Variationsin InelasticFailure of SubductingContinental Lithosphere and TectonicDevelopment: -Banda Arc Convergence

KushTandon •, Juan M. Lorenzo2, SriWidiyantoro 3,and Geoffrey W. O'Brien4

Factorsthat controlthe geometryof forelandbasins and accretionaryprisms, back thrusting,and changesin structuralstyle duringthe continentalcollision can be linkedto variationsin EffectiveElastic Thickness (EET). A variableEET map is computedat an incipient continentalcollision (Pliocene-Recent)site in the northernAustralian continental lithosphere along the Bandaorogen. Incorpora- tion of almostall the forearcbasin within the centralTimor Island andtomogram in the easternTimor Island suggesta more rigid northernAustralian lithosphere indenting between •125ø-127 ø E longitude. Modeled flexure deflection is matchedto seafloorbathymetry and Banda orogentopography, Pliocene conti- nental shelf, and marine complete 3D Bouguer gravity anomalies.Calculated EET of the northernAustralian continental lithosphere using an elastichalf-beam model show a sharpdecrease in EET from 230-180 km on the continentalshelf (from Roti to west of Am Island) down to •40 km on the continentalslope and beneathBanda orogen (from • 121ø-132ø E longitude)favoring an hypothesisof inelasticfailure at the start of continentalsubduction. The decreasein EET along the continental slope is not entirely uniform and parallel'to the subduction boundary,especially south of centralTimor Island.In the southof centralTimor Island, the EET remainsclose to 55 km on the continentalslope and beneaththe central Island. An increase of •15 km in EET calculations on the continentalslope south of centralTimor Island shiftsthe majority of present-day strainpartitioning to the forearcregion as comparedto regionsin westernTimor Island and Tanimbar Island. A pre-existingstronger of continentallitho- spherepresent within the transferzone between Malita andValcun Graben(con- tinental shelf) is a likely causefor lesserinelastic failure in that region during subduction.Results show that continentallithosphere is inherentlystrong on the continentalshelf under normal crustal thicknessand in absenceof significant weakeningfrom past rifling. Variations in the gradientof EET changeon the continentalslope can determinethe evolutionof many featuresin continental collision tectonics.

•Departmentof Geology and Energy Institute, University of Louisiana, Lafayette, Louisiana 2DepartmentofGeology and Geophysics, Louisiana State Uni- versity, Baton Rouge, Louisiana PlateBoundary Zones 3Departmentof Geophysics andMeteorology, Bandung Insti- GeodynamicsSeries 30 tute of Technology,Bandung, Indonesia Copyright2002 by theAmerican Geophysical Union 4AustralianGeological Survey Organization, GPO Box 378, 10/1029/030GD20 Canberra, Australia

341 342 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

121 E 123 125 127 129 131 133 135 137 -4 S

-6

-'- ß -8 lore • - imor •• Australian continental • - Savu •, • •, lithosphere -10• Basin z -• X•. ArafuraBasin

Australi,. , />' Bona•"'-.•t-•. ," • " ': ' Oceanic• e :as-• • • Australia ' -14 Lithpspshe - Legend ---- = ExtensionalFaults :-',•= Depocenter(Paleozoic & Mesozoic) ß = ActiveVolcano o = InactiveVolcano •, = HeatFlow (mW/m 2) AP = Ashmore Platform MG = Malita Graben PS = Petrel Sub-basin SP = Sahul Platform SS = Sahul Syncline VG = Vulcan Graben

Figure 1. Generaltectonic map for thenorthern Australian continental lithosphere across the Banda Arc [Hamilton, 1974;Veevers et al. 1974;Stagg, 1993; AGSO North West Shelf Study Group, 1994; Snyder et al., 1996].Heat flow valuesare from Bowin et al. [1980].The thick line with blacktriangles (point to overthrustplate) shows the subduc- tion boundarybetween the northernAustralian continental lithosphere and the Eurasianlithosphere in Timor- Tanimbar-AruTrough; north of Flores,in Alor and Wetar,it denotesthe back-thrusting.Thick black line lettered "AGSO98R07" marks the locationof seismictrack for AGSO line 98R07.The 200 rn bathymetrycontour marks the continentalshelf-slope boundary from ETOPO-5 [National Geophysical Data Center, 1988].

1. INTRODUCTION especiallyin the downdip directionshould play an impor- tant role in determiningthe progressof the continental Local area Airy isostasy(hydrostatic equilibrium) is collision tectonics and how foreland basins evolve with generallyfound only in large-scalegeological features, time. such as Tibetan Plateau [Lyon-Caen and Molnar, 1985; EffectiveElastic Thickness (EET) measuresthe ability Burovand Diament,1992 and 1996].In Himalayasand to supportloads and transmit bending stresses regionally as othermountains chains, the flanksof the mountainranges a rigidlithosphere rather than mass compensation being in showrigid behavioras opposedto localarea Airy isostasy a local Airy isostaticmanner (zero EET) [McAdoo et al., [Lyon-Caen and Molnar, 1983 and 1985; Royden and 1978;Watts, 1983]. There has been a long-standingdebate Karner, 1983; Stewart and Watts, 1997]. This enablesa on factors that control the value of EET in the continental part of the continentalcollision margin to supportthe lithosphere,like inelasticfailure, crustal thickening, and weight of the thrust sheetsand lead to formationof fore- tectono-thermalage (the time elapsedbetween the lastrift- landbasins [Lyon-Caen and Molnar, 1983and 1985;Roy- ing eventand the startof formationof a forelandbasin) den and Kamer, 1983; Stewartand Watts, 1997]. Within [McNutt et al., 1988;Burov and Diament, 1995 and 1996; the samecontinental collisional margin, areas with similar Stewartand Watts, 1997; Poudjom Djomani et al., 1999]. convergencevector also display markedly different tec- Wattset al. [ 1995] advocatethat the highestthe valueof tonicdevelopment (Figure 1). Differencesin the rigid part EET should be found where the 'collisionhas maximum of the continentallithosphere along a collisionalmargin, shortening.Pre-collisional high EET facilitates thin- TANDON ET AL. 343

