Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
6 The stateof stresswithin the Australiancontinent KURT LAMBECK (t), H.W. S. McQUEEN ('), R. A. STEPHENSON('), D. DENHAM (t) (L) ResearchSchool of Earth Sciences,Australian National University, Canberra ACT, Australia (2) GeologicalSurvey of Canada, Calgary, Alberta, Cqnada (3) Bureau of Mineral Resources,Geology and Geophysics,Canberra ACT, Australia Received15103184,accepted02107184. pointto a predominantly compresseismicity and<in situ> stress measurements ABSTRACT.Faultplanesolutions, sive state ofthe Australian continent. This stress is being relieved by brittle failure in the upper and middle regions of the crust only. The orientation of the axes of maximum compression varies conside¡ably across the continent. Most Australian earthquakes occur in tectonic provinces that range in age from Palaeozoìc to A¡chaean. There is no evidence that these regions have been actively rejuvenated in more ¡ecent times. In eastern Australia much of the seismicity is associated with the Lachlan Fold Belt which was cratonized by Late Palaeozoic time. Earthquakes occur down to about 20 km depth but no recent surface faulting has been observed. Deep events fail predominantly by thrust faulting and shallow events by strike-slip faulting. The seismicity is consistent with the superposition of a local stress field caused by the erosion and rebound ofthe highlands, on a predominantly southeast-northwest regional field Seismicity in South Australia is confined to the Cambrian Adelaide Geosyncline and its vicinity. Earthquakes occur here at generally shallower depths than in eastern Australia. The predominant direction of the axis of compression is northeast-southwest. In western Australia the earthquake activity is diffuse and not obviously related with major tectonic features. Seismicity here is well documented only in the Archaean Yilgarn Block where much of the activity is very shallow and associated with surface faulting. This activity may also be the consequence of a local stress freld. related to the nearby Perth Basin and Darling faull superimposed on a regional east-west field. That seismicity occu¡s in these old structures reflects the relaxation of deviatoric stress associated with past tectonic upheavals. As relaxation takes place at depth, the effective elastic thickness of the iithosphere is reduced and deviatoric stress in the upper crust increases until failure occurs, possibly triggered by the imposition of the regional stress field. Where non-hydrostatic stressesare greatest, as in the centre of the continent, levels of seismicity are exceedingly low. This suggeststhat such regions are held in mechanical equilibrium by loading buoyancy, elastic and viscoelastic forces and by horizontal compression. In the case of central Australia the required compression is approximately north-south. The continent-wide stress f,reld could be associated with the plate tectonic movement of the Aust¡alian Plate or it could be the result of temperature and density differences between continental and oceanic lithosphere. The observations and models suggest that this regional force is variable in azimuth with a magnitude of the order 100-200 MPa. Key words: tectonicstress,seismicity, continentallithosphere,Australia. AnnalesGeophysicae,7984,2, 6, 723-742. INTRODUCTION The studyofpresentand paststressstatesofa continent is of relevanceto the understandingof a number of tectonic problems that have been much discussedin the recent geologicaland geophysicalliterature. These problems include : (i) the interpretation of < in situ >> stress measurements and intraplate seismicity, (ii) the study of tectonic processeswithin the lithospheric plate and along plate boundaries,(iii) the quantitative evaluationof continental lithosphericrheology,and (iv) the evaluationof driving mechanismsof plate tectonics. Evidencefor present and past stressstatesis of three andgeological main types: (i) that basedon geophysical geòphysical (ii) that basedon modelfield observations, ling of tectonic processesin the lithosphereor of convectionin the mantle, and (iii) that basedon laboratory experimentsof the rheologicalbehaviourof lithospheric and mantle rocks. The volumes of papers edited by tilyss (1977)and Hanks and Raleigh (1980)contain 0755 0685l84l06723 lg $ 3,gOO AnnalesGeophysicae, EGS-Gauthier-Villars good examplesof both the geophysicalrelevanceand observationsof stressin the lithosphere. Geological evidenceincludes the f,reldobservationsof folding and faulting on both the micro- and macroscales. The former includes induced microstructures,such as changesin dislocation densities and grain sizês and the recrystallization of grains along faults and in kimberlite intrusions.Thç interpretation of thesepalaeostress indicators remains uncertain. For examplg stressdifferences of i00-200MPa (1-2kbars)havebeen determinedfrom dislocationdensitiesalong fault zones, but these yalues are þenerally 2-3 times larger than estimatesdeducedfrom recrystallizedgrain sizeswithin the samerock (e.g.Christieand Or4 1980). Geophysical observational evidence includes < in situ )) stressmeasurements,present and past evidence of seismicity, the seismologicalevidencefor faulting and lateral variations in crustal structure,and the analysis of topography and gravity. The major difliculty with the < in situ > measurementsis that these are K U R T L A M B E C KH . . W . S M C O U E E NR, A - S T E P H E N S O D N, DENHAN/ made near the surfaceso that they may not be very representativeof the state of stress in the crust, or lithosphere,asa whole.Near surface< in situ ))measurements usually result in inferred stressdifferencesthat are of the order of a few tensof MPa but it is expected that deeper in the crust these may increase several fold (McGarr and Gay, 1978). Seismicityis indicativeof where stresses have concentrated so asto exceedthe brittle failure limit of the crust. However,becausethe strengthof the crustis unknown, earthquakes cannotbe usedto estimatethe magnitudes ofthe regionalstress.Fault planesolutionscan be used to estimatethe directions of the principal stressesbu! becausefailure usually takes place along pre-existing zones of weakness,the estimatesof principal stress directions from individual solutions may have errors of up to 90o (the principal compressionalstressneed only be in the samequadrant as the principal compressional axis of the focal mechanismsolution; see,for example,McKenzie,1969; Raleighet al.,1972).Estimates of seismicstress-dropsinferred from crustal earthquakesindicatethe changein the stress-state that occurs at the time of the earthquake rather than the actual value of stress(e.g.Hanks, 1977). Analysis of gravity data gives information on stress but only if specificmechanicalmodels,such as a statement on the nature of isostatic compensation, are introduced.Even then, the deducedmagnitudesof the stressesare uncertain and rheology dependent(e.g. Lambeckand Nakiboglq 1980).Estimatesof minimum stress-difference are obtained if the assumptionof local isostasyis made (Jeffreys,1970)but if compensation is regional the stress-differences may be severaltimes greater than these minima. Topography may better represent past and present stress states - because the load is more clearly defined- but the stressestimates remain model dependent : on the mechanism .bywhichthe topographyformed,on the stateof isostasy attained,on the crustal rheologyand on past erosion (e.g.Stephensonand Lambeck, 1984).Lateralvariations in crustal structure as observedby seismicrefraction and reflection surveys or from the analysisof traveltime anomaliesare also indicative of a non-hydrostatic stress state. But even when such observations are combinedwith gravity and topographicdata it remains diffrcult to separatethe assumptionsmade about rheology from any statementson the stressstateand tectonic processitself. Specific examples of geophysical modelling for the stress-state include the analysisof passiveloading of the lithosphere by volcanic loads, sediments,ice or water. In these casesa history of loading is required. Observationsthat constrainthe modelsinclude gravity, deflectionsof the crust and the time-dependentbehaviour of deformation. Less direct examplesof geophysical modelling for estimating tectonic stressinclude the analysisof subduction tectonicsand of sedimentary basin formation by processesother than passiveloading. Here the resultant stressestimatesare intimately relatedto the assumptionsmadeabout the forcesassociated with the geophysical process. These brief introductory comments serve onlv to 724 indicate that the measurementand interpretationof stressin the lithosphere,whethercontinentalor oceanic, is not unique.Only by combiningthe various lines of evidenceis a coherentpicture likely to emerge,a picture that should relate to the structure, evolution and motion of a continent.Thesecommentsalso serveto illustrate that any statementsthat we make in this paper about the stresswithin the Australiancontinent should be consideredas preliminary and that more questionsare likely to be raised than answersgiven. Thereforg what we attempt t.o do is twofold. Firs! we survey some of the geophysicalevidencefor the presentstressstate within the continent and discuss someof the implicationsof this evidence.Second,we presentsome preliminary analysesof the stressstate associated with certainaspectsof the presentAustralian continentaltopographyand gravity.Threetopographic and gravity featuresin particular will be discussed; the stressstate implied by the large gravity anomalies in the centre of the continent, the stressstate near the Darling Fault and Yilgarn Block in westernAustralia and the stressstate associatedwith the eastern highlands. OBSERVATIONAL EVIDENCE FOR THE STRESS STATE IN THE CRUST Seismicity The first permanentseismograph stationswereinstalled in Australia early in the twentieth century and much of the equipmentoperating from this time up to about 1960 was of relatively low gain and sensitiveto low frequencysignalsonly. Instrumentationgraduallyimproved and from about 1960onward eventsof Richter magnitude M > 5 occurring anywhere in Australia would havebeenlocated(Doyleand Underwood,i965). The network of seismographstations has continued to improve (e.g.Cleary, 1,977),but there are still areas of northern Queenslandwhere earthquakesof magnitude 4 cannotbe located.Thus only a very shortrecord of Australian-wide quantitative seismicity is available at this time. Surveyshave beengiven by Burke-Gaffney (1951), Doyle et al. (1968); Drake (1974); and Denham et al. (1979\.From thesestudiesit becomesclear that contrary to what is often supposed and despite the fact that the land mass lies far from active plate boundaries,the Australian continent is not particularly aseismic. The largestrecenteventwasthe 1968Meckering earthquakeof magnitudeMr: 6.8(Gordon"t97I; Gordon and Lewis, 1980). Other large earthquakes include an Mr:7.7 earlhquakewhich occurred off the northwest coast in 1906 and an M, - 7.0 event which occurred northeast of Geralton in western Australia in 1941. The known Australian seismicity is characterizedby three main regionsof activity within each of which the distribution of earthquakesis diffuse (fig. 1). The southern part of the easternregion is well covered from about1960onwardsandmosteventsof magnitudeM > 3 would have been located after about 1965(e.g.Jaeger and Read,1.969). 