to generate a gradual increase in conductivity accompanied by a decrease in geothermal gradient. Representative values of thermal conductivity can be assigned if minor changes in lithology are ignored. The compaction model predicts a mid-range value of 2.0 W m^K" 1 for thermal conductivity of sediments at depths less than 1000m and 2.5 W m^K' 1 for greater depths (Cull & Conley 1983). The selection of representative values for thermal conductivity in the mantle becomes complicated by systematic trends in temperature and pressure. Even if a uniform mineralogy is assumed at any depth within the Earth the corresponding physical constants are normally determined from samples at the surface. MacDonald (1963) acknowledges that thermal conductivity varies inversely with temperature and consequently with depth. However he has emphasised the contribution from radiation; this mechanism is partly suppressed by optical absorption but in general it dominates the total conductivity in profiles of the mantle (Schatz & Simmons 1972). Thermal conductivities are further modified by variations in pressure with depth (eg. Cull 1975; Kieffer 1976) and correlations have been suggested linking conductivity and seismic velocity (Horai & Simmons 1969). Consequently complex profiles are required to describe the thermal conductivity and all estimates are subject to errors of at least ±20% (Fig. 2). A mean value can be selected at any depth for calculations of the geothermal gradient but systematic deviations must be included to investigate the consequences of macroscopic heterogeneity.

Geol. Soc. Aust. Spec. Publ. 17,147-156.

149

Fig. 2. Variations in thermal conductivity with depth. Dots indicate acceptable models with random error of + 20%. Solid line describes one model based on systematic trends related to temperature and pressure (Schatz & Simmons 1972; Cull 1975).

Heat Production Heat is generated within the Earth principally by the decay of radioactive trace elements contained in crustal rocks (Roy et al. 1968). There are additional sources at greater depth associated with global accretion, core segregation, tidal deformation, and core/mantle rotation (eg. MacDonald 1963). Equilibrium is maintained in a conductive regime by radiation to space detected as a constant heat flow at the surface. All heat flow data obtained in Australia have been compiled in a standard format rating the principal facts of each determination (Cull 1982). Three heat flow provinces have been defined (Fig. 1, Sass & Lachenbruch 1979). Each is considered to have a characteristic composition and tectonic history. Representative geotherms can be constructed for each on the assumption of an exponential decrease

in heat production as a function of depth (Fig. 3). Simple solutions conform to the expression

A(z) = A(0) exp(- j-)

where A(o) |iW m"3 is the heat production at the surface and D(m) is a measure of the depth extent of the radiation for any heat flow province. Exponential distributions are consistent with the model proposed by Lachenbruch (1970) to explain observed linear relationships between surface heat flow and surface heat production in basement rocks. This relationship is described by the expression Q(0) = q* + A(0) * D

(3)

150

Geothermal Gradients in Australia.

where Q(o) W m" is the surface heat flow and q is the residual heat flowing into the crust from mantle sources unrelated to radioelement distributions. 2

Simple analytical expressions describing vertical distributions in heat production fail to accommodate some regional geochemical models (eg. Hyndman et al 1968). These complications have been emphasised by Smithson & Decker (1974). Consequently Alliss (1979) has adopted an alternative approach; correlations are observed relating density, seismic velocity, and heat production. As a

result gradients in seismic velocity are assumed to be accompanied by gradients in heat production. Profiles of heat production for Australia can be based on the seismic sections presented by Finlayson et al. (1979). The results incorporating an uncertainty of 20% are illustrated in Fig. 4.

Temperature Profiles ^ Representative values can be selected for D and q in each heat flow province defined in Fig. 1. These together with observed values of A(o) determine the geothermal gradient. Sass and Lachenbruch (1979) assumed that these parameters were well determined and they were able to present a range of solutions for different values of A(o) in the composite expression T(z) = [q*z + D A(0)(1 - exp(-^)]/ X 2

Depth (km)

25/30

Fig. 3. Exponential distributions of radiogenic heat production (Lachenbruch 1970). Random error of + 20% on surface values of 5.0 W m" and characteristic depths of 10000 m. 3

Depth (km)

(4)

Errors were incorporated in the corresponding values of Q. The sensitivity of the exponential model can be demonstrated more generally by considering a typical range for each of the physical parameters in equation (4). In particular all models are subject to error in the boundary conditions Q and A(0). These values can be established by direct observation and

25/31

Fig. 4. Profiles of heat production based on correlations with seismic velocity (Alliss 1979; Finlayson & Others 1979).

Fig. 5. Geothermal gradients calculated for models with exponential distributions of heat production (Fig. 3) and constant thermal conductivity (3.0 W rn^K" ). 1

151

Geol. Soc. Aust. Spec. Publ. 17,147-156. estimates of temperature are readily calculated according to the expression T(z) = I[Q - A(z)] / X(z) dz

(5)

There is no need to specify q and a range of estimates can be incorporated in A(z) or D with summation by digital computer. Temperatures based on an exponential distribution for A with constant X are shown in Fig 5. Random selections for D (10000 ± 20%) generate

variations of ±50 °C (6.5%) at depths of 50 km. In contrast a similar range for Q generates an error of +160 °C (26%). Consequently separate geotherms must be constructed for anomalous values of Q in each heat flow province. Concepts of crustal evolution based on a regional geotherm for each province can only be retained as an approximation. Constant values are adopted for X in most geothermal models. However lateral equilibration is required along the boundaries of each province. These transition zones are complicated by minor changes in X consistent with Fig. 2. More generally any abrupt changes in mineralogy are accompanied by segmentations in the geothermal gradient. Each 20 18 -

h 20

r 40 Depth (km)

16 -

14 -

(b)

12 -

X o 10 0 a. 8 E

1 |

o o

6 4 5 2 "

4 c o " 3 CO

—i 1 1

o -

140

£_g 12 0 -1 1—i—i—i 1—i—i 1 1—i 1—i 1—i E

20 -

80 100 Depth (km)

120

140

18 -

Depth (km)

160

25/34

160

(c)

16 " 14 -

o o

12 -

X o

B/G 10 -

8 -

4 -

—i 1 1 1 1 1 1 1 1 1 1 60

80 100 Depth (km)

120

140

160 25/33

Fig. 6. Geothermal gradients for compound models witn vertical geological heterogeneity. Multiple inflections with Q = 90 mW m" generated by random error of +20% in A and X. Xenolith data presented as B/G geotherm (O'Reilly & Griffin 1985); resistivities indicate values in the range pl/p2 (Hermance & Grillot 1970; Shankland & Waff 1972). 2

80 100 Depth (km)

140

160

25/35

Fig. 7. Geothermal gradients and homologous temperatures ( Tc/Tm = 0.60 - 0.75 ) for (a) Proterozoic/Cenozoic, and (b) Archaean/Precambrian domains. Critical temperatures (Tc) indicated by hachured zone marking transition from brittle/plastic deformation (Vetter & Meissner 1979). No plastic deformation in Archaean lithosphere (Q <40 mW m~ ). 2

152

Geothermal Gradients in Australia.

segment contains an additional low level noise reflecting minor changes in lithology (or lattice structure). Corresponding estimates for A have been considered to be independant of A,but a similar segmentation can be anticipated according to the geological profile. Temperatures calculated for a compound model are illustrated in Fig. 6. There are numerous inflections in each gradient and the range of error is doubled to ±100 °C at depths of 50 km. However the general form of the exponential model is preserved and the levels are set primarily by variations in Q. The results are summarised in Fig. 7 and a second suite is presented representing conditions in the Precambrian. Similar values are used for X but estimates for Q and A are reduced by 50% (approx) consistent with the observations of Hyndman et al (1968).

External Constraints Each variant of the exponential trend causes cumulative error obscuring fine detail in calculations of temperature. None of the major parameters are well constrained and many solutions diverge at moderate depth. Consequently other independant data must be considered prior to selecting an appropriate combination. Petrological data can be used to verify the results of extrapolation at the surface. Garnet lherzolite nodules found in kimberlite are assumed to be derived from the upper mantle. The temperature and pressure at the depth of origin can be estimated for each nodule from the composition of coexisting mineral assemblages (Davis & Boyd, 1966). Consequently a paleogeotherm can be constructed independent of the more detailed conduction models. Data obtained from xenoliths in Australia have been summarised by Ferguson & others (1979) and O'Reilly & Griffin (1985). Their temperature/depth estimates are illustated in Fig. 6. Extreme near-surface gradients are required to satisfy individual xenolith suites. However the construction of a single geotherm represents a gross approximation. Variations in heat flow for different regions are accompanied by changes in the geothermal gradient. Consequently individual suites must

be segregated. Estimates for an isolated group considered by O'Reilly & Griffin (1985) can be approximated using conduction models based on a surface heat flow of 90 mW m"2 (B/G, Fig. 6). The B/G data were obtained from samples in Western Victoria. Geological mapping indicates anomalous volcanism with heat flow in the range 90-120 mW m"2 (Cull, 1982). These results confirm the validity of the xenolith technique. However Ferguson & others (1979) emphasise that nodules are generated in periods of anomalous tectonic activity characterised by major thermal transients. Consequently the xenolith data may represent maximum values at the depth of origin possibly reflecting mechanisms of crustal underplating/intrusion (Sass & Lachenbruch 1979; Cull 1990). Electrical soundings provide one alternative to seismic profiling for structural analysis. Galvanic methods (eg D C Schlumberger) are used routinely for mineral exploration and ground water mapping. However except for a few isolated expanded spreads (eg. Van Zijl 1978; Constable et al. 1984) the depth of penetration is less than 1000 m. Normally long period EM techniques are used for investigations of the crust and upper mantle. Magnetotelluric data can be analysed for estimates of resistivity at depths from 1-100 km and magnetometer arrays are used to extend the results to depths of 500 km (eg Gough 1983). Near surface resistivities are determined primarily by water content (Olhoeft 1981). However a semi-conductor mechanism becomes significant at depths exceeding 5 km. This mechanism is strongly dependent on temperature (eg. Garland 1981) and consequently variations in resistivity can be used to calculate a geothermal gradient. In general the resistivity data are presented as layered models after inversion; all physical properties must be interpolated from major discontinuities and few details can be retained. Furthermore Constable et al. (1984) have emphasised the practical problems associated with sampling in an area of complex geology. As a result any estimates of temperature must be treated with caution. Most regional surveys of electrical resistivity are dominated by lateral contacts (eg Cull 1985). Layered models are often inappropriate and es-

Geol Soc. Aust. Spec. Publ. 17,147-156.

153

timates of temperature are complicated by anisotropy. However one simple profile has been reported by Spence & Finlayson (1982) for Central Queensland. Magnetotelluric soundings reveal a sequence of conductive sediments on a uniform basement. Consequently apparent resistivities can be directly related to the activation energies described by Garland (1981) and these can be used as an indication of temperature. The results in Fig. 6 suggest a gradient consistent with the conduction models but there is an offset indicating high temperatures similar to those observed by Hermance & Grillot (1970) for Iceland.

Variations in the rheological response can be used to describe the nature of regional tectonic blocks (Oxburgh & Turcotte 1976). Stresses in each block are accommodated by brittle fracture or by plastic deformation. Melting is not essential for mobility but viscosity is strongly dependent on temperature. Vetter and Meissner (1979) demonstrate an abrupt transition for earthquake zones consistent with the concept of a rigid outer layer. They define a critical homologous temperature (T/Tm; temperature/melting point) in the range 0.6 - 0.65. However values of 0.75 may be required for complete decoupling (Oxburgh & Turcotte 1976) and a broad transition zone is indicated in Fig. 7.

Conclusions

Onuoha (1981) has calculated several viscosity profiles for Europe from estimates of the homologous temperature. However in view of the approximations involved some more general observations may be sufficient for Australia. In particular the Precambrian suite must be distinguished from the younger regimes in the east. In the Archaean (heat flow <40 m W m"2) the homologous temperature always remains less than the critical value (0.6). Elsewhere in the Precambrian the brittle zone extends to depths of 80 ± 30 km (Fig. 7). In contrast temperatures are sufficient in the east to ensure plastic deformation at depths of 45 ± 15 km. Any defect in crustal structure may then result in melt emplacement or intrusion consistent with the models of crustal underplating/intrusion proposed by Sass and Lachenbruch (1979).

Correlations between surface heat flow and heat production are normally used to define a uniform heat flow province. Representative geothermal gradients are then suggested independent of any identified anomalies in heat flow data. Minor changes in lithology can be included in any model as a second level noise confusing the systematic trends related to crustal evolution. However abrupt changes in mineralogy present a more serious problem. Several solutions can be generated for each province using random segmentation but the parameters adopted for any geothermal model can be further modified by gradients in temperature and pressure. Consequently there are major errors in any estimates of temperature and the distinctions for each province may become obscured by lateral equilibration. Geothermal anomalies are consistent with other regional geophysical data characterising the Australian crust. Correlations have been demonstrated by Cull & Denham (1979) linking heat flow and P-wave travel time residuals. This trend was confirmed in further studies demonstrating a functional dependence on P-wave velocity in the upper mantle. Variations of 0.4 km s"1 from west to east can be accommodated by conduction models indicating temperature differences of 600 °C at the Moho (assuming dV/dt = -0.0006 km s^K' 1 , Anderson et al 1968). Variations of this magnitude are attributed to erosion and depletion of the radiogenic elements in the Precambrian crust compared to younger distributions in the east.

References 1979. A heat production model for stable continental crust. Tectonophysics 57, 151-165. ALLISS R . G .

ANDERSON D . L . , SCHREIBER E , LIEBERMANN R . L . & SOGA N. 1968. Some elastic constant data on minerals relevant to geophysics. Reviews of Geophysics 6, 491-524.

C.L. & TURCOTTE D.L. 1981. Thermal subsidence and compaction in sedimentary basins: applications to Baltimore Canyon trough. American Association of Petroleum Geologists Bulletin 65, 219-225. ANGEVINE

154

Geothermal Gradients in Australia.

CONSTABLE S . C . , MCELHINNY, M . W . & MCFADDEN P.L.

HORAI K. & SIMMONS G. 1969. Thermal conductivity of

1984. Deep schlumberger sounding and the crustal resistivity structure of central Australia. Geophysical Journal of the Royal astronomical Society 79, 893-930.

Letters 6, 359-368.

CULL J.P. 1985. Magnetotelluric soundings over a Precambrian contact in Australia. Geophysical Journal of the Royal astronomical Society 80, 661-675.

rock forming minerals. Earth and Planetary

Science

HYNDMAN R.D. 1967. Heat flow in Queensland and Northern Territory Australia. Journal of Geophysical Research 72, 527-539. HYNDMAN R . D . , LAMBERT I.B., HEIER K . S . JAEGER, J . C .

CULL J.P. 1975. The pressure and temperature dependence of thermal conductivity within the Earth. PhD. thesis, Oxford University, unpublished. CULL J.P. 1990. Underplating of the crust and xenolith geotherm in Australia. Geophysical Research Letters (in press) CULL J.P. & CONLEY D. 1983. Geothermal gradients and

heat flow in Australian sedimentary basins. BMR Journal of Australian Geology & Geophysics 8, 329-337. CULL J.P. & DENHAM D. 1979. Regional variations in

Australian heat flow. BMR Journal Geology & Geophysics 4, 1-13.

Australian

DAVIS B.T.C. & BOYD F.R. 1966. The join MgSi04 CaMgSiC>6 at 30 kb pressure and its application to pyroxenes from kimberlite. Journal of Geophysical Research 71, 3567-3576.

& RING WOOD A.E. 1968. Heat flow and surface radioactivity measurements in the Precambrian Shield of Western Australia. Physics of the Earth and Planetary Interiors 1, 129-135. KIEFFER S.W. 1976. Lattice thermal conductivity within the Earth and considerations of a relationship between the pressure dependance of the thermal diffusivity and the volume dependance of the Gruneisen parameter. Journal of Geophysical Research 81, 3025-3029. LACHENBRUCH A.H. 1970. Crustal temperature and heat productivity: implications of the linear heat flow relation. Journal of Geophysical Research 75, 3291-3300. LAMBECK K. 1984. Structure and evolution of the Amadeus, Officer and Ngalia basins of central Australia. Australian Journal of Earth Sciences 31, 25-48. MACDONALD G.J.F.

1963. The

deep

structure

continents. Reviews of Geophysics 1, 587-665. FERGUSON J., ARCULUS R . J . & JOYCE J. 1 9 7 9 . K i m b e r l i t e

and kimberlitic intrusives of southeastern Australia: a review. BMR Journal of Australian Geology & Geophysics

4,227-241.

OLHOEFT G.R. 1981. Electrical properties of granite with implications for the lower crust. Journal of Geophysical Research 86, 931-936.

FINLAYSON D . M . , PRODEHL C . & COLLINS C . D . N . 1 9 7 9 .

Explosion seismic profiles and implications for crustal evolution in southeastern Australia. BMR Journal of Australian Geology & Geophysics 4, 243-252.

ONUOHA K.M. 1981. A comparison of the thermal and mechanical structure of the lithosphere beneath the Bohemian Massif and the Pannonian Basin. Geophysics 49, 212-216.

GARLAND G.D. 1981. The significance of terrestrial electrical conductivity variations. Annual Reviews of Earth and Planetary Science 9, 147-174. GOUGH D.I. 1983. Electromagnetic geophysics and global tectonics. Journal of Geophysical Research 88, 3367-3377.

O'REILLY S.Y. & GRIFFIN W . L . 1 9 8 5 . A x e n o l i t h - d e r i v e d

geotherm for southeastern Australia and its geophysical implications. Tectonophysics 111, 41-63. OXBURGH

E.R.

TURCOTTE

D.L.

physico-chemical behaviour of the lithosphere. Tectonophysics 32, 107-128.

1976.

The

descending

HERMANCE J.F. & GRILLOT L . R . 1970. C o r r e l a t i o n o f

magnetotelluric, seismic, and temperature data from southwest Iceland. Journal of Geophysical Research 75, 6582-6591.

ROY R.F., BLACKWELL D . D . & DECKER E . R . 1 9 6 8 . H e a t

generation of plutonic rocks and continental heat flow provinces. Earth and Planetary Science Letters 5, 1-12.

Geol. Soc. Aust. Spec. Publ. 17,147-156. SASS J.H. & LACHENBRUCH A.H. 1979. The thermal regime of the Australian continental crust. In M.W. McElhinny ed The Earth - Its Origin, Structure and Evolution. Academic Press. SCHATZ J.F. & SIMMONS G. 1972. Thermal conductivity of earth materials at high temperatures. Journal of Geophysical Research 77, 6966-6983. SHANKLAND T.T. & WAFF H.S. 1972. Partial melting and electrical conductivity anomalies in the upper mantle. Journal of Geophysical Research 82, 5409-5417. SMITHSON S.B. 1978. Modelling continental

crust: structural and chemical constraints. Geophysical Research Letters 5,749-752. SMITHSON S.B. & DECKER E.R. 1974. A continental crustal model and its geothermal implications. Earth and Planetary Science Letters 22, 215-225.

155

SPENCE A.G. & FINLAYSON D.M. 1983. The resistivity structure of the crust and upper mantle in the central Eromanga basin, Queensland, using magnetotelluric techniques. Geological Society ofAustralia, Journal 30, 1-16.

VAN ZUL J.S.V. 1978. The relationship between the deep electrical resistivity structure and tectonic provinces in southern Africa, I. Results obtained by Schlumberger soundings. Geological Society South Africa, Transactions 81, 129-142. VETTER U.R. & MEISSNER R.O. 1979. Rheological properties of the lithosphere and applications to passive continental margins. Tectonophysics 59, 367-380. J.P. Cull Department of Earth Sciences Monash University Clayton VIC 3168 AUSTRALIA

Geothermal Gradients in Australia.

Geol. Soc. Aust. Spec. Publ. 17,157-162.

157

Petrologic Constraints on Geophysical Modelling, Southeastern Australia Suzanne Y. O'Reilly 1 and W.L. Griffin 2 L

School of Earth Sciences, Macquarie University, Sydney, NSW, Australia

2 ' Division of Exploration Geoscience, CSIRO, North Ryde, NSW, Australia This study aims to show that petrologic data on unaltered deep-seated rock types, which occur as xenoliths in basaltic rocks, can provide constraints for geophysical modelling. The petrologic data are especially relevant to seismic, heat-flow and MAGSAT interpretations. Xenoliths derived from the lower crust and upper mantle occur in basaltic flows and cinder cones in eastern Australia. The host basaltic rocks range from Jurassic to Recent but are mainly Tertiary to Recent. These high pressure xenoliths belong to three broad suites: 1. Cr-diopside Iherzolites represent mantle wall-rock (see summaries in O'Reilly, 1984; O'Reilly and Griffin, 1987). These dominantly comprise the four-phase assemblage of olivine + clinopyroxene + orthopyroxene + spinel. Rare garnet-bearing Iherzolites are known from four localities (Wass & Irving, 1976; Day, 1983; Ferguson etaU 1977; Sutherland & Hollis, 1982). The spinel Iherzolites may contain volatile-bearing minerals such as amphibole, apatite, mica and calcite in varying proportions (e.g. Griffin et al., 1984; Andersen et al., 1984). 2. Al-augite suite xenoliths originate as frozen basaltic liquids or comulates. These may be subdivided into two further groups: (i)

The wehrlite suite shows dominantly igneous microstructures. The constituent rock types contain varying modes of clinopyroxene, olivine, spinel, ilmenite, orthopyroxene, amphibole, mica, feldspar, and sulphides. Some amphibole- and apatite-rich xenoliths also belong to this suite and probably have affinities with the MARID suite xenoliths from

kimberlites (Wass, 1979; Dawson & Smith, 1977; O'Reilly, 1984). (ii) Spinel ± garnet meta-pyroxenites appear to be extensively recrystallized rocks formed by reequilibration of original wehrlite suite pyroxenites. The microstructures of these pyroxenites ranges from lamellar (providing evidence of exsolution origin) to granoblastic mosaic (Griffin et al., 1984). 3. Granulite xenoliths are dominantly mafic and probably represent lower crustal material, although rare examples may derive from the upper mantle (Wass & Hollis, 1984). Mineral assemblages are commonly combinations of plagioclase ± clinopyroxene ± spinel ± garnet ± olivine ± orthopyroxene. Rare garnet-spinel websterites may also belong to this suite. Full reviews of granulite suite xenoliths from eastern Australia are given in Griffin & O'Reilly (1986) and Griffin & O'Reilly (1987a). Petrologic information including geothermobarmetry calculations and recognition of composite xenoliths with lithological contacts, has allowed the interpretation of a probable deep crust/upper mantle stratigraphy for eastern Australia (O'Reilly & Griffin, 1985). This comprises a lower crust dominated by mafic granulites (Griffin & O'Reilly, 1987b), then a crust-mantle transition zone (starting at about 25 km) consisting of spinel lherzolite wall-rock interleaved with basaltic cumulates, granulites and spinel ± garnet pyroxenites. The proportion of spinel lherzolite increases gradationally with depth until about 55 km where it is the main rock type. Multiple intrusion of basaltic magmas near the crust/mantle interface over a long time span is the probable mechanism for producing this lithological profile.