skinned, low-angle fold and thrust tectonics, and back- thrustingon a low-angle flexurally more rigid plate [Watts et al., 1995]. EET values differing by 1.5 times within a 200 kilometers range are noticed in the Himalayas and Ganga basin [Lyon-Caenand Molnar, 1985]. In the Apen- nine foreland basin system,Royden et al. [1987] showed that the differenceswithin subductionprocesses, foreland basin and thrust sheetgeometry can be related to the seg- mentationof the lithospherewith uniform EET. In the cen- tral Brazilian Shield, Stewart and Watts [1997] calculated variable EET of > 85 km on the craton that decreases to < 25 km beneath the thrust sheets. In the central Brazilian Shield [Stewart and Watts, 1987], the EET variations are not uniform and not strikeparallel to the thrusting. At a site of a Pliocene-Recent continental-arc collision, Figure 2. Uninterpretedmigrated seismic reflection AGSO line the northern Australian continental lithospherealong the 98R07 on the northern Australian continental shelf southwest of Banda orogen [Warris, 1973] one observes significant Timor Island. Age abbreviations: Plioc(ene)-Rec(ent), variationswithin different elementsof plate-boundaryin- Mioc(cene), Pal(eocene), Eo(cene), J(urassic)-K(Cretaceous), teractions,e.g., strainpartitioning, structural style, geome- Tr(iassic), and P(ermian).The seismicsection shows how the ex- tent of normal faulting changeswith time. Some faults appearto try of the foreland basin and the adjoining accretionary detach within Cretaceous units. Most reactivated faults terminate prism, and presenceof back-thrusting.In this study, we above, or near, a regional basePliocene unconformity [Boehme, attempt to find a relationshipbetween variationsin EET 1996; Curry et al., 2000] that constrainsthe end of the dominant (down dip and along the strike of the thrust sheets)and extensional stressregime to Latest Miocene/EarliestPliocene. different plate boundary observationsfound during the The locationfor AGSO line 98R07 is shownin Figure 1. juvenile stagesof continentalcollision tectonics. We per- form elastichalf-beam modelingalong 15 profiles (Figure 6) to calculate variable EET from Roti to the Kai Plateau (west of Aru Island) traversingthrough Timor and Tanim- ton and Bowin, 1981]. Regionally,the timing of termina- bar Islands(- 121ø-137ø E longitude)(Figure 1). Modeled tion of the majority of normal faults migratesfrom Late flexure from elastichalf-beam modeling is matchedto the Miocene-Early Pliocene-Recentas one moves from west seafloor, the Banda orogen, the continental shelf from of Timor Island toward Tanimbar Island [Curry et al., Pliocene, and the marine complete 3D Bouguer gravity 2000] (Figure 3). Interestingly,the strain partitioningin anomalies.Variable EET on the continentalslope accounts the forearc basin north of central and eastern Timor Island for the changein the mechanicalstrength of the continental is under compressionunlike north of western Timor and lithospheredown dip causedby inelasticyielding [McNutt Tanimbar Island [Charlton, 1997] (Figure 3). Foreland et al., 1988]. vergance has stopped at present in central and eastern Timor Island but is still active in western Timor and 2. DIFFERENT FEATURES OF CONTINENTAL Tanimbar Islands [Charlton, 1997] (Figure 3). The width COLLISION TECTONICS of the forearc basin north of central Timor Island is also the least comparedto other north of the Banda At the start of continental subduction, there are normal orogen (Figure 1). In the vicinity of Timor Island, the faults due to bendingstresses on the continentalshelf and bendingstresses caused by continentalsubduction are neu- slope[Bradley and Kidd, 1991], thrustingwithin the accre- tralized by horizontal compressionas soon as the subduc- tionary prism [Karig et al., 1987] and extensionin the tion started. On the contrary, near Tanimbar Island, the forearcbasin [Uyeda and Kanomori, 1979]. However, the bending stressesstill exceed the horizontal compression seismic reflection data from the northern Australian conti- (Figure 3). nental shelf-slope(Figure 1) show that the extent of nor- The P-seismicwave tomographyindicates that the north- mal faultingis reducedsignificantly in Late Miocene-Early northernAustralian continental lithosphere north of eastern Pliocene strata on the continental shelf south of Timor Is- Timor Island is more detachedfrom the leading edge of land (Figure 2). The Late Miocene-EarlyPliocene is gen- the subductingAustralian lithosphere than near western erally thoughtof as the time when continentalsubduction Timor and TanimbarIsland (Figures3 and 4). In addition, started in northern Australia [Veevers et al., 1978; Johns- central Timor Island has highest topographicelevation 344 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

121 E 123 125 127 129 131 133 135 137 -4 S

-6

•' • •li.•w•al-•.-, : -, -,•-'" Tanimbar -8 - . .... • Alor •ck-thrusts '• ' i ß. '" •' "" ß ' .. f, -land ;• ß -• •,• Jmo ve ,.-nce Savu Basi -10 " t I N, , al faulting • forel"'•• I not.ant .12I =•j•e t•••l•!i;! EarlyIIAt ornear .. ßcane io I P!iocene the seafloor I I I I •, napart• •" '" . , ! !B..in Australia -14

Figure 3. Present-daydominant structural style on the continentalshelf, near the subductionboundary and in the forearcbasin. Age of terminationof majority of normal faulting on the continentalshelf of the northernAustralia is mappedby thousandsof kilometersof seismicreflection data by Curry et al. [2000] (shownas italic on the continental shelfwith verticaldashed line as boundaries).There is still evidencefor present-dayextensional faulting (caused by bendingstresses) as seafloorscarps on the northernAustralian continental slope near Timor Islandbut with a reduced arealextent in theshelf-slope compared to LateMiocene-Early Pliocene [Boehme, 1996]. Different regions of present- day strain partitioning near the subductionboundary and in the forearc basin are shown in underlinedbold italics [Charlton,1997]. The thick NW-SE lines(a, b, c in Figure4) arethe profilesfor seismictomograms.

comparedto the other islandsin the Bandaorogen (Figure Even the decreasedbending stresses need to be neutral- 5). These observationsseen on the northern Australian ized by increasedhorizontal compression within northern continental shelf, Timor-Tanimbar-Aru Trough, Banda Australianshelf and Timor-Tanimbartrough for normal orogen and Banda Arc could be respondingto regional faultingto stop(Figure 2). A morerigid northernAustra- changesin EET. lian continentallithosphere on the continentalslope and beneaththe orogen(higher EET values)will promotethin- skinnedtectonics on an indentingsurface that supports 3. WORKING HYPOTHESIS more loading. That may causeincreased horizontal com- pression,as seen in the vicinity of central Timor Island. Quite possibly,the cessationof normal faulting seenat However,higher EET valueswill alsohave more bending Miocene-Plioceneon seismicreflection data (Figure 2) is stressescompared to areas of lesserEET values. Another becausethe extensionalstresses were greateron the north- mechanismfor increasinghorizontal compression during em Australian continental lithosphere when it was far collision is break-off of the lithospherewithin the sub- away from the continental subductioncompared to its ductedpart [Chemendaet el., 2000]. position today near the shelf-slopeboundary. One of the Higher EET valueson the continentalslope will resist possiblemechanisms for reducing extensionalstresses is subductionand can shift the locus of present-daystrain decreasingthe bending stresses[Turcotte and Schubert, accommodationto the forearcregion duringthe continen- 1982]. Bending stressescan be reducedby decreasingthe tal-arc collision. Similarly, higher EET continentallitho- EET (less rigid) of continentallithosphere at the start of spherebeneath the orogencan also aid in the detachment the continental subduction and/or the decrease the curva- of the shallow parts of continentallithosphere from the ture of the lithosphere(less bent) [Turcotteand Schubert, oceanic lithospherealready subducted.A buoyant, de- 1982]. tachedcontinental lithosphere could stop the subduction. TANDON ET AL. 345

1989]. The northernAustralian continental lithosphere also underwentNW-SE striking Late Devonian-EarlyCarbon- iferous rifling (Petrel sub-basinin Figure 1) and ENE- 410km WSW strikingJurassic (Malita grabenin Figure 1) [Harris, 1991]. At DSDP 262 site in the Timor Trough (Figure 1), 660km the transitionfrom shallow-waterto deep-watersedimenta- tion in the Early Plioceneindicates that the subductionof (a) northernAustralian continentallithosphere started at least 1075km 1543km 3 Ma ago near westernTimor Island [Veeverset al., 1978; ...... -.e•--?:'::-:----:'-' :i•<'"'"":%•.: '/':'-'711)- '...... •...... Johnstonand Bowin, 1981]. ii:::;":"•':• if':.:.::: .!%o.. -.: :• ....•:•..:..;•)..:.::::..:...::..(•:;.•;:...... The northernAustralian continental lithosphere south of ...... •;•i:•& •:::•. -::-:-:-:.•;;;-..•::..,-. Roti near AshmorePlatform (Figure 1) is now estimatedto be converging with the Eurasian lithosphere[Warris,