7 earthquakesof magnitude M >- 5.5 " CONTINENT W I T H I NT H E A U S T R A L I A N S T A T EO F S T R E S S a' Eastern Region ¡l oI ¡,rru j I fownsville .rrgzo / ..lr ( '1 fËh, western 'ääit,. i'tt¡' o o . RTH o a o o o a o r o ú ;;.,¡Tt WITH MAGNITUDE EARTHOUAKES 4 . OO R G R E A T E1R8 7 3 - 1 9 8 0 . Magnitude < 5.0 O Magnitude 5.O-5.9 )Magn¡tude> 1946 - Year of earthquake w¡th magn¡tude 6 O or greater 5.9 Figure 1 Earthquakeswith magnitudes 4 or greaterrecordedin Aus*aliaîrom 1873ta 1980.The year of the eventis indicatedforeventsof magnitude6 or greater. 16 stations,including three stations run by the Sydney Metropolitan Water Board. It is supplemented by stations establishedin Victoria from 1976by the Preston Institute of Technology(PIT) (Gibson et al., l98l) and by two stations operated by the Bureau of Mineral Resources(BMR). The recordedseismicityfrom 1960to September1983is illustrated in figure 2ø. Someof the smaller earthquakeslocated in Victoria by the PIT network only have not been included. No attempt has beenmadehere to distinguishbetweenmainshocks and aftershocksand some of the clusteringis a consequenceof aftershocksequences.Only reliably located events,where a minimum of five stations were used to determine the hypocentres,are illustrated. The locations of all events of M" ) 3 recorded from 1960 to September1983are illustrated in figure 2b.Superimposed upon this distribution of seismicityare the station locations and the limits beyond which eventsof magnitude less than 3 cannot, as a rulg be reliably located from records at a minimum of 5 stations.In figure 2c all events of M" < 3 located by the southeastern network are illustrated. have occurredin this region from 1959to 1982.The central region coincides apprôximately with the Adelaide Geosynclinein South Australia and a northward extensioninto the Simpson Desert of the Northern Territory. This northern part is the most active area in Australia and at least five earthquakesof M"2 6 haveoccurredthere in the last 100years.The location of the smaller (M" < 4) earthquakesin the central region did not becomepossibleuntil-about 1962(Sutton and White, i968). The third region includesmuch of westernAustralia where only sinceabout 1980has it been possibleto locate eventsof magnitude4 or greater throughout the region. Much of the better known seismicity in this region occurs in the southwestern regior¡ east of Perth, and is also referred to as the southwestseismiczone (Doyle, l91l). SoutheasternAustralia The principal information on the seismicity of this region comesfrom the network operatedby the Australian National University (ANU) since1958.The network has been gradually improved and now comprises The accuracywith which the earthquakesare located varies considerablywith their position and magnitude, with the state of the network at the time of the evenl 725 K U R TL A M B E C KH , . W . S M C O U E E NR,. A S T E P H E N S O D N ., D E N H A M co, o YarB e Sychey ' i x 'C+ o " o ^ x o * .'t* o o o { Robertsor¡ Picton o x x c ':* ,;-Ì*åi_:;i. . Talktlìgo-----_i=.-: o o ôo t t o ^ o ô o " .| 4-itç x ;. &" I'" x<^ ' o xv /x x c i x o x x x Dalton- Gunring t x v , ) ' o x o o o o @ xA fMelboure o x x o o ô " o tr M>s O 4<M<5 o 3(M<4 x 2.5<M<3 M <2'5 . [] Gippsland l# o r48 r50 152 2a) Figure 2 (a) Seismicity oÍ southeastern Australia from 1960-1983. (b) Earthquakes of magnituìle 3 or greater in southeastern Australia from 1960-1983. The topography of the highlands is iniJicated by the 500 anil 7500 m contours. The 1000 m contour is shown anly partly ( dotted line). This ilefines approximately the topographic low in the dividing range near 35o latitude. Elevations above 1500 m are shaded. The 4000 m depth contour defines the lower edge of the continental shelf. Solid triangles refer to the seismograph network. The lines I anil 2 ilefine the region beyond which magnitu¿le3 events cannot be located from 5 or more stations. (c) Earthquakes of magnituile less than 3. AIso indicateil are known surfacefaults anil the limits of the major basins and fold belts. and by the state of knowledgeof the seismicvelocities of the crust and upper mantle. Using known quarry blasts as a calibrator, it appears that epicentres of eventsoccurring within the perimeter of the network are usually determined to within I 5 km, and that these locations are not very sensitiveto the adopted crustal model. Events outside the network perimeter are progressivelylesswell determinedas their distance from the seismographstations increases.Earthquakes near the limits depictedin figure 2a have an epicentral precision that is typically about 10 km. Accuraciesof these locations may be less satisfactory becauseof systematicerrors resulting from the simplicity of the adopted twoJayer crustal model. Depth determinations are much more problematical. Typical uncertainties for events within the network perimeter are 5 km with the better events having a precision of 2-3km. Figure 3 illustrates the depth 726 W I T H I NT H E A U S T R A L I A N CONTINENT S T A T EO F S T R E S S n U c 2\ qo o o o t/2 \ ôI oI I / l " I I I I I þ .1" \ t t ¡ \ l-/ /l < - J q \ \ \ o o o o M>s o i;&n O /t< M<5 o 3(M<¿ o \--_ l0O km r50 t1g r52 2b\ distribution of events for which the solutions for depth, based on P and later phases,yield precisions that are better than 2 km. The mean depth of all events is about 11-12km and is about the samefor all magnitude intervals. As noted by previous investigatorsof Australian seismicity(e.g.Doyle et al.; L968)few events occur at depthsgreaterthan about 25 km and no event can be attributed with any degree of certainty as occurring in the mantle. The depth distributions are suggestive of concentrations of seismicity at two depths, near 6 km and near 15 km, with reduced levels of seismicity in the interval 8-12km. This may reflect the presenceof a low velocity layer in the crust (e.g.Finlaysonet a1.,1979),in which stressis released partly by creep and partly by brittle failure. It may 727 also indicate that stressdifferencesin this part of the crust are reduced, as would be the case if flexural stressesare significant and the middle plane of the effective elastic or viscoelastic lithosphere is near and Lambecþ 1984).It must be 10 km (e.g.Stephenson remembered,however,that becauseof the uncertainty of the depth determinations,and becauseof the dependence of these depths on the chosen crustal model, this distribution may be a consequenceof the method of analysisand this needsto be examinedmore closely before either of the above interpretations can be accepted. The horizontal distribution of seismicity does not exhibit any dependenceon depth : deep and shallow events have essentially the same geographic distributions. K U R TL A M B E C KH, W S N / C O U E ERN ,A S T E P H E N S ODN , D E N H A M t,, \ \ ' N \ \)/ ' ' t \ ,ì /t \ '[ , |( \ \ _rr' i"í'i ""'* t-li') "'', SV$neY \r-.\-i ' '-i{M ,$ '#f,þ(' \\ tl t.-___ 'r..' Surat Basin \ (,;, \ x I Lachlan \ "\ \ .ù \ t'l , Fold t V lt ra \, \ \\ x '*' J L r ' , t Ì I i (t r T .ñ 'fu ' , "ì \ I',';# I t- ' r \ , ! r \ . I \\'\ "jÍ \.-i - ( f \'1 " .\ l * I :r, \\ ' ) - - - z/' " I ,\\ fffi*ì , /r \ I , \\ " ' / o x / - k X r x r I \ x x (x \ \ rt) \ \\ . ' .l ) \. / ^7-'ai' < '' -7:;æ, . - J /: * ; . ¿ x 2.5<M<3 M<2.5 , Gippslahd Basin/ '"1 ) },,-'-1--' \-'.. /-Jffi I I , l x \ 'f^l i r#"+. ru:!\i\ ' l '(! .\l 1. , ) Basin l l New England ' Fold Belt f --l Murray Basin l ., lO0 km 152 The seismicityof Victoria has beenreviewedby Underwood (1972)and Gibson et at. (1981.). ievels of seismicity are greatest in the eastern part of the state wherg apart from a concentration of activity in the westernpart of the Gippsland Basin (frg.2a) seismicity remains diffuse.Any event of M" < 3 that occurs in the westernGippsland Basin would not be recorded on the ANU network but Gibson et al.'s results leave no doubt that such eventsoccur frequently both here and in the region immediately to the north and east of Melbourne. Seismicityin the New England Fold Belt to the north is only poorly documentedand the paucityof recorded eventsin this regionis probablymore a consequence of the lack of seismographstations than of low levels of seismicity.The Lachlan Fold Belt was shapedduring the Palaeozoic by intense folding and deformatior¡ by major igneousintrusions, and by volcanic activity, and cratonized by Early Permian time. The New England Fold Belt stabilized later, by Early Mesozoic time. Subsequenttectonic activity in both fold belts appearsto have been minor. Present-day seismicity does not exhibit any clear relation with either the continental topography (hg.2b) or gravity. Nor is there a clear correlation with geology The ANU network is capable of locating M,4 3 eventsfor much of the Lachlan Fold Belt (f,rg.2c) and activity is relatively diffuse throughout the region. 