158

Petrologic constraints on geophysical models.

25 • B / G , THIS WORK * VICTORIA, N.S.W. (Irving) • DELEGATE • JUGIONG

DELEGATE EXPT.

GT. LHERZOLITE .a ?? '5 CL-

SR LHERZOLITE

GT+ PX+PL 2PX + PL

2 P X + SP OL + PL

600

800

1000

TEMPERATURE :°C

1200

1400

Fig. 1. The constructed geotherm for southeastern Australia based mainly on P-T data from garnet websterites from Bullenmeri and Gnotuk (Griffin et al., 1984). All other data sources are given in that reference. The stippled area is based on T or T+P estimated from data for 2-pyroxene + garnet granulites from Delegate (Lovering & White, 1969) and Gloucester (Wilkinson, 1974). The steady-state geotherm for the B/G region (1.8 hfu) is from Sass & Lachenbruch (1979). Reproduced from Kimberlite Origin and Occurrence (Glover & Harris, eds.) with permission of the Geology Department and University Extension, the University of Western Australia.

159

Geol. Soc. Aust. Spec. Publ. 17,157-162.

Velocity ( km s H )

Velocity ( k m s"1)

Q_ CD Q

Fig. 2. Seismic compressional velocity/depth models for (a) the Pilbara Craton and (b) the Lachlan Fold Belt of southeastern Australia. Seismic velocities have been adjusted to 25°C and IGPa. Taken from Drummond (1982). Reproduced from Kimberlite Origin and Occurrence (Glover & Harris, eds.) with permission of the Geology Department and University Extension, the University of Western Australia. Geothermobarometry calculations on a suite of well-equilibrated garnet websterite xenoliths from Bullenmerri and Gnotuk (B/G) Maars (Victoria) have allowed construction of an empirical geotherm for that mantle/crust column at the time of eruption (Griffin etal., 1984; O'Reilly & Griffin, 1985). The resulting curve suggests a high geothermal gradient (Fig. 1) lying well above a steady-state model geotherm based on high surface heat flow data from that region, and resulting in high T at shallow depths. The B/G geotherm suggests that advective heat transfer (probably due to volcanism) was the dominant heat input. The xenolith-derived geotherm is empirically derived and is not modeldependent as are geotherms constructed from surface heat flow data.

The high geothermal gradient and rock-type distributions have important implications for the interpretation of seismic refraction and reflection profiles. The elevated temperatures significantly decrease the seismic velocities for given rock types so that the seismically-defined Moho (Vp 8.0km/s) would be interpreted to lie at a deeper level than the "petrologic" moho change from mafic granulites to dominant spinel (or garnet) lherzolite. In addition, the inferred intercalation of mafic lenses in spinel lherzolite over a transition zone from about 25 to 55 km, correlates with an observed gradational increase in Vp over that depth range for seismic traverses in that region (Finlayson et al., 1979). Vp's have been calculated for specific xenolith rock types using measured densities, compositions and calculated T's. Averaged Vp's for a model lower crust/upper mantle section agree well with

160

Petrologic constraints on geophysical models.

values of Vp inferred from seismic refraction profiles and show an analogous gradational increase with depth (O'Reilly & Griffin, 1985). It is suggested that in this region of relatively young and recently active crust, the seismically-defined moho may represent the spinel lherzolite-garnet lherzolite transition at about 65 km. This contrasts with crustal seismic profiles in cratonic western Australia (Drummond, 1982) where the seismic moho is defined by a distinct Vp increment showing none of the transitional features from about 25 to 55 km characteristic of the eastern Australian seismic profiles (Fig. 2). This probably reflects differences both in the lower crust/upper mantle stratigraphies and in the geothermal gradients of these contrasting tectonic regimes. Seismic reflection studies in eastern Australia (e.g. Mathur, 1983a, b, c) suggest the lower crust/upper mantle region is strongly layered and are consistent with the inferred stratigraphy in this technically active regime. The xenolith geotherm is consistent with the high regional heat flow measured in western Victoria. According to this geotherm, the Curie-point for magnetite is reached at a depth of about 12 km. This could explain the magnetic lows evident from MAGSAT data for southeastern Australia. The empirical geotherm was constructed using data from one restricted crust/mantle column only. However, data from similar rocks from other localities throughout Tasmania, Victoria, New South Wales and Queensland (O'Reilly & Griffin, 1985) also lie on the geotherm. This suggests that the geotherm reflects an equilibrium thermal state reached beneath volcanic provinces in eastern Australia at the time of the magmatic activity. The opening of the Tasman Sea approximately 80 Ma ago coincided to a large extent with tectonic uplift and the onset of the major Tertiary volcanism along the rifted coastline. The empirical geotherm appears to reflect the thermal effect of these events in contrast to the inferred steady-state geotherm of the western cratonic region of the Australian continent.

REFERENCES ANDERSEN, T., O'REILLY, S.Y. & GRIFFIN, W . L . 1 9 8 4 .

The trapped fluid phase in upper mantle xenoliths from Victoria, Australia: implications for mantle metasomatism. Contributions to Mineralogy and Petrology 88, 72-85. DAWSON, J . B .

SMITH, J.V.

1977.

The

MARID

(mica-amphibole-rutile-ilmenite-diopside) suite of xenoliths in kimberlite. Geochimica CosmochimicaActa 41, 309-323. DAY, R.A. 1983. Petrology and geochemistry of the Older Volcanics, Victoria. Unpub. PhD Thesis, Monash Univ., Melbourne, Australia. DRUMMOND, B.J. 1982. Seismic constraints of the chemical composition of the crust of the Pilbara Craton, northwest Australia. Revista Brasileira Geosiencias 12, 113-120.

FERGUSON, J., ELLIS, D . J . & ENGLAND, R . N .

1977.

Unique spinel-garnet lherzolite inclusion in kimberlite from Australia. Geology 5, 278-280. FINLAYSON, D . M . , PRODEHL, C . & COLLINS, C . D . N .

1979. Explosion seismic profiles, and implications for crustal evolution, in southeastern Australia. BMR Journal of Australian Geology & Geophysics 4, 243-252. GRIFFIN, W . L . WASS,

S.Y.,

HOLLIS, J . D .

1984.

Ultramafic xenoliths from Bullenmerri and Gnotuk maars, Victoria, Australia, petrology of a sub-continental crust-mantle transition. Journal of Petrology 25, 53-87. GRIFFIN, W.L. & O'REILLY, S.Y. 1986. The lower crust in eastern Australia, xenolith evidence. In Dawson, J.B., Carswell, D.A., Hall, J. & Wedepohl, K.H. eds. Nature of the Lower Crust, Geological Society of London Special Publication 24, 263-274 GRIFFIN, W.L. & O'REILLY, S.Y. 1987a. The composition

of the lower crust and the nature of the continental Moho. In Nixon, P.H. ed., Mantle Xenoliths, Wiley, London. GRIFFIN, W . L . & O'REILLY, S.Y. 1 9 8 7 b . Is the c o n t i n e n t a l

Moho the crust/mantle boundary? Geology 15, 241-244. LOVERING, J.F. & WHITE, A . J . R . 1 9 6 9 . G r a n u l i t i c a n d

eclogitic inclusions from basic pipes at Delegate, Australia. Contributions to Minerology and Petrology 21, 9 - 5 2 .

Geol. Soc. Aust. Spec. Publ 17,157-162.

MATHUR, S.P. 1983a. Deep reflection experiments in northeastern Australia, 1976-1978. Geophysics 48, 1588-1597. MATHUR, S.P. 1983b. Deep reflection probes in eastern Australia reveal differences in the nature of the crust. First Break, July, 9-16. MATHUR, S.P. 1983c. Deep crustal reflection results from the central Eromanga Basin, Australia. Tectonophysics 100, 163-173. O'REILLY, S.Y. 1984. The mantle environment. In Glover, J.E. & Harris, P.G. eds., Kimberlite Occurrence and Origin, Geology Dept and University Extension, University of Western Australia, Publication, 8, 63-102. O'REILLY,

S.Y.

GRIFFIN,

W.L.

1985.

161

WASS, S.Y. 1979a. Fractional crystallisation in the mantle of late-stage kimberlitic liquids - evidence in xenoliths from the Kiama area, N.S.W., Australia. In Boyd, F.R., & Meyer, H.O.A. eds., The Mantle Sample: Inclusions in Kimberlites and Other Volcanics. Proceedings of the 2nd International Kimberlite Conference, 2. American Geophysical Union, Washington, 366-373. WASS, S.Y. & HOLLIS, J.D., 1983. C r u s t a l g r o w t h in

southeastern Australia -evidence from lower crustal eclogitic and granulitic xenoliths. Journal of Metamorphic Geology 1, 25-45. WASS, S.Y. & IRVING, A.J., 1976. XENMEG: A Catalogue of Occurrences of Xenoliths and Megacrysts in Volcanic Rocks of Eastern Australia. Australian Museum, Sydney. 441pp

xenolith-derived geotherm for southeastern Australia and its geophysical implications. Tectonophysics, 111, 41-64. O'REILLY, S.Y. & GRIFFIN, W . L . 1987. E a s t e r n A u s t r a l i a :

WILKINSON, J.F.G. 1974. Garret clinopyroxenite inclusions from diatremes in the Gloucester area, New South Wales, Australia. Contributions to Minerology and Petrology, 46, 275-299.

4,000 km of mantle samples. In Nixon, P.H. ed., Mantle Xenoliths, Wiley, London. 267-280 SASS, J.H. & LACHENBRUCH, A.H. 1979. Thermal regime of the Australian continental crust. In McElhinney, M. W. ed., The Earth, It's Origin, Structure and Evolution, Academic Press, London, 301-352. SUTHERLAND, E L . & HOLLIS, J . D . 1982. M a n t l e - l o w e r

crust petrology from inclusions in basaltic rocks in eastern Australia - an outline. Journal ofVolcanological and Geothermal Research 14, 1-29.

Suzanne Y. O'Reilly School of Earth Sciences Macquarie University Sydney, NSW, 2109 AUSTRALIA W.L. Griffin Division of Exploration Geoscience,CSIRO North Ryde, NSW, 2113 AUSTRALIA

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

163

The Chemical and Isotopic Composition of the Lower Eastern Australian Lithosphere: A Review W.F. McDonough, R.L. Rudnick and M.T. McCulloch

Research School of Earth Sciences, Australian National Univ., Canberra, A.C.T., Australia Petrological, chemical and isotopic data for granulite facies xenoliths entrained in Mesozoic and Cenozoic basalts from eastern Australia indicate that the lower crust is generally mafic, though intermediate to felsic compositions are observed and may represent common lithologies in restricted zones. The majority of the mafic xenoliths are cumulates from mafic magmas and a few may be restite material left after the extraction of granitic melts from mafic-intermediate precursors; others represent crystallized basaltic melts. Isotopic data suggest that crustal underplating by mafic magmas occurred during the Palaeozoic and Cenozoic. Seismic data for eastern Australia indicate that the depth to the Moho is variable (35-55 km) and that in some regions the Moho is transitional. Similarly, integrated petrological, geochemical and isotopic data for spinel-bearing lherzolite and harzburgite xenoliths allow development of an internally consistent, multistage evolutionary model for the stabilisation and growth of the eastern Australian lithospheric mantle. The bulk of this region consists of refractory peridotite which has experienced multiple episodes of melt extraction and incompatible element enrichment, producing significant chemical and isotopic heterogeneities. Growth of the eastern Australia lithospheric mantle is believed to have occurred by the accretion of intrinsically buoyant, refractory peridotite diapirs. Later tectono-magmatic reactivation episodes are recorded in the xenoliths, these events occurred in an intraplate and/or divergent plate tectonic environment. The presently available Sr and Nd isotopic data for upper mantle xenoliths from eastern Australia do not provide definitive constraints on the age of lithosphere formation.

Introduction Petrological, chemical and isotopic studies of upper crustal rocks have led to a self consistent model for the bulk composition and growth of the upper continental crust through time (McCulloch and Wasserburg, 1978; Taylor and McLennan, 1985). In contrast, the lower crust and subadjacent lithospheric mantle are isotopically, geochemically and geophysically less well known. Samples of deep crust are limited to xenolith fragments entrained in basalts and kimberlites, and rocks of tectonically emplaced high-grade metamorphic terrains. In eastern Australia numerous Cenozoic alkali basalts contain a variety of granulitic, pyroxenitic and peridotitic xenoliths which provide crucial information on the composition and origin of the lower crust and lithospheric mantle. In addition, the basalts themselves can provide information about the averaged regional composition of the lowermost lithospheric mantle (Chen and Frey, 1985; McDonough et al., 1985). Lower crustal xenoliths throughout the world are characterised by dominantly mafic lithologies (Kay and Kay, 1981), although felsic and

metasedimentary lithologies are locally abundant (Padovani and Carter, 1977; Leyreloup et al., 1977). This is in marked contrast to the typical lithologies of granulite facies terrains, which tend to be dominated by intermediate to felsic rocks (Kay and Kay, 1981). Whereas granulite xenoliths are demonstrably fragments of present-day lower crust (as evidenced by their high pressure mineralogies and decompression features), the relationship of granulite terrains to the lower crust is uncertain. If granulite terrains form primarily in continental collision zones (Newton and Perkins, 1982), they simply represent supracrustal rocks which have been transiently buried to great depths. Therefore, the study of granulite facies xenoliths may provide the most important constraints on lower crustal composition. Studies of upper mantle xenolith suites worldwide show that the subcontinental lithospheric mantle is composed of ultramafic rocks, dominated by lherzolites and harzburgites, although other rock types, mainly pyroxene-rich and of cumulate origin, are found (Frey and Green, 1974; Wilshire and Shervais, 1975; Frey and Prinz, 1978; Irving, 1980; Nixon et al., 1981). Direct

164

Chemical and isotopic composition.

sampling of the subcontinental lithospheric mantle is usually restricted to xenoliths in alkali basalts and kimberlites; the only other source of this material being alpine peridotites which are dominantly composed of lherzolite and harzburgite with lesser amounts of dunite and pyroxenite (Frey, 1984). These bodies are more probably fragments of the suboceanic or continental margin lithospheric mantle. Studies on eastern Australian ultramafic xenolith suites have been important in establishing models on the growth and composition of the subcontinental lithospheric mantle (eg., Frey and Green, 1974; Irving, 1980; Nickel and Green, 1984). Through systematic studies combining petrographic, chemical and isotopic data on these fragments of the eastern Australian deep lithosphere, we can develop models for the composition, structure and growth history of this region. This contribution examines the available data on lower crustal and upper mantle xenoliths in eastern Australia and their host basalts and develops a framework for understanding this region. New data are presented from several ongoing studies of lower crustal and upper mantle xenoliths and their host basalts. The Lower Crust Xenoliths with granulite facies mineralogies occur in a variety of young basaltic scoria cones, maars and tuff rings and older kimberlitic diatremes that intrude rocks of variable crustal ages throughout eastern Australia (Table 1; see summaries by Ferguson et al., 1979; Sutherland and Hollis, 1982; Griffin and O'Reilly, 1986). Thermobarometry for coexisting phases in these xenoliths yield high pressures and temperatures (in the range of 0.7 to 1.2 GPa and 600-1100°C, see Table 1 for references), which, coupled with the commonly observed decompression features, suggests a deep crustal origin for these rocks. Most of the xenolith-bearing basaltic and kimberlitic occurrences were emplaced through rocks of the Tasman Fold Belt, which have Palaeozoic depositional and crystallisation ages, but Proterozoic mantle extraction ages (McCulloch and Chappell, 1982; Hensel et al., 1985). Several xenolith localities lie in regions of older crust; the Calcutteroo and Kayrunnera kimberlitic pipes in South Australia and New

South Wales, respectively, were emplaced through Proterozoic sediments; and xenolith-bearing basalts from the McBride province, N. Queensland, were erupted through metamorphic rocks of the Proterozoic Georgetown Inlier. Lithologies of the granulite xenoliths are predominantly mafic, but intermediate to felsic granulite xenoliths have been observed at Delegate, N.S.W. (Lovering and White, 1969), Calcutteroo and White Cliffs, S.A. (Ferguson et al., 1979; McCulloch etal., 1982; Arculus etal., 1987), McBride province, north Queensland (Rudnick and Taylor, 1991) and the Newer volcanic province (Rudnick, unpubl.). Felsic xenoliths are abundant only in the case of the Newer volcanic province, where they are the only crustal xenolith observed in several vents (eg., Mt. Leura, Mt. Noorat). The felsic xenoliths generally show more severe decompression effects (for example, sillimanite and garnetbearing felsic xenoliths from the Newer volcanics have pumiceous textures due to melting) and consequently, they have generally not been included in geochemical studies. Table 1 summarises published petrologic studies on lower crustal xenoliths from eastern Australia and the following paragraphs summarise the chemical and isotopic data available on these suites. The major conclusion reached by more recent petrologic studies are that the eastern Australian lower crust is mafic, and interleaving of this material with mantle lherzolite creates a transitional Moho, which extends over approximately 30 km depth range (Ferguson et al., 1979; Wass and Hollis, 1983; O'Reilly and Griffin, 1985). Debate remains as to the depth at which spinel lherzolites become dominant lithologies (see below). A mafic lower crust may be the result of two genetically distinct processes: (1) the crystallisation of, and/or crystal accumulation from mafic magmas intruded into the deep crust, or (2) removal of partial melts from mafic to intermediate rocks in the deep crust, leaving a mafic residue. In the absence of obvious igneous textures, distinguishing between these two origins for mafic xenoliths is only possible by detailed geochemical and petrologic studies, and to date few of these have been completed.

165

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

Locality

Country Rocks

Age of Pipe

Xenolith Lithologies

T-P Conditions

Reference

SOUTH A U S T R A L I A Culcutteroo

ProL Marine Sediments

164-175 Ma

GG, 2 PxG KyGar-Ksp G

600-900" C 1.1-1.6 GPa

McCulloch et al., 1982 Arculus et al., 1987

El Alamein

ProL Marine Sediments

164-175 Ma

2 PxG, 2 PxGG

700-870'C

Arculus et al., 1987

Anakies

Cenozoic Basalts, Palaeozoic Seds.

Quaternary

2 PxG, 2 PxGG

880-980'C

Wass & Hollis, 1983

Mt. Leura Mt. Noorat Mt. Shadwell

Cenozoic Basalts Palaeozoic Seds.

Quaternary

Felsic Sil-GG

NEW SOUTH

WALES

White Cliffs

Precambrian Chlorite Schists

260 Ma

630"C 1.4 GPa

Arculus et al., 1987

Kayrunnera

Precambrian Chorite Schists

260 Ma

850-900'C 1.8-2.8 GPa

Edwards et al., 1987

Jugiong

Basalts

< 17 Ma

2 PxG, 2 PxGG

850-960"C 1.6 GPa

Arculus et al., 1987

Palaeozoic Sediments and Granites

170 Ma

2 PxG, 2 PxGG

850-950'C 0.6-1.8 GPa

Lovering and White, 1969 Irving. 1976 Arculus et al. 1987 Griffin & O'Reilly, 1986

Gloucester

Carboniferous Sediments

1000'C 1.0-1.4 GPa

Wilkinson, 1974 Griffin & O'Reilly, 1986

Boomi Creek

Carboniferous Sediments

2 PxG

950'C 0.8-1.0 GPa

Wilkinson, 1975 Wilkinson & Taylor, 1980

Walcha

Tertiary Gravels

Teriary

2 PxGG

1000'C 1.4 GPa

McBride Province

Pro tero zoic Metamorphics

< 3 Ma

2 PxG, GG 2 Px GG, Charn Metasediments

700-1000'C 0.8-1.0 GPa

Kay & Kay, 1983 Rudnick & Taylor, this vol Rudnick & Williams, 1987 Rudnick & Taylor, 1987

Chudleigh Province

Palaeozoic Sediments and Granites

< 2 Ma

2 PxG, 2 PxGG GG

700-1000'C 0.6-1.2 GPa

Kay & Kay, 1983 Rudnick et al., 1986a Rudnick & Taylor, this vol.

VICTORIA

Delegate

GG, Charn

This paper

Stolz, 1984

QUEENSLAND

Where 2 PxG = two pyroxene granulite; 2 PxGG = two pyroxene garnet granulite, GG = garnet granulite, Charn = Charnockite. Ky = kyanite; Ksp = K feldspar, Gar = garnet. Table 1. Summary of Studies on Lower Crustal Xenoliths from Eastern Australia

166

Chemical and isotopic composition.