660km 1973] at a rate of 74 +/- 2 mm/yr in a N 17 øE +/- 3ø direction[DeMets et al., 1994] (Figure 1). Analysisof GPS [Genrich et al., 1996] and earthquake seismology (b) --• 1075• [McCaffrey and Nabelek, 1986] data alsoshow that at pre- sent the subductionwithin the Timor Trough has stopped and the majority of strain is being accommodatedwithin the forearc basin and the back-thrustfaults [Kreemer et al., 2000] (Figure 1). Recent GPS calculationsnear Darwin (Australian Shelf) indicate that the northern Australian continentallithosphere-Eurasian lithosphere as a block is convergingwith regardto the Sundashelf at a 7% slower rate comparedto DeMets et al.'s [1994] estimationbut in (c) • -'• 1075km the samedirection [Genrich et al., 1996]. Earthquakefoci northof the Bandaorogen at intermedi- Wave speedrelative to ak135 ate depths(50-100 km) alsoindicate that the leadingedge •'* };i;:::•i•i,•&::,,;....:•p•,•i•;;...... :.•,•.•:•½:.:.:•...... of the Australianlithosphere is presentlydetaching at those depths[McCaffrey et al., 1985; McCaffrey, 1988; Charl- ton, 1991], as also seen in tomograms(Figure 4). The -1.5% slow fast +1.5% isostaticrebound caused by partial decouplingof the Aus- tralianlithosphere is evidencedby uplift of the Bandaoro- Figure 4. Vertical tomogramsacross: a) westof Timor Island,b) gen and the BandaArc islands[DeSmet et al, 1990]. The throughthe easternTimor Island,and c) TanimbarIsland from a tomograms(Figure 4) do not indicatecomplete detachment P-wave tomographicmodel by Widiyantoro and van der Hilst of the leadingedge of the northernAustralian lithosphere. [1996]. The tomogramsare plotted from the north (left) to the south (right) in terms of wave speedperturbations relative to ak135, a radially stratifiedvelocity model [Kennettet al., 1995]. 5. DATA The blue color is the inferred subductingnorthern Australian lithosphere.Locations are shownin Figure3. Modeled flexure is matched to ETOPO-5 bathymetry data that is gridded at 5 minute (• 9 km in the area of study) intervals[National GeophysicalData Center, 1988] 4. GEOLOGICAL BACKGROUND (Figure 1). We did not use Smith and Sandwell's[1997] bathymetrydatabase as a very high-resolutionbathymetry The northernAustralian continentallithosphere started convertedfrom gravity can run into nonlinearproblems on continental subductionacross the Banda orogen in Late the continentalshelf [Mackenzie, 1997]. More importantly, Miocene-Early Pliocene [Veevers et al., 1978; Johnston we want to, keep the bathymetry[National Geophysical and Bowin, 1981]. The earliestrecord of collisionbetween Data Center, 1988] and gravity database[Sandwell and Australianand Eurasianlithosphere in Timor Island is Smith, 1997] independentof each other. A sedimentary markedby the presenceof •38 Ma (Late Eocene)meta- surface marking the base of Plioquaternarysediments morphicoverprint on ophiolites[Sopaheluwaken et al., [Curry et al. 2000] from Australian Geological Survey 346 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

lOOO Above Organization(AGSO) seismicreflection surveys(Figures ß central Timor Island 8oo sea-level ß easternTimor Island 6 and 7) is also used. Top of the 600 westernTJmorelsland e Satellite-derived marine free-air gravity anomaly data Banda orogen accretionary400 [Sandwi•11and Smith, 1997] across the BandaArc areused prism(m) 200 to obtain complete3D Bouguergravity anomalies(Figure e Rotl Island ß TanlmbarIsland 0 , , 8). A 3D Bouguer gravity anomaly correctsfor limited '100 RotiIsland ,'}rtrt•..vv Kai Pl•eau -2OO Below length-scaleof the density contrastalong the strike direc- sea-level : -4OO (Bandaomgen Islands •n T•lm•risla• a• •1 Plateau) tion [Blakely, 1995]. Whereas,2D Bouguer anomaly as- -6OO sumesthat the anomaly being causedby the densitycon- -8OO trast is infinite in extentalong the strikedirection [Blakely, Widthof the Bandaorogen accretionary prism (kin) 1995]. Sandwell and Smith's[ 1997] data is preparedfrom Geosat Geodetic Mission and ERS 1 Geodetic Phase with a Figure 5. The width of the Banda orogen measuredfrom the horizontal resolution of-10 km. The accuracy of the back-stopto the Timor-Tanimbar-AruTrough and the maximum worldwide satellite-derivedfree-air gravity anomalyis 3-7 topographicelevation within the Banda orogen measuredfrom mgal [Sandwell and Smith, 1997]. Marine complete3D the 15 profiles(transects shown in Figure 7) usedfor elastichalf- Bouguergravity anomaliesare createdfrom the satellite- beam modeling.Note that the central and easternTimor Island derived free-air gravity anomalies[Sandwell and Smith, has the maximum elevation comparedto other areas and also shows evidence that the area (the Wetar thrust) is becominga 1997] by replacing the seawaterwith Tertiary sediments doubly-vergentorogen [Kreemer et al., 2000]. (Table 1). The 3D bathymetriccorrections [Blakely, 1995; Tandon,2000] are appliedusing ETOPO-5 [NationalGeo- physicalData Center, 1988]. A seismictomogram based on the regionalmodel for In- donesia developed by Widiyantoro and van der Hilst

121 E 123 125 127 129 131 133 135 137 -4 S

-6

-8

-10

-12

' ?• B•asin•, '• Australia -• -14 ' -;/ ' .... ';'."'

PlateBoundary __ = Extensional Faults

Figure 6. The dashedlines locate numbered transects used for the elastichalf-beam modeling for variableEffective ElasticThickness (EET) modeling.The box showsthe regionwhere we reconstructthe continentalshelf during Plio- ceneto be usedfor flexure modelingin additionto seafloorbathymetry, Banda orogen topography and marinecom- plete3D Bouguergravity anomalies. TANDON ET AL. 347

124 E 125 126 127 128 129 130 131 132 i i Timor

. ::,•:•:.:::..::.::.:.-:'•:..;;:....•.=•.'•:...... :.::.;.:...... -...?.', •...... -'.

:•.•:.:;••m..... •. _ ..•'.:• •.:• ...... •"•'"'••••••••••••••:'•:• ....•:•. ? •:•?•'•½•::•m•...... ;' '

-13

100 200 300 400 500 600 700 800 900 1000

mbsf

Figure7. Griddedmap of the Plioquaternary sedimentary surface in mbsf(meters below sea floor) created by mapping thousandsof kilometersof seismicreflection data [Curry et al., 2000].Numbered dashed lines are transects for elastic half-beammodeling within that region. The contours are for seafloorbathymetry. The thickness of Plioquaternary sedimentsisdetermined bysubtracting thebase Plioquatemary (sedimentary) surface from the seafloor bathymetry.