728 W I T H I NT H E A U S T R A L I A N CONTINENT S T A T EO F S T R E S S approximately with regions where the density of known surface faults is also greater (fig. 2c). Cleary (1967)suggestedthat much of the activity occurs on or at the edgesof the abundant granite intrusions in the Lachlan Fold Belt and many of the events do appear to occur near granite outcrops. However, the precisionof their location is inadequateto determine the nature of this association.What is curious is that this relation is limited to the S-type granites; very little seismicity is associatedwith the I-type granites of southernNew South Wales and easternVictoria. ( ) The off-shoreseismicityoccurslargelywithin about 70 km of the shoreline,below the narrow continental shelf and slope (see figs. 2b and 2e). The depths of theseeventsare typically in the range 1.6-20kmbelow sealevel.It has not beenpossibleto obtain fault plane solutions for these events. AII Magnitud.es M 2,5-2.9 237 euents depth depth Figure 3 Depthdistributionof earthquakesin southeastern Australiafrom 7960 to 7983whoseapparentdepthsare locatedto within 2 km. South Australia or tectonicstructure(fig. 2c).Nevertheless, somegeneral commentscan be made. (1) The M"2 3 events are distributed quite uniformly through much of the Lachlan Fold Belt. There is a decreasein recordedactivity towardsthe northwest and offshore beyond the continental shelf. These decreasesdo not appear to be a network effect since the smallerevents(M < 3) are observedout to about the same distances(compare figs. 2b and 2c). There is alsoa seismicallyquietzoneto theeastof the southern part of the Dividing Range (below Eucumbene in frg. 2a). This was previously noted by Drake (1.974) and again,is not a network effect. (2) The M,"< 3 events (frg. 2c) have a very similar geographic distribution to the larger earthquakes, with similar areaswhere activity is lower than average. Small eventsbecomeproportionally less frequent in the northern part of the SydneyBasinand in the Murrây Basinandthis is probablya consequence ofthe decreasing radius of detectability at smaller magnitudes.In several localities the levels of seismicity are much greater than average.Some of thesereflect aftershocks of larger events(e.g.Picton and Robertson).Others reflectinducedseismicitysuch as that associated\/ith the Talbingo Dam (Muirhea{ 1981) and possibly the EucumbeneDam (Cleary,1977;Bock and Denham, 1983).Other areas of high levels of seismicactivity include the Dalton-Gunning,Crookwell and Young areas which coincide approximately with the topographic low in the southeasternhighlands (fig. 2b, seealso frg. 9 below).This region hashad a consistent record of seismicity since the first settlers arrived (Burke-Gaffney,l95l; Cleary,1.967),butno successful explanationhas yet been proposed. (3) Some of the seismicity appears to be associated with known faultg particularly those'occurring in the Snowy Mountains (Cleary et al-, 1964; Bock and Denharr¡ 1983). For most other seismic regions, notably the Dalton-Gunning seismiczone,no obvious correlation with faults has been found" other than that regionsof higher than averageseismicitycoincides 729 Recent seismicity in southern Australia has been discussedby Sutton and White (1968),Stewart et al. (1973),Sutton et al. (1977),and McCue and Sutton (1979).Much of the seismicityis conhnedto the Adelaide Geosynclineand the adjacentgulf grabenregions (frg. 4). This geosyncline,formed during a prolonged orogeny in Late Proterozoic-Cambriantimg consists of a thick sequenceof sedimentsthat have been folded tectonic down to depthsof 5 km or more. Subsequent activity appearsto have been minimal. Much of the seismicity has occurred in the Mt. Lofty-Flinders Rangespart of the syncline,but this may be a consequenceof the restrictednature of the network prior to about 1969.In that" year three additional stations were set up in the northern part of the synclineand a ,l <M<5 O Àt>5 Fig. 4 Seismicityfrom 1960-1983in South Australia. K U R TL A M B E C KH. W . S M C O U E E NR,. A S T E P H E N S O D N, DENHAN¡ these indicated that there is considerableseismicity in this region as well (Stewart and Mount, 1972). To the south west the seismicityoccurspredominantly to the west of the Lincoln Fault on the Eyre Peninsula. Seismicity within the syncline is generally shallow, with more than 501 of zl located eventsbeing at depthsof lessthan 5 km (Stewartand Mount 1972). These are also of small magnitude (M" < 3) and appearto be confined mainly to the thick sedimentary folded into the syncline.They may therefore sequences reflect more a local responseto a regional stressfield than the regional stressstate itself. Events of M" > 3 occur down to greaterdepths but few eventsoccurring within the perimeter of the network of stations that recorded the event are located at depths greater than about 20 km. McCue and Sutton (1979)estimatean averagedepth ofabout 10 km for sucheventsrecorded in the two year interval of 1976-19'77. Persistent seismicity of large magnitude (MrÞ 6) has been recordedsince 1900 in the SimpsonDesert in the south east corner of the Northern Territory (Burke-Gaffney,1951) (lig. 1). However, only a few precise locations and two fault plane solutions are available (Stewart and Denham, 1974; Everingham and Smith, 1979). Ãny relation with deep crustal structure is maskedby recentsedimentsand the origin of this activity remainsan enigma. Western Australia The seismicityof westernAustralia has been discussed by Everingham(1966)and summarizedin Denhamet al. (1979). This area experiencesthe highest levels of earthquake activity throughout the continent (fig. 1). There does not appear to be a clear relation between seismicityand the major tectonicfeatures.For examplg there is no known seismicity on the nearly 1000km long Darling Fault (total vertical throw - 10 km) separating the Perth Basin from the Yilgarn Block, even though movementson it have occurred continuously sincethe Permian (fig. 5). The offshoreseismicity extendsfarther out to seathan it doesalong the eastern and southernmargins but the shelf and slope are also much wider here and as elsewhere,much of the seismicity appearsto lie below the shelfand sloperegion; betweenthe coastline and the 5000m isobath. tI . calinsir¡ t",SYti ' '.ÌX-' Fig. s Seismicity in the southwest seismic zone of western Australia. Also illustrated are the surJaceJaulting ( light lines) and the directions of the compressive axes as determined for the Cadoux, Calingiri and Meckering earthquakes (heavy lines) andfrom < in situ ) stress measurements (meilium lines). The maximum < in situ > sfress is 23 MPa near Cadoux. 730 CONTINENT W I T H I NT H E A U S T R A L I A N S T A T EO F S T R E S S Depth control on the earthquakesis very poor but the surface faulting associatedwith the larger events in the south-westseismiczone. as well as the isoseismal patterns, indicates that most if not all of the earth quakes here take place near the surface and in the upper crust (e.g.Everingham et al., 1982). Fault plane solutions A number of reliable fault plane solutions have been obtained for the larger earthquakesthat have occurred in Australia since about 1960. Solutions have also been possible for some of the smaller events that occurred within the regional networks. Figure 6 and table 1 summarize the results. In general, these solutions point to compressiveforces throughout the continent (see,for example,Cleary, 1967; Fitch et al., 1973; Mills and Fitch, 1,977 ; Denham et al., 1979) but considerable variation in the directions of the compressive axesdoes occur. Reliable fault plane solutions for earthquakes in New South Wales have been obtained by Mills and Fitch (1977),Everingham and Smith (1979),Denham et al. (1979,1981, 1982)and Bock and Denham(1983). All solutions yield compressionaxes that are nearly horizontal but whose azimtthal orientations vary to souththrough about 90o,from northeast-southwest east-northwest(fig. 6a). The different azimuths may be the result of faulting being controlled by the geometry of pre-existingcrustal faults or zonesof weakness. Most deeper events, where the depth (D) is greater than about 15 kr& are associatedwith thrust-faulting mechanisms and most shallow events (D I 5 km) indicate failure by predominantly strike-slip faulting (table 1). The two intermediate depth earthquakes near Bowning are indicative of strike-slip faulting with a major thrusting component. Earlier, Underwood (1967) had suggestedthat some of the Dalton-Gunning earthquakeswere the result of tensional stressbut his solutions were poorly constrained and the more recent work has not confirmed this interpretation. Fault plane solutions for eventsin Victoria have been discussedby Underwood (1972) and Denham et aI. (1981).The 1977Balliang event about 50 km to the west of Melbourne (f,rg.6b), is indicative of thrust faulting at a depth of 21 km but the depth is poorly constrained. Two nearby events in the Victorian highlands yield quite different fault plane solutions. A re-examinationby Denham et al. (1984)of the 1966 Mt. Hotham event suggeststhat failure is by normal tensionaxis(fig.6a). faulting with a northeast-southwest This represents the only well documented case of tensionalstressin the continent. The nearby Wonnangatta event of 1982indicated failure by thrusting with a northwest-southeastorientation of the compressive axis. Depths of the events are poorly determinedbut both earthquakesare believedto have occurredin the middle crust. Table 1 Summary offautt plane solutions.All events except the Mt. Hotham are compressional.'l denotes thrust fault, SS strike-slip fa¡¿l¿,SS-T strike-slip with a significant thrust component, N normal fault. Year Earthquakelocation Magnitude ML Snowy Mountains New South Wales 59 7l 76 81 Berridale Middlingbank Pilot SugganBuggan Dalton Gunnins Seismic Zone New South Wales 7l Dalton 73 Dalton 74a Dalton 74b Dalton 77a Bowning 77b Bowning 5.2 4.0 3.8 3.7 Depth km T ] J SS T? SS 5 6 6 Appin Appin Victoria Mt. Hotham Wonnangatta Balliang SouthAustralia SimpsonDesert SimpsonDesert Quorn Blinman Melton WestAustralia Meckering Calingiri Lake Mackay Halls Creek SS SS SS-T? SS SS-T SS-T Compression (or teniion) axis '(trend plúnge) Denham et al., l98l Bock and Denhar4 1983 Bock and Denharn,1983 Bock and Denharn,1983 262,12 77,24 320,5 271,t0 Denham et al., 198I Denham et al., I98l Denhamet al.,798t Denhamet a1.,7981 Denham et al., l98l Denham et al., 1981. 3 2 4 5 l2 13 5.6 5.5 19 21 T T 81 Aftershock 4.6 15 12 T T ) 1 <? 280,5 66 82 77 5.5 5.4 4.4 15 N l t T T 39,2 11 3 , 1 72 78 77 77 77 6.2 4.7 3.8 4.0 7 ? 