Wilkinson (1975) and Wilkinson and Taylor (1980) reported on the petrology and composition of a suite of ultramafic to mafic granulite xenoliths from Boomi Creek, New South Wales (NSW). Additionally, Stolz (1984) reported similar information on comparable xenoliths from Walcha, NSW. These studies showed coherent geochemical correlations for both compatible and incompatible trace elements in each xenolith suite, and concluded that the xenolith suites represent cogenetic cumulates which crystallised from tholeiitic melts in the deep crust; no cumulate textures were preserved. Arculus etal. (1987) summarise petrologic and geochemical data for a variety of mafic granulite and garnet pyroxenite xenoliths from 7 widespread localities in South Australia, Victoria and New South Wales. They note that most of the xenoliths have basaltic compositions, and display a wide range in silica saturation from nepheline-normative to quartz-normative. Trace element compositions of the lower crustal xenoliths are highly variable; some have apparent basaltic characteristics whereas others show enrichments in certain trace elements (eg., Ba, Sr, Pb and Nb), which are not attributable to igneous processes such as crystal accumulation alone. Arculus et al. (1987) conclude that these xenoliths represent basaltic melts, some locally enriched in cumulus phases or contaminated by interactions with pre-existing crust, that were later metamorphosed at higher pressures than the original fractionation events. Additional processes, as yet uncharacterised, are needed to explain the peculiar enrichments of alkaline earth elements. In contrast with the highly variable major and trace element compositions of the xenoliths studied by Arculus et al., mafic lower crustal xenoliths from the Chudleigh volcanic province, north Queensland show coherent chemical and isotopic correlations (Rudnick et al, 1986). These correlations allow genetic interpretations to be made of the xenolith chemistries. The xenoliths represent a co-genetic suite of cumulates from basaltic magma(s) which progressively assimilated isotopically evolved country rock as fractionation proceeded. In addition, the good correlation between isotope ratios and major and trace element concentrations degrades as the isotopic ratios are

back-calculated to earlier times, suggesting the xenoliths are less than 100 Ma old, and are probably related to the Tertiary volcanic activity. The mineral assemblages and corona textures of the Chudleigh province xenoliths were interpreted to reflect a relatively simple P-T history: intrusion and fractional crystallisation of the basaltic magma between 20 to 40 km depth with subsequent isobaric cooling (Rudnick and Taylor, 1991). Therefore, this section of the lower crust seems to be dominated by young basaltic cumulates. The proximity of the Chudleigh volcanic vents to a major fault zone (Burdekin fault zone) suggests that the unusual abundance and freshness of xenoliths from these vents may be the result of basaltic intrusion at variable depths along the fault zone. The geochemistry of lower crustal xenoliths from Hill 32 in the McBride province of north Queensland points to diverse origins for these rocks (Rudnick and Taylor, 1987). The majority of the xenoliths are mafic, but they formed by a variety of processes including crystallisation of mafic magmas, ciystal accumulation from mafic and felsic magmas and partial melt extraction from intermediate source rocks. Intermediate and felsic granulite facies xenoliths make up about 20% of the observed rock types and come from sedimentary and felsic igneous protoliths. Ion microprobe U-Pb zircon ages show that all these xenoliths experienced high-grade metamorphism in the late Palaeozoic, which coincides with the eruption of extensive felsic ash flows and emplacement of high-level felsic intrusives. Most of the xenoliths formed during this orogeny, although several originally crystallised in the Proterozoic (Rudnick and Williams, 1987). It is noteworthy that the two zircon-bearing mafic xenoliths with melt-like compositions appear to have formed during the Palaeozoic granite-forming events. They may thus represent mantle-derived heat sources which may have caused the crustal melting. Figure 1 shows the available (early 1987) Nd and Sr isotopic compositions of eastern Australian lower crustal granulite xenoliths. Several garnet pyroxenite xenoliths (eg., cpx-gt and no plagioclase) are also included, although they may represent upper mantle material. The two intermediate to felsic granulites plot far into the lower

167

Geol. Soc. Aust. Spec. Publ. 17,163-188.

20 • • • •

Chudleigh Delegate Calcutterroo McBride

-10 0.850 -20

0.700

0.705

0.710 87

Sr/

0.715 86

0.720

0.725

Fig. 1. 87Sr/86Sr versus £Nd values of lower crustal xenoliths from eastern Australia. Data sources are from McCulloch et al (1982), Rudnick et al (1986) and this paper (Table 2). eNd are normalized to 143Nd/144Nd = 0.511836 (CHUR). Samples from the Delegate breccia pipe are plotted at 170 Ma (the time of eruption). Open symbols are for garnet pyroxenites, filled symbols are for mafic granulites and half-filled symbols are for felsic to intermediategranulites.Open field encompasses Chudleigh xenoliths and the patterned field encompasses Delegate xenoliths.

right quadrant, reflecting long term Rb and LREE enrichments and their supracrustal origin. The mafic granulite xenoliths fall along, and to the right of, the present-day mantle array. For the Chudleigh province mafic granulites the divergence to the right of the mantle array can be explained by assimilation of crustal materials with high and negative £Nd (Rudnick et al., 1986). The isotopic composition of mafic granulites from the Delegate breccia pipe, N.S.W. are given in Table 2, along with the composition of the host nephelinite. These samples have LREE-enriched patterns and plot along an extension of the mantle array, albeit many at negative £Nd values; they exhibit a relatively large range of £Nd values (£Nd at 170 Ma, the age of the Delegate pipe [Compston and Lovering, 1969], ranges from +13 to -8), but limited range of

87 Sr/ 86 Sr. It has been postulated that these xenoliths formed as basaltic melts which intruded the base of the crust and contain variable proportions of cumulate phases (Arculus et al1987). Therefore, the spread of these data on the isotope diagram (Fig. 1) may reflect their derivation from isotopically distinct mantle source regions, variable residence times within the deep crust and/or mixing with a crustal component. The garnet pyroxenite with the very high £Nd value is strongly LREE-depleted ( 1 4 7 Sm/ 1 4 4 Nd = 0.337), so that if this sample originally plotted along the mantle array at the time of its ciystallisation, it has a minimum age of 395 Ma. Interestingly, this model age is similar to the crystallisation age of granites in this region of the Tasman Fold Belt (Williams et al1983), suggesting that igneous activity which

Rb (ppm)

Sr (ppm)

87Rb

86Sr

R 877 Host

15.2

1043

0.0742

0.70415

-8

9.8

50.4

0.1182

0.511948±14

+2.2

+3.9

69-27 GPx

2.53

226

0.0322

0.70563

+15

1.0

1.9

0.3367

0.512639±24

+15.7

+ 12.6

R 698 2PxG

10.0

367

0.0788

0.70702

+32

4.2

13.7

0.1866

0.511447±14

- 7.6

-7.3

R 130 GG

25.6

798

0.0928

0.70561

+12

5.6

20.9

0.1614

0.511909124

+1.4

+2.2

Sr Sr

£Sr(170Ma)

,47 Sm l*>Nd

"3Nd l*»Nd

£Nd(°M»)

E[s( ( j(170Ma)

Delegate xenoliths and host nephelinite (170 Ma)

R 960 GPx

4.30

165

0.0420

0.70566

+14

0.9

4.2

0.1297

0.511781±28

-1.1

+0.4

R 967 GG

11.4

1008

0.0329

0.70584

+17

1.1

3.7

0.1700

0.511566114

-5.3

-4.7

Lake Bullenmerri mineral separates from garnet pyroxenites (Q uaternary) 84-99 Garnet

0.06

2.0

0.0956

0.70687

0.5

0.9

0.3653

0.511549144

-5.6

84-99 Cpx

0.20

181

0.0029

0.70654

—

1.9

12.9

0.0871

0.511607116

-4.5

84-98 Cpx

0.14

220

0.0019

0.70648

—

3.5

20.8

0.1005

0.511501126

-6.5

—

niiOS t T 11 1116 2 a u n c c r U i n t 87 mCItaM4CS in ^ f ^ 0 0 1i V n d f O T N d y ,cvcl S r / 8 6 S r ratios were normalized to 8 8 S r / 8 6 S r = 8 37520 and Nd/ 1 Nd ratios were normalized to 1 4 & N d / 1 4 2 N d - 0.636151. Sample dissolution methods are described in McDonough and McColloch (1987) and mass spectrometry techniques are described in McDonough et al. (1985). Sample descriptions for the Delegate samples are given in Irving (1974); the garnet pyroxene xenoliths from Lake Bullenmerri are similar to those described by Griffin et al. (1984). GPx - garnet pyroxenite, 2PxG = 2 pyroxene eranulite. G G = garnet granultite. Host = nephelinite host rock. Epsilon values were calculated using the following values: 8 7 S r / 8 6 S r - 0.70475, 8 7 R b / 8 6 S r = 0.0876, 1 4 3 N d / 1 4 4 N d - 0.511836 1 4 7 S m / 1 4 4 N d = 0 1967 Decay constants used are 8 7 R b « 1.42 x 10" 1 1 and 1 4 7 S m = 6.54 x 1 0 1 2 -

Table 2. Isotopic compositions of mafic and pyroxenitic xenoliths from Southeastern Australia

Geol. Soc. Aust. Spec. Publ. 17,163-188.

formed the garnet pyroxenite may have provided a heat source for crustal melting. The Calcutteroo mafic granulites generally fall in a scattered field to the right of the mantle array, with about half of the samples plotting in the upper right quadrant. The REE patterns of these xenoliths are quite variable, only one or two show geochemical characteristics of cumulates (McCulloch etal., 1982) and their low Rb contents are unlikely to reflect the original concentrations. If the isotopic ratios of these xenoliths are back-calculated to earlier times, the data still fall far to the right of the mantle array, suggesting that these samples originally had higher Rb/Sr ratios and that they have experienced a longterm Rb depletion. In summary, the highly variable Sr and Nd isotopic compositions of lower crustal xenoliths from eastern Australia reflect the source regions of the rocks (i.e., mantle-derived melts and/or cumulates versus supracrustal rock types), and the processes which affected them during and after their original crystallisation. Some xenoliths have relatively primitive isotopic compositions and appear to represent mantle-derived melts or cumulates, whereas others, with more evolved isotopic compositions, represent mixtures of mantlederived melts and supracrustal material. Still other xenoliths are high-grade supracrustal material. Many of the eastern Australian granulite facies xenoliths can be shown to have experienced Rb depletion after crystallisation, probably at lower crustal pressures and temperatures. The age of lower crust formation varies from province to province. The age of crystallisation for one suite of mafic xenoliths (Chudleigh province) is inferred to be relatively young (<100 Ma) and these xenoliths are associated with the Cenozoic volcanism, whereas depleted mantle Nd model ages (Table 2) for other xenoliths are older (400 Ma for the garnet pyroxenite from Delegate and 2170 Ma for a felsic granulite from Calcutteroo). The lower crust beneath the McBride province, in the Proterozoic Georgetown Inlier, grew mainly during the late Paleozoic (270-320 Ma), but contains some proportion of re-worked Proterozoic crust. Until additional detailed geochemical and isotopic studies are performed on spatially associated suites of lower crustal xenoliths, it is impossible to determine whether the mafic lower crust

169

prevalent throughout eastern Australia is primarily the result of cumulate processes, or whether a significant proportion of the xenoliths represent restite material left after extraction of granitic melts. The abundant granites throughout the Tasman Fold Belt require mafic residual rocks to be present in the deep crust (Compston and Chappell, 1979; Chappell, 1984), so it appears likely that some of this residual material would be sampled as xenoliths by the Tertiary basalts. The Subcontinental Lithospheric Mantle The most common types of ultramafic xenoliths found in eastern Australia are the spinel-bearing lherzolites and harzburgites, which are presumed to represent the bulk of the lithospheric mantle. Other upper mantle materials, which are considered to comprise only a minor volume of this region, are the ultramafic cumulate suites (eg., Al-Ti-rich pyroxenites and related rocks) and megacrysts. The major and trace element compositions of ultramafic cumulate suites and megacrysts have provided data on the local passage of basaltic magmas through the lithospheric mantle. The chemical compositions of spinel-bearing lherzolite and harzburgite xenoliths (and the deeper, garnet-bearing varieties) have been used to characterise the overall nature and composition of the eastern Australian lithospheric mantle. Ultramafic xenoliths were subdivided into 2 broad groups based on their composition and presumed origin by Wilshire and Shervais (1975). Frey and Prinz (1978) suggested an alternative classification which does not emphasize the xenolith origin, and avoids characterising a suite of xenoliths by a single lithologic name. The Cr-diopside group (Wilshire and Shervais, 1975) (or type I of Frey and Prinz, 1978) is dominated by lherzolite and harzburgite xenoliths, although less abundant Cr and Mg-rich dunites, wehrlites, olivine websterites, orthopyroxenites and clinopyroxenites are also included in this group. The second group, the Al-augite group (Wilshire and Shervais, 1975) (or type II of Frey and Prinz, 1978), includes Al-TiFe-rich xenoliths: pyroxenites, wehrlites, websterites and the like. It is commonly suggested that these xenoliths are products of the passage of basalts through the lithospheric mantle, either in association with the present igneous activity or

170

Chemical and isotopic composition.

related to earlier events. Composite xenoliths of type I and II lithologies generally show type II assemblages crosscutting type I lithologies, suggesting an intrusive relationship (Wilshire et al., 1980; Irving, 1980). The following sections summarise published and ongoing geochemical and isotopic studies of eastern Australian ultramafic xenoliths based on these subdivisions. Lherzolites and harzburgites (type I lithologies) Throughout eastern Australia Cenozoic basalts cany lherzolite and harzburgite xenoliths, which have been the focus of numerous penological and geochemical studies (Wilshire and Binns, 1961; Frey and Green, 1974; Wilkinson, 1975; Wilkinson and Binns, 1977; Ferguson et al., 1977; Varne, 1977; MacRae, 1979; Ferguson and Sheraton,

1979; Irving, 1980; BVSP, 1981; Mitchell and Keays, 1981; Dal Negro et al., 1984; Nickel and Green, 1984; Sutherland et al., 1984; Griffin et al., 1984). Many of these studies have characterised the spectrum of petrographic and mineral/whole rock compositions for these xenoliths. These xenoliths show a restricted range in mineralogies and major element compositions. Eastern Australian lherzolite and harzburgite xenoliths usually have a 4 phase mineralogy of olivine (-50 to -85%), orthopyroxene (-35 to -5%), clinopyroxene (-20 to -2%) and spinel (<5%). Some also have limited quantities (<5%) of amphibole, which is pargasitic in composition, or phlogopitic mica. A variety of textures are found in these xenoliths, although coarse and porphyroclastic varieties (Harte, 1977) are most common (eg., Frey and Green, 1974; Dasch and Green, 1975; Nickel and Green, 1984). Rare foliated peridotites, in which the planar fabrics are defined by tabular olivine and/or elongate grains of spinel and clinopyroxenes, are found throughout eastern Australia. The mineral chemistry of the lherzolite and harzburgite xenoliths is similar to that of other such xenolith suites worldwide: olivines display a limited range of Mg/(Mg + Fe) values, usually F088-92, and pyroxenes show only minor compositional variations, mostly in their AI2O3 and Cr203 contents. The whole rock major element composi-

tions of these xenoliths suggest they are residual peridotites, formed through extraction of variable amounts of a basaltic melt. Trace element compositions reflect subsequent addition of an incompatible element enriched component (Frey and Green, 1974). The estimated equilibration pressures and temperatures for these xenoliths indicate an upper mantle origin. Most equilibrated between 900 and 1100°C, based upon two pyroxene thermometry (Frey and Green, 1974; Nickel and Green, 1984; Griffin et al., 1984), with the anhydrous lherzolite and harzburgite xenoliths generally having higher temperatures (>1000°C) than the hydrous bearing xenoliths (<1000°C); see also Fig. 1 of Nickel and Green (1984). Studies on the deeper origin, rare garnet-bearing lherzolite and harzburgite xenoliths from eastern Australia show them to have equilibration temperatures of 980-1300°C, based on garnet-clinopyroxene thermometry (Sutherland et al., 1984). Estimates of the depths of origin of spinel-bearing lherzolite and harzburgite xenoliths have been mainly based on mineral stability studies. There is no mineral pair in the spinel-bearing xenoliths which can provide well constrained pressure estimates for these rocks, excepting the olivine clinopyroxene geobarometer which requires high quality Ca data, and such data are not available for most of these xenoliths (Nickel and Green, 1984). The experimental study of O'Neill (1981) and related earlier studies, suggest that spinel peridotite xenoliths are stable at depths less than 50 to 65 kms (1.6 to 2.0 GPa), and are underlain by garnet lherzolites. The garnet-orthopyroxene geobarometer (Harley and Green, 1982), when applied to eastern Australian garnet-bearing lherzolite and harzburgite xenoliths using clinopyroxene-garnet temperatures, yields pressures of between 1.5 to 2.4 GPa (50-80 km) (Ferguson et al., 1977; Ferguson and Sheraton, 1979; Sutherland et al., 1984). Recently Griffin et al. (1984) and O'Reilly and Griffin (1985) derived a geotherm based upon thermobarometric estimates from mineral phases in garnet pyroxenite xenoliths from Bullenmerri and Gnotuk maars, Victoria. By projecting the 2 pyroxene temperatures of spinel lherzolites on to this geotherm, they estimated the depth of origin

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

for these rocks to be between 25 and 50 km (0.8-1.6 GPa). They then compared these results with seismic refraction profiles from the Lachlan Fold Belt, N.S.W., and suggested that throughout southeastern Australia the crust is only 25-30 km thick and that it is underlain by a mixture of peridotite, pyroxenite and granulite, which, when combined with the very hot temperatures inferred from their geotherm, accounts for the relatively low Vp (6.4-7.6 km/s) observed at 25-45 km depths and the transitional nature of the Moho as observed in seismic profiles. They also suggested that the previously defined seismic Moho at 55 km represents the transition from spinel lherzolite to garnet lherzolite. This contrasts with previous crustal profiles (i.e., Ferguson et al, 1979; Finlayson et al., 1979) primarily by assigning a shallow depth to spinel lherzolites. Because of the implications of this model, and the pitfalls which are often associated with derivation of regional geotherms from xenolith suites, the assumptions used in their model bear examination. Firstly, although Griffin et al (1984) used primary mineral assemblages of garnet py roxenites to define their geotherm, they showed that many of the pyroxenites contain secondary mineral assemblages, often in the form of symplectic coronas surrounding garnets. They interpreted these coronas as forming due to isothermal uplift in a slow-moving (10 cm/s) magma. However, recent fluid dynamic calculations suggest that such a magma would need to be at least 15% crystalline in order to carry a 20 cm peridotite xenolith (Spera, 1984). Given the very low percentages of phenocrysts in the basanitic lavas in the Newer Province (Ellis, 1976; Irving and Green, 1976; Frey et al., 1978), we regard this scenario for the formation of the coronas as unlikely. The observation that these secondaiy assemblages yield different (mainly higher but some lower) equilibration temperatures than the primary assemblages by 15 to 40°C, using the Wood and Banno (1973) 2 pyroxene thermometer, or 25 to 77°C using the Wells thermometer (Table 7 of Griffin et al, 1984) suggests that the garnet pyroxenites may not have fully equilibrated to a geotherm. Secondly, it is necessary to consider the precision with which geothermometers can be applied. The two pyroxene and garnet-clinopyroxene

171

thermometers have uncertainties of ±5-7%, or about ±50-70C at 1000°C, based just on the thermodynamic calibrations (Wells, 1977; Ellis and Green, 1979). Because of the extreme temperature dependence of the geobarometer used in formulating the geotherm (0.4 GPa per 100°C, Harley and Green (1982)), a temperature uncertainty of ±5070°C leads to pressure uncertainties of ±0.2-0.3 GPa. Due to the steepness of the postulated geotherm, if the temperatures are lowered by only 50°C, then the corresponding pressures for spinel lherzolite xenoliths increase from 0.8-1.6 GPa (2550 km) to 1.1-1.9 GPa (35 to 60 km). These uncertainties must be considered when comparing xenolith data with refraction profiles. Thirdly, agreement of equilibration temperatures between lherzolite and pyroxenite portions of composite xenoliths was used by Griffin et al. (1984) to justify plotting lherzolite xenoliths on their pyroxenite-derived geotherm. However, numerous studies of such composite xenoliths have shown that lherzolites adjacent to pyroxenite dykes are chemically zoned in their bulk rock and mineral compositions (see below; Wilshire and Shervais, 1975; Wilshire et. al., 1980; Irving, 1980). The Fe-rich nature of the composite lherzolites described by Griffin et al. (Fo82 versus Fo86-9i for most type I spinel lherzolites) attests that similar effects occurred in these lherzolites. Therefore, it is likely that the composite lherzolite xenoliths were equilibrated by the very events(s) that formed them and should not be used as evidence of pervasive equilibration of spinel lherzolite to the calculated geotherm. Finally, the depth of the spinel-garnet transition for lherzolites is not clear cut. Griffin et al. place this transition at 55 km depth, based on the experimental data of O'Neill (1981) (note that the spinel-garnet phase boundary of O'Neill is not plotted correctly in Fig. 8 of Griffin et al. ~ the boundary is steeper than shown, but it crosses their geotherm at the same position). However, if one uses thermobarometry of xenoliths, a much deeper phase boundary is suggested. A spinel-gamet lherzolite xenolith from Jugiong, N.S.W., is a sample from this transition zone (Ferguson et al., 1977; Ferguson and Sheraton, 1979). Using the same thermobarometers employed by Griffin et al. to generate their geotherm, this xenolith yields equi-

172

Chemical and isotopic composition.

libration conditions of 2.2 GPa (75 km) and 1150°C, which is 20 km deeper than the phase boundary shown by Griffin et al (1984). Nickel and Green (1985), who have recently modified the Harley and Green geobarometer, concluded that this xenolith was equilibrated at about 2.0 GPa (65 Km) and 1050C. Regardless as to which of these estimates are used, the calculated equilibration temperature and pressure of this spinel-garnet lherzolite does not coincide with the univariant curve derived from O'Neill's experimental studies. This implies that one of the following is true: if the spinel-garnet transition is univariant (i.e., low amounts of &2O3 in the whole rock), as plotted by Griffin et al (1984), then this transition either occurs deeper in the mantle than predicted by the experimental data, or the garnet-orthopyroxene barometer gives systematically high pressures (which may be due to overestimated temperatures). Alternatively, if the spinel-garnet transition is divariant (due to the presence of significant C n 0 3 , O'Neill, 1981), then the spinel-garnet transition occurs over a considerable depth range and cannot be responsible for a jump in seismic velocities. The latter is considered more likely, given that Al/(A1+Cr) ratios observed in spinels are considerably below 1.0 (Griffin et al., 1984). However, irrespective of which of these cases is true, these observations cast doubt on the accuracy of the mantle profile derived from the Bullenmerri samples and the ability with which it can be correlated with velocity profiles from other areas of southeastern Australia. Other observations concerning geothermometry measurements (McGuire et al., 1989) further demonstrate the shortcomings associated with the assumption made in Griffin et al (1984). We conclude that the geotherm derived by Griffin et al (1984) is probably representative of crustal areas in which active volcanism occurs and is equivalent to the alkaline province geotherm of Jones et al (1983). However, such high temperatures are transient phenomena (Irving, 1976; Harte et al, 1981) and not representative for all of present-day eastern Australia, especially the eastern Lachlan Fold Belt (site of the refraction surveys) where the last volcanism occurred over 20 Ma ago (Wellman and McDougall, 1974). In addition, the inferences on crust-mantle structure

derived from this geotherm are equivocal. In particular, there is no thermobarometric or geophysical evidence that spinel lherzolites are the dominant rock types at 25 to 40 kms depth and that the velocity increase observed at 55 km depth in the Lachlan Fold Belt coincides with the spinel-garnet phase transition. Finally, in agreement with earlier studies (Ferguson et al, 1979; Finlayson et al, 1981), we suggest that the seismic velocity profile for the Lachlan Fold Belt in N.S.W. reflects a dominantly mafic lower crust between 20 and 40 kms depth (see below), with only minimal amounts of spinel lherzolite present. Major element compositions of the spinel-bearing lherzolite and harzburgite xenoliths from eastern Australia (eg., BVSP, 1981; Nickel and Green, 1984) are similar to those of spinel lherzolites worldwide (see Frey, 1984 for a review). The refractory e l e m e n t s (eg., M g O , Cr and Ni) d e f i n e

positive correlations with one another and the readily fusible major and minor elements (eg., CaO, FeO, AI2O3, Ti02, Na20) displaying negative correlations with MgO. The trace element compositions of some of these xenoliths have been characterised in detail (eg., Frey and Green, 1974; Irving, 1980; BVSP, 1981; Mitchell and Keays, 1981). The peridotite xenoliths from eastern Australia display a range of REE patterns from LREE-enriched to LREE-depleted (Fig. 2). The relative abundances of Ir, Os and Pd in these xenoliths are close to the chondritic values, whereas Re and Au contents are more variable and depleted relative to the primitive mantle (BVSP, 1981; Mitchell and Keays, 1981). These siderophile element abundances are typical for many garnet- and spinel-bearing lherzolite and harzburgite xenoliths from other localities around the world (BVSP, 1981). From the various studies on spinel-bearing lherzolite and harzburgite xenoliths from eastern Australia, internally consistent geochemical and petrological models have been developed for the origin and evolution of the lithospheric mantle. Foremost among these is the classic study by Frey Fig. 2. REE pattern for spinel-bearing lherzolite and harzburgite xenoliths from eastern Australia. Data from Frey and Green (1974) and Irving (1980).