121 E 123 125 127 129 131 133 135 137 I i i i i i i i i i i i i -4S ':•.4½,•:•...,,.,,• ' ..... :• .• ' '

-6 • "• B;

<-100-100 -80 -60 -40 -20 0 20 40 60 80 100 >100 mgal

Figure 8. Marinecomplete 3D Bouguergravity anomalies for thenorthern Australian continental lithosphere across the Bandaorogen. The gray shadedarea corresponds to the portionof the BandaArc abovethe sealevel with no ma- rinesatellite data coverage [Sandwell and Smith, 1997]. Geological information overlain is fromFigure 1. 348 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

Table 1 High EET values on the continentalshelf combinedwith Parameters for Flexure Modeling EET values decreasingdown dip on the slopewill have a forebulge with lower amplitude comparedto a constant Name Value high EET estimate.As it is, a constanthigh EET estimate Parameters for Elastic Half-Beam Modeling Elastic Beam (> 80 km) createsa shallowbut long wavelengthforebulge Young'sModulus (E) 5 X 1010kg/ms 2 on the continental shelf that can be further decreased if the Poisson'sRatio (¾) 0.25 value of EET decreasesdown dip. DecreasingEET on the continentalslope due to inelasticfailure decreasesthe ca- Timor-Tanimbar-Aru Trough (Sea water) pacity of the continental lithosphereto transmit stresses Density 1030kg/m3 and doesnot produceidealized forebulges [McAdoo et al., 1978]. In contrast,a constantlow EET (-20-40 km) value Supporting Fluid (Mantle) createsa low wavelength,high amplitudeforebulge on the Density 3300kg/m3 shelf. Our observationshows lack of high amplitudefore- Triangular Wedge (Banda orogen) bulge seen on the seafloor bathymetry [National Geo- Density 2670kg/m 3 physicalData Center, 1988] and from Pliocenecontinental shelf on the continental shelf of northern Australia Parameters for Gravity Modeling [Lorenzo et al., 1998; Tandon et al., 2000]. Evidencefor Density significantreduction from high EET valueson shelfto low on the slopeis the presenceof high curvature(•--10 -7 m -l) Seawater 1030kg/m3 on the slope along with no clear evidence for a small Tertiarysediments 2200kg/m3 Bandaorogen 2670kg/m3 Elastic half-beam flexure model with Australiancontinental crust 2800kg/m3 Mantle 3300kg/m3 varying effective elastic thickness (EET)

EET Table 1. Parameters[Chamalaun and Grady, 1976; Turcotteand (km) Schubert,1982; Richardson,2001] usedfor a) elastichalf-beam, and b) gravity modeling.All the unitsare definedin SI. •==" N"-- IIIOrogen Bancla Aru Timor-Tanimbar-Trough (sea water)AustraliJ [ 1996] is shownin Figure4. The modelhas been produced [i?•F•;':: ..:..;"',"..'i...... Triangular ...... by a cellular representationof the mantle structure.The !i:.•i..::?•.:•"Poi:nt :•' Load ...... :::::LOad.. mantle volume inside and outside the study region was o O O '"" Elastic parameterizedusingcells of 1.0ø x 1.0 and 5.0 x5.0 , Beam

respectively.The abovementioned model parameterization ...... allows the resolutionof relatively small-scalestructure in the region under investigationand minimizes contamina- tion from outside.The P-wave delay time data set inverted to constructthe tomographicmodel is from the global data Supporting Fluid processedby Engdahlet al. [ 1998].

6. ASSUMPTIONS 10km i 100 km 6.1 Elastic Half BeamModels Figure 9. An elastic half-beam model with laterally variable Ef- fective Elastic Thickness(EET) is used to calculatethe EET of We approximate the northern Australian continental the northern Australian continental lithosphere.The Banda oro- lithospherewith a 2D elastichalf-beam model of variable gen in the elastchalf-beam model is approximatedby a triangular effective elastic thickness(EET) embeddedin an inviscid load. The surfacial deflection created by the elastic half-beam fluid [Turcotteand Schubert,1982] (Figure 9). A laterally model is filled with seawater. The end-point load at the northern variableEET model incorporatesthe decreasein EET due end of elastic half-beam approximatesthe subsurfacecomponent to inelasticprocesses, such as brittle and plastic failure, of loading.Laterally variableEET model has the EET decreasing and mechanicaldecoupling between upper-lower crust and from Australia to Timor-Tanimbar-Am trough to Banda orogen mantle [McNutt et al., 1988]. assuminginelastic failure. TANDON ET AL. 349 wavelength forebulge on the shelf [Lorenzo et al., 1998]. 6.2 ReconstructingContinental Shelf in Pliocene Lack of significantforebulge is not due to horizontalten- sile forces [Bowin et al., 2000] becausethe stressregime The thickness of the Plioquaternarysediments and the on continental shelf does not show extensionbeing domi- flexure causedby the weight of Plioquaternarysedimenta- nant today (Figures2 and 3). tion is removed from the seafloor to reconstruct the conti- The northern Australian continental lithosphere since nental shelf that existed in Pliocene at the start off conti- Plioceneis assumedto bend underthe weight of the Banda nental subduction.Based on present-dayconvergence rates orogen acting as a surfaceload (Figure 9). Ideally, using between Australia and (Figure 1), in Early Plio- the Plioquaternary sedimentsfrom shelf to slope along cene the current shelf/slope transitionwas-350 km fur- with Banda orogen sedimentswould be a more accurate ther south and therefore was not significantly affected by descriptionof surfaceload. However, Plioquaternarysedi- the bendingof the oceanicpart of the northernAustralian ments are only few hundredmeters thick [Veevers, 1974] lithosphere.The thicknessof the Plioquaternarysediments comparedto 1Os of kilometer thick Bandaorogen [Snyder is calculatedby subtractingthe basePliocene surface con- et al., 1996]. The Banda orogenacts as a major contributor structedfrom seismicsurveys (Figure 7) with the seafloor to the surface loading. We only have limited information bathymetry.Since, we do not have regional information regardingthe thicknessof Plioquaternarysediments on the aboutthe porosity-depthrelationship, therefore we do not shelfwhen comparedto the total areacovered by our study decompactthe older sediments(Miocene onwards) as done (Figure 6). We do not have any detailedsedimentary thick- in other studies[van der Meulen et al., 2000]. We use EET nessinformation on the slope.To perform a regionalEET = 170 km as the EET of the Australian continental shelf analysisfrom Roti to Kai plateau,we need to be consistent prior to the continentalsubduction which we load with for comparison. In elastic half-beam modeling, a Plioquaternarysediments. The choice of pre-continental subsurface loading component is only invoked when subductionEET is basedpresent-day Australian continen- surface loads alone cannot account for the deflection tal shelf EET values that range from 230-125 km travers- observed of reference horizon data. These subsurface loads ing from Roti to Kai Plateau(see the Resultssection). The originate in deeper parts of the lithosphere,as found in thicknessof the Plioquaternarysediments and the flexure some other foreland basins, e.g., Apennine foreland causedby it is subtractedfrom the seafloorto constructa [Royden and Karner, 1984]. A slab-pullcan act as hidden Pliocene continental shelf. subsurfaceload uponth e northernAustralian continental lithosphere.On the other hand, a completedetachment of 6.3 Marine Complete3D BouguerGravity Anomalies the northern Australian continental lithospherefrom the oceaniclithosphere would mean that there is no slabpull. Marine complete3D Bouguergravity anomalies (Figure A partial detachment or buoyancy of the continental 8) remove the gravity effect of seafloor-waterinterface lithospherewould cause the effect of slab pull to be [Blakely, 1995] and allows us to delineateintra-crustal buffered. A buffered slab pull means that the full down- heterogeneitiesand the shapeof the Moho. The subduction ward pull by the denseoceanic lithosphere is not felt by of the flexed Moho at the outer trench slope and beneath the northernAustralian continental lithosphere. We invoke the accretionaryprism createsa negativeBouguer gravity the conceptof a bufferedslab because the northernAustra- anomaly on the order of 100 mgals spreadover several lian continentallithosphere is assumedto be detaching hundred kilometers. Beneath the orogen, a high-density from the oceanic lithosphere[Chamalaun and Grady, subsurface load can create a positive Bouguer gravity 1978; McCaffrey et al., 1985; McCaffrey, 1988; Charlton, anomaly, thereby making a positive/ negative Bouguer 1991]. Other subsurfaceloads include ophiolites,like the gravity anomaly couple from the orogen to the craton onesfound in Timor Island [Sopaheluwakenet al., 1989]. [Karner and Watts, 1983]. Seafloorbathymetry data allows us to match elastichalf- Small-wavelengthgravity anomaliesover the shelf of beam modeling on a regional basisfor northernAustralian the northern Australian continentallithosphere arise from continental lithosphere.In an under-filled foreland basin the presenceof horstand grabenstructures (Figure 1) and where few hundred meters sedimentation has occurred igneousrocks relatedto Late Devonian-EarlyCarbonifer- sincePliocene, one canuse seafloorbathymetry as the bent ous and Jurassicrifling [Anfiloff, 1988]. Also, crustalhet- horizon for flexure modeling. In the regions where we erogeneitiesat longer wavelengthsand tectonicdeforma- have the informationabout Plioquaternarysedimentation, tion other than continental subductionmask the gravity it is usedin the modeling. signatureof the flexed Moho. The seismicrefraction data 350 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