5 10 15 68 70 70 78 6.9 5.9 6.7 6.2 5 1 15 l't À 1 ol IJ Á 1 2l 731 SS-T SS SS SS ss T T T T Reference 327,tO 136,r4 326,8 95,1 0 2.5 3.0 3.8 Ã') 4.8 SydneyBasin New South Wales Robertson Picton r'åulr Prane 117 ?5 154,0 46,34 646 DenhanL1980 Mills and Fitch, 1977; Denham et al.,1980 Denham et al., 1982 Denham et al., 1982 )q) )5 Denhamet a1.,7984 Denham et aI., L984 Denham et al., I98l 181,1 125,O )\a r1 41,0 60,25 Stewa¡tand Denham, 1974 Everinghamand Smith, 1979 McCue and Sutto4 1974 McCue and Suttorl 1974 McCue and Suttoq 1974 )71 17 Denham et aI., 1980 Denham et al., 1980 Denham et al., 1980 Everinghamand Smith, 1979 282,t6 218,24 7)\ A K U R TL A M B E C KH, . W S N / U O U E E R N ., A . S T E P H E N S O D N ., D E N H A M Figure 6 (a) Fault plane solutions for some southeast Australian earthquakes.(b) Fault plane solutions for same earthquakes in other regions of Australia. Inward pointing arrows indicate inferred direction of principal compressional stress in the horizontal plane. 6a) Fault plane solutions for South Australia have been givenby McCue and Sutton (1979)for 3 eventsin the Adelaide Geosyncline,and by Stewart and Denham (L974) and Everingham and Smith (1979) for two eventsin the SimpsonDesert.All solutionsare indicative of failure by strike-slip faulting irrespectiveof the focal depth (Fig. 6b). The AdelaideGeosynclinesolutions are for eventsthat arelocatedbelow the sediments and the three solutions give very similar northeastsouthwest azimuths for the axis of compression. The Simpson Desert earthquake of 1972 has a fault plane solution that is indicative of north-south compressionwith failure along a steeplydipping strike-slip fault (Stewart and Denham, 1974). A nearby event, while not producing a good focal mechanism, did yield a well constrainedhorizontalaxisof compression but with an azimuth that differs by nearly 600from the 1972 event (Everinghamand Smith, 1979). Fault plane solutions for western Australia have been publishedby Fitch et ø1.(L973),Everinghamand Smith (1979) and Denham et al. (1980).All events indicate that failure occurred by thrust faulting irrespectiveof the focal depth of the earthquake (table 1). Azimuths of the compressiveaxes are again quite variable. The two events in the Canning Basin in the northwest of Australia are separated by only about 275 km and yield compressiveaxes that differ by 90o in azimuth. The solutions for the two events in the south-west seismic zong the Meckering (1968) and Calingiri (1.97L) earthquakes, are very consistent with each other and with the evidencefor surfacefaulting from theseand other events(e.g.Lewis et al., 1,98I;Gordon and Lewis, 1980). This brief summary of the fault plane mechanisms 732 confirms that the predominant stress state within the Australiancontinentis one of compression. There are however,regional variations in the failure mechanisms and the directionsof the principal stressaxes. In south-easternAustralia" the near-surfacefaulting is predominantly by strike-slip motion while the apparently deeperearthquakesreflect failure by thrust faulting. In South Australi4 failure is by strike-slip faulting even for the deeper events. In contrast all solutions for western Australia indicate failure by thrust faulting from surfaceeventsto depthsof 17 km. The axesof compressionlie along all possibleazimuths and this suggeststhat failure is partly controlled by past geological structures. The different azimuths obtained for the Hall's Creek and Lake Mackay earthquakesin the northern part of the Canning Basin may, for example,be a consequenceof a local stressfield associatedwith the basin and basin-margin superimposedupon a more regional stressfield. Likewise, the different azimuths obtained in eastern Australia may be controlled by older faults or it may be the result of a local, topographically related stress field superimposedupon a regional stressfield. The different azimuthsfor the two SimpsonDesert earthquakes are less readily associatedwith local structure, even in the qualitativemanner usedfor the other solutions. One trend that does occur is that the compressive axesare approximatelyperpendicularto the continental margins for those events within a few hundred kilometersof the margin. The compressiveaxesin southeast Australia are predominantly east-westor northwestsoutheast, in South Australia they are northeastsouthwest,in the southwest they are again east-west, and the Hall's Creek earthquake has a northwestsoutheast axis of compression. CONTINENT W I T H I NT H E A U S T R A L I A N S T A T EO F S T R E S S tï'ü.tl ' Picronl9Z3 1977ht Appin l98l Berridole t959 Roberlson l9ól lDolton-Gunn¡ng lgzl l n r t 9 7 3 x t 1971o I \ r n 197Ä6 Mt Hothom 19óó Wonnongotto 1982 6b) <(In situ > stressmeasurements Since 1957 over 1500 measurementsof < in situ > stress have been made in Australi4 predominantly using flat jacks and, since 1963,bore-holeovercoring methods(Worotnicki and Denham, 1.976).Geophysically useful and reliable results are, however,few and thesehave been summarizedby Denham et al. (1979, 1980).Thesemeasurementsare in generalagreement with the results from the fault plane solutions in that (i) the predominant stressstate is one of compressior¡ and (ii) there is considerableregional variation in the direction of the compressive axis. In southeastern Australia, the overcoring methods indicate that the maximum principal stress is approximately east-west in the westernpart of the Lachland Fold Belt and more nearly north-south in the easternpart (seealso Gray, 1982).Local geologicalconditionsand nearby topography may strongly influence the total stress field so that significant changesin stressmay occur from one locality to another.Analysisof stressinduced by topography of small slope (McTigue and Mei, 1981)indicatesthat surfacehorizontal stressvariations of up to 5-10MPa may be expectedin the neighbourmay therehood of 1 km topography.Closeagreement fore not occur between the < in situ r> measurements taken at the surlâceand the principal stressorientations deducedfrom the fault plane solutions for earthquakes occurring at considerable depths below the surface. < In situ )) stress measurementsof up to 23 MPa made near the Meckerine and Cadoux earthquakes 733 Bolliong 1977 in westernAustralia, a region of particularly low relief show a much more consistent relation to the faultplane solutions,with both sets of resultspointing to east-westcompression(fig. 5). EVIDENCE GRAPHY FROM GRAVITY AND TOPO- Central Australian gravity anomalies In the absenceof expressionsof active or recent tectonism, seismicitymay be anticipated in regions where stress differencesdepart significantly from a hydrostatic state, that is, in regions of large variations in topography or gravity. The largest gravity anomalies within continental Australia occur in the centre,where changesof 200 mgals occur in a distanceof lessthan 100km (BMR, 1976).Theseanomaliesare associated with Late Proterozoic and PalaeozoicBasins and with intervening uplifted arches (fi5. 7). Negative gravity anomaliesoccur over the basins and positive anomaliesover the arches.The elevation is relatively constant over the region as a whole and thiq plus the density contrasts between the basins and arches, contributes only a small amount to the observed anomalies.Arrival times'of P-wavesfrom teleseismic sourcesare late at stations situated in the basins and early at stations on the arches(Lambeck and Penney, 1984).This, plus the gravity observations,points to sienificant low densitv and low velocitv material N ., D E N H A M K U R TL A M B E C KH, W S M C O U E E NR, A . S T E P H E N S O D Musgrove Amocleus Arunto tlgolio E ¿ Figure 7 Predícted crass-section of the crust in central Australiafrom the OIJìcer Basin in the south to the northern Arunta Block in the narth (from Lambeck, 1984). A hypothetical upperlower crust boundary, originally at a depth of 15 km, is also ind.icated. below the basinsand to high density and high velocity material beneath the arches. The simplest interpretation is in terms of an undulating Moho (Lambecþ 1983a,b ; Lambeckand Penney,1984)and undulations of about 20 km are predicted. The mechanism by which this stateevolved is believedto be one of crustal folding through horizontal compressionof an inhomogeneousviscoelasticlithosphere.The process is facilitated by the erosion of the uplifting areasand the deposition of sediments into the down-warps. As bending stressesreach critical levels, major thrust faults developnear someof the basinmarginsand these are observedat several localities. To developand maintain this structure,the predominant stressstate must have been one of horizontal compressionwith an approximately north-south azimuth. The magnitude of the requisite stressto support the present structure is difficult to evaluate since it is dependent on the assumed rheology and effective thicknessof the lithosphere but a minimum value of about 100-200MPa appearsto benecessary to maintain a balancebetweenthe various forces operating (Lambecþ 1983ø). The presentbasin structuresevolvedfrom Late Proterozoic to Late Palaeozoicwith only relatively minor evolution having occurred since then. A regional compressive stress must, therefore, have been in existencesince at least Carboniferous time in order for the structureto be maintainedagainstthe buoyancy forces. In the absence of such a compressior¡ the basinswould be uplifted and the archeswould subside until isostatic equilibrium was achieved. It appears from the near surface geology and geomorphology that there may have been a period of rebound of the basins and this implies reduced overall compressior¡ subsidenceof the arches and possibly a transport of some sediments from the basins onto the blocks. Evidencefor this is seenin the outcropsof Mesozoic and Tertiary erosionsurfacesabovethe presentQuaternary surfaceof the basins and in the down cutting of rivers through some of the hills within the Amadeus Basin.It would appeaqhowevet that the compression was again effectivein Cainozoic time since rhany of the Carboniferous and older faults appear to have been reactivatedover large distancesand since many Cainozoic structural features appear to be a conse734 (Bureket a1.,I978). quenceof a north-southcompression can be expectedto occur The largeststress-differences near the basin margins where the gravity gradientsare greatest.Yet despitethe nearby location of two highly sensitive seismometerarrays (at Alice Springs and Tennant Creek), the recorded seismicity for the past 15 or 20 yearsis limited to a few small eventspossibly located on some of the old faults to the north of the Ngalia Basin. There seemsto be no history of recent faulting on the margins of the Amadeus Basin. In a region wherestress-differences are probably the greatest that occur anvwhere in the continental Australian lithosphereseismicity is insignificant. The SimpsonDesert earthquakeslie about 300 km to the eastof the AmadeusBasinand the directionof the compressiveaxisfor the L972earthquakeis in agreement with what is implied by the structure to the west. Yet there is no obviousreasonwhy the earthquakesoccur here and not to the west where the inferred stressdifferencesare much greater. Darling Fault, West Australia As first noted by Vening Meinesz(1948),substantial gravity anomaliesoccur to the east of Perth in West Australia. These coincide with the ancient Darling Fault bounding the Archaean Yilgarn Block and the Perth Basin. If this fault definesa near-verticalboundary betweencrustal blocks of different thickness,then the gravity anomaly Âg acrossthe fault has an approximate amplitude of LS - 2nG(p^ - p")LH (1) where G is the gravitational constant p* is the density of the mantlg p" the density of the crustal and LH the differencein crustal thickness of the two blocks. The associated maximum stress difference ø*u* is (Jeffreys,1932). o-u* ! þ* - p) S AH (2) and with (1) 6 ^ u *j Ls' slz"O :'jLs pR (3a) S T A T EO F S T R E S S W I T H I NT H E A U S T R A L I A N CONTINENT where þ is the mean density and R the mean radius of the Earth (seealso Lambeck, 1980).For Âg in mgals 6^^*/ 0.23A,g(mgal)MPa . Indion Oceon Shelf Perth Bosin o) (3b) The Darling Fault separatesthe Yilgarn Block from the Late Palaeozoic-Mesozoic Perth Basin.The change in gravityacrossthis fault is of the orderof 100-120mgal with. the positive anomaly being on the Yilgarn side of the fault. Then, with (p^- p") - 0.3 I cffi-3, LH - 8-9 km accordingto (1) and this is similar to what is observedon the fault (Jones,1976).With equation (3) o*", ! 25-30MPa. In the absenQeof forces other than buoyancy, there shouid be a tendencyfor the basin to be uplifting with rnovement occurring on the fault, but nearly all of the seismicactivity lies some50-100km inland (Doyle et al., 1.968;Denham et al., 1980)(fig. 5). I[ through the passageof time, movementon the fault is no longer permitted becauseof some annealing process,the effect of the buoyancy force is to create a more regional uplift involving the westernlimit of the Yilgarn Block. Maximum compressive forces will develop to the east of the Darling fault but more important in magnitude will be the near-surfacetensional forces in the vicinity of this fault. The absence ofevidencefor the latter suggests that thereis a regional compressiveforce which annuls the local tensional force in the basin but magnifiesthe local compressional force on the Yilgarn Block (fig. 8). It may be fortuitous, but where this maximum is predicted to occur by the model is actually where compressive seismicity is observed. Thetensionalforceattainsabout100-200 MPa and the compressiveforce to the eastof the fault does not exceed10 MPa. To preventfailureon the fault, the regional compressionalforce must also be near 100200MPa. Failureto the eastthenimpliesthat the stressstate within the crust is quite close to the limits at which brittle failure occurs sincethe seismicityappears to occur in an area where stressesexceedsurrounding valuesby lessthan 10 MPa. Southeast Australian highlands The major topography in Australia is the eastern highlands which coincides with the more intensely folded parts of the Lachlan Fold Belt of Palaeozoic age.Many geologistsview the landscapeas one that is considerablyyounger than this and have argued that a peneplainexistedin Late Mesozoicor Cainozoictime which was subsequentlyuplifted to form the present topography (e.g. Ollier, 1978; Wellmar¡ 1979). Stephensonand Lambeck(1984)on the other hand,argue that the highlandsarean erosionalresidueof the Palaeozoic orogeny and that the geomorphologicalevidence for uplift is not so much indicative of a constructive phaseof mountain building as of a destructivephase; that the uplift of the baseof the highlandsis occurring in response to theunloadingof thehighlandsby erosion. gravity data indicate that the highlandsare nearly The in isostatic equilibrium. The maximum stress differencesthat will occur in the underlyingcrubt and upper mantle will thereforebe of the order p"gh or greater. 735 Yilgorn L 200m Block o ô I I | 32km <+ Figure I Schematic model of the crust in western Australio across the Perth Basin and Yilgarn Block. The buoyancy force results in uplíft of the basin and in tensional stress near the basin surface and compression stress in the Yilgarn Block. Here p. is the density of the topography and å is the elevation. For h - 1500 m, o^u*t 50 MPa. This estimatecorrespondsto the state of local isostasyand representsa lower bound, but if compensationis regional then the stressdifferencesmay be severaltimeslarger than this. Most authors assumethat it is the presentday topography that controls the isostatic state of continents (e.g. Lewis and Dormar¡ 1970; Banks et al., 1977),but in the erosion-reboundmodel of and Lambeck(1984)it is the long term and Stephenson large scale erosion of topography that controls the stress state and flexural deformations of old continental lithosphere (see also Stephensor¡1984). A quantitative evaluation of this model is possible. In particular, it is possibleto predict the associatedstress state. In the model adopted here, and discussedat length in Stephensonand Lambeck (1984),the lithosphereis taken asa linear viscoelasticsolid whoseelastic and viscoelasticproperties are characterizedby an effectiveflexural rigidity D and a relaxation time constant r. The topography created during the orogeny is assumedto be in local isostaticequilibrium at the time that the fold belt stabilized.Erosion of this topography at any time I is assumedto be proportional to the remaining topographic height at this time. From a knowledgeof the presenttopography and an assumed erosion time constant the palaeotopographycan be computed and, starting with this and now extrapolating forwards to the present,it becomespossible to compute the stressstate arising from the erosional unloading the viscoelasticresponseof the lithosphere to this unloading and the change in, buoyancy force. Figure 9 illustrates the predicted horizontal surface stressbased on an assumption of continuous erosion sincethe stabilizationof the fold belt about 180Ma ago. The effectiveparametersusedareD : 1023Nm (equivaIent plate thickness: 30 km), r - 25 Ma (equivalent viscosity- I02s Pa s or 10'u p) and an erosiontime constantof 150 Ma. Surfacestressesbeneaththe uplifting areasare tensional but, due to the regional nature of the rebound, become compressionalat the periphery. Stresseswithin the lower half of the plate will K U R TL A M B E C KH. W . S M C O U E E NR,. A S T E P H E N S O D N. DENHAIV 3ó 20OMPo u6 t48 t50 152 Figure 9 resultingfrom the erosíonal Present-day horizontalprincipalstresses reboundmadelof Stephenson and Lambeck(1984).No regionalstress Shadedareasrcpresentthepredictednormal Jìeldhasbeensuperimposed. 750m elevatíons. faulting regimes.Thecantoursrepresent be compressivebeneath the elevatedterrain. Stress magnitudesare difÏicult to quantify with precision but maximum values of 200 MPa are predictedfor the Victorian part of the highlands. The stressmodel illustrated in hgure 9 is not wholly satisfactorybecauseit neglectsthe stressstateassociated with the continental margin. In particular, it does not take into account the offshore deposition of sediments which load the ocean crust at the same time that the continent is being unloaded by erosion. This aspectof the erosion-reboundmodel will be discussed elsewhere.The flexural stressesare a function of the effective flexural rigidity and of the erosional and viscoustime constantsas well as of the time at which the erosion and rebound is assumedto becomethe dominant landform shapingprocess.The generalstress pattern remains unchanged except that the position of the line separating tensional from compressional stressmay shift inwards oÍ outwards from the topographicaxis. The erosion-rebound model explains adequately the overall geomorphologicaland geologicalevidencefor the uplift that has occurred since at least early Cainozoic time, yet it does not explain the present-day seismicity.Areas where the predicted stressdifferences are greatest do not correspond with the regions of greatestseismicactivity (comparefigs.2 and 9). Furthermore, the predictedtensional failure in the upper crust is not seenin the few availablefault plane solutions or < in situ > stressmeasurements.This discrepancy can be overcome if a strong compressionalforce is superimposedon the local stressfield. This will, for example,increasecompressionalstressin the southern part of the SydneyBasin. It will also reduceor annul 736 the tensionalstresses in the highlandsimmediatelyto compression the westof the basinand resultin increased further to the west.Much of the seismicityin the upper half of the plate would thereforebe associated with the flanks of the elevated region. This pattern should reversewithin the lower half of the plate where the flexural stressesshould be reducedin those flanking areasand increasedunder the highland.Much of the seismicitydoes occur on the flanks of the Dividing Range $9. 2) but there is no clear differencebetween the patterns of seismicity below and above about 10 km. This is possibly a consequence of the uncertainty of the depthdeterminationof many of the events. Of somepossiblesignificanceis that the depth distribution of seismicityrevealsa marginally reducedlevel of activity near 10-12km corresponding approximately to the middle-planeof the plate. I[ as we believe,the presenttopographyis essentially an erosional residue of the Palaeozoicorogeny, then is it inescapablethat the pattern of flexural stresses illustrated in hgure 9 occurs.The magnitude of the stressesare less certain, being model and rheology dependent.The absenceof evidencefor tensionalfailure then requiresthe imposition of a regional stress field, either one that is nearly uniform over the area as a wholeif it is associated with plate tectonicsor one that is variable if it is associatedwith different mechanical and thermal propertiesof oceanicand continental lithosphere.To remove the predictedtensionalstresses west of Sydneyrequiresan almost east-westcompression in centralNew South Wales,and a nearly northwest-southeast compressionin Victoria. In both instances this requisite force appears to be normal to the coastline. 'We previously noted that the near-surfacefault plane solutions are predominantly indicative of failure by strike-slip faulting while the deeper events failed by thrust faulting. In the highland area all three principal stressesdue to erosion are tensionalwith the vertical stress ø,, being the minimum. A uniaxial regional stressmay annulone of the horizontalerosionalstresses leaving o"" as the intermediatestress.Fault plane solutions would, therefore, be strike-slip mechanisms.At greaterdepth below the highlandsthe horizontal forces are compressional while o,, remains tensional and cannot be an intermediatestress.The fault plane solutions are thereforeconsistentwith the superpositioning of a regionalstressfield on the erosion-reboundstress field. DISCUSSION : MORE QUESTIONS THAN ANSWERS Much of the Australian seismicity is associatedwith tectonicfeaturesof Palaeozoicor older age and there is little evidenceto suggestthat thesestructureshave been activated in more recent times. Some activity occursin youngerbasins,particularly the PalaeozoicMesozoic Gippsland Basin of Victoria. This activity appearsto be relatedmore to the underlying basement than to the basins themselves.Possibly here old faults are being reactivated by the superpositioningof the CONTINENT W I T H I NT H E A U S T R A L I A N S T A T EO F S T R E S S basin-generatedstressfield on a more regional stress f,reld.One possibleinterpretation of the seismicityin the older provinces is that it is a reflection of the release of tectonic stressby a slow migration of non-hydrostaticstressfrom the lower part of the lithosphereinto the middle and upper crust (e.g.Kusznir and Bott, 197'7;Lambeckand Nakiboglu, 1981).As the stresses in the lower regions relax, the maximum stressdifferencesin the colder and elasticpart of the upper crust increasewith time and this may lead to brittle failure long after the introduction of the initial stress field. Implicit in this argument is that continental crust that has been subject to signiftcanttectonism is near a critical stresslimit for much of its history, and that seismicityis initiated by the superpositionof stress fields that, in themselves,may be inadequateto produce failure. This additional stress f,reld could be causedby sedimentationand erosion or by a change in the global plate-tectonicstressregime. Some support, albeit very tentativg for this concept of stressrelaxation comes from the depth distribution of the seismicity.As emphasizedabove,depthsfor the Australian seismicity are generally poorly determined but there is a suggestion that seismicity decreases in depth acrossthe continent from eastto lvest,corresponding to increasingage of the major tectonic events. In easternAustrali4 the depths of eventsare distributedfairly uniformly down to about 20km (flrg.3),although there are no known eventsthat have given rise to surface faulting. In South Australia, the small magnitude seismicity occurs predominantly at depths shallower than about 5-10 km, but the larger events(M" > 3) may occurmore uniformlydown to about20 km depth. In the southwestseismiczone all the evidenceis consistentwith near surfacefailure.A closerexamination of the data is warranted, in particular as the above suggestion contradicts the argument presented by Chen and Molnar (1983),basedon equallyinadequate evidence,for increasingdepthsof continentalseismicity with tectonic age.Chen and Molnar argue that, as the lithosphereages,it also strengthensand becomesmore at depth. What supportive of greater stress-differences we suggesthere is t}:'at,at least within the Australian continent the present-daycontinental seismicity may be the resultof relaxationof stresses that wereimplanted into the lithosphereat the time of tectonism,possibly triggered by minor stressfields of more recent origin. An interestingaspectof this propositionis that regions of past tectonicactivity may remain seismicallyactive long afterthe originaltectoniceventceased. The upward migration of stress may be sufficient to keep nearsurfacefaults active and so result in a distribution of in the original seismicitythat followszonesof weakness structure. Such delayed stress build-up is predicted, model of Stefor example,by the erosional-rebound phensonand Lambeck (1984)in which the combinastress tion of the erosionthrough time,plus viscoelastic relaxation,resultsin maximum stressdifferencesoccurring some 100 to 200 Ma after the initiation of the erosionalphase.Conceivablythis offersan explanation for the widespreadbut small-volume Cainozoic volcanism of eastern Australia. Most of the better constrainedsolutions for earth737 quake depthsindicate an absenceof seismicityin the crustbelowabout 20 km and no earthquakes havebeen observedin the upper mantle. Models of tectonic processes within the Australiancontinentindicatethat the effectiveflexural rigidity of the lithosphereis of the and Lambeck,L984; order (1-5)1023Nm (Stephenson Lambeck, I983a) and the corresponding equivalent thicknessof the layer is about 30-50km. Stress-difference estimateswithin this layer remain model dependent but it appearsinevitable that magnitudesof 50200 MPa occur in the lower crust and upper mantle. That stress differencesin the lower crust cannot be negligibly small is also indicated by the inferenceof a variabledepth Moho in severalparts of Australia(e.g. Drummond, 1981;Lambeckand Penney,1984)requiring stressdifferencesof the order of 50 MPa. The implication is that thesestressesin the lower crust do not relax by brittle failure but by a creepprocesswith a very long time constant.This is largely in agreement with presentconceptsof lithosphericrheology based on laboratory results of a brittle upper crust and a ductile lower crust (e.g.Brace and Kohlstedt, 1980). That earthquakesdo not appear to occur in the upper mantle beneaththe Australian continent suggeststhat this part of the lithosphereis alsoductile,or wasductile in the past when temperatureswere higher than today, or that stressesthere have already relaxed by brittle failure in theserelatively old cratons. Chen and Molnar (1983)discusssomeevidencefor mantle seismicity in other parts of the world and suggestthat the uppermost part of the mantle is relatively strong and that deformation by brittle failure occurs therein. Possibly what distinguishesthe Australian continentalseismicity by Chenand Molnar from that in the regionsdiscussed is that, in the latter, tectonic processesare operating today, primarily in the form of collisional tectonics. In contras! the absenceof deeper seismicity in the Australian continent could mean that the shallower activity is not predominantly the result of present-day driving forces. There are, however, some difhculties with this interpretation. Firs! the predominant mode of failure near the surfacedue to the stressrelaxationand the readjustment to erosionwould be one of failure by normal faulting. Secondwherethe processis anticipatedto be most important, such as in central Australi4 present-day levels of seismicityare also low. Thesedifficulties can be overcomeif a regional compressivestressfield acts on much of the continent. Indeed.the evidencefor the predominant state of compressionof the Australian continent does suggestthat there may be such a field with the consequencethat some of the local tensional forces are cancelledand local compressiveforces are reinforced.Examplesof where this is inferred include the southwestseismiczone,the southeasternhighlands and the intracratonic basins of central Australia. The nature of this regional compressionalforce remains obscure.Severalattemptshave been made to interpret both the stressheld and the seismicitywithin the framework of the plate tectonicshypothesis.Sykes (1970) speculatedthat the seismicityin westernAustralia may be part of a nascentsubductionzone; Cleary and Simpson (1971)suggestedthat the Indian-Australianplate K U R TL A M B E C KH, W . S M C O U E E NR, A S T E P H E N S O D N ., D E N H A I \ 4 given by Richardsonet al. (1979)show remarkably good agreementbetweenpredictedand observeddirections of axesof maximum compressionover the Australian continent(lig. 10)if not alwayseverywhereelse. Their model E1 (fig. 10a)drivesthe platessolelyby a ridge-push force and, in the region in question, the horizontal deviatoric stressesare compressive.Similar stressstates are given by their models 827 and E3t (fig. 10b) in which the plates are driven by different combinationsof ridge-pushforces,forcesat subduction and continental convergencemargins, and viscous drag forceson the baseof the plate. >-- The Richardson el ø/. models produce deviatoric stresses that are of the order of 10 MPa or lessbut this is a consequence of the adoption of nominal valuesfor the various forces. Boundary stressesare typically setat about 10 MPa whereasDavies(1978)and Hanks (1977)arguethat thesestresses arethe orderofhundreds of MPa. Theselatter estimatesare of a magnitudewhere they may support or annul the stressfields generated by the local tectonic structure and generateseismicity that may not otherwiseoccur. One of the important aspectsof thesemodelsis that they canreproduceregional variations in the orientation of the stressfield that are of comparablemagnitudeto those observed. a) b) Figure 10 (a) Principal horizontal stress in the lithosphere far model El (ridge push only) of Richardson et al. (1979). Principal stress axes v)ithout arrows iÌenote compressían. (b) Principal horizontal tleyiatoric stress in the lithosphere for model E3l (ridge push, continental convergence, slab pull anil yiscous drag on the base of the lithosphere) of Richardson et al. (1979).Principal stressaxes without arraws denote compression. Axes with arrows denole tension. is not a singleentity but that the Australasianpart may consist of severalplates whose boundariesare def,ined by three broad zones of seismicity seen within the continent; Fitch et al. (1973)argued that the lithosphere acts as an efficient stressguide such that the seismicity within the continent reflects the response of the continent to variable forces acting on the plate boundaries. Estimatesof the stressesgeneratedat plate margins are little more than guessesand stressesassociatedwith density and structural anomalieswithin the plate interior are likely to be at leastasimportant. The separation of the local and regional stressf,reldsremains problematical as the above-discussed examplesshow. In the east and west of Australia this regional field appears to be predominantly east-west but it is more nearly north-south in south and central Australia. This regional or global stressfield must thereforeexhibit some azimuthal variation. Also, the superpositioningof a regional stresscould be at variance with the previous argument, that the Australian continent is relatively free from a contemporaryforce f,reldunlessthesestress levelsare below the brittle strengthof the upper mantle. Nevertheless,it is interesting that some of the models The Indian-Australian plate experienceshigher levels of seismicitythan do other intraplate regionssuchasthe Antarctic and American plates. From an analysis of plate motions, Minster and Jordan (1978)deduceda compressional deformationof about 1 cm yr- 1 occurring in a northwest-southeastdirection across the Ninety East Ridge. Weisselet al. (1980)also showed from the deformation of recent sediments.that the plate has been undergoing deformation since the late Miocene,possiblyin responseto the changingboundary conditions along the northern margin of the plate. They suggestthat an increasein resistiveforces along the Himalayanboundary could resultin the transmission of horizontal compressioninto the plate interior and that this could qualitativelyexplainthe east-west and northwest-southeastcompressivedirections in westernAustralia. We havenot examinedherethe geologicalevidencefor the past stressstate of the Australian continent but there appears to be very considerableevidence that suggeststhat compressionwas more the rule than the exceptionsince at least Palaeozoictime (e.g.Powell et al., 1984; Burek et al., 1978),both before and after the Mesozoicbreak up of Gondwanaland.If correct, the ridge-pushforcesare unlikely to play a major role in determining the overall stressfield. One platetectonicsforcenot considered in the Richardson et ø1. model is the stress associatedwith the change in curvature of the Earth's surface as the continents change latitude (Turcotte and Oxburgh, 1973).As a plate movesfrom high to low latitudesthe radius of curvature decreases and the associatednorthsouth stresses in the upper half of the plate should be tensional near its centre and possibly compressional near its edge.Magrritudesof thesestresses have been estimated to approach several kbars (Turcotte and Oxburgh,i973). Dooley (1983)suggested that the Aus- 738 S T A T EO F S T R E S S W I T H I NT H E A U S T R A L I A N CONTINENT tralian continent as part of Gondwanaland,may have spentmuch of its time moving from low to high latitudesand that this may explain the geologicalevidence in latePalaeozoicand for the predominantcompression Mesozoic time. In more recent time the continent has been moving equatorward,ye| compressionremains the dominant stress.The associatedstressespostulated by Turcotte and Oxburgh are basedon elastic shell theory. This is inappropriate for processesoperating and the actualstressstate on thesevery long timescales is likely to be much less than the above estimates. In addition to the stressfield associatedwith plate tectonics there is the possibility that the regional stressis in part an inherent property of the continent's mechanicaland thermal properties.This is suggested by the observationthat the compressionaxesare nearly have orthogonalto the coastline. Similar observations beenmade elsewhere;for example,along the Atlantic coast of the United States(Zoback and Zoback, 1980) and along the south and south-easternshore of China (Ding Yuanzhangper comm.,1983).The mechanismis not the lateral structural contrasts betweencontinent and ocean as discussedby Bott and Dean (1972), nor the flexure resulting from sedimentloading (Walcotl. 1972;Turcotte et al., 1977).Rather,a more appropriate mechanismmay be the one proposedby Fleitout and Froidevaux(1982)in which cold lower lithospheric material, in this case the subcontinentallithosphere, inducesa strongregional compressivestressthat is sufficient to support near-surfaceloads. Fleitout and Froidevaux cite the geophysicalevidencefor such a cold lower lithosphereunder the EuropeanAlps in support of their model. Similar evidencehas not been found under the easternAustralian highlands,but neither has the limited geophysicaldata been examinedwith such a model in mind for this or any other provinceof the Australiancontinent. CONCLUSIONS The seismig and < in situ > stressmeasurements, as emphasizedby most previous writers on the subject, provide overwhelmingevidencefor the predominantly compressivestate of the Australian continent,and that this stressis being releasedby brittle failure only in the upper and middle regionsof the crust.Geologicalevidence suggeststhat this compressivestate may have existedthrough much of geologicaltime since at least the early Palaeozoic. The seismicityof Australia occursin threemain zones. primarily with the In easternAustralia it is associated most deformed parts of the PalaeozoicFold Belt. Seismicity occurs down to depths of about 20 km and no eventshave been located with certainty in the lower crust below the Moho at a depth of about 4050 km (Finlayson et aI., 1979).The South Australian seismicityis restricted mainly to the late Proterozoic Adelaide Geosynclineand most of the eventsoccur in the upper crust within the thick sedimentarysequences folded into the syncline.The reasonsfor the seismicity in the Simpson Desert remain obscure, particularly becausethe solid geology is obscuredby the Quater739 nary cover which blankets the area. The seismicity of westernAustralia is diffuseand is well documentedonly for the southwesternarea. Here, the activity is located in the ArchaeanYilgarn Block and there is no obvious correlationwith any major tectonicfeatures. Much of the Australianseismicityappeaßto delineate broad zones of past weakness,such as the Adelaide Geosynclineor the Lachlan Fold Belt. That seismicity still occurs in these areas may reflect the relaxation of stress-differences associatedwith thesepast tectonic upheavals.As relaxation occurs at depth and the effective elastic thickness of the lithosphereis reduced, deviatoric stressesin the upper crust actually increase with time. During this relaxation process,however, the lithospherecools,viscouscreeprates decrease,and the layer increasesin strength.It remains uncertain which of the two competing effectsdominates; the migration of deviatoric stressinto the upper crust or the increasein strength of the lower crust with time. The Australian data does suggestthat the stress-differences may ultimately exceedthe strengthof the crust and that failure takes place long after the original teetonic process occurred. Superimposedupon this relaxation model will be changesin the ambientstressstateinduced by changingsurfaceloads suchas erosionand sedimentatior¡ or by changes in the plate tectonics regime such as the changing boundary conditions along the plate margin to the north of Australia. Relativelysmall changesin the overall stresspattern could be enough to trigger seismicity in regions where the relaxation processhas resulted in stresslevels that are near the critical failure limit. It is curiousthat wherethe non-hydrostaticstresses are greatest- as deducedfrom the gravity field - levels of seismicity are exceedinglylow. This indicates that featuressuch as the basin and arch of central Australia or the Perth Basin and Darling Fault of westernAustralia representstates of mechanical equilibrium between buoyancy,load, elastic and viscoelasticforces on the one hand and horizontal compressiveforces on the other hand and that theselast-mentionedforces must be of the order of 100-200MPa. In the model of the Perth Basin and the Darling Fault the resultant of the local and regional stressfieldsleadsto maximum compressionin the Yilgarn Block, east of the fault and this is wheresomeof the major seismicityin western Australiaoccurs. The erosionof the easternAustralianhighlandsleads to regionalisostaticrebound and the associatedstress field can be evaluated quantitatively. The results are tension near the surface of the central highlands and compressionnear the surfaceon the flanks. At depth this pattern is reversedwith compressionbelow the highlandsand tension below the flanks. Instead the observedstressstateis one of compressionthroughout the upper crust. Again a regional compressivestress must be invoked. The natureof the regionalcompressionremainsobscure. The observationsand modelssuggestthat the azimuth of this compressiveforce is variable and that this force may be nearly orthogonal to the coastline. Arguments can be marshalled for and against a plate tectonics K U R TL A M B E C KH, W S M ' O U E E N R , A . S T E P H E N S OD N, DENHAM observations. Theseassertions, like many othersin this paper,remainto be tested. cause,where the compression is a consequenceof forces acting on the plate margins and on the base of the lithosphere. Another possibility is that this regional field is a consequence of thermal and mechanical properties of old continents along the lines proposed by Fleitout and Froidevaux (1982), particularly if the compressive force has operated through much of geological time as is suggested by numerous geological A,cknowledgements David Denham publisheswith the permissionof the Director, Bureau of Mineral Resources, Geology and Geophysics. REFERENCES Banks R. J.o Parker R. L., Huestis 5.P.,7917.Isostatic compensation on a continental scale '. local ve-rsusregional mechanisms, Geaphys. J. Roy. astr. Soc., 51, 431-452. B. M. R., 1976.Gravity Map of Australi4 1 : 5,000,000Aust Min. Res..Canberra Australia. Bock G., Denham D., 1983. Recent earthquake activity in the Snowy Mountains region and its relationship to major faults, J. Geol. Soc. Aust., 30, 423-429. Bott M. H. P., Dean D.5., 1972. Stresssystemsat young continental margins, N ature Physical Science, 235, 23-25. Brace W. F., Kohlstedt D, L., 1980. Limits on lithospheric stress imposed by laboratory experiments, J. Geophys. Res., 85, 6248-6252. Burek P. J., Wells A. T., Loeffler 8.,1978. Tertiary tectonomagnetism in the Southe¡n Georgina and Northern Ngalia Basins, Res. School Earth Sci.,Aust. Nat. Univ., Annual Report, 59-62. Burke-Gaffney T. N., 1951. Seismicity of Australi4 Prac. of the Royal Soc. N ew South W ales,85,47-52. Journal and Chen W. P., Molnar P., 1983. Focal depths of intracontinental and intraplate earthquakes and their impliõations for the thermal and mechanical properties of the lithosphere, J. Geophys.Res., 88, 41834214. Christie J. M., Ord 4., 1980. Flow stress from microst¡uctures of mylonìtes : examples and current assessmenq J. Geophys. Res., 85, 6253-6262. ClearyJ. R., 1967.The seismicityof the Gunning and surrounding areas,1958-196I, J . Geol.Soc.Aust.,14,23-29. Cleary J, R., 7977. Australian seismicitystudies: a historical survey, EarthquakeI nformationBull., 9, 4-8. Cleary J. R., Doyle H. A,.,Moye D. G., 7964.Seismicactivity in the SnowyMountainsregionand its relationshipto geological structures, J. Geol.Soc.Aust.,11,89-106. ClearyJ. R., SimpsonD. W., 1971.Seismotectonics of the Australian continent, Nature, 230, 239-241. DaviesG. F., 1978.The rolesofboundary friction,basalshearstress and deep mantle convection in plate tectonics,Geophys.Res.Lett., 5.161-164. Denham D., AlexanderL. G., Worotnicki G., 1979.Stressesin the Australian crust : evidencefrom earthquakesand < in situ ) stress Bur. Min. Res.J. Aust. Geol.Geophys.,4,289-295. measurements, Denham D., ÄlexanderL. G., Worotnicki G., 1980.The stressfield nearthesitesoftheMeckering(1968)andCalingiri(1970)earthquakes, WesternAustralia, T ectonophysics, 67, 283-317. Denham D., Gibson G., Smith R. L., UnderwoodR., 1984.Source mechanisms and strongground motion from the 1982Wonnangatta and the 1966Mt. Hotham earthquakes, Aust.J. Earth Sci.(in press). DenhamD., WeekesJ., KrayshekC., 1981.Earthquakeevidencefor compressivestressin the southeastAustralian crust "I. Geol. Soc. Aust.,28, 323-332. DenhamD.; BockG., SmithR. S.,1982.The Appin(NewSouthWales) earthquakeof 15 November 1981,Bur. Min. Res.J. Aust. Geol. Geophys.,7, 219-223. DooleyJ. C., 1983.The Australianpolar wanderpath asan indicato¡ of stressdue to curvature variation in the earth'ssurface,G¿ol.Soc. Aust.Abstractsseries,9, 85-86. DoyleH.,4..,1971.Seismicityandstructu¡ein Australi4 NewZealand Roy.Soc.Bull.,9, 149-152. Doyle H. ,A.,EveringhamI. 8., SuttonD. J., 1968.Seismicityof the Australiancontinent,J. Geol.Soc.Aust.,15,295-312. DoyleH..{., Underwood R., 1965.Seismological stationsin Australi4 Aust. J. Sci, 28, 40-43. 740 DrakeL., 1974.The seismicityof New SouthWales,J. Proc.Roy.Soc. N ew SouthWales,107, 35-40. DrummondB. J., 1981.C¡ustalstructureof the Precambriante¡rains of northwestAustralia from seismicrefraction dafu Bur. Min. Res.J. Aust. Geol.Geophys., 6, 723-135. Everingham I. 8., 1966.Seismicityof WesternAustrali4 Bur. Res.Rept 132, Canberr4 Australia. Everingham L B., McEwinA. J., DenhamD., 1982.Atlasof isoseismal mapsof Australianearthquakes, Bur.Min. Res.Bulletin2l4. Everingham I. 8., SmithR. S., 1979.Implicationsof fault planesolutions for Australianearthquakeson 4 July 1977,6 l:|4ay1978and 25 November1978,Bur.Min. Res.J. Aust.Geol.Geophys.,4,297-301. FinlaysonD. M., ProdehlC., CollinsC. D. N., 1979.Explosionseismic profiles, and implications for crustal evolution, in southeastern 4,243-252. Australia,Bur. Min. Res.J . Aust. Geol.Geophys., I. 8., 1973.Mechanisms Fitch T. J., WorthingtonM. H., Everingham of Aust¡alian earthquakesand contemporary stressin the Indian Oceanplate,Earth Planet.Sci.Lett.,18, 345-356. Fleitout L., Froidevaux C., 1.982.Tectonics and topography for a Tectonics,l, 2I-56. lithospherecontainingdensityheterogeneities, FleitoutL., FroidevauxC., YuenD. ,{., i983. Dynamicsof the continentallithosphereand its associated stressfields,Inter-UnionCommissionof theLithosphereProgrammeand Abstracts,26-27,IUGG, Hamburg. GibsonG., WessonV., Cuthbertson R., 1981.Seismicityof Victoria to 1980,J. Geol.Soc.Aust.,28,34l-356. GordonF. R., 1971.Faulting during the earthquakeat Meckering, Roy.Soc.Bull.,9, WesternAustralia: 14 October1968,New ZealanrL 85-93. GordonF. R., LewisJ. D., 1980.The Meckeringand CaiingiriearthquakesOctober1968and March 197QGeol.Surveyof WestAustralia Bulletin, 126. Gray N. M., 1982.Di¡ection of stress,southernSydneyBasin,J. Geol.Soc.Aust.,29, 271-284. HanksT. C., 1977.Earthquakestressdrops,ambienttectonicstresses and stressesthat drive plate motions, Pure Applied Geophys.,ll5, 441-458. Hanks T. C., Raleigh C. 8., 1980.The conferenceon magnitude of deviatoric stressesin the Earth's crust and uppermost mantle, J. Geophys. Res.,85, 6083-6085. JaegerJ. C., ReadL., 1.969.Seismicityof southeastAustrali4 Am. Geophys.Un. monograph 13 Ed. P. J. Hart, 145-147 . Jeffreys H' 1932.On the stressesin the Earth's c¡ust required to Month. Not. Roy.Astron.Soc.,Geophys. supportsurfaceinequaliTies, Suppl., 3, 30-41. JeffreysH., 1970.The Earth, 4th editior\ CambridgeUniversityPress. Jones D. K., 1976.Perth Basin. In : Leslie,R. B., Evans, H. J., Knight C. L. Eds.,EconornicGeologyof Australiaand PapuaNew Guinea,3, Monogr. Aust. Inst. Mining Metallurgy,7, 1,08-126. KusznirN. J., Bott M. H. P., 1977.Stressconcent¡ationin the upper lithospherecausedby underlying viscoelasticcreep,Tectonophysics, 43.247-256. LambeckK., 1980.Estimatesof stressdifferences in the crust from isostaticconsiderations,J. Geophys.Res.,85, 6397-6402. Lambeck K.. 1983a.Structureand evolution of the intracratonic basinsofcentral Australi4 Geophys. J. Roy.astr.Íoc.,74,843-886. Lambeck K., 1983b.The role of compressiveforcesin int¡ac¡atonic basin formation and mid-plate orogenies,Geophys.Res.,Lett., 10, 845-848. S T A T EO F S T R E S SW I T H I N T H E A U S T R A L I A NC O N Ï I N E N T Lambeck K., 1984.Structure and evolution of the Amadeus,Officer and Ngalia Basinsof centralAustralia, Aust.J . Earth 9ci.,31,25-48. LambeckK., NakibogluS. M., 1980.Seamountloadingand stressin the oceanlithospherg J. Geophys.Res.,85, 6403-6418. LambeckK., NakibogluS. M., 1981.Seamountloading and stress models. in theoceanlithospherg2.Viscoelasticandelastic-viscoelastic J. Geophys.Res.,86, 6961-6984. LambeckK., PenneyC., 1984.Teleseismictravel time anomaliesand crustal structure in central Australia, Phys. Earth Planet. Int., !, 46-56. Lewis B. T. R., Dorman L. M., 1970.Experimentalisostasy.2. An isostaticmodel for the USA derived from gravity and topographic Res.,75, 3367-3386. data.J. Geophys. LewisJ. D., DaetwylerN..{'., BuntingJ.,{., MoncrieffJ. S., 1981.The Cadoux earthquake,2 Jtne 1979,Geol. Survey of West Australia Report11. during McCueK. F., SuttonD. J., 1979.SouthAustralianearthquakes 1976and1977,J . GeoI.Soc.Aust.,26'237-236. McGarr 4., Gay N. C., 1978.Stateof stressin the earth'scrtsl Ann. Rev.Earth Planet. Sci.,6, 405-436. McKenzieD. P., 1969.The relation betweenfault plane solutions for earthquakesand the directions of the principal stress,Ball. Seismol. Soc.4m., 59, 591-601. stresses neartopoMcTigueD. F., Mei C. C., 1981.Gravity-induced graphy of small slope,J. Geaphys.Res.,8ó, 9268-92'78. Mills J. M., Fitch T. J., 1977.Thrust faulting and crust-uppermantle J. Ray.astr.Soc.,48,351-384. structuresin eastAust¡ali4 Geophys. Minste¡ J. 8., Jordan T. H., 1978. Present-dayplate motions, J. Geophys.Res., 83, 5331-5354. MuirheadK. J., 1981.Seismicityinducedby thefilling of the Talbingo Reservoir,J. Geol.Soc.Aust.,28,297-298. Ollier C. D., 1978.Tectonicsand geomorphologyof the eastern highlands.In : Davies,J. L., Williams, M. A. J. Eds.,Landform Evolution in Australasi4 Aust. Nat. Univ. Press,Canberr4 5-47. PowellC. McA., ColeJ., CudahyI., 1984.Megakinkingin the Lachlan Fold Belt (Abstract),Int. Conf. on Multiple Deformation and Foliation Development 135-136,Bermagui. Raleigh C. 8., Healy J. G., Bredgehoeft,J. D,, 1972.Faulting and crustal stressat Rangely,Colorado, Am. Geophys.Un., Monogr., 16,275-284. RichardsonR. M., SolomonS. C., SleepN. H,, 1979.Tectonic stress SpacePhys.,17,981-1019. in the plates,Rev.Geophys. Stephenson R. Ä.., 1984.Flexur¿l models of continental lithosphere 741 basedon the long term decay of topography,Geophys.J. Roy. astr. Soc.77,385-413. R.,4'.,LambeckK., 1984.Erosion-isostaticreboundmodels Stephenson for uplift : An application to southeasternAustrali4 Geophys.J. Roy. astr. Soc. (in press). StewartI. C. F., DenhamD., 1974.SimpsonDesertearthquakeCentral Australi4 August 1972,Geophys. J. Roy. astr.9oc.,39,335-341. StewartI. C. F., Mount T. J., 1972.Earthquakemechanismsin South Australiain relationto plate tectonics,J. Geol.Soc.Aust.,19,4l-52. StewartI. C. F., SladeÄ..,SuttonD. J., 7973.South Australianseismicity,J. GeoLSoc.Aust.,19,441-452. SuttonD. J., McCueK. F., BugejaÄ., 1977.Seismicityof the southeast of South Australi4 J. Geol. Soc.Ãust.,25, 357-364. Sutton D. J,, White R. 8., 1968.The seismicityof South Australia, J. Geol.Soc.Aust.,15,25-32. Sykes L. R., 1970.Seismicityof the Indian Ocean and a possible nascentisland arc betweenCeylon and Australia J. Geophys.Res., 75, 5041-5055. TurcotteD. L., Âhern J. L., Bird J,M., 1977.The stateof stressat 42, l-28. continentalmargins,T ectonophysics, Turcotte D. L., Oxburgh E. R., 1973.Mid-plate fectonics,Nature, 24,337-339. UnderwoodR., 1967.The seismicnetwork and its applications,Ph. D Thesis,Aust. Nat. Univ., unpublished. Underwood R., 1972. Studies of Victorian seismicity, Proc. Roy. Soc. Victoria, 85, 27-48. Vening MeineszF. ,4'., 1948.Gravity expeditionsaT se4 1923-1918, Delf! Mulde¡ 233 p. Walcott R. I., 1972.Gravity, flexure,and the growth of sedimentary basinsat a continentaledge,BuIl. Geol.Soc.4m.,83, 1845-1848. WeisselJ. K., AndersonR. N., Gelle¡ C. A.,,1980.Deformation of the Indo-Australian Plate, N ature, 287, 284-291. WelfmanP,, 1979.On the Cainozoic uplift of the southeasternAustralian highland,J. Geol.Soc.Aust.,26, l-9. Worotnicki G., DenhamD., 1916.The stateof stressin the upper part of the ea¡th'scrust in Australia accordingto measurements in mines and tunnels and from seismicobservàtions.Proceedingsof ISRM Symposium,Investigation of Stress in Rock, Advances in Stress Measuremen! Sydney, Australi4 71-82. Wyss M., Ed, 19'/7.Stressin the Earth, Pure Appl. Geaphys.,ll5, Zoback M. L., Zoback M,, 1980.Stressin the conterminousUnited Res.,85,6113-6156. States,J. Geophys.