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

SH76-2L LE76-1L Q76-5

La Ce Pr Nd

Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu

Lherzolites (Victoria, SE Australia) ."2 K? •S 1 h.

I |

I .1 u

•

• •

.01

2604 2669 2700

La Ce Pr Nd

Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu

m 2640

Lherzolites (Victoria, SE Australia)

• •

2728 2642

La Ce Pr Nd

Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu

Chemical and isotopic composition.

174

Fig. 3. 87Sr/86Sr versus Esd values for spinel-bearing lherzolite and harzburgite xenoliths from southeastern Australia and garnet and clinopyroxene separates from garnet pyroxenite xenoliths from Lake Bullenmerri. Data from McDonough and McCulloch (1987) and Table 2. Data field for the Newer basalts, the host basalts for all these xenoliths, is from McDonough et al (1985). Lherzolite and harzburgite localities include Mt. Porndon. Mt. Shadwell. Mt. Noorat and Anakies. western Victoria. For additional details see Fig. 1. Sample with tie line represent analyses of coexisting garnet and clinopyroxene from a Lake Bullenmerri garnet pyroxenite xenolith.Host basalts small filled squares and peridotites large filled squares.

and

clinopyroxene, and depletions in Ca, Al, Na, Fe. Sc.

geochemistry of a representative spectrum of lher-

V, Cr and the heavy R E E (elements that charac-

zolites and harzburgites from Victoria and South

terise the clinopyroxene component) relative to

Australia. A later study (Nickel and Green, 1984),

pyrolite. This event also produced depletions in

and

Green

(1974)

the

mineralogy

using a larger number of samples, has confirmed

incompatible trace elements (elements which par-

these

tition strongly into melts).

earlier

observations

and

interpretations.

Major and trace element data from these studies

Later, the residual

peridotites were enriched by the addition of a

showed that the underlying lithospheric mantle has

volumetrically minor, incompatible element-rich

experienced at least a two stage history. The first

component. This enrichment was found to affect

stage involved the extraction of basaltic melt from

all of the samples to varying degrees, but surpris-

a homogeneous peridotite (commonly assumed to

ingly, as noted by Frey and Green (1974), the most

be pyrolite (Ringwood, 1966)), and the second

refractory inclusions in terms of depletion in Ca.

stage consisted of an enrichment event (or events),

Al, Fe, etc., contain the highest relative abundance

in which

of incompatible trace elements. This second stage

various incompatible elements

were

added to the refractory peridotite. Support for this model comes from Sr and Nd isotope studies (discussed below).

zolite and harzburgite xenolith suites worldwide (Frey, 1984) and is commonly referred to as a

The early removal of a basaltic melt component is reflected bv low and variable amounts

enrichment event has been identified in many lher-

metasomatic , event. Frey and Green (1974) sug-

gested the two evolutionarv stages were not geneti-

Geol. Soc. Aust. Spec. Publ. 17, 163-188. cally related and that there may be a significant time difference between them.

175

Data from Sr, Nd and Pb isotope studies of southeastern Australian lherzolite and harzburgite xenoliths also show that these rocks have developed through a complex, multistage evolution. The Pb isotopic compositions of some of these xenoliths require a minimum of two stages (Cooper and Green, 1969). Many of the southeastern Australian lherzolites and harzburgites studied possess chondrite-normalized LREE enriched patterns and positive SNd values. Since positive £Nd values reflect long-term LREE depletion (i.e., Sm/Nd>0.31), these peridotites must have experienced a relatively recent LREE enrichment.

variation is greater than that previously reported for a single suite of spinel lherzolite xenoliths (Jagoutz et al., 1980; Menzies and Murthy, 1980; Stoschef al., 1980; Mengel etal., 1984; Roden, M.F. etal., 1984; Roden, M.K. etal., 1984; Betton and Civetta, 1984; Menzies et al., 1985). In addition, analyses of many samples from Mt Leura and Mt Gambier show that a wide range in isotopic compositions may be present at a single centre (Burwell, 1975; Chen and Frey, 1981; McDonough and McCulloch, 1987). Enriched isotopic compositions are also found for clinopyroxene separates from Lake Bullenmeiri garnet pyroxenite xenoliths (Fig.3, Table 2). These xenoliths have high Sr/ Sr ratios and negative ENd values. The

lherzolite xenoliths from the same region (Chen and Frey, 1980; 1981) also have been used to argue for an early melting event and at least two "metasomatic" events.

southeast Australian lithospheric mantle contrasts with the restricted isotopic compositions measured for the xenolith host, Newer basalts (McDonough etal., 1985).

Similarly, combined chemical and isotopic data on

Sr and Nd isotopic compositions vary widely for southeastern Australian spinel-bearing lherzolite and harzburgite xenoliths (Fig. 3). This

diversity of Sr and Nd isotopic compositions in the

Studies of sheared garnet lherzolites from South Africa showed these rocks to be fertile in terms of major element compositions, with

Nd (ppm) Fig. 4. Concentrations of Nd versus Sr for southeastern Australian spinel-bearing lherzolite and harzburgite xenoliths and some of their clinopyroxene mineral separates (McDonough and McCulloch, 1987).

176

Chemical and isotopic composition.

chondritic REE patterns and depleted Nd and Sr isotopic compositions. In contrast, granular lherzolites have depleted major element compositions, LREE enriched patterns and evolved Nd and Sr isotopic compositions (Nixon et al., 1981; Richardson et al1986). Sheared lherzolites have not been reported from eastern Australia. Foliated peridotite xenoliths from eastern Australia have Sr and Nd isotopic compositions indistinguishable from those of the non-foliated equivalents. Interestingly however, foliated lherzolites from Mt Gambier (see samples 2730, and 2736 of Dasch and Green, 1975) possess the highest 87 Sr/ 86 Sr ratios measured for the upper mantle beneath southeastern Australia, although they do not possess the lowest eNd values. Good correlation (r = 0.99) between Sr and Nd concentrations in the peridotite xenoliths and their clinopyroxene separates gives a nearly constant Sr/Nd ratio of 15 (Fig. 4) (McDonough and McCulloch, 1987). Stosch and Lugmair (1986) reported a similar correlation for spinel peridotite xenoliths from Europe. The average Sr/Nd ratio of these peridotites is similar to the bulk Earth value of 17 (Sun, 1982), and Sr/Nd ratios of primitive (i.e., unfractionated) basalts from mid-ocean ridges (MORBs) (Sr/Nd = 10-15), and intraplate settings (ocean island basalts (OIBs) and continental intraplate basalts) (Sr/Nd = 15-20) (McDonough and McCulloch, 1987). In contrast, island arc basalts have much higher Sr/Nd ratios of >30. These higher Sr/Nd ratios indicate a high Sr/Nd ratio in the source of island arc basalts (consistent with preferential enrichment of Sr from the slab). The Newer basalts, host for the southeastern Australian xenoliths, have a typical intraplate Sr/Nd ratio of 19.5 (McDonough etal., 1985). These data suggest that the constant Sr/Nd ratio in peridotite xenoliths reflects an intraplate origin for the added, incompatible element-enriched component (McDonough and McCulloch, 1987); the data are not consistent with the signature expected for the sources of island arc magmas. However, this does not necessarily exclude the possibility of such a tectonic setting existing in this region at some time in the past. Early Rb-Sr isotopic studies on southeast Australian lherzolite and harzburgite xenoliths found correlations between 87 Rb/ 86 Sr and 87 Sr/*6Sr ratios for whole rocks (Burwell, 1975)

and mineral phases (Dasch and Green, 1975), and both studies speculated that these correlations had age significance. The proposed ages (600 to 700 Ma) were suggested to represent the time of the last melt extraction experienced by the peridotites, and were related to an early Cambrian orogenic event in the region (Dasch and Green, 1975; Burwell, 1975). Recently, combined Sr and Nd isotopic studies of lherzolite and harzburgite xenoliths (Chen and Frey, 1981; McDonough and McCulloch, 1986; 1987), in some cases for the same samples as the earlier studies, showed that the two isotope systems yield different ages, suggesting that these ages are not significant. These Sr and Nd isotope data have been used to support a model of recent mixing in the mantle and to suggest that these isotope systems do not record a precise timing of mantle events (McDonough and McCulloch, 1987). As garnet-bearing lherzolites are relatively rare in eastern Australia ( Ferguson et al., 1977; Ferguson and Sheraton, 1979; Sutherland etal, 1984) very little is known about the trace element and isotopic compositions of the deeper parts of the eastern Australian lithospheric mantle. Al-rich Pyroxenites and related rocks (type II lithologies) The Al-augite group, or type II ultramafic xenolith suites, are less abundant in eastern Australia. Most studies of these xenoliths have been restricted to descriptions of the relative abundance and variety of rock types found at given localities and aspects of their mineral compositions (eg., Sutherland and Hollis, 1982). Often these studies draw petrographic and chemical analogies with specimens described in more detailed petrological/geochemical studies (eg., Irving, 1974b, Knutson and Green, 1975, Ellis, 1976, Frey and Prinz, 1978, Irving, 1980). These xenoliths are considered to represent cumulates which formed by dynamic flow crystallisation of basaltic magmas within the mantle (Irving, 1980) or minerals separated from basaltic magmas by filter pressing, or related processes, at mantle pressures (Wilshire etal., 1980). Irving (1980) reported petrographic and geochemical data for a variety of composite

177

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

xenoliths from Cenozoic basalts in northern Queensland and Victoria. These xenoliths consist of lherzolite or harzburgite veined by type II material (clinopyroxenite, wehrlite, websterite or orthopyroxenite). He interpreted these xenoliths as fragments of subcontinental lithospheric mantle intruded by pyroxene-rich cumulate material which separated from basaltic magmas during ascent. Similar composite xenoliths are reported from numerous other worldwide localities ( Wilshire et al., 1980) and provide important information on the physical and chemical response of the lithospheric mantle to ascending basaltic magmas. Wilshire and Shervais (1975) and Irving (1980) suggested that pronounced compositional gradients across the contacts of type II veins and lherzolite wall rock reflect the thermal and chemical modification of

the mantle produced by the passage of basaltic melts. Irving (1980) proposed that Fe-rich spinel lherzolite xenoliths, purported to be common in the subcontinental lithospheric mantle of eastern Australia (Wilkinson and Binns, 1977), are products of such wall rock-magma interaction. Though not conclusive, this model suggests that the fertile Fe-rich spinel lherzolite xenoliths are a small scale (on the order of tens of centimeters) localised reaction product, and therefore are of limited regional significance. Wass and coworkers (Wass et al., 1980; Menzies and Wass, 1983) have documented the occurrence of a notable group of xenoliths which are rich in apatite and amphibole. They consider the apatite/amphibole xenolith suite to be distinct from type I and II xenoliths and analogous to pegmatite

Chudlelgh/UcBrlde Clermont Tweed

V v 7

(NSW)

Comboyne

NSW

Leucltltes

Warrumbungle So.

Highlands

Newer Cosgrove Tasmania Norfolk

Is.

Lord Howe

Is.

- 4

0.7038 - 8

0.7044

0.7020

0.7040

0.7060

0.7080

,860

Sr/

Fig. 5. ^ S r / ^ S r versus £Nd values of Cenozoic, tholeiitic and alkalic basalts from eastern Australia. Data sources include Menzies and Wass (1983), McDonough etal .(1985), Nelson etal.{ 1986), Knutson et al (1986) and McDonough (1987). The inset diagram details the variation in 8 7 Si/ 6 Sr versus £Nd values for tholeiitic and alkalic basalts from the eastern 2/3 (Victoria region) of the Newer volcanic province; compare with Fig. 6, which shows the total variation in ^ S r / ^ S r versus £Nd value in the Newer basalts.

178

Chemical and isotopic composition.

veins developed within the mantle under a relatively high partial pressure of CO2 (Menzies and Wass, 1983). These volatile-rich xenoliths represent one end member of a spectrum of lithologies observed for mantle-derived xenoliths. Sr and Nd isotopic studies on eastern Australian type II xenoliths, apatite/amphibole xenoliths and various megacrysts related to these, show that many have isotopic compositions similar to those of their host basalts or regionally related magmatism (Stuckless and Irving, 1978; Menzies, 1983; Menzies and Wass, 1983). This suggests that they are genetically related to the contemporaneous magmatism, though in some cases not specifically related to their host magma. This situation contrasts strongly with the lack of evidence for genetic links between most type I lherzolite and harzburgite xenoliths and their hosts. Finally, there are other type II xenoliths (eg., the Lake Bullenmerri garnet pyroxenites) with rather evolved and high Sr/* Sr ratios and negative 8Nd values (Table 2). These xenoliths have isotopic compositions that are markedly different from their host basalts (McDonough et al, 1985) and are interpreted as pyroxene-rich cumulates formed during earlier magmatic events, which are not related to the present basaltic magmatism. 87

Chemical and Isotopic Systematics of Intraplate Basalts Basalts provide an additional source of information about the composition of the upper mantle. They are usually grouped by their mineralogy and composition as either tholeiitic, transitional or alkalic. Extensive efforts have been invested in trying to understand the origin of these different basalt types and to characterise their source regions. The combined data of experimental petrology and trace element geochemistry have given us greater insight into the differing origins of these basalt types and have shown that by varying the melting conditions (eg., temperature, pressure, mantle volatile content and composition, degree of partial melting and source mineralogy) a variety of basalts can be produced (eg., Ringwood, 1975; Frey et al, 1978; Takahashi and Kushiro, 1983). Frey et al (1978) identified a series of relatively unfractionated, tholeiitic to alkalic basalts from

southeast Australia and characterised the mineralogy and composition of their mantle source regions. More recently McDonough et al (1985) measured their Sr and Nd isotopic compositions and showed that the tholeiitic and alkalic basalts are from different sources. Primary and near primary eastern Australian Cenozoic tholeiitic and alkalic basalts display a large range of Sr and Nd isotopic compositions, which fall within the 'oceanic mantle array' (Fig. 5). The NSW leucitites have Sr/ Sr ratios of 0.7050 to 0.7055 and £Nd values of 0 to -4 (Nelson et al., 1986), whereas Tasmanian (McDonough et al., 1985) and Norfolk Island alkalic basalts (McDonough, 1987) have Sr/ Sr ratios of 0.7027 to 0.7033 and SNd values of +8 to +5. If the isotopic composition of the alkalic basalts represents an averaged regional melt composition from the lowermost lithospheric mantle, then this suggests significant isotopic heterogeneity exists in the lithospheric mantle throughout eastern Australia. 87

The systematic chemical and isotopic differences observed in the tholeiitic to alkalic basalts from southeastern Australia has been attributed to mixing between plume-derived and lithospheric mantle-derived melt components (McDonough et al., 1985). In this model the isotopic compositions of tholeiitic basalts are considered to be plumedominated, since tholeiitic basalts require higher melting temperatures and larger degrees of melting for their genesis. In contrast alkalic basalts require lower degrees of partial melting and may contain a dominantly lithospheric mantle-derived component. Therefore, the isotopic composition of the primitive alkalic Newer basalts may represent an averaged isotopic composition of the regional subcontinental lithospheric mantle component (McDonough et al, 1985). More evolved basaltic compositions may have been affected by crustal assimilation. Identical systematic trends in the Sr and Nd isotopic compositions of alkalic and tholeiitic basalts have been recognized in Hawaiian basalts (Fig. 6). Chen and Frey (1985) have proposed a similar mixing model to explain the origin of these differing basalt types. In contrast, a recent study on French Polynesian basalts (Duncan et al, 1986) has identified a case where the Sr and Nd isotopic composition of spatially and temporal-

179

Geol. Soc. Aust. Spec. Publ. 17,163-188.

The NSW leucite-bearing lavas outcrop well to the west of the continental dividing range and are considerably further inland than any of the other Cenozoic eastern Australian basalts (Cundari, 1974). These lavas have more evolved Sr and Nd isotopic compositions than those of the Newer volcanics (Fig. 5). In contrast to the NSW leucitebearing lavas, the Cosgrove olivine leucitite in northern Victoria (Birch, 1978) outcrops only about 100 km from the nearest centers of the Newer volcanic field and is isotopically similar to the basalts of this province. It is conceivable that isotopic compositions of the NSW potassic lavas reflects a distinctive and possibly older composition of the lowermost lithospheric mantle beneath this region. Sr / Sr

ly associated tholeiitic and alkalic basalts show an opposite shift in isotopic composition from that observed in Hawaii and southeastern Australia (Fig. 6). Duncan et al. (1986) suggested that the isotope data for French Polynesian basalts could be explained either by the above model, or as an alternative, they suggested a model involving a heterogeneous plume which mixes isotopically different components within the torus of the convening plume. In either case all of these models are consistent with thermal and mechanical considerations of melts interacting with the base of the lithosphere. The intrusion of hot mantle plumes into the base of the lithosphere during intraplate magmatism would result in large scale melting of the lowermost lithospheric mantle. Sr / Sr 0.7020 0.7030 0.7040 8 7

8 8

HALEAKALA

0.7030

;

A Hana • Kula • Honomanu

i i

UA POU

•

'Nd

*•

i i

NEWER

• ALKALIC • THOLEIITIC

0.7030

0.7040 0.7050

Q7Sr /

8 8

•

i I

• •

HONOLULU - KOOLAU ~

• 1

0.7050

rg tgl

LP • L

"Nd

0.7040

•

0.7030 B7Sr

0.7040

/ Sr

•

i 0.7050

Fig. 6. Sr/ Sr versus 8Nd values of Cenozoic, tholeiitic and alkalic basalts from Hawaii (Chen and Frey, 1985; Roden et al., 1985; Stille et al., 1984), French Polynesia (Duncan et al1986) and Newer volcanic province, southeastern Australian (McDonough et al1985). 87

180

Chemical and isotopic composition.

Origin and Growth of the Lower Eastern Australian Lithosphere Combining these petrological and geochemical data on eastern Australian xenoliths and basalts with geophysical information for the region, a model for the origin and evolution of the lower eastern Australian lithosphere may be constructed. This region of Australia is a relatively young part of the Australian continent (McCulloch, 1987) and may therefore have distinct characteristics from other regions of the continent which are part of the stable Archaean craton.

Considerable controversy exists concerning the nature of continental growth through time. There appears to be several processes involved, including: lateral accretion via subduction-related magmatism (Taylor and White, 1965; Taylor, 1967), continental addition during intracratonic rifting, which is suggested to be an important Archean growth process (Kroner, 1981), and crustal underplating due to conductive cooling (Oxburgh and Parmentier, 1978). Of these three mechanisms, the underplating model involving mantle growth through conduc-

Western Victoria

New South Wales

Mafic Granulite Pyroxenite

cn 3 u

CJ c S

<D -c

cx cn O

pC r^T"?"?"?"3 If' |i rI' r. I', f I 'Tyry-r;T,T,-^77V777l<7it7l<"ic7?7T7f71 ,1 I I >TyX yXyJtyJtyJt yXyX yXyiyX y * y * y * y * ^ $ t f t f $ ^ ^ ^ ^ r J r J $ y J ^ ^ Jf^f* < *<*<^ <^<

Fig. 7. A schematic model for eastern Australia emphasizing the chemically and isotopically heterogeneous nature of the crust and mantle and the variable depth of the Moho. The depth to the base of the lithosphere is not specified. The lithospheric mantle is defined as that portion of the lithosphere which is mechanically coupled to the continental crust. Compositional heterogeneities in the lithospheric mantle (discrete, textured areas) are idealised. These regions would have different chemical and isotopic compositions, which would reflect variations in the amount of melt extracted and the degree of secondary enrichment in incompatible elements. Regions in the lithospheric mantle with type II, Al-Fe-Ti-rich material are not depicted, but are considered to be distributed throughout the region as discrete vein-like bodies. The interaction of a hot mantle plume with the base of the lithospheric mantle is from the model presented in McDonough et al (1985). The considerable volumes of basaltic cumulate materials in the lower continental crust and uppermost lithospheric mantle is based on seismic data (Finlayson et al., 1979; Drummond and Collins, 1986), thermobarometry data (Griffin et al., 1984; O'Reilly and Griffin, 1985; Rudnick and Taylor, 1991), and petrological data (Ewart et al., 1980; Kay and Kay, 1983; Wass and Hollis, 1983; Rudnick et al., 1986).

Geol. Soc. Aust. Spec. Publ. 17, 163-188.

tive cooling may be the least likely to contribute to the long term stability of the lithospheric mantle. A lithosphere which has grown by passive underplating or thermal accretion is not consistent with various penological, seismological, thermal, gravity and tectonic considerations (Clark and Ringwood, 1964; O'Hara, 1975; Oxburgh andParmentier, 1978; Davies, 1979; Jordan, 1981; Ringwood, 1982). The chemically unmodified asthenospheric mantle peridotite accreted in this manner would be gravitationally unstable and would likely be detached from the lithospheric mantle and sink into the asthenosphere (Houseman etal, 1981). The refractory nature of most lherzolites suggests that the subcontinental lithospheric mantle

may grow dominantly by the accretion of thermally driven, intrinsically buoyant refractory peridotite diapirs (Clark and Ringwood, 1964; Oxburgh and Parmentier, 1978; Jordan, 1981; Ringwood, 1982). In this model, refractory peridotite diapirs are accreted onto the base of the lithospheric mantle during subduction zone and hotspot related magmatism, where they are permanently trapped beneath continents. Subsequent enrichment events change the parent/daughter isotopic ratios so that the isotopic compositons of the lherzolites change significantly with time, leading to an isotopically heterogeneous upper mantle. In contrast, in the oceanic lithosphere refractory peridotite is recycled back into the mantle at subduction zones (Ringwood, 1982), and the short time scales involved do not allow for development of significant isotopic heterogeneity. It has been suggested that regions of the lithospheric mantle which have grown in this manner are also susceptible to being detached from the lithospheric mantle and recycled back into the asthenosphere (McKenzie and O'Nions, 1983). However, the temperature gradients and density contrasts required (Houseman et al., 1981) are such that this would be a rare fate for such chemically buoyant material. The early basaltic melt extraction event, so commonly identified in lherzolite and harzburgite xenoliths, may record the initial development and stabilisation of the subcontinental lithospheric mantle. A once fertile peridotite source region that undergoes partial melting would become less dense, due to extraction of a dense Fe-rich com-

181

ponent, and buoyant with respect to the surrounding mantle. The second stage enrichment event(s), identified in the incompatible trace elements and Sr and Nd isotope compositions of these xenoliths (Frey and Green, 1974; Chen and Frey, 1981; McDonough and McCulloch, 1987), document later tectono-magmatic reactivation in the subcontinental lithospheric mantle. The chemical signature recorded in the lherzolites from these later events (in particular the Sr/Nd ratios) are similar to those predicted to occur in an intraplate or divergent plate tectonic setting. Similarly, the composition and growth of the lower continental crust is related to a series of tectonothermal events. Geochronologic evidence shows that the mafic lower crustal xenoliths, which are the dominant lower crustal rock type, formed through both basaltic underplating of the lower crust (which occurred during both the Palaeozoic granite-forming orogenies and the Cenozoic basaltic magmatic events) and partial melt extraction from pre-existing mafic to intermediate rock types. In addition, minor quantities of supracrustal rocks and felsic igneous rocks exist in the lower crust. The supracrustals presumably were tectonically emplaced during continental margin orogenic events and the felsic meta-igneous rocks may represent near in situ melts of intermediate to mafic rock types which formed during the Palaeozoic orogenies. Seismic refraction profiles coupled with petrological/geochemical data for xenoliths in eastern Australia provide a picture of a heterogeneous deep lithosphere throughout the region (Fig. 7). Although still equivocal, the seismic data suggest that the Moho may vary considerably in depth and structure, from a distinct Moho at 35 km in Victoria to a transitional Moho between 50-55 kilometers in the Lachlan Fold Belt of N.S.W. (Finlayson et al 1979; Finlayson, 1982; Wesson and Gibson, unpublished data). Refraction profiles through the eastern Lachland Fold Belt show that Vp ranges from 6.4 to 7.6 km/s over a depth interval of 25 to 45 km. Such velocities suggest that anhydrous spinel lherzolite (with>60% olivine) can only be a dominant rock type (ie., >50% by volume) at depths greater than about 45 km, assuming it is interlayered mafic granulites (based on the laboratory measurements of Bezant (1986), Jackson and Ar-

182

Chemical and isotopic composition.

cuius (1984), and Rudnick and Jackson (1987) extrapolated to ~1000°C). The xenolith data are therefore consistent with a model of a thick, predominantly mafic granulite lower crust, with spinel lherzolite becoming a volumetrically significant rock type at depths greater than about 45 km in the eastern Lachlan Fold Belt, as postulated by Ferguson et al (1979), Finlayson et al. (1979) and Drummond and Collins (1986). ACKNOWLEDGEMENTS We thank Drs. I. Jackson, H. O'Neill, D.J. Ellis, W. Cameron, S R. Taylor and G.F. Davies for very fruitful discussions. We are most grateful to Prof. D.H. Green for generously providing us with very well documented samples of basalts, pyroxenites and peridotites from eastern Australia. We thank Hans Hensel and Andy Haynes for assistance with isotopic measurements. The comments of Dr Ian Nicholls, Dave Nelson and an anonymous reviewer are appreciated; they improved the manuscript. REFERENCES ARCULUS R.J., FERGUSON J., CHAPPELL B.W., SMITH D., MCCULLOCH M.T., JACKSON I., HENSEL H.D., TAYLOR S.R., KNUTSON J. & GUST D.A. 1987. Eclogites and

granulites in the lower continental crust: examples from eastern Australia and southwestern U.S.A. In Smith D.C., ed. Eclogites and Related Rocks. Elsevier, (in press).

BVSP, 1981. Basaltic Volcanism Study Project. Basaltic Volcanism on the Terrestrial Planets. Pergamon Press, Inc., New York, 1286 pp. BETTON P.J. & CIVETTA L. 1984. Strontium and neodymium isotopic evidence for the heterogenous nature and development of the mantle beneath Afar (Ethiopia). Earth and Planetary Science Letters 71, 59-70.

BEZANT C. 1986. Geothermometry and seismic properties of upper mantle lherzolites from eastern Australia. B.S. Thesis, Macquarie University, 65 pp.

BIRCH W.D. 1978. Mineralogy and geochemistry of the leucitite at Cosgrove, Victoria. Geological Society of Australia, Journal 25, 369-385. BURWELL A.D.M. 1975. Rb-Sr isotope geochemistry of lherzolites and their constituent minerals from Victoria, Australia. Earth and Planetary Science Letters 28,69-78. CHAPPELL B.W. 1984. Source rocks of I- and S-type

granites in the Lachlan Fold Belt, southeastern Australia. Philosophical Transactions of the Royal Society of London A310, 693-707. CHEN C. Y. & FREY F. A. 1980. Geochemistry of lherzolite

inclusions from Mt. Leura, Victoria, Australia. EOS 61,

413.

CHEN C.Y. & FREY F.A. 1981. Multi-stage geochemical events in the upper mantle: evidence from geochemical studies of spinel lherzolites from Mount Leura, Australia. EOS 62,415. CHEN C.Y. & FREY F.A 1985. Trace element and isotopic geochemistry of lavas from Haleakala volcano east Maui, Hawaii: implications for the origin of Hawaiian basalts. Journal of Geophysical Research 90, 8743-8768. CLARK S.P. &RINGWOOD A.E. 1964. Density distribution

and constitution of the mantle. Reviews of Geophysics 2, 35-88.

COMPSTON W. & LOVERING J.F. 1969. The strontium

isotopic geochemistry of granulitic and eclogitic inclusions from the basic pipes at Delegate, eastern Australia.. GeochimicaCosmochimcaActa 33, 691-699. COMPSTON W. & CHAPPELL B.W. 1979. Sr-isotope evolution of granitoid source rocks. In M. W. McElhinny ed. The Earth: Its Origin, Structure and Evolution. Academic Press, 377-426. COOPER J.A. & GREEN D.H. 1969. Lead isotope measurements on lherzolite inclusions and host basalts from western Victoria, Australia. Earth and Planetary Science Letters 6, 69-76. CUNDARI A. 1974. Petrology of the leucite-bearing lavas

in New South Wales. Geological Society of Australia, Journal 20,465-492.

183

Geol. Soc. Aust. Spec. Publ. 17, 163-188. DAL NEGRO A.,CARBONIN S., DOMENEGHETTI C., MOLIN G.M., CUNDARI A. & PICCIRILLO E.M. 1984. Crystal

chemistry and evolution of the clinopyroxene in a suite of high pressure ultramafic nodules from the Newer volcanics of Victoria, Australia. Contributions to Mineralogy and Petrology 86, 221-229. DASCH E.J. & GREEN D.H. 1975. Strontium isotope

geochemistry of lherzolite inclusions and host basaltic rocks, Victoria, Australia. American Journal of Science 21S, 461-469.

DAVIES G.F. 1979. Thickness and thermal history of

continental crust and root zones. Earth and Planetary Science Letters 44, 231-238. DRUMMOND B.J. & COLLINS C.D.N. 1986. Seismic

evidence for underplating of the lower continental crust of Australia. Earth and Planetary Science Letters 79, 361-372. DUNCAN R.A., MCCULLOCH M.T., BARSCZUS H.G. & NELSON D.R. 1986. Plume versus lithospheric sources

for melts at Ua Pou, Marquesas Islands. Nature 322, 534-538.

EDWARDS A.C., LOVERING J.F. & FERGUSON J. 1979.

High pressure basic inclusions from the Kayrunnera kimberlitic diatreme in New South Wales. Contributions to Mineralogy and Petrology 69, 185-192.

ELLIS D.J. 1976. High pressure cognate inclusions in the Newer volcanics of Victoria. Contributions to Mineralogy and Petrology 58, 149-180. ELLIS D.J. & GREEN D.H. 1979. An experimental study

of the effect of Ca upon garnet-clinopyroxene Fe-Mg exchange equilibrium. Contributions to Mineralogy and Petrology 71, 13-22. EWART A., BAXTER K. & Ross J.A. 1980. The petrology

and pedogenesis of the Tertiary anorogenic mafic lavas of southern and central Queensland, Australia - possible implications for crustal thickening. Contributions to Mineralogy and Petrology 75, 129-152. FERGUSON J. & SHERATON J.S. 1979. Petrogenesis of

kimberlitic rocks and associated xenoliths of southeastern Australia In F.R. Boyd and H.O.A. Meyer eds. Kimberlites, Diatremes and Diamonds: Their Geology, Petrology and Geochemistry. American Geophysical Union, 140-160.

FERGUSON J., ELLIS D.J. & ENGLAND R.N. 1977. Unique

spinel-garnet lherzolite inclusion in kimberlite from Australia. Geology 5,278-280. FERGUSON J., ARCULUS R.J. & JOYCE J. 1979. Kimberlite

and kimberlitic intrusives of southeastern Australia: a review. BMR Journal of Australian Geology & Geophysics4,227-241.

FINLAYSON D.M. 1982. Geophysical differences in the

lithosphere between Phanerozoic and Precambrian Australia. Tectonophysics 84, 287-312. FINLAYSON D.M., COLLINS C.D.N. & DENHAM D. 1979.

Crustal structure under the Lachlan Fold Belt, southeastern Australia. Physics of the Earth and Planetary Interiors 21, 321-342. FREY F.A. 1984. Rare earth element abundances in upper

mantle rocks. In Henderson P. ed. Rare Earth Element Geochemistry Elsevier, 153-175.

FREY F.A. & GREEN D.H. 1974. The mineralogy,

geochemistry and origin of lherzolite inclusions in Victorian basanites. Geochimica Cosmochimica Acta 38, 1023-1059.

FREY F.A. & PRINZ M. 1978. Ultramafic inclusions from

San Carlos, Arizonia: petrologic and geochemical data bearing on their petrogenesis. Earth and Planetary Science Letters 38,129-176.

FREY F.A., GREEN D.H. & ROY S.D. 1978. Integrated

models of basalt petrogenesis: a study of quartz tholeiites to olivine melilitites from south eastern Australia utilizing geochemical and experimental penological data. Journal of Petrology 19, 463-513. GRIFFIN W.L., WASS S.Y. & HOLLIS J.D. 1984.

Ultramafic xenoliths from Bullenmerri and Gnotuk maars, Victoria, Australia: petrology of a subcontinental crust-mantle transition. Journal of Petrology 25, 53-87. GRIFFIN W.L. & O'REILLY S.Y. 1986. The lower crust in eastern Australia: xenolith evidence. In Dawson J.B., Carswell D.A., Hall J. & Wedepohl K.H., eds. The Nature of the Lower Continental Crust. Geological Society of London, Special Publication 24, 363-374. HARLEY

S.L.

GREEN

D.H.

1982.

Garnet-orthopyroxene barometry for granulites and periodtites. Nature 300, 697-701.

184

Chemical and isotopic composition.

HARTE B. 1977. Rock nomenclature with particular

relation to deformation and reciystallisation textures in olivine bearing xenoliths. Journal of Geology 85, 279-288.

JAGOUTZ E., CARLSON R.W. & LUGMAIR G.W. 1980.

Equilibrated Nd-unequilibrated Sr isotopes in mantle xenoliths. Nature 286, 708-710. JONES A.P., SMITH J.V. & DAWSON J.B. 1983.

of mineral equilibria in granulite facies nodules from kimberlites. Nature 291,147-148.

Metamorphism, partial melting, and K-metasomatism of garnet-scapolite-kyanite granulite xenoliths from Lashaine, Tanzania. Journal of Geology 91,143-165.

HENSEL H.-D., MCCULLOCH M.T. & CHAPPELL B.W.

JORDAN T.H. 1981. Continents as a chemical boundary

HARTE B., JACKSON P.M. & MACINTYRE R.M. 1981. Age

1985. The New England Batholith: constraints on its derivation from Nd and Sr isotopic studies of granitoids and country rocks. Geochimica Cosmochimica Acta 49, 369-384. HOLLIS J.D. 1981. Ultramafic and gabbroic nodules from

the Bullenmerri and Gnotuk maars, Camperdonw, Victoria. Royal Society of Victoria, Proceedings 92, 155-167.

HOUSEMAN G.A., MCKENZIE D.P. & MOLNAR P. 1981.

Convective instability of a thickened boundary layer and its relevance for the thermal evolution of continental convergent belts. Journal of Geophysical Research 86, 6115-6132.

IRVING A.J. 1974a. Geochemical and high pressure experimental studies of garnet pyroxenite and pyroxene granulite xenoliths from the Delegate basaltic pipes, Australia. Journal of Petrology 15, 1-40. IRVING A.J. 1974b. Pyroxene-rich ultramafic xenoliths in

the Newer basalts of Victoria, Australia. Neues Jahrbuch fur Mineralogie Abhandlungen 120, 147-167.

IRVING A.J. 1976. On the validity of paleogeotherms

determined from xenolith suites in basalts and kimberlites. American Mineralalogist 61, 638-642.

1980. Petrology and geochemistry of composite ultramfic xenoliths in alkalic basalts and implications for magmatic processes within the mantle. American Journal of Science 280-A, 389-426. IRVING A.J.

IRVING A.J. & GREEN D.H. 1976. Geochemistry and pedogenesis of the Newer basalts of Victoria and South Australia. Geological Society ofAustralia 23, 45-66. JACKSON I. & ARCULUS R.J. 1984. Laboratory wave

veolcity measurements on lower crustal xenoliths from Calcutteroo, South Australia. Tectonophysics 101, 185-197.

layer. Philosophical Transactions of the Royal Society of London A301, 359-373.

KAY R.W. & KAY S.M. 1981. The nature of the continental crust: inferences from geophysics, surface geology, and crustal xenoliths. Reviews of Geophysics and Space Physics 19, 271-297. KAY S.M. & KAY R.W. 1983. Thermal history of the deep crust inferred from granulite xenoliths, Queensland, Australia. American Journal of Science 283-A, 486-513. KNUTSON J. & GREEN T.H. 1975. Experimental duplication of a high-pressure megacryst/cumulate assemblage in a near-saturated hawaiite. Contributions to Mineralogy and Petrology 52, 121-132. KNUTSON J., MCDONOUGH W.F., DUGGAN M.B. & CHAPPELL B.W. 1986. Geochemical and isotopic

characteristics of eastern Australian Cainozoic "Central" volcanoes. International Volcanological Conference Abstracts, 174.

KRONER A., ed. 1981. Precambrian Plate Tectonics. Elsevier, Amsterdam, 781 p. LEYRELOUP A., DUPUY C. & ANDRIAMBOLOLONA R.

1977. Catazonal xenoliths in French Neogene volcanic rocks: constitution of the lower crust. 2. Chemical composition and consequences of the evolution of the French Massif Central Precambrian Crust. Contributions to Mineralogy and Petrology 62, 283-300. LOVERING J.F. & WHITE A.J.R. 1969. Granulitic and

eclogitic inclusions from basic pipes at Delegate, Australia. Contributions to Mineralogy and Petrology 21, 9-52. MACRAE N.D. 1979. Silicate glass and sulfides in ultramafic xenoliths, Newer basalts, Victoria, Australia. Conributions to Mineralogy and Petrology 68, 275-280.

Geol. Soc. Aust. Spec. Publ. 17, 163-188. MCCULLOCH M.T. 1987. SM-ND isotopic constraints on

the evolution of Precambrian crust in the Australian continent. In Kroner, A., ed. Proterozoic Lithosphere Evolution, AGU Press, 115-130. MCCULLOCH M.T. & WASSERBURG G.J. 1978. Sm-Nd

and Rb-Sr chronology of continental crust formation.

Science 200,1003-1011.

MCCULLOCH M.T. & CHAPPELL B.W. 1982. Nd isotopic characteristics of S- and I-type granites. Earth and Planetary Science Letters 58, 51-64. MCCULLOCH M.T., ARCULUS R.J., CHAPPELL B.W. & FERGUSON J. 1982. Isotopic and geochemical studies of

nodules in kimberlite have implications for the lower continental crust. Nature 300, 166-169. Chemical and isotopic systematics of basalts and peridotite xenoliths: implications for the composition and evolution of the earth's mantle. PhD thesis, Australian National University, 204 pp.

MCDONOUGH W.F. 1987.

MCDONOUGH W.F. & MCCULLOCH M.T.

1985.

Geochemical and isotopic systematics of spinel (sp) lherzolites from SE Australia. EOS 66, 1110.

MCDONOUGH W.F. & MCCULLOCH M.T. 1987. Isotopic heterogeneity in the southeast Australian subcontinental lithosphere mantle. Earth and Planetary Science Letters 86, 327-340.

MCDONOUGH W.F., MCCULLOCH M.T. & SUN S.-S. 1985.

Isotopic arid geochemical systematics in Tertiary-Recent basalts from southeastern Australia and implications for the evolution of the sub-continental lithosphere. Geochimica Cosmochimica Acta 49, 2051-2067. MCGUIRE A.V., DYAR M.D. &WARD K.A. 1989. The

neglected Fe /Fe ratio: a study of Fe contents of megacrysts from alkali basalts. Geology 17, 687-690. 3+

185

MENZIES M. 1983. Mantle ultramafic xenoliths in alkaline magmas: evidence for mantle heterogeneity modified by magmatic activity. In Hawkesworth, C.J. & Norry, M.J eds. Continental Basalts and Mantle Xenoliths, Shiva, 92-110. MENZIES M. & MURTHY V.R. 1980. Nd and Sr isotope geochemistry of hydrous mantle nodules and their alkali basalts: implications for local heterogeneities in metasomatically veined mantle. Earth and Planetary Science Letters 46, 323-334. MENZIES M.A. & WASS S.Y. 1983. CO2- and LREE-rich

mantle below eastern Australia: a REE and isotopic study of alkaline magmas and apatite-rich mantle xenoliths from the Southern Highlands Province, Australia. Earth and Planetary Science Letters 65, 287-302. MENZIES M., KEMPTON P. & DUNGAN M. 1985.

Interaction of continental lithosphere and asthenospheric melts below the Geronimo Volcanic Field, Arizonia, U.S.A. Journal of Petrology 26, 663-693. MITCHELL R.H. & KEAYS R.R. 1981. Abundance and

distribution of gold, palladium and iridium in some spinel and garnet lherzolites: implications for the nature and origin of precioius metal-rich intergranular components in the upper mantle. Geochimica Cosmochimica Acta 45, 2425-2442.

NELSON D.R., MCCULLOCH M.T. & SUN S.-S. 1986. The

origins of ultrapotassic rocks as inferred from Sr, Nd and Pb isotopes. Geochimica Cosmochimica Acta 50, 231-246.

NEWTON R.C. & PERKINS D., Ill 1982. Thermodynamic

calibration of geobarometers based on the assemblages garnet-plagioclase-orthopyroxene (clinopyroxene) -quartz. American Mineralogist 67, 203-222. NICKEL K.G. & GREEN D.H. 1984. The nature of the

MCKENZIE D. & O'NIONS R.K. 1983. Mantle reservoirs and ocean island basalts. Nature 301, 229-231.

upper-most mantle beneath Victoria, Australia as deduced from ultramafic xenoliths. In Kornprobst J. ed. Kimberlites. II: The Mantle and Crust-Mantle Relationships. Elsevier, 161-178.

MENGEL K., KRAMM U., WEDEPOHL K.H. & GOHN E.

NICKEL K.G. & GREEN

87, 369-375.

158-170.

1984. Sr isotopes in peridotite xenoliths and their basaltic host rocks from the northern Hessian depression (NW Germany). Contributions to Mineralogy and Petrology

D.H. 1985. Empirical geothermobarometry for garnet peridotites and implications for the nature of the lithosphere, kimberlites and diamonds. Earth and Planetary Science Letters 73,

186

Chemical and isotopic composition.

NIXON P.H., ROGERS N.W., GIBSON I.L. & GREY A. 1981.

Depleted and fertile mantle xenoliths from southern African kimberlites. Annual Reviews of Earth and Planetary Science 9, 285-309. O'HARA M.J. 1975. Is there an Icelandic mantle plume? Nature 253,708-710. O'NEILL H.ST.C. 1981. The transition between spinel

lherzolite and garnet lherzolite, and its use as a geobarometer. Contributions to Mineralogy and Petrology 77, 185-194. O'REILLY S.Y. & GRIFFIN W.L. 1985. Axenolith-derived geotherm for southeastern Australia and its geophysical implications. Tectonophysics 111, 41-63.

OXBURGH E.R. & PARMENTIER E.M. 1978. Thermal processes in the formation of continental lithopshere. Philosophical Transactions of the Royal Society of

London A288,415-429.

PADOVANI E.R. & CARTER J.L. 1977. Aspects of deep crustal evolution beneath south central New Mexico. In Heacock J.G. ed. The Earth's Crust: Its Nature and Physical Properties. American Geophysical Union, Geophysical Monograph 20, 19-55. RICHARDSON S.H., ERLANK A.J. & HART S.R. 1985.

Kimberlite-borne garnet peridotite xenoliths from old enriched subcontinental lithosphere. Earth and Planetary Science Letters 75, 116-128. RINGWOOD A.E. 1966. The chemical composition and

origin of the earth. In Hurley P.M. ed. Advances in Earth Science M.I.T. Press, 287-356.

RODEN M.K., HART S.R., FREY F.A. & MELSON W.G.

1984. Sr, Nd and Pb isotopic and REE geochemistry of St. Paul's Rocks: the metamorphic and metasomatic development of an alkali basalt mantle source. Contributions to Mineralogy and Petrology 85,376-390. RUDNICK R.L. and JACKSON 1.1987. Physical properties of a magmatically underplated lower continental crust, [abst.] Terra Cognita 7. RUDNICK R.L. & TAYLOR S.R. 1987. The composition and petrogenesis of the lower crust: a xenolith study. Journal of Geophysical Research 92, 13981-14005. RUDNICK R.L. & TAYLOR S.R. 1991. Petrology and

geochemistry of lower crustal xenoliths from northern Queensland and inferences on lower crustal composition. In Drummond B.J. ed. The Australian Lithosphere. Geo I. Soc. Aust. Spec. 17, 189-208. RUDNICK R.L. & WILLIAMS I.S. 1987. Dating the lower crust by ion microprobe. Earth and Planetary Science Letters 85, 145-161. RUDNICK R.L., MCDONOUGH W.F., MCCULLOCH M.T. & TAYLOR S.R. 1986. The chemical and isotopic

composition of lower crustal xenoliths from Queensland, Australia: evidence for deep crustal assimilation and fractionation of continental basalts. Geochimica Cosmochimica Acta 50, 1099-1115.

SPERA F.J. 1984. Carbon dioxide in petrogenesis III: role of volatiles in the ascent of alkaline magma with special reference to xenolith-bearing mafic lavas. Contributions to Mineralogy and Petrology 88, 217-232. STILLE P., UNRUH D.M. & TATSUMOTO M. 1983. Pb, Sr,

RINGWOOD A.E. 1975. Composition and Petrology of the Earth's Mantle. McGraw-Hill, New York, 618 pp.

Nd and Hf isotopic evidence of multiple sources for Oahu, Hawaii basalts. Nature 304, 25-29.

Phase transformations and differentiation in subducted lithosphere: implications for mantle dynamics, basalt pedogenesis, and crustal evolution. Journal of Geology 90, 611-643.

STOLZ A.J. 1984. Garnet websterites and associated ultramafic inclusions from a nepheline mugearite in the Walcha area, New South Wales, Australia. Mineralog.

RINGWOOD A.E. 1982.

RODEN M.F., FREY F.A. & FRANCIS D.M. 1984. An

example of consequent mantle metasomatism in peridotite inclusions from Nunivak Island, Alaska. Journal of Petrology 25, 546-577.

Mag. 48. 167-180.

STOSCH H.G., CARLSON R.W. & LUGMAIR G.W. 1980.

Episodic mantle differentiation: Nd and Sr isotopic evidence. Earth and Planetary Science Letters 47, 263-271.

187

Geol. Soc. Aust. Spec. Publ. 17,163-188.

STUCKLESS J.S. & IRVING A.J. 1976. Strontium isotope

WASS S.Y. & HOLLIS J.D. 1983. Crustal growth in

geochemistry of megacrysts and host basalts from southeastern Australia. Geochimica Cosmochimica Acta

south-eastern Australia-evidence from lower crustal eclogite and granulite xenoliths. Journal ofMetamorphic Geology 1, 25-45.

40, 2 0 9 - 2 1 3 .

SUN S.-S. 1982. Chemical composition and origin of the earth's primitive mantle. Geochimica Cosmochimica Acta 46,179-192.

WELLMAN R & MCDOUGALL 1 . 1 9 7 4 . P o t a s s i u m - A r g o n

ages on the Cainozoic volcanic rocks of New South Wales. Geological Society of Australia, Journal 21, 247-272.

SUTHERLAND F.L. & HOLLIS J.D. 1982. M a n t l e - l o w e r

crust petrology from inclusions in basaltic rocks in eastern Australia - an outline. Journal ofVolcanological and Geothermal Research 14, 1 -29.

WELLS P.R.A. 1977. Pyroxene thermometry in simple and complex systems. Contributions to Mineralogy and Petrology 62, 129-139.

SUTHERLAND F.L., HOLLIS J . D . & BARRON L . M . 1984.

WILKINSON J.F.G. 1975. An Al-spinel ultramafic-mafic

Garnet lherzolite and other inclusions form a basalt flow, Bow Hill, Tasmania. In Kornprobst J. ed. Kimberlites II: The Mantle and Crust-Mantle Relationships, Elsevier, 145-160.

inclusion suite and high pressure megacrysts in an analcimite and their bearing on basaltic magma fractionation at elevated pressure. Contributions to Mineralogy and Petrology 53, 71-104.

TAKAHASHI E . & KUSHIRO I. 1983. M e l t i n g of a d r y

WILKINSON J.F.G. &

peridotite at high pressures and basalt magma genesis. American Mineralogist 68, 859-879.

iron-rich lherzolite xenoliths of the Cr-diopside suite: a guide to the primary nature of anorogenic tholeiitic andesite magmas. Contributions to Mineralogy and Petrology 65, 199-212.

TAYLOR S.R. 1967. The origin and growth of continents. Tectonophysics 4, 17-34.

BINNS R . A .

1977.

Relatively

WILKINSON J.F.G. & TAYLOR S.R. 1980. Trace element TAYLOR S . R . & MCLENNAN S . M . 1985. The

Continental

Crust: Its composition Oxford, U.K., 312 pp.

Blackwell,

and evolution.

fractionation trends of tholeiitic magma at moderate pressure: evidence from an Al-spinel ultramafic-mafic inclusion suite. Contributions to Mineralogy and Petrology 75, 225-233.

TAYLOR S . R . & WHITE, A . J . R . 1965. G e o c h e m i s t r y of

andesites and the growth of continents. Nature 208,

WILLIAMS I.S., COMPSTON W. & CHAPPELL B . W . 1 9 8 3 .

271-273.

Zircon and monazite U-Pb systems and the histories of I-type magmas, Berridale Batholith, Australia. Journal of Petrology 24, 76-97.

VARNE R. 1977. On the origin of spinel lherzolite inclusions in basaltic rocks from Tasmania and elsewhere. Journal of Petrology 18, 1-23. WASS S.Y. & IRVING A.J. 1976. X E N M E G A catalogue

of occurrences of xenoliths and megacrysts in basic volcanic rocks of eastern Australia. The Australain Museum, Sydney, 441 pp. WASS S.Y., HENDERSON P. &

ELLIOTT C . J .

1980.

Chemical heterogeneity and metasomatism in the upper mantle: evidence from rare earth and other elements in apatite-rich xenoliths in basaltic rocks from eastern Australia. Philosophical Transactions of the Royal Society of London A297, 333-346.

WILSHIRE H.G. & BINNS R.A. 1961. Basic and ultrabasic xenoliths from volcanic rocks of New South Wales. Journal of Petrology 2, 185-208. WILSHIRE H . G . & SHERVAIS J.W. 1 9 7 5 . A l - a u g i t e a n d

Cr-diopside ultramafic xenoliths in basaltic rocks from western United States. Physics and Chemistry of the Earth 9, 257-272. WILSHIRE H . G . , NIELSON PIKE J.E., MEYER C . E .

SCHWARZMAN E.C. 1980. Amphibole-rich veins in lherzolite xenoliths, Dish Hill and Deadman Lake, California. American Journal of Science 280A, 576-593.

188

Chemical and isotopic composition.

WOOD B . J . & BANNO S. 1 9 7 3 . G a r n e t - o r t h o p y r o x e n e a n d

W.F. McDonough

orthopyroxene- clinopyroxene relationships in simple and complex systems. Contributions to Mineralogy and Petrology 42, 109-121.

R.L. Rudnick M.T. McCulloch Research School of Earth Sciences Australian National University G P O Box 4 Canberra A C T 2601 AUSTRALIA.

Geol. Soc. Aust. Spec. Publ. 11, 189-208.

189

Petrology and Geochemistry of Lower Crustal Xenoliths from Northern Queensland and Inferences on Lower Crustal Composition Roberta L. Rudnick and S.R. Taylor Research School of Earth Sciences, Australian National University, Canberra, A.C.T. Australia Lower crustal xenoliths from northern Queensland have highly variable compositions, which correlate with their geographic positions. Xenoliths from the Chudleigh province, carried in <2Ma basalts which erupted through Palaeozoic rocks of the Tasman fold belt, are exclusively mafic. Three types of xenoliths are present: plagioclase-rich, pyroxene-rich and xenoliths transitional between these two types. Mineralogy of the plagioclase-rich xenoliths varies from moderate- to high-pressure assemblages (20 to 45 km depth), at a relatively constant whole rock composition. Coronas of (1) olivine rimmed by orthopyroxene, which is in turn rimmed by symplectic pyroxene-spinel and (2) spinel rimmed by garnet, point to isobaric cooling of the rocks at different crustal levels. Major and trace element chemistry of these xenoliths suggests they are genetically related cumulates from basaltic liquids, but are not related to their host basalts. Crustal xenoliths from the McBride province are carried in <3Ma basalts which erupt through the Proterozoic Georgetown Inlier. These xenoliths range from mafic through felsic compositions, with one possible metasedimentary type. All of these xenoliths have equilibrated at lower crustal conditions. The compositional diversity of the xenolith suites from different volcanic provinces in north Queensland suggests variations in lower crust composition, which may be related to the age and tectonic environment of the different crustal blocks from which they come. The mafic xenoliths from the Chudleigh province represent young additions to the crust through intrusion of basaltic magmas at depth. The mafic McBride province xenoliths may reflect a similar process, or, like the felsic xenoliths, may be fragments from an older lower crust, possibly associated with the Proterozoic rocks exposed at the surface.

Introduction

ing of mafic magmas may be an important lower crust forming process in these areas.

The composition of the lower crust is the least known parameter used in estimates of bulk crust composition. Geophysical surveys incorporating seismic reflection and refraction techniques provide information on the structure and bulk composition of the lower crust over wide areas, but are unable to provide petrologic details on crust forming processes. Crustal xenoliths carried in alkali basalts occur in continental areas of a variety of tectonic settings and ages and provide direct evidence on the characteristics of at least a portion of the lower crust in these areas. Studies show that many of these xenoliths throughout the world have mafic compositions (see compilation by Griffin and O'Reilly, 1986a), but metasedimentary lithologies are locally present (e.g., Massif Central, France [Leyreloup et al., 1977; Dupuy et al, 1979], Kilbourne Hole, New Mexico [Padovani and Carter, 1977]). Many mafic lower crustal xenoliths have trace element characteristics typical of basalts or cumulates therefrom (Rogers and Hawkesworth, 1982; Leyreloup et al., 1982; Arculus et al., 1985; Stosch etal., 1985), which suggests that underplat-

P-wave velocities obtained from seismic refraction profiles in eastern Australia range between 6.7 to 7.7 km/sec at 20 to 40 km depth, indicative of a mafic lower crust (Finlayson et al., 1980; 1984). This may be due to basaltic underplating near the crust-mantle boundary, which has been suggested to be a significant crust forming process in eastern Australia (Ewart et al, 1980; Griffin and O'Reilly, 1986b). Basaltic underplating is also supported by the mafic composition of many eastern Australian lower crustal xenoliths (Edwards et al., 1979; Wilkinson and Taylor, 1980; Kay and Kay, 1983; Wass and Hollis, 1983; Arculus et al, 1985; Griffin and O'Reilly, 1986b). Geochemical studies of Australian lower crustal xenoliths can help to define the processes involved in the generation of a mafic lower crust. Using trace element geochemistry, Wilkinson and Taylor (1980) concluded that the mafic, two pyroxene granulite xenoliths from Boomi Creek, New South Wales, represent oogenetic cumulates

190

Lower crustal xenoliths, northern Australia.

from a layered tholeiitic pluton at a depth of 25 km. Arculus et al., (1986) present major and trace element analyses for a variety of xenolith types from various localities in South Australia, Victoria and New South Wales and conclude that these xenoliths represent crystallized mafic melts, some with variable proportions of cumulate phases, which have been metamorphosed and variably contaminated by preexisting crust. Integrated studies of spatially associated xenolith suites are the best means of unraveling lower crustal history. Here we summarise petrographic, mineralogic and geochemical data for lower crustal xenoliths from two northern Queensland volcanic provinces in order to define the composition and origin of the lower crust in this region. 144

QUEENSLAND XENOLITHS

Geological Setting In northern Queensland, crustal xenoliths are found in many of the Cenozoic ( <3 Ma) basaltic lavas which cover an area of ~22,000 km (Stephenson and Griffin, 1976). These lavas consist mainly of alkali olivine basalts, hawaiites or basanites and have been divided into 5 geographic provinces: Nulla, Sturgeon, Chudleigh, McBride and Atherton (Fig 1). The crustal xenoliths described here are from two of these provinces: Chudleigh and McBride. The lavas in these provinces erupt through varied crustal types. The xenolith-bearing vents in the Chudleigh province (Batchelors Crater, Airstrip Crater and Sapphire Hill) crop out just south of the extension of the 146

19°

Fig. 1. Location of volcanic provinces (striped) and xenolith-bearing vents in north Queensland. Proterozoic rocks of the Georgetown Inlier shown in shading.

Geol. Soc. Aust. Spec. Publ. 17,189-208. surface expression of the Burdekin fault zone, separating the medium- to high-grade metamorphics of the Proterozoic Georgetown Inlier from the Phanerozoic lavas and sediments of the Broken River province of the Tasman fold belt. Basalts from Hill 32, in the McBride province, erupted through the centre of the Georgetown Inlier (Stephenson et al., 1980). These varied tectonic settings may be reflected in the crustal xenoliths, which show marked compositional differences between provinces. Crustal xenoliths of the Chudleigh province are exclusively mafic, whereas crustal xenoliths in the McBride province exhibit a range in compositions from mafic to felsic, with one metasedimentary type.

Previous Work Petrologic and geochemical investigations of crustal xenoliths from the Chudleigh and McBride provinces have been reported by Kay and Kay (1983) and O'Brien (1983). Kay and Kay (1983) report mineral analyses and whole rock chemistry

191

for mafic xenoliths from Hill 32 in the McBride province and Sapphire Crater and Batchelor Crater in the Chudleigh province. They consider the xenoliths to represent two basaltic suites: a quartz tholeiite suite (McBride province) and an olivine tholeiite suite (Chudleigh province). Kay and Kay (1983) interpret the complex textures observed in the Chudleigh xenoliths as reflecting two pressuretemperature histories: (1) igneous crystallisation at high crustal pressures (in spinel stability field) with subsequent isobaric cooling and recrystallisation, and (2) high-pressure crystallisation followed by tectonic uplift and recrystallisation. They interpret the McBride xenoliths as basaltic intrusions which crystallised at depth during the Tasman Orogeny. O'Brien (1983) reports REE analyses of xenoliths for which major element analyses are given by Kay and Kay (1983). The following sections summarise the mineralogy, petrology, thermobarometry and geochemistry of the Chudleigh and McBride province xenoliths. Details of the whole-rock major, trace element and

Fig. 2. Extended basalt tetrahedron showing Chudleigh province mafic xenoliths and several experimentally studied basalt compositions. Note lack of correlation between composition and presence of garnet.

Lower crustal xenoliths, northern Australia.

192

Mineralogy

Samples

Textures

Chudleigh Province Plagioclase-rich Ol-Sp-Opx-Cpx-Pc

106 107 109

Relict igneous textures: poikilitic Opx and lath-shaped Pc Coronas: 01 rimmed by Px-Sp symplectites P£ composition: labradorite

Sp-Opx-Cpx-Pc

112 126 127 139

Relict igneous textures: laree broken Pc? Coronas: Sp/Opx svmplectites rimmed bv Cpx. Pc comp: labradorite rimmed by andesine

Sp-Gar-Opx-Cpx-Pc

114 131 138

Relict igneous textures: None? Coronas: Sp rimmed bv Gar. Pc comp: labradorite or labradorite rimmed by andesine

Gar-Opx-Cpx-Pc rare Sp cores

117 133 140

Relict igneous textures: none. Meta textures: polygonal crystals. Pc comp: andesine or labradorite rimmed by andesine

Gar-Cpx-Pc rare Opx

125 BC

Relict igneous textures: none. Meta textures: polygonal crystals: euhedral Gar. Pc comp: andesine

Pyroxene-rich Cpx-0px-Pc-Rut±01

110 115

Relict igneous textures: possible cumulate layering Meta textures:polvgonal Pc crystals. Pc comp: labradorite

McBride Province Mafic Xenoliths Pc-Cpx-Gar-Amph ±Qtz Bio Opx Zir

158 159

Relict igneous textures: none. Meta textures: polygonal crystals.

Felsic Xenoliths Pc-Qtz-Gar ±Cpx Opx Kf Bio Ap Zir

160 162

Relict igneous textures: none. Meta textures: polygonal crystals

Metasedimentary Xenoliths Pc-Qtz-Gar-Opx-BioRut-Ap-Zir

157

Relict primary textures: layering? Meta textures: polygonal crystals.

Where Cpx = clinopyroxene, Opx = orthopyroxene, 01 = olivine, Pc = plagioclase, Kf = Kfeldspar, Sp = spinel, Gar = garnet, Amph = amphibole, Qtz = quartz, Bio = biotite, Rut = rutile, II = ilmenite, Ap = apatite, Zir = zircon.

Table 1 Summary of Mineralogy and Textures in Northern Queensland Xenoliths

Geol. Soc. Aust. Spec. Publ. 17,189-208.

isotope chemistry for the northern Queensland xenoliths are presented elsewhere (Rudnick et al., 1986). Mineralogy and Equilibration Conditions Chudleigh Province The mafic xenoliths from the Chudleigh province fall into three general categories: plagioclase-rich xenoliths, pyroxene-rich xenoliths and xenoliths with chemical and mineralogical features between these two types (transitional xenoliths). The mineralogy and textures of the xenoliths are generally complex, reflecting the whole-rock composition as well as the variable pressure and temperature conditions which these rocks experienced. Metamorphic textures are common, whereas relict igneous textures (i.e., large, poikilitic pyroxenes and tabular plagioclases) occur in only a few xenoliths. C02-rich fluid inclusions are present in all xenoliths in varying proportions. The plagioclase-rich xenoliths often contain coronas, reflecting incomplete mineral

§s

0 0 900

CD "2 § 0o

800

700

0 0

- /

• Gar-free Plagioclase-rich ° Gar-bearing • Pyroxene-rich

• <*° ^ 0

•

1 •

m • •

Many of the plagioclase-rich xenoliths have appropriate mineral assemblages for simultaneous determination of pressure and temperature of equilibration through cation exchange thermobarometry. Representative mineral analyses

? o° 800 -

The bulk-rock composition of the plagioclaserich xenoliths is veiy constant with respect to Si02 and AI2O3 contents (Table 2), and the presence of garnet does not appear to correlate with normative mineralogies, hence composition (Fig 2), as found for other Australian lower crustal xenoliths (Griffin and O'Reilly, 1986b). Therefore the large variation in mineralogy probably reflects differences in P-T conditions experienced by these rocks.

900

reactions associated with changes in T and/or P„ Table 1 summaries the mineral assemblages and coronal textures observed in the plagioclase-rich and pyroxene-rich xenoliths. Here the xenoliths are grouped according to mineralogy; the xenoliths with transitional chemical characteristics are grouped with the plagioclase xenoliths because of their high modal plagioclase content.

/ /

| Uncertainty

1 /

/ /

193

/ 0 A>

700

•

D '

kL—1——1 600

700

800

900

1000

600

700

800

900

1000

Lindsley and Anderson °C

•

Ellis and Green Or\

Fig. 3. Comparison of results from different cation exchange thermometers for Chudleigh province xenoliths. Diagonal line in each graph has slope = 1.0. Uncertainty box indicates those observed for 5 adjacent rim analyses in each rock.

Lower crustal xenoliths, northern Australia.

194

tween thermometers. The Lindsley and Anderson (1983) orthopyroxene thermometer yields a fairly narrow temperature range between 600 and 750°C for the plagioclase-rich xenoliths. The orthopyroxenes in these xenoliths generally have less than 6 wt. % non-quadrilateral components, making this thermometer suitable for use on these pyroxenes. As noted by Kay and Kay (1983), the Lindsley and Anderson clinopyroxene thermometer yields temperatures near 1000°C and these temperatures correlate with AI2O3 content of the clinopyroxene. For these reasons, the Lindsley and Anderson clinopyroxene temperatures are not reported here. The Lindsley and Anderson orthopyroxene temperatures fall below the Wells temperatures by 40-200°C. The Wells temperatures are highly variable between samples and do not appear to correlate with mineralogy. The Ellis and Green temperatures are also higher than the Lindsley and Anderson temperatures and generally overlap the Wells temperatures (Fig. 3). While the

are given in the Appendix along with analytical techniques. Table 3 lists the results from two pyroxene thermometry (both Wells [1977] and Lindsley and Anderson [1983] methods), garnetclinopyroxene thermometry (Ellis and Green, 1979) and garnet-orthopyroxene barometry (Harley and Green, 1982). In all cases, the estimates in Table 3 represent the range of pressures and temperatures calculated from analyses of 5 pairs of adjacent mineral rims for each sample. In many samples the garnets are altered to kelyphite (as described by Kay and Kay [1983]), with a glass forming the outermost rim. This glass is detectable by the presence of Na, K and/or P and the inability of the analysis to fit a garnet structure. In these samples the garnet analyses were made more towards the centre of the garnet and may not represent the outermost rim composition. The most striking feature of the calculated temperatures in Table 3 is the large variation be-

50.3 0.70 17.0 8.05 0.17 8.3 11.2 2.8 0.33 0.08

49.93 1.34 16.75 11.40 0.18 7.59 9.33 2.92 0.37 0.19

50.3 1.7 17.0 8.27 0.16 7.8 11.4 2.8 0.18 -

52.16 1.86 14.60 9.50 0.14 7.36 9.44 2.68 0.73 0.18

100.23

98.93

100.00

99.60

98.65

69.7

64.8

54.2

62.7

58.0

1 ~X

Si02 Ti02 M2O3 FeO* MnO MgO CaO Na20

K2O P2O5

50.29 0.42 19.69 7.69 0.13 9.31 9.42 2.97 0.23 0.08

0.56 0.56 0.78 1.12 0.02 2.08 0.84 0.72 0.08 0.04

TOTAL Mg*

1 = Average of 10 plagioclase-rich xenoliths from the Chudleigh volcanic province; 2 = Delegate 2 pyroxene lower crustal granulite xenolith (R698 of Irving, 1974); 3 = Olivine tholeiite (Ito and Kennedy, 1971); 4 = High-Al basalt (Green, 1967); 5 = Quartz tholeiite (Green and Ringwood, 1971). * Total Fe as FeO. Mg # = 100 (Mg/Mg + SFe)

Table 2 Comparison of plagioclase-rich xenolith compositions with experimentally studied basaltic compositions

195

Geol. Soc. Aust. Spec. Publ. 17,189-208.

Temperature (°C) Land A EandG Wells Chudleigh Province Plagioclase-rich

Sample

Pipe*

83-106 83-107 83-109 83-112 BC 83-114 83-117 83-125 83-126 83-127 83-131 83-133 83-136 83-137 83-138 83-139 83-140

AC AC AC AC BC BC BC BC SH SH SH SH SH SH SH SH SH

860-890 800-850 870-910 810-870

610-680 600-680 670-700 600-640

750-800 900-940 850-910 820-865 850-900 875-930 840-900 840-860 880-910 790-840 770-830 770-865

680-710 690-750 700-750 600-650 650-700 690-710 680-750 680-700 650-690 670-700 600-610 650-690

83-110 83-115

AC BC

870-910 875-930

83-158 83-159

H32 H32

865

83-160 83-162

H32 H32

915-930

83-157

H32

—

Pressure (GPa) P and N Hand G 2

—

950-1040 750-830 940-1000 910-930 — —

0.5-1.05 1.0-1.8 1.1-1.5

— — —

—

0.9-1.4 1.0-1.2

830-910 963

—

820-875 910-970

— — —

1.0-1.5

— —

—

845-855

7-11

—

940-1000 840-920

1.1-1.5

—

700-800

—

630$

4-16

8-10

Pyroxene-rich 750-820 800-880

McBride Province Mafic Xenoliths —

—

Felsic Xenoliths

Metasedimentary Xenoliths

—

--- means rock is missing appropriate assemblage for given thermometer/barometer. L and A = Lindsley and Anderson (1983); E and G = Ellis and Green (1979); H and G = Harley and Green (1983); P and N = Perkins and Newton (1981) orthopyroxene. In all thermometers and barometers in pyroxenes calculated from stoichiometry. *AC = Airstrip Crater, BC = Batchelors Crater, SH - Sapphire Hill, H32 = Hill 32. Assuming P = 1.0 GPa. -^Using Wells or Ellis and Green temperature, or Ferry and Spear (1978) thermometer for sample 83-157. tOnly one cpx grain, adjacent to garnet, in thin section. This sample yields Ferry and Spear, garnet-biotite temperature of 800°C. 1

Table 3 Thermobarometry of N. Queensland crustal xenoliths from adjacent mineral rim pairs

196

Lower crustal xenoliths, northern Australia.

Wells and Ellis and Green temperatures fall within error of a line of slope 1.0 in Fig. 3 (signifying the two thermometers yield the same temperature for a given sample), the Lindsley and Anderson temperatures are consistently lower than the Wells and Ellis and Green temperatures. In addition, if it is valid to assume that the presence of garnet is a function of the T and P of equilibration, then the garnet-bearing xenoliths would be expected to yield higher temperatures, if the xenoliths are equilibrated to the present-day geotherm. Temperatures determined by the Wells thermometer clearly do not show such correlations, whereas the Lindsley and Anderson temperatures do appear to be lower in the garnetfree samples, although the uncertainty is large (Fig 3). This may suggest that the temperatures T

Uncertainty 16

recorded by the thermometers reflect the "blocking temperature" for a given xenolith, i.e., the temperature below which the cation exchange between coexisting minerals was halted or greatly slowed (Harte et al., 1981). This temperature may be different for different xenoliths depending upon the presence or absence of a fluid and the stress field experienced by each rock. Certainly, the compositional zoning common in the pyroxenes and plagioclases indicates these minerals were not in equilibrium at the time of eruption. Given the above uncertainties associated with application of cation exchange thermobarometry, the best method of determining the relative level in the crust from which the xenoliths were plucked is T

: Wells, Ellis and

Green Thermometers

i [0

Harley and Green Barometer

Uncertainty: Lindsley and Anderson Thermometer

Harley and Green Barometer

400

600

800

TEMPERATURE (°C)

1000

1200

Fig. 4. P-T plot showing experimentally determined stability fields for Delegate xenolith (Irving, 1974) and calculated pressures and temperatures for gamet-2 pyroxene-plagioclase xenoliths from the Chudleigh province. Dotted line represents boundary between olivine-bearing 2 pyroxene-plagioclase rocks (labeled "01") and spinel-bearing 2 pyroxene-plagioclase rocks (labeled "sp") as determined by Herzberg (1978). Solid circles represent P and T calculated using Wells (1977) and Ellis and Green (1979) thermometers and Harley and Green (1982) barometer. Open circles represent P-T determinations for same set of rocks using Lindsley and Anderson (1983) 2 pyroxene temperatures and Harley and Green (1982) barometer. Rectangles at left show uncertainty in P-T determinations due to mineral compositional heterogeneity. Cpx = clinopyroxene, Opx = orthopyroxene, Sp = spinel, 01 = olivine, Pc = plagioclase, Gar = garnet.

Geol. Soc. Aust. Spec. Publ. 17,189-208. through comparison of their mineralogy with experimentally determined mineral stability fields. In columns 2-5 in Table 2 are listed some basaltic compositions for which experimental determinations of phase stabilities have been made. The high-alumina basalt (Green, 1967) and the Delegate two pyroxene granulite (Irving, 1974) closely resemble the Chudleigh province plagioclase-rich xenoliths in SiC>2, AI2O3 and Mg numbers. These experimental studies are therefore useful in defining the relative P and T conditions experienced by the xenoliths. Green's (1967) results are similar to those of Irving (1974) and are considered together here. With increasing pressure at constant temperature the stable mineral assemblages change from 2 pyroxene-olivineplagioclase, to 2 pyroxene-plagioclase, to garnet-2 pyroxene-plagioclase, and finally to garnetclinopyroxene-plagioclase. These experiments were run at or above 1000°C and the phase boundaries are linearly extrapolated to lower temperature conditions in Fig. 4. Figure 4 shows the stability fields and P-T estimates of Chudleigh 2 pyroxene-garnet (plagioclase) xenoliths calculated from cation exchange thermometers and barometers. Also shown in Fig 4 is the boundary separating olivine + plagioclase bearing rocks from spinel-2 pyroxene rocks (Herzberg, 1978) and a geotherm derived from xenolith studies in southeastern Australia (Griffin et al, 1984). The Chudleigh province plagioclase-rich xenoliths possess all the mineral assemblages shown in Fig. 4 (Table 1). Therefore, a first approximation to their P-T history is that they equilibrated within all the stability fields shown, i.e., they have come from variable depths. The P-T conditions experienced by the pyroxene-rich xenoliths cannot be evaluated using these experimental studies, due to their different compositions. The coronas present in many of the plagioclaserich xenoliths provide further insight into the changing P-T conditions experienced by these xenoliths. There are two types of coronas present: (1) olivine rimmed by orthopyroxene, which is in turn rimmed by symplectically intergrown pyroxene and spinel;

197

(2) spinel rimmed by garnet. Type 1 coronas reflect the reaction between olivine and plagioclase to form orthopyroxene, clinopyroxene and spinel (Kushiro and Yoder, 1966; Griffin, 1971; McBirney and Aoki, 1973; Herzberg, 1978). In this case, the rock has moved from the olivine-plagioclase-pyroxene field to the spinel-2 pyroxene-plagioclase field in Fig 4, either due to a decrease in temperature and/or an increase in pressure. The second type of corona reflects the reaction between spinel, orthopyroxene, clinopyroxene and plagioclase to form garnet (equation 9 of Lovering and White, 1969; Johnson and Essene, 1982). These rocks have moved from the spinel-2 pyroxene-plagioclase field to the garnet-clinopyroxene-plagioclase field in Fig 4, again, due to a temperature decrease and/or a pressure increase. Since all the mafic xenoliths in the Chudleigh province have trace element characteristics typical of cumulates (see next section), the simplest P-T history for these xenoliths is the following: intrusion of a mafic magma at variable depths in the crust (between 20 to 40 km, based on the phase assemblages) with subsequent crystallisation and crystal accumulation. These cumulates cooled isobarically, some passed outside their original mineral stability fields and partially reacted subsolidously to form the coronal textures. Cation exchange between adjacent minerals was halted or greatly slowed at different temperatures in different rocks, possibly depending upon the differences in stress field and fluid content of the rocks. This leads to compositionally zoned minerals, variable temperatures recorded by cation exchange thermometers for single samples and the apparent lack of correlation between mineral assemblages and calculated temperatures in rocks of the same composition. This scenario differs from that pictured by Kay and Kay (1983) in that they interpreted the symplectic spinels and pyroxenes as reflecting garnet breakdown due to a pressure decrease. Some of the xenoliths studied here contain olivine coronas, and the symplectic spinel and pyroxenes in these rocks are a product of the reaction between olivine and plagioclase; these coronas were not reported in the samples examined by Kay and Kay

Lower crustal xenoliths, northern Australia.

198

(1983). The presence of these symplectites suggests that the spinel-pyroxene symplectites may have formed by the isobaric olivine breakdown reaction (Fig 4), even in the samples where no olivine remains. Therefore, no pressure decrease (which only some of the xenoliths experienced) is required. Alternatively, the symplectic spinel and pyroxenes in olivine-free samples may have formed by the breakdown of a complex clinopyroxene to spinel, 2 pyroxenes and plagioclase which again, is caused by isobaric cooling (Lovering and White, 1969). McBride Province Fewer xenoliths have been investigated from the McBride province, but despite this, the compositional diversity is considerable. A summary of the mineralogy of the 5 samples is presented in Table 1. Table 3 contains thermobarometric determinations for these xenoliths. No coronal textures were observed in the Hill 32 xenoliths so clues to the P-T history must be obtained from mineral assemblages. All of the Hill 32 xenoliths examined here contain garnets, which, unlike the Chudleigh province xenoliths, are generally fresh, with only thin kelyphitic rims. High-grade mineral assemblages and decompression features (glass along grain boundaries and kelyphitic rims on garnets), which are present in all Hill 32 xenoliths, suggest these rocks were entrained by the host basalts at lower crustal depths. They do not represent fragments of the high-grade metamorphics which are exposed on the surface within the Georgetown Inlier. The mafic xenoliths of the McBride province have different mineralogical characteristics than the Chudleigh province xenoliths. First, hydrous phases are common in the McBride province xenoliths, and are rare in the Chudleigh province samples. Secondly, mineral chemistries in the McBride xenoliths are distinct from those of the Chudleigh province. Garnets and clinopyroxenes are generally more Fe-rich in the McBride province mafic xenoliths. These mineral compositional differences between provinces are due, in part, to whole-rock compositional differences between Chudleigh and McBride xenoliths (see below). Equilibration temperatures derived from the Ellis

Sm Eu

Mixed

—I

Xenoliths

Sm Eu

Fig. 5. Chondrite-normalized REE and Ba patterns for Chudleigh province (a) plagioclase-rich xenoliths, (b) pyroxene-rich xenoliths, and (c) transitional xenoliths.

Geol. Soc. Aust. Spec. Publ. 17,189-208. and Green (1979) thermometer range from 840 to 1000°C (Table 3). Only one of the felsic Hill 32 xenoliths (83-160) allows calculation of temperature from cation exchange thermometry. This sample yields two pyroxene temperatures between 915-930°C (Wells method) or 700-800°C (Lindsley and Anderson method), comparable to the equilibration temperatures for the mafic xenoliths. Sample 83-157, a possible metasediment (see below), contains orthopyroxene in equilibrium with garnet and biotite: a single clinopyroxene grain has been found adjacent to garnet. The estimated P-T conditions for this sample show considerable range from 630°C, 0.4 GPa to 800°C, 1.6 GPa (Table 3). However, the very low temperature yielded by the garnetclinopyroxene thermometer may be unrealistic, if the rare occurrence of clinopyroxene indicates it is

199

not in equilibrium. Because of the temperature dependence of the Harley and Green (1982) barometer, the Perkins and Newton (1981) barometer may provide the best pressure estimate for this sample. Therefore, this sample has equilibrated near the base of the crust, at about 800°C, 1.0 GPa. Since most of the Hill 32 samples do not contain the appropriate assemblages for cation exchange pressure determinations, and because no relevant experimental investigations exist for the felsic lithologies, another means of estimating pressures of equilibration is required. This may be done by relating the equilibration temperatures to an inferred geotherm. The Chudleigh province samples yield P and T's that fall slightly lower than, but within error of, the relatively steep eastern Australian geotherm proposed by Griffin et al.

Fig. 6. Calculated primitive mantle-normalized trace element concentrations for melt coexisting with plagioclase-rich xenoliths 83-107 (circles) and 83-127 (squares). Inset: Chondrite-normalized REE patterns for melt coexisting with plagioclase-rich xenoliths. Melt compositions calculated by multiplying bulk partition coefficients (calculated using normative mineralogies) by whole rock composition. Open symbols represent melts calculated assuming interstitial melt present, solid symbols represent melts calculated assuming no interstitial melt. Trapped melt is treated as phase with D = 1.0 and is assumed to be less than 2% in both rocks (Rudnick et al., 1986). Source of partition coefficients: Schnetzler and Philpotts, 1970; Irving, 1978; Dunn and McCallum, 1982; Fujimaki et al., 1984. Primitive mantle values from McDonough et al., (1985).

200

Lower crustal xenoliths, northern Australia.

(1984) (Fig 4). Using this geotherm, then, the two mafic McBride xenoliths have equilibrated between 0.8 to 1.4 GPa. This is consistent with the garnet-in line of olivine tholeiite studied by Ito and Kennedy (1970). By the same reasoning, the felsic xenolith has equilibrated between 1.0 to 1.1 GPa (using the Wells (1977) temperature estimates). These pressure estimates represent minima if the geotherm is actually lower than that of Griffin et al, (1984). Geochemical Features and Inferred Origin Chudleigh Province Major and trace element and isotopic results are presented in Rudnick et al. (1986) and only a summary of their findings are reported here. The three classes of xenoliths are characterised by major and trace element contents that reflect their modal and normative mineralogies. Plagioclase-rich xenoliths have high AI2O3 and low Ti02 contents at nearly constant SiC>2. REE (rare earth elements) patterns of these rocks show low overall abundances with LREE (Light REE)-enrichments and large positive Eu anomalies (Fig 5a). Pyroxene-rich xenoliths have lower AI2O3 and higher Ti02 contents than plagioclase-rich xenoliths, at similar Si02 contents. REE patterns of these rocks are LREE depleted at about 10 times chondrite (Fig 5b). The transitional xenoliths have major and trace element compositions transitional between the two end member types; REE patterns are flat to LREE depleted with positive Eu anomalies (Fig 5c). Correlations between Mg# (100Mg/Mg+ EFeO), an indicator of differentiation, and compatible and incompatible trace elements suggest the xenoliths are genetically related to one another. Correlations between AI2O3 and Ti, V, Zr, Hf and Y reflect the original proportions of pyroxene to plagioclase in the rocks. The variable Sr and Nd isotopic compositions of the xenoliths are distinct from those of the host basalts, excluding a genetic relationship between the two, and good correlations between isotopic composition and Mg# indicate open system behaviour. These xenoliths are interpreted to represent a genetically related suite of cumulates, crystallised from a basaltic melt which evolved through simultaneous fractionation and

assimilation of more enriched crustal materials (Rudnick et al., 1986). It is possible to use the trace element chemistry, along with published partition coefficients, to set limits on the chemical characteristics of the melts which were in equilibrium with these cumulates. Figure 6 presents the coexisting melt's primitive mantle normalized trace element characteristics and REE patterns. Although the calculated compositions vary depending on whether trapped melt is assumed, several conclusions can be drawn. The coexisting melt had LREE enriched over HREE (heavy REE) and a (Ba/La)N ratio greater than 1.0. La/Nb ratio, which is suggested to be a good discriminant of alkaline versus tholeiitic lavas (Thompson et al., 1984), cannot be used as such in this case. The La/Nb ratio is both greater and less than 1.0, depending on whether trapped melt is McBRlDE

PROVINCE

XENOLITHS

Fig. 7. Chondrite-normalized REE and Ba patterns for McBride province xenoliths. Top, possible metasedimentary xenolith, 83-157 (open squares) and calc-alkaline xenolith, 83-160 (solid squares). Bottom, mafic xenoliths 83-158 (open circles) and 83-159 (solid circles).

Geol. Soc. Aust. Spec. Publ. 17,189-208. assumed. The LREE enrichment, (Ba/La)N ratio >1.0 are characteristics of basaltic to andesitic magmas from island arcs (Perfit et al 1980; Gill, 1981; Thompson et al., 1984) or tholeiitic and alkaline basaltsfromcontinental environments (Thompson et al., 1984; Dupuy and Dostal, 1984) and are unlike basalts from mid-ocean ridges. A Sr enrichment and strong Nb depletion, characteristics of island arc tholeiites, are not present in these coexisting melts, suggesting the melts were not related to island arc volcanism. Therefore, the xenoliths probably represent cumulates from intraplate continental alkalic or tholeiitic igneous activity. y

Because of the highly variable Sm/Nd ratios of the xenoliths, the present-day correlation between isotopic composition and major and trace elements scatter as the isotopic ratios are back-calculated to earlier times. This is interpreted to mean that the xenoliths are relatively young (less than 100 Ma) and probably represent deep-seated intrusions associated with the Cenozoic igneous activity in eastern Australia (Rudnick et al., 1986). McBride Province The McBride province xenoliths are chemically distinct from those of the Chudleigh province. The two mafic xenoliths have lower Si02 and lower normative feldspar contents than the plagioclaserich Chudleigh province xenoliths. REE patterns of these xenoliths are shown in Fig 7. Sample 83-159 has high AI2O3 and a positive Eu anomaly, features common with the plagioclase-rich xenoliths from the Chudleigh province. This sample may therefore represent a cumulate or restite; although the concentrations of incompatible trace elements in this sample are about an order of magnitude higher than in the Chudleigh xenoliths. Sample 83-158 has a LREE depleted pattern, similar to the Chudleigh province pyroxene-rich xenoliths. However, the AI2O3 content of this sample (14.7%) is distinctly higher than that of the Chudleigh pyroxene-rich xenoliths and other pyroxenites thought to have formed by cumulate processes (Frey, 1980). Other major element ratios in this sample are similar to that of mid-ocean ridge basalts (MORB) (i.e., Mg#, Ca0/Ti02 and Al203/Ti02)(Sun et al1979; Basaltic Volcanism

201

Study Project, 1981). However, CaO and Na20 contents are higher and lower, respecitvely, than those observed in MORB. Trace element concentrations and ratios in 83-158 are nearly identical to those of MORB (e.g., Zr/Nb, La/Sm), but Sr/ Sr (0.7052) is distinctly higher than Sr/ Sr of MORB. Therefore, this sample may represent a melt derivedfroma MORB-like source, but with a higher Rb/Sr, or may represent a pyroxene cumulate with significant interstitial melt. REE patterns of the intermediate and felsic xenoliths are also shown in Fig 7a, along with their Si02 contents. Sample 83-157 has high AI2O3 (19.8 wt. %) and may be metasedimentary. It possesses the typical REE pattern of many post-Archean shales: LREE-enriched, with a negative Eu anomaly (Taylor and McLennan, 1981), although it has high Ti02 and Na20 contents for a shale. The two more felsic samples are LREE-enriched with a slight negative to no Eu anomaly (Fig 7b), similar to calc-alkaline rocks of all ages. The large compositional range in this small number of samples from Hill 32 implies a very heterogeneous lower crust in this region, with no apparent genetic link between the individual xenoliths. 87

Inferences on Lower Crustal Composition The crustal xenoliths from the Chudleigh and McBride volcanic provinces of northern Queensland have all equilibrated at lower crustal pressures and temperatures. Thus, the lower crust in this region is very heterogeneous, with the composition correlating with geographic position. Crustal xenoliths from the Chudleigh province are all mafic cumulates which were emplaced between 20 and 45 km depth. However, this does not require mafic rocks make up such a great thickness of the crust in this region. The basalts which contain these xenoliths erupted very near the projected extension of the Burdekin fault zone (Fig 1), which separates the rocks of the Tasman fold belt from those of the Georgetown Inlier. Therefore, these xenoliths may be intimately associated with this fault zone, i.e., if this structure represents a longlived crustal zone of weakness through which basaltic magmas could rise, the Chudleigh province xenoliths may represent mafic cumulates plated out

202

Lower crustal xenoliths, northern Australia.

along this fracture zone during an earlier stage of the Cenozoic basaltic activity. In addition, the presence of intermediate to felsic lithologies in the mid to lower crust is implied by the isotopic results, which require assimilation of evolved crustal material by the magma parental to the Chudleigh cumulates. Laboratory density and elastic wave velocity measurements are currently underway on the xenolith suite and preliminary results indicate a their densities are close to 2.8 to 3.0 g/cm and their compressional wave velocities vary from 6.75 to 7.6 km/sec at room temperature (Rudnick and Jackson, in prep). Such high Vp material is similar to that observed at similar depths in other regions of the Tasman fold belt (Finlayson et al., 1980; Finlayson et al., 1984). A reconnaissance gravity survey in this part of north Queensland shows a Bouguer gravity anomaly of up to 15 mGal amplitude (calculated with a Bouguer density of 2.2 km/m ) situated directly over the Chudleigh and McBride basaltic provinces (Shirley, 1979). This gravity data has been attributed to a slightly thicker crust in this region (Moho depth = 42.6 km versus a Moho depth of 40 km in the surrounding regions) (Shirley, 1979). This, coupled with the inference that the xenoliths represent young additions to the crust, and the interpretations of seismic refraction profiles from other parts of eastern Australia, suggests that basaltic underplating is responsible for the increased crustal thickness beneath the Cenozoic volcanics of north Queensland. The xenoliths from the McBride province range from felsic to mafic compositions and have equilibrated near 1.0 GPa. Decompression features in the intermediate to felsic xenoliths suggest they represent fragments of the present day lower crust. The marked compositional diversity of the McBride xenolith suite indicates a heterogeneous lower crust in this region, which may have formed from a variety of processes (e.g., underthrusting and subsequent metamorphism of supracrustal material, basaltic underplating and/or intracrustal melting) operating at different times since the Proterozoic. 3

Acknowledgements Several people provided enlightening discussion on various aspects of this work: D.J. Ellis, W.L. Griffin, H. O'Neill and S.Y. O'Reilly on thermal histories; W.F. McDonough and S.-s. Sun on geochemistry and K.L. Gallagher on geophysical data. A.J. Irving, A.E. Ringwood, H.-G. Stosch, S.-s. Sun provided constructive comments on the manuscript. We gratefully acknowledge the assistance and expertise of Nick Ware in obtaining microprobe analyses and Bruce Chappell in obtaining XRF trace element analyses.

Appendix Microprobe Techniques and Mineral Chemistries All mineral analyses were obtained by energydispersive X-ray analysis using a Technisch Physische Dienst (TPD) electron probe at the Australian National University, with an accelerating voltage of 15 kV and beam current of 3 nA. Data reduction was performed using peak integration with background subtraction and ZAF corrections as outlined in Ware (1981). Fe+ in the pyroxenes was calculated using the charge balance method (Papike et al., 1974). All analyses reported here represent coexisting mineral rim analyses; P and T calculations presented in the text represent the range produced from five sets of adjacent mineral rim analyses in each sample. The mineral compositions given in tables 4-8 were selected as falling towards the middle of the range of compositions for given mineral rims in each sample. 3

Note added in proof: Since the writing of this paper in 1985, many papers on the north Queensland xenoliths have appeared. The reader is referred to these for a more up-to-date review of the topic: Rudnick & Taylor (1987); Rudnick & Williams (1987), Stolz & Davies (1989), Rudnick & Goldstein (1990) - full citations are given in the reference list.

203

Geol. Soc. Aust. Spec. Publ. 17,189-208.

CLINOPYROXENES BC 83-107 83-125 83-126 83-127 83-131 83-140 52.08 49.53 51.84 52.45 51.34 52.39 51.73 Si0 0.84 0.76 0.44 0.23 0.81 0.37 0.33 TiC>2 9.16 5.93 6.56 5.59 5.61 4.92 5.47 AI O 2.89 7.00 4.54 4.14 5.26 4.90 6.54 FeO* 13.23 12.52 14.14 13.43 14.58 14.70 13.49 MgO 19.45 21.41 21.50 22.10 21.98 22.23 22.14 CaO 2.35 1.26 0.98 0.88 0.90 0.57 0.93 Na 0 TOTAL 100.63 100.08 99.30 99.99 100.00 99.42 100.00 Number of Ions on the basis of 6 oxygens 1.887 1.865 1.873 1.895 1.912 1.894 1.909 0.113 Si 0.135 0.127 0.091 Al(rv) 0.105 0.088 0.106 0.128 0.261 0.131 0.124 0.138 0.149 0.394 A1(VI) 0.022 0.023 0.012 0.006 0.022 0.010 0.009 TI 0.703 0.709 0.737 0.799 0.738 0.791 0.767 Mg 0.220 0.087 0.138 0.126 0.200 0.150 0.162 Fe* 0.863 0.749 0.839 0.862 0.869 0.869 0.869 Ca 0.066 0.040 0.063 0.063 0.069 0.092 0.164 Na TOTAL 4.011 3.992 3.993 3.997 3.994 4.009 3.992 ORTHOPYROXENES 52.42 55.21 53.46 53.70 53.80 51.91 Si0 3.40 4.90 3.99 3.83 3.76 3.54 A1 0 11.64 17.79 FeO* 17.78 12.30 15.60 12.87 0.16 0.30 MnO 29.12 29.52 25.81 MgO 25.42 30.17 26.40 0.38 0.45 0.32 0.23 0.42 0.32 CaO TOTAL 100.00 101.63 99.53 100.00 100.00 99.44 Number of ions on the basis of 6 oxygens 1.902 1.908 1.922 1.928 1.907 1.899 0.098 Si 0.092 0.078 0.072 0.093 0.101 0.049 Al(rv) 0.103 0.074 0.091 0.069 0.067 A1(VI) 1.379 1.545 1.419 1.541 1.537 1.410 Mg 0.541 0.358 0.470 0.382 0.343 0.545 Fe* 0.005 0.009 Mn 0.012 0.016 0.009 0.012 0.017 0.015 Ca TOTAL 4.011 4.006 3.990 4.010 4.000 4.024 2

—

83-110 49.41 0.76 7.79 6.87 12.82 21.00 0.97 99.77 1.828 0.172 0.168 0.021 0.707 0.213 0.833 0.070 4.016 51.63 5.45 16.37 0.25 25.07 0.61 99.38 1.880 0.120 0.114 1.360 0.498 0.008 0.024 4.003

* Total Fe as FeO - Not detected

Table 4 Representative pyroxene rim analyses for selected Chudleigh Province xenoliths

204

Lower crustal xenoliths, northern Australia.

Si0 TiOz AI2O3 FeO* MnO MgO CaO Na 0 TOTAL

83-157 52.65 3.28 4.04 15.00 23.30 0.24 98.51

Si Al(rv) Al(VI) Ti Mg Fe* Mn Ca Na TOTAL

1.950 0.050 0.093 0.828 0.125 0.924 0.017 3.987

* Total Fe as FeO - Not detected

—

ORTHOPYROXENES CLINOPYROXENES 83-157 83-159 83-160 83-160 83-158 54.98 51.91 51.30 49.60 49.71 0.16 0.57 1.01 2.34 3.33 1.79 6.36 7.31 17.53 7.07 9.66 23.68 8.20 0.19 26.86 11.65 12.22 12.83 11.75 0.26 20.87 0.78 20.76 21.91 0.45 1.37 0.65 99.54 101.42 99.70 100.00 98.48 Number of ions on the basis of 6 oxygens 1.962 1.926 1.949 1.842 1.867 0.074 0.038 0.051 0.158 0.133 0.037 0.052 0.162 0.073 0.148 0.028 0.057 0.005 1.429 0.684 0.718 1.165 0.650 0.744 0.303 0.523 0.255 0.222 0.006 0.839 0.010 0.031 0.826 0.882 0.033 0.099 0.047 4.000 3.998 3.999 4.020 4.000 -

—

- -

Table 5 Representative pyroxene rim analyses for selected McBride Province xenoliths

CHUDLEIGH PROVINCE 83-131 83-140 BC 83-157 40.96 39.47 41.61 39.40 22.85 22.18 22.15 23.57 15.98 21.74 11.20 21.93 0.21 0.61 0.27 0.33 14.90 11.29 11.67 18.13 5.44 5.60 3.81 5.33 100.34 100.86 100.11 99.31 Number of ions on the basis of 24 oxygens 5.998 5.934 5.972 5.981 Si 5.896 0.002 Al(rv) 0.104 0.066 0.019 0.028 3.941 Al(vi) 3.935 3.860 3.949 3.959 1.956 2.734 2.783 Fe 1.895 1.345 0.027 0.042 Mn 0.026 0.078 0.033 Mg 3.253 2.529 2.640 3.376 3.879 Ca 0.852 0.853 0.901 0.820 0.619 TOTAL 16.084 16.030 16.103 16.035 16.035

83-125 41.16 Si0 23.92 A1 0 15.81 FeO* 0.22 MnO 15.81 MgO CaO 5.55 TOTAL 102.47 2

* Total Fe as FeO

™ Not detected

HILL 32 83-158 83-159 40.24 39.41 22.62 22.06 20.41 21.52 0.57 0.48 11.80 9.84 5.99 7.33 101.63 100.66 5.961 0.039 3.909 2.528 0.072 2.606 0.950 16.065

5.958 0.042 3.889 2.721 0.062 2.217 1.188 16.077

Table 6 Representative garnet analyses for selected xenoliths

205

Geol. Soc. Aust. Spec. Publ. 17,189-208.

83-107

83-125

83-126

83-127

83-131

83-140

83-110

Si0 2 AI2O3 CaO Na 2 0 K2O

53.17 29.99 11.89 4.66 0.26

58.16 26.37 8.51 6.44 0.51

56.21 27.53 9.28 5.96 0.34

58.39 26.32 7.89 7.05 0.35

58.66 26.26 7.72 6.89 0.48

59.84 25.29 6.93 7.25 0.69

59.70 26.19 7.50 7.06 0.07

55.24 28.45 10.49 5.47 0.34

TOTAL

99.97

100.00

99.33

100.00

100.53

99.98

Number of ions on the basis of 32 oxygens Si A1 Ca Na K

9.625 6.397 2.308 1.634 0.058

10.418 5.568 1.633 2.238 0.118

10.160 5.865 1.798 2.090 0.078

10.449 5.551 1.512 2.446 0.080

10.487 5.533 1.478 2.388 0.109

10.681 5.319 1.326 2.508 0.157

10.577 5.468 1.424 2.424 0.016

9.957 6.044 2.026 1.910 0.078

TOTAL

20.022

19.975

19.991

20.038

19.995

19.991

19.909

20.015

Table 7 Representative plagioclase rim analyses for selected Chudleigh Province xenoliths

83-157

83-158

83-159

83-160

Si02 A1203 CaO Na 2 0 K2O

56.34 27.76 9.99 5.45 0.45

59.83 25.55 7.20 6.97 0.74

56.63 28.27 10.16 5.66 0.11

59.72 25.78 6.87 7.18 0.45

TOTAL

100.00

100.28

100.83

100.00

Number of ions on the basis of 32 oxygens Si A1 Ca Na K TOTAL

10.127 5.880 1.924 1.900 0.104 19.935

10.650 5.360 1.372 2.406 0.169 19.957

10.086 5.934 1.938 1.954 0.025

10.640 5.412 1.311 2.481 0.102

19.937

19.945

Table 8 Representative plagioclase rim analyses for selected McBride Province xenoliths

206

Lower crustal xenoliths, northern Australia.

References ARCULUS R.J., FERGUSON J., CHAPPELL B.W., SMITH D., MCCULLOCH M.T., JACKSON I., HENSEL H.D., TAYLOR S.R. KNUTSON J. & GUST DA. 1988. Eclogited and

granulites in the lower continental crust: examples from eastern Australia and southwestern U.S.A. In Smith D.C. ed. Eclogites and Related Rocks. Elsevier 335-386.

Basaltic Volcanism Study Project (1981) Basaltic Volcanism on the Terrestrial Planets. Pergamon Press, Inc., New York. 1286 pp. DUNN T. & MCCALLUM I.S. 1982. The partitioning of Zr

and Nb between diopside and melts in the system diopside-albite-anorthite. Geochimica Cosmochimica Acta 46, 623-629. Trace element geochemistry of some continental tholeiites. Earth and Planetary Science Letters 67, 61-69. DUPUY C. & DOSTAL J. 1984.

DUPUY C., LEYRELOUP A. & VERNIERES J. 1979. The

lower continental crust of the Massif Central (Bournac, France) - with special references to REE, U and Th composition, evolution, heat-flow production. In Ahrens L.H. ed. Origin and Distribution of the Elements Vol. 2, Pergammon Press. EDWARDS A.C., LOVERING J.F. & FERGUSON J. 1979.

High pressure basic inclusions from the Kayrunnera kimberlitic diatreme in New South Wales, Australia. Contributions to Mineralogy & Petrology 69, 185-192.

ELLIS D.J. & GREEN D.H. 1979. An experimental study

of the effect of Ca upon garnet-clinpyroxene Fe-Mg exchange equilibria. Contributions to Mineralogy & Petrology 71, 13-22. EWART A., BAXTER K. & Ross J.A. 1980. The petrology and pedogenesis of the Tertirary anorogenic mafic lavas of southern and central Queensland, Australia - possible implications for crustal thickening. Contributions to Mineralogy & Petrology 7 5 , 1 2 9 - 1 5 2 . FERRY J.M. & SPEAR F.S. 1978. Experimental calibration

of the partitioning of Fe and Mg between biotite and garnet. Contributions to Mineralogy & Petrology 66, 113-117. FINLAYSON D.M., COLLINS C.D.N. & DENHAM D. 1980.

Crustal structure under the Lachlan fold belt, southeastern Australia. Physics of the Earth and Planetary Interiors 21, 321-342.

FINLAYSON D.M., COLLINS C.D.N. & LOCK J. 1984.

P-wave velocity features of the lithosphere under the Eromanga Basin, eastern Australia, including a prominent mid-crustal (Conrad?) discontinuity. Tectonophysics 101, 267-291. FREY F.A. 1980. The origin of pyroxenites and garnet pyroxenites from Salt Lake Crater, Oahu, Hawaii: trace element evidence. American Journal of Science 280A, 427-449. FUJIMAKI H., TATSUMOTO M . & AOKI K-I. 1984.

Partition coefficients of Hf, Zr, and REE between phenocrysts and groundmasses. Proceedings of the Fourteenth Lunar and Planetary Science Conference, Part 2. Journal ofGeophysical Research 89, B662-B672. GILL J. 1981. Orogenic Andesites and Plate Tectonics.

Springer-Verlag, Berlin, 390p.

GREEN D.H. & RINGWOOD A.E. 1972. A comparison of

recent experimental data on the gabbro-garnet granulite-eclogite transition. Journal of Geology 80, 277-288. GREEN T.H. 1967. An experimental investigation of sub-solidus assemblages formed at high pressure in high-alumina basalt, kyanite eclogite and grosspydite compositions. Contributions to Mineralogy & Petrology 16, 84-114. GRIFFIN W.L. 1971. Genesis of coronas in anorthosites

of the Upper Jotun Nappe, Indre Sogn, Norway. Journal of Petrology 12, 219-243.

GRIFFIN W.L. & O'REILLY S.Y. 1986a. The composition

of the lower crust and the nature of the continental Moho - xenolith evidence. In Nixon P.H. ed. Mantle Xenoliths, Elsevier, Amsterdam, 413-430. GRIFFIN W.L. & O'REILLY S.Y. 1986b. The lower crust

in eastern Australia. In : Dawson, J.B., et al., eds. The Nature of the Lower Continental Crust. Journal of the Geological Society of London Special Publication 24, 363-374. GRIFFIN W.L., WASS S.Y. & HOLLIS J.D. 1984.

Ultramafic xenoliths from Bullenmerri and Gnotuk maars, Victoria, Australia: petrology of a sub-continental crust-mantle transition. Journal of Petrology 25, 53-87. HARLEY

S.L.

GREEN

D.H.

1982.

Garnet-orthopyroxene barometry for granulites and peridotites. Nature 300, 697-701.

207

Geol. Soc. Aust. Spec. Publ. 17,189-208.

HARTE B.JACKSON P.M. &MACINTYRER.M. 1981. A g e

LINDSLEYD.H.& ANDERSON D . J . 1 9 8 3 . Atwo-pyroxene

of mineral equilibria in granulite facies nodules from kimberlites. Nature 291,147-148.

thermometer. Proceedings of the Thirteenth Lunar and Planetary Science Conference Part 2. Journal of Geophysical Research 8 8 Supplement, A 8 8 7 - A 9 0 6 .

HERZBERG C.T. 1978.

Pyroxene geothermometry and geobarometry: experimental and thermodynamic evaluation of some subsolidus phase relations involving pyroxenes in the system Ca0-Mg0-Al203-Si02. Geochemica Cosmochimica Acta 42, 9 4 5 - 9 5 7 .

Granulitic and eclogitic inclusions from basic pipes at Delegate, Australia. Contributions to Mineralogy and Petrology 21, 9 - 5 2 .

LOVERING J.F. & WHITE A . J . R . 1 9 6 9 .

IRVING A.J. 1 9 7 4 . Geochemical and high pressure experimental studies of garnet pyroxenite and pyroxene granulite xenoliths from the Delegate basaltic pipes, Australia. Journal of Petrology 15, 1-40.

Factors governing the stability of plagioclase at high pressures as shown by spinel-gabbro xenoliths from the Kerguelen archipelago. American Mineralogist, 58, 271-276.

A review of experimental studies of crystal/liquid trace element partitioning. Geochimica Cosmochimica Acta 42,743-770.

MCDONOUGH W.F. MCCULLOCH M . T . & SUN, S.-S. 1 9 8 5 .

IRVING A . J . 1978.

ITO K . & KENNEDY G.C. 1971.

An experimental study of the basalt-garnet granulite-eclogite transition. In Heacock J.G. ed. The structure and physical properties of the earth's crust. American Geophysical Union, Washington. Geophysical Monograph 14, 3 0 3 - 3 1 4 . The formation of garnet in olivine-bearing metagabbros from the Adirondacks. Contributions to Mineralogy and Petrology SI, 240-251.

MCBIRNEY A . R . & AOKI K-I. 1973.

Isotopic and geochemical systematics of Tertiary-Recent basalts from southeastern Australia and implications for the evolution of the sub-continental lithosphere. Geochimica Cosmochimica Acta 49, 2051-2068. O'BRIEN T.F. 1983. Evidence for the nature of the lower

crust beneath the central Colorado Plateau as derived from xenoliths in the Buell Park-Green Knobs diatremes. PhD Thesis, Cornell University (unpublished), 205p.

JOHNSON C.A. & ESSENE E . J . 1982.

KAY S.M. & KAY R.W. 1983. Thermal history of the deep crust inferred from granulite xenoliths, Queensland, Australia. American Journal ofScience 283-A, 486-513. KUSHIRO 1. & YODER H . S . Jr. 1966.

Anorthite-forsterite and anorthite-enstatite reactions and their bearing on the basalt-eclogite transformation. Journal of Petrology 7,

PADOVANIE.R.& CARTER J.L. 1977. Aspects of the deep

crustal evolution beneath south central New Mexico. In Heacock J.G. ed. The Earth's Crust: Its nature and physical properties. American Geophysical Union, Washington. Geophysical Monograph 20, 19-55. PAPIKE J.J., CAMERON K . L . &

BALDWIN K .

1974.

Amphiboles and pyroxenes: characterization of other than quadrilateral components and estimates of ferric iron from micropribe data. Geolical Society of America, Abstract Progs 6, 1053-1054.

337-362. PERFIT M . R . , GUST D . A . , BENCE A . E . , ARCULUS R.J. & LEYRELOUP A., DUPUY C . & ANDRIAMBOLOLONA R .

1977. Catazonal xenoliths in French Neogene volcanic rocks: constitution of the lower crust. 2. Chemical composition and consequences of the evolution of the French Massif Central Precambrian crust. Contributions to Mineralogy and Petrology 62, 283-300. LEYRELOUP A., BODINIER J.L. DUPUY, C . & DOSTAL J.

1982. Petrology and geochemistry of granulite xenoliths form central Hoggar (Algeria) - implications for the lower crust. Contributions to Mineralogy and Petrology 79, 68-75.

S.R. 1980. Chemical characteristics of island-arc basalts: implications for mantle sources. Chem. Geol 30, 227-256.

TAYLOR

D. in & NEWTON R.C. 1981. Chamockite geobarometers based on coexisting garnet-pyroxene-plagioclase-quartz. Nature 292, 144-146.

PERKINS

Proterozoic age and cumulate origin for granulite xenoliths, Lesotho. Nature 299,409-413. ROGERS N . W . & HAWKESWORTH C . J . 1982.

208

Lower crustal xenoliths, northern Australia.

RUDNICK R.L., MCDONOUGH W.F., MCCULLOCH M.T. & TAYLOR S.R. 1986. Lower crustal xenoliths from

Queensland, Australia: evidence for deep crustal assimilation and fractionation of continental basalts. Geochimica Cosmochimica Acta 50,1099-1115.

RUDNICK R.L. & TAYLOR S.R. 1987. The composition and pedogenesis of the lower crust: a xenolith study. Journal of Geophysical Research 92,13981 -14005. RUDNICK R.L. & WILLIAMS I.S. 1987. Dating the lower

crust by ion microprobe. Earth Planetary Science Letters

85, 145-161.

RUDNICK R.L. & GOLDSTEIN S.L. 1990. The Pb isotopic

compositions of lower crustal xenoliths and the evolution of lower crustal Pb. Earth and Planetary Science Letters 98, 192-207. RUDNICK R.L. 1990. Nd and Sr isotropic compositions of lower crustal xenoliths from north Queensland, Australia: Implications for Nd model ages and crustal growth processes. Chemical Geology 83 (in press).

Partition coefficients of rare-earth elements between igneous matrix material and rock-forming mineral phenocrysts — II. Geochimica Cosmochimica Acta 34, 331 -340. SCHNETZLER C.C. & PHILPOTTS J.A. 1970.

SHIRLEY J.E. 1979. Crustal structure of north Queensland from gravity anomalies. BMR Journal of Australian Geology & Geophysics 4, 309-321. STEPHENSON P.J. & GRIFFITH T.J. 1976. Cainozoic volcanicity north Queensland. Excursion Guide No. 7A, 25th International Geological Congress, 39p. STEPHENSON P.J. GRIFFIN, T.J. & SUTHERLAND F.L. 1980.

Cainozoic volcanism in Northeastern Australia. In Henderson R.A. and Stephenson P.J. eds. The Geology and Geophysics of Northeastern Australia. Geological Society of Australia, Queensland Division, 349-374.

STOLZ A.J. & DAVIES G. R. 1989. Metasomatized lower crust and upper mantle xenoliths from north Queensland: chemical and isotopic evidence bearing on the composition and source of the fluid phase. Geochimica Cosmochima Acta 53, 649-660.

STOSCH H.-G., LUGMAIR G.W. & SECK H.A. 1985. Geochemistry of granulite facies lower crustal xenoliths: implications for the geological history of the lower continental crust underneath the Eifel, West Germany. Journal of the Geological Society ofLondon. In Dawson J.B., et al, Ids. The Nature of the Lower Crust: Special Publication 24: 331-350.. SUN S.-S., NESBITT R.W. & SHARASKIN A.Y. 1979.

Geochemical characteristic of mid-ocean ridge basalts. Earth and Planetary Science Letters 4 4 , 1 1 9 - 1 3 8 . TAYLOR S.R. & MCLENNAN S.M. 1981.

The

composition and evolution of the continental crust: rare earth element evidence from sedimentary rocks. Philosophical Transactions of the Royal Society of London A301, 381-399. THOMPSON R.N., MORRISON M.A., HENDRY G.L. &

PARRY S.J. 1984. An assessment of the relative roles of crust and mantle in magma genesis: an elemental approach. Philosophical Transactions of the Royal Society of London A310, 549-590.

WARE N.G. 1981. Computer programs and calibration

with the PIBs technique for quantitative electron probe analysis using a lithium-drifted silicon detector. Computers and Geosciences 7, 167-184.

Crustal growth in southeastern Australia ~ evidence from lower crustal eclogite and granulitic xenoliths. Journal of Metamorphic Geology 1, 25-45. WASS S.Y. & HOLLIS J.D. 1983.

WELLS P.R.A. 1977. Pyroxene thermometry in simple and complex systems. Contributions to Mineralogy and Petrol 62, 129-139. WILKINSON J.F.G. & TAYLOR S.R. 1980. Trace element fractionation trends of tholeiitic magma at moderate pressure: evidence from an Al-spinel ultramafic-mafic inclusion suite. Contributions to Mineralogy and Petrology 75, 225-233.

Roberta L. Rudnick S. R. Taylor Reseach School of Earth Sciences Australian National University GPO Box 4 Canberra ACT 2601 AUSTRALIA.