NW SE [Bowin et al., 1980] show that the crustalstructure below A) 200 the northernAustralian continental shelf and slopeis ho- 0.00 mogeneous.We assumethat the changesin dip angle in E the continentalslope seen on the marine 3D complete • -400....: Bouguergravity anomaliesare mostly due to variationsin [ i I / Model4 EET due to a simple crustalmodel of a flexed Moho. For X -8oo / ! ...... / i...... i...... ?...... ••..Od...?' 3 intra-crustaldensity contrastto be incorporatedin our / I! ' _l•1o_?elI __ , study, one has to have a detailed structural and strati- 1:-1200 1/ -.-[...... •;...... •...... --•.a..t..•.¾ metric graphicknowledge of the entirestudy area. 0 !J[";r ...... [...... •...... "•""rot•i'}e" #.1

> -1600 7. MODELING i/

0.00 200.00 400.00 600.00 800.00 7.1 Modeling Seafloorand Continental$helf Jbom Distance (km) VE -421 Pliocene B)190.00 180.00 Northern Australian shelf

We attemptto match the seafloorand the unloadedbase 160.00

Pliocenesurface to elastichalf-beam modeling (Figures 9 continental slope and 10) as follows: 140.00 1) Match the predictedflexure to the northernAustralian E 12o.oo continentalslope using 15 transects(Figure 6) and to the continentalshelf from Pliocene.However, we avoid any 1oo.oo statisticalcriteria for our model fit, such as RMS fit or lin- 80.00 ear correlation because these error measures are dominated by small-wavelengthvariations in the seafloorbathymetry, 60.00 Banda orogentopography and Buoguergravity anomalies. 40.00 We are attemptingto matchthe shelf-slope-orogenon hun- dreds of kilometers scale to the modeled response.The EET modeling for Profile I choice of our variablesEET values is qualitativeand ex- Figure 10. Increasein EET valuesby a factor of 2 on the conti- periments on the uncertaintiesin the EET serve as error nental shelf of the northern Australian continental lithosphere bounds. further decreasesthe amplitudeof the forebulgerelative to previ- 2) Minimize the modeled forebulge. Lack of evidence ous calculations [Lorenzo et al., 1998; Tandon et al., 2000]. for a small wavelength,high amplitudeforebulge on the Model 1 assumesconstant EET down dip, Model 2 usesan EET shelf in the data provides an important observationto profile similar to the ones presentedby Tandon et al. [2000]. match. Models 3-5 are resultsfrom sensitivityanalysis for the regional 3) Match the shapeof the triangular load on the flexed EET calculationspresented in Figure 11. A) The top portion is beam with the shape of the accretionaryprism of the the elastichalf-beam response compared to seafloorbathymetry Banda orogen. The flexed profile from the elastic half- andBanda orogen topography and B) the bottompart showsEET beam model is addedto the triangularload to match the variations for different models. Bandaorogen topography. 4) Determine the northern limit of the non-zero EET significantly less than the observedwidth of the Banda elastichalf-beam by the total width of the triangularload orogen,like partsof Timor Islandand Kai Plateau. that the elastichalf-beam model can support.In our model- 5) We startour modelingwith applyinga triangularload ing effort, we initially attemptto match the width of the and only use an end-pointload (Figure 9) at the northern triangularload to the width of the Bandaorogen loading a end for elastichalf-beam modeling if the triangularload is non-zero EET beam. The back-stopis assumedto be the unable match the data. The magnitude of the end-point northernboundary for the accrectionaryprism and demar- load is determinedby the closestvisual fit of the model. cation from the forearc basin north of Banda orogen(Fig- ure 1). That is how we estimatethe width of the Banda 7.2 Uncertaintyand SensitivityAnalyses orogenand the startingestimate for the northernlimit for our elastic half-beam model. In some cases the width of An order of magnitude decreasein Young's Modulus the triangularload usedon the elastichalf-beam model is (E) of the elastic half-beam resultswill increasethe EET TANDON ET AL. 351

121 E 123 125 127 129 131 133 135 137 -4 S

-6

' • •w•t.ar, • .•'*'•animbar .... Alor ,•1•- • .... •tlr•J• ...... •:•:?::-:.'"

ß.:..:.:. .• • . AUStrali:a

i i

Range of variable ...... :: 160 - 200 km //= Extensionalfaulting EET values ':• 120- 160 krn •;•.•••.... 80 - 120 km •i 40- 80km

20- 40 km

Figure 11. Contourmap for variable EET calculationsfor the northernAustralian continental lithosphere across the Banda orogen.EET valuesare determinedby bendingof the elasticlithosphere via triangularloading and end-point loading (Figure 9). The thick solid line lettered"northern limit" is the edge of the non-zeroEET half-beamas deter- mined by our model (Figure 9). The detailsof the loadingforces are given in Figure 12. The contourinterval is every 20 km. The uncertaintyin EET calculationsbased on our sensitivityanalysis is 50 km and 5 km on the slopeand be- neath the orogen.

by two-fold for the same flexural rigidity. However, we EET on the Australian continentalslope and beneaththe keep most of the parametersfixed in our modeling(Table Banda orogen (Figure 11) changesthe modeled response 1). We assumethat the elastic propertiesof the northern appreciably compared to the data. In the modeled re- Australian continental lithospherewere nearly uniform sponse,the triangular load sinks beneath the observed along the Banda orogenprior to continentalsubduction. If Banda orogen if the slope values of EET are decreased there were significant changesin the propertiesof the beyond 5 km. The supportof Banda orogenis very sensi- northern Australian continental lithosphere, one would tive to EET values on the slopeand below the orogen.Our noticethose changes in the morphologyof the today'scon- error bars have 50 km of uncertaintyin EET calculations tinentalshelf-slope (Figure 1) or at the baseof the Plioqua- on the continentalshelf but only 5 km on the slope and ternary sediments(Figure 7). beneaththe accretionaryprism. In one of the experimentsfor sensitivityanalysis, we The sensitivityanalysis for point load used in elastic varied the EET values but not the triangular load and the half-beammodeling shows that a variationwithin a range point load. On the Australiancontinental shelf, even after of + 0.05X 10•2 N/m 2to - 0.05X 10•2 N/m 2 canstill pro- adding +40 km to-10 km to the EET shelf values pre- duces an acceptablefit to the data. A zero point load in sentedin Figure 11, there is still a reasonablematch to the Profile 5 (FigUre 12) is believedto be due to lack of slab seafloor. Interestingly,a variation of even-5 km from pull rather than modeling artifact. In other profiles, the 352 INELASTIC FAILURE AND TECTONICS: TIMOKE-TANIMBAR-ARU TROUGH

121 E 123 125 127 129 131 133 135 137 -4 400 300 200 100 600 500 '" "': "... .. WB' _ 3._7L -6 Aru

-8

-10 ,•d (WidthX • Maximum Heic

-12

-14

Figure12. Forces(surface and subsurface) loading the northern Australian continental lithosphere leading to thefor- mationof a œor½landbasin within the Timor-Tanimbar-Aru trough. These œorc½s arc used to computethe variable EET (km)in Figure11. Shownwithin parenthesis onthe northern Australian continental shclœ arc: a) length(km) and b) maximumwidth (km) for thetriangular load that approximates the Banda orogcn. End-point loads arc used for subsur- œac½loading arc to bc multiplied by 1012N/m 2 (shownnear the Banda orogcn). SB = SavuBasin and WB = Weber Basin.The thick dashed lines arc the 100km contoursfor the Bcnioff seismic zone determined by earthquakelocations [Hamilton,1974]. Note that the point load does not correlate to thedip of theBcnioff zone implying direct correlation to slab pull.

point load neededfor modeledresponse to match the data thicknessas a background).A depressionin the flexure is numerically larger than the uncertaintyassociated with profile from elastichalf-beam modeling produces an area the point load estimation. wherethe crustreplaces the mantleon the slopeand be- On the easternmargin of the Banda orogen,the orocline neaththe orogenand a forebulgecreates a correspondingly ßcan causethree-dimensional loading of the northernAus- slightcrustal thinning on the shelf. Seismicrefraction data tralian continentallithosphere (Figure 1). A 2D formula- in the northern Australian continentalcrust along the tion of flexuremay resultin an over-estimationof EET by Bandaorogen [Bowin et al., 1980] providesus an estimate 30 % [Wessel,1996]. The width of the triangularload used on the flexed Moho for gravity modeling. The crustal for elastichalf-beam modeling near the easternmargin of thicknessin our model is ---31km on the continentalshelf, the Banda orogenis lessthan the width of the triangular a lower limit from the refractionexperiment [Bowin et al., loads used for modeling in other regions (Figure 12). 1980]. Our motivation is to investigatewhether the Moho Therefore,the contributionof the Bandaorogen near Aru in the northernAustralian shelf-slope is flexed similar to Island as a surfaceload is perhapsnot as a cylindricalmass the seafloor.Therefore, the mass compensationis not in but more like a thin shellvalidating 2D approximation. localAiry isostasybut respondsto the rigidity of the conti- nentallithosphere. 7.3 Modeling Marine Complete$D BouguerGravity Assuming that the Moho is also flexed, theoretical Anomalies Bouguergravity anomaliesare derivedfrom the EET pro- files (Figure 11) using an assumptionof homogeneous The theoreticalBouguer gravity anomaliesfor a flexed crustalin outerNW Australia.Calculated Bouguer gravity Moho are createdfrom the densitycontrast given in Table anomaliesare then matchedto correspondingprofiles from 1. The flexure topographyis added to constantthickness the marine complete 3D Bouguer gravity anomalies.A Moho creating crustal thickening and thinning (flexed good fit meansthat the Bouguergravity anomaliescalcu- Moho = bending of the crustby flexure + constantMoho lated from the elastic half-beam model matches the ob- TANDON ET AL. 353 servedmarine complete Bouguer gravity anomalies at the A/•) NW SE slopeof the flexedcontinental lithosphere. VE1 OOO I•" ....i i i i 'Sea floor • t1• •, Unloaded 8. RESULTS O.O Modeled response i• - 1000 i••• i .i...... i...... '•.. Pliocenesurface 8.1 Elastic Half-Beam Deflectionand Data : We show that the amplitudeof the forebulgedecreases -2OO ...... further in our modelingif the EET of the northernAustra- I• 0.00 200.00 400.00 600.00 800.00 Distance (km) VS - 136 lian continentalshelf is increasedfrom -80 km [Lorenzo et al., 1998; Tandon et al., 2000] to 190 km (Figure 10). Our B) basis for using such a high EET on northern Australian 80.0 continentalshelf is to minimize the modeledforebulge to Observed Bouguer gravity match the data as much we can in the elastic half-beam 40.0 anomalies modeling. Such high EET value on the shelf is plausible Theoretical '•':'•'•'•"•'•''" anomaliesBouguer gravity for normal crustalrheology [Simonset al., 2000]. Our cal- culationsobey the thin-beam approximation(one of the -40.0 assumptionsfor elastichalf-beam modeling) sincethe cal- -80.0

culatedEET value on the shelf is numericallyless than the Northern Australian shelf wavelength of the entire flexed elastic half-beam (Figure C) 19o.o 10) [van der Meulen et al., 2000]. ^ 16o.o ...... The variable EET calculations by matching seafloor • ,oo.o 140.0 ...... ic.=..i:..,._X.....:;,i: ...... v 120.0 bathymetrygenerally decreasesfrom EET = 230-180 km • 100.0 on the shelf to less than EET =- 40 km on the continental IlJ 8o.o ...... slope from Roti to west of Am Island (Figure 11). Near 60.0 Kai Plateau, the EET decreasesfrom 155-125 km on the EET modeling for Profile 5 shelf to -20 km on the slope(Figure 11). Southof central Timor Island, the decreasein EET in northern Australian Figure 13. Variable EET valuesfrom profile 5 (Figure 6) are lithosphereis from 190 km to 55 km (Figures 11 and 13). determinedby matchingthe flexural profile qualitativelyto the Profile 5 (Figure 13c) and Profile 14 (Figure 14c) show A) seafloorbathymetry and Bandaorogen topography [National examplesof shallowerand steeperfailure slope compared GeophysicalData Center,1988) and continentalshelf from Plio- to Profile 1 (Figure 10a). The valuesof EET on the slope cene.The triangularload is addedto the flexedresponse to match and beneaththe Banda orogenfor Profiles 1 (Figure 10b), the topographyof the Banda orogen.B) TheoreticalBouguer gravityanomaly created by a flexedMoho is thenmatched to the 5 (Figure 13c), and 14 (Figure 14c) are.beyondthe uncer- observedmarine complete3D Bouguergravity anomalies.The tainty of our modelingusing our sensitivityanalysis. shaded area shows that there is no observedmarine Bouguer Regional base Plioquaternarysurface (Figure 7) deter- gravity anomalydata. The origin of the profile is the northern mined by the seismicreflection data does not show any limit of the non-zero EET elastic beam for that transect(Figure systematicchanges (migrating sedimentarydepocenters) 11). C) VariableEET profile.VE = VerticalExaggeration. +/- = on the continentalshelf-slope which may indicateevidence positive-negativeBouguer anomaly couple found traversing from for deeperlithospheric processes, such as forelandsegmen- the Banda orogento the northernAustralian continental slope tation [Royden et al., 1987] and/or slab detachment[van indicatesthe presenceof a high-densitysubsurface load beneath Meulen et al., 2000]. The Pliocene continentalshelf was the orogenadjacent to the subductingcontinental crust. not completelysub-horizontal as it might be a part of a long wavelength,low amplitudeforebulge present during the Miocene-Plioceneboundary. That would indicatehigh gravityanomalies clearly show that Australiancontinental EET values at the start of continental subduction. A small lithosphereis not in localAiry isostasy(Figures 10, 13 and wavelength, high amplitude paleo-forebulgeif formed 14). The heterogeneityat differentscales with the crust duringthe Plioceneon the continentalshelf has eitherto- doesnot allow Bouguergravity anomalies to be usedquan- day moved to the continentalslope or been subductedbe- titatively to determinevariations in EET. Matching3D neaththe Banda orogen. Bouguergravity anomalies serves only as a qualitativetool Matching of the Bouguer gravity anomaliesfrom a testingour EET estimatesusing surfacial data. Figure 13b flexed Moho to the observedcomplete 3D marineBouguer showsthe limitationof simpletheoretical Bouguer gravity 354 INELASTIC FAILURE AND TECTONICS' TIMORE-TANIMBAR-ARU TROUGH

A) 200 NW ? ; • •...... SE Malita and Vulcan Graben (Figure 11) correspondto the Sea floor relatively higher EET corridorfrom continentalshelf to the Modeled slope. The Malita and Vulcan Graben were formed due to response -1000 Jurassicrifling [Stagg, 1993; AGSO North West Shelf

U, -2000 StudyGroup, 1994] (Figure 1). However, we do not notice regions of lesser inelastic failure at other transfer zones 0 ,,,, within northern Australian shelf. In absence of detailed • -3000 •1 0.00 200.00 400.00 600.00 800.00 publicly available structural geology map, small wave- • Distance (km) VE -- 155 length Bouguergravity anomaliescan help estimateposi- tions of rift structuresand transferzones (Figure 8). Near Aru Island, the decreasein EET to 20 km (Figure 14) can B)80.040.0 ""'•"••11.*•i' '...... ObservedBouguer gravity anomalies be explained by an increasedinelastic failure due to the ...... Theoretical oroclinal bending [McNutt et al., 1988] of the northern E• Bouguergravity -40.0 anomalies Australiancontinental lithosphere. We notice reductions in EET close to 4-5 times in our -80.0 c) results whereas modeling of the decoupling within the

120.0 lithosphereonly explainsreductions of EET up to a factor of 2 [Brown and Phillips, 2000]. The bathymetricprofiles E 130.01 oo.o (Figures 10, 13, and 14) have similar curvatureseen on • 80.0 Tonga and Kermadactrenches [Garcia-Castellanos et al., 60.0 2000] favoring the hypothesisof a multilayeredelastic- plastic rheology lithospherethat underwentinelastic fail- 20.0 ...... ure. The resultspresented here are in contrastto the con-

EET modeling for Profile 14 clusions that the continental lithospherehave inherently low EET values of-20 km [McKenzie and Fairhead, Figure 14. Variable EET valuesfrom profile 14 (Figure 6) with 1997]. samedescription as Figure 13. Decreasein EET values from shelf to slope (Figure 11) along the entire collisionalmargin will decreasethe bend- ing stressesbut doesnot explainwhy normal faultinghas calculations.On the other hand, Figure 14b illustratesa ceasednear Timor but not near Tanimbar Island (Figure good match betweentheoretical and observedBouguer 3). Supportingsurface loads more buoyantlynear central gravityanomalies even using a simplecrustal model. Also, Timor Island due to higher EET values on the slope and presenceof positive/ negative Bouguer anomaly couple beneaththe orogencould increasehorizontal compression. due to ophiolitesin centralTimor Island [Sopaheluwaken Numerical modelsfor slab break-off producea time lag of et al., 1989] is evidentin Figure 13b indicatesthat buffered 1.7-3.9 Ma between the start of continental subduction and slab pull is not the only subsurfaceloading component in the actual break-off [Wong A Ton and Wortel, 1997]. our studyarea There is no evidence for complete slab break-off of the Australianlithosphere on a regionalbasis that would have 9. DISCUSSION resulted in a cessationof arc volcanism along the Banda Arc (Figure 1). A partial slab break-offon a regionalscale The changesin EET on the continentalslope of the is possible based on the evidence from seismictomo- northernAustralian continental lithosphere are not uniform graphicimages (Figure 4). and parallel to the subductionboundary (Figure 11). Our If the subductingcontinental lithosphere has lesserine- hypothesisis that the Jurassicrifting with- 20 % strain lastic failure and retains higher EET values, it not only [AGSO North West Shelf StudyGroup, 1994] only predis- resistssubduction but perhapsalso aid in the detachment posedthe northernAustralian continental lithosphere to of the shallowerparts of continentallithosphere from the weakening rather than actual weaken the lithosphere oceaniclithosphere after the continentalsubduction (Fig- [Stewartand Watts, 1997]. Lesserdecrease in EET of the ures 4). Continentalnecking noticed on tomogramssouth northern Australian continentallithosphere on the conti- of eastern Timor Island indicate higher possibility of nentalslope (south of centralTimor Island)is dueto litho- detachmentoccurring in that area(Figure 4). sphere'sresistance to inelasticfailure to somedegree dur- Accordingto the calculationsdone by Molnar and Gray ing the continentalsubduction. The transferzone between [ 1979],a pointload of theorder of 1011-1012N/m 2 (Figure TANDON ET AL. 355

13) showsthat the slab pull's contributionto the formation the present-daystrain partitioning to the forearc basin. of the Timor-Tanimbar-Arutrough (foreland basin system) Comparingend-point load in elastic half-beam modeling is insignificant.Establishing a link betweenend-point load and seismictomography leads us to speculatethat slabpull in the elastichalf-beam model and deeper subduction proc- doesnot contributeto the formationof Timor Trough in essesusing seismictomography is still speculative.Slab the vicinity of centralTimor Island. pull has been identified as one of the major mechanism driving [Jurdy and $tefanick, 1991; Garcia- Acknowledgments.EET modeling was kindly given to us by Castellanoset al., 2000]. We do not seea clearrelationship JonathanStewart (code originally written by Bodine). Some of betweenthe slabpull and its contributionto the loadingof the figures are made usingthe GMT software[Wessel and Smith, the foreland basin (there is no correlation between the 1995]. Commentsfrom anonymousreviewers and the editor Seth point load and the dip of the Benioff zone in Figure 12). Stein greatly improved this manuscript.Thoughtful readingsof the draftsby Uwe Streckerwere very beneficial.We acknowledge Thrustingwithin the prowedge and the retrowedgeis no- JamieCurry for providingus the interpretedtop of Plioquaternary ticed at all stagesof the collisionin clay models[Storti et sedimentarysurface for our study.AOA GeophysicsInc. and LR al., 2000] as well as in ourt studyarea (Flores,Alor and softwarehelped KT with gifting their softwarefor gravity model- Wetar back-thrustshown in Figure 1). Regionalvariations ing. Tommy Smith from Louisiana State University helped im- in EET may play an importantrole in causingthe back- menselyin letting use their workstationand computerprograms thrustingto be a dominantmeans of strainpartitioning remotely required for the researchpresented in this manuscript. duringthe progressof continentalcollision (Figure 1). The KT acknowledgesthe facilitiesused for the writing of this manu- height of the accretionaryprism is one of the factorsthat script and computersupport by Kyle Hubbard during his stay at controlsthe formationof doublyvergent orogens [Storti et the Laboratory for ComputationalGeodynamics, Department of al., 2000]. Higher EET values beneaththe central Timor Chemistry,Indiana University. Islandincreases the elevationof the island(Fugure 5) and that helpsin the developmentof back-thrusting.However, the role of EET is not clearin westernpart of Timor Island (Figure 11) where back-thrustingis also dominantwithin REFERENCES the forearc and the volcanic arc (Figure 1) [Chaffton, 1997]. AGSO North West Shelf Study Group, Deep reflectionson the north west shelf: changingperceptions of basin formation,in: The North WestShelf, Australia: proceedings of petroleumex- 10. CONCLUSIONS ploration societyof Australia symposium,edited by P. G., and R. R. Purcell, 63-76, 1994. The EET changesin the down dip directionplay a sig- Anfiloff, V., Polycyclicrifting-an interpretationof gravity and nificant role in the evolution of continental collision tec- magneticsin the North West Shelf, in: The North WestShelf, tonics. Our results show a similar relationshipbetween the Australia:proceedings of petroleum exploration society of Aus- tralia symposium,edited by P. G., and R. R. Purcell, 443-455, variable EET valuesand shortening,as also conceptualized 1988. in the central Brazilian Shield [Watts et al., 1995; Stewart Blakely, R. J., Potentialtheory in gravityand magneticapplica- and Watts, 1997]. Inherent structureof the lithosphere tions, CambridgeUniversity Press,New York, 1995. prior to subduction,like the presenceof transfer zones Boehme, R. S., Stratigraphicresponse to Neogene tectonismon from previousrifting can affectthe degreeof inelasticfail- the Australian Northwest Shelf, M.S. thesis, Louisiana State ure. The Australian continentallithosphere prior to conti- University, Baton Rouge, USA, 1996. nental subductionhad an EET of at least230-180 km (Roti Bowin, C., G. M. Purdy, C. Johnston,G. Shor, L. Lawver, H. M. Island-westof Aru Island) that decreased4-5 times due to S. Hartono, and P. Jezek, arc- collision in inelastic failure at that start of continental subduction. region,The American Association of Petroleu.m GeologistBul- Continentallithosphere under normal crustalthickness and letin, 64, 868-915, 1980. not weakened significantly by prior rifting can be very Bradley,D.C., and W. S. F. Kidd, Flexural extensionof the up- per continentalcrust in collisionalforedeeps, Geological Soci- rigid. Almost the lack of forearc basin north of central ety of America Bulletin, 103, 1416-1438, 1991. Timor Island is due to increasedshortening by higher EET Brown, C. D., and R. J. Phillips, Crust-mantledecoupling by lithosphereon the slope and beneaththe orogen. Higher flexure of continentallithosphere, dournal of GeophysicalRe- EET on the slope and beneaththe orogen is capableof search, 105, 13,221-13, 237, 2000. supportingmore surfaceload and facilitatingthe formation Burov, E. B., and M. Diament, Flexure of the continental litho- of doubly-vergentorogens. On the continentalslope and spherewith multilayeredrheology, Geophys.d. Int., 109, 449- beneaththe orogen,even an EET increaseof-15 km shifts 468, 1992. 356 INELASTIC FAILURE AND TECTONICS: TIMORE-TANIMBAR-ARU TROUGH

Burov, E. B., and M. Diament, The effective elastic thickness Jurdy,D. M., andM. Stefanick,The forcesdriving the plates: (EET) of continentallithosphere: What does it reallymean?, constraintsfrom kinematicsand stressobservations, Phil. dournal of GeophysicalResearch, 100, 3905-3927, 1995. Trans. R. Soc. Lond., 337, 127-139, 1991. Burov,E. B., andM. Diament,Isostasy, equivalent elastic thick- Karig,D. E., A. J. Barber,T. R. Charlton,S. Klemperer,and D. ness,and inelastic rheology of continentsand oceans, Geology, H. Hussong,Nature and distributionof deformationacross the 24, 419-422, 1996. BandaArc-Australian collision zone at Timor,Geological So- Chamalaun,F. H., andA. E. Grady,The tectonic development of ciety of America Bulletin, 98, 18-32, 1987. Timor:a newmodel and its implications for petroleum explora- Karner,G. D., andA. B. Watts,Gravity anomalies and flexure of tion, TheAustralian Petroleum Exploration Association, 102- the lithosphereat mountainranges, dournal of GeophysicalRe- 108, 1978. search, 88, 10,449-10,477, 1983. Charlton, T. R., Post collision extensionin arc-continentcollision Karner,G. D., N. W. Driscoll,J. K. Weissel,Response of the zone, easternIndonesia, Geology, 19, 28-31, 1991. lithosphereto in-planeforce variations, Earth and Planeta•3, Charlton,T. R., Back-thrustingon theBIRPS deep seismic reflec- ScienceLetters, 114, 397-416, 1993. tion profiles,Banda Arc, Indonesia,a responseto changing Kennett,B.L.N., E. R. Engdahl,E.R. and R. Buland,Constraints slabinclination, dournal of the GeologicalSociety, London, on seismicvelocities in the Earthfrom travel times, Geophysi- 154, 169-172, 1997. caldournal International, 122, 108-124, 1995. Chemenda,A. I., J-P,Burg, M. Mattauer,Evolutionary model of Kreemer, C., W. E. Holt, S. Goes, and R. Govers,Active defor- theHimalaya-Tibet system: Geopoem based on new modelling, mationin easternIndonesia and the Philippinesfrom GPS and geologicaland geophysical data, Earth andPlanetary Science seismicitydata, dournal of GeophysicalResearch, 105, 663- Letters, 174, 397-409, 2000. 680, 2000. Curry,J. S., J. M. Lorenzo,and Geoff W. O' Brien,Polarity of Lorenzo,J. M., G. W. O'Brien, J. Stewart,and K. Tandon,Inelas- continent-islandarc collisionsince late Miocene: ,N. tic yielding and forebulgeshape across a modernforeland ba- W. Shelf, TheAustralia, American Association of Petroleum sin:north west shelf of Australia,Timor Sea,Geophysical Re- Geologists,A35, 2000. search Letters, 9, 1455-1458,1998. DeMets,C., R. G. Gordon,D. F. Argus,and S. Stein,Effects of Lyon-Caen, H., and P. Molnar, Constraintson the structureof recentrevisions to the geomagneticreversal time scaleon esti- Himalayafrom an analysisof gravityanomalies and a flexural matesof currentplate motions, Geophysical Research Letters, model of lithosphere,dournal of GeophysicalResearch, 88, 21, 2191-2194, 1994. 8171-8191, 1983. DeSmet, M. E. M., A. R. Fortuin, S. R. Troelstra, L. J. Van Lyon-Caen,H., and P. Molnar, Gravity anomalies,flexure of the Marle, M. Karmini, S. Tjokrospoetro,and S. Hadiwasastra,De- Indian Plate and the structure,support and evolutionof the tection of collision-related vertical movements in the outer Himalaya abd Gangabasins, Tectonics, 4, 513-538, 1985. Banda Arc (Timor, Indonesia),using micropaleontological Mackenzie,D., Oceanfloor is laid bare by new satellitedata, data, dournal of SoutheastAsian Earth Sciences,4, 337-356, Science,277, 1921, 1997. 1990. McAdoo,D.C., J. G. Caldwell,and D. L. Turcotte,On the elas- Engdahl, E.R., R. D. van der Hilst, and R. Buland, Global tele- tic-perfectlyplastic bending of the lithosphereunder general- seismicearthquake relocation with improvedtravel times and izedloading with applicationto the Kuril Trench,Geophys. proceduresfor depthdetermination, Bull. Seism.Soc. Am., 88, riourn.Roy. Astron. Soc., 54, 11-26, 1978. 722-743, 1998. McCaffrey, R., Active tectonicsof the easternSunda and Banda Garcia-Castellanos,D., M. Torne, and M. Fernandez,Slab pull Arcs, dournalof GeophysicalResearch, 93, 15,163-15,182, effects from flexural analysisof the Tonga and Kermadec 1988. trenches(), Geophys. d. Int., 141,479-484, 2000. McCaffrey,R., andJ. Nabelek,Seismological evidence for shal- Genrich,J. F., Y. Bock, R. McCaffrey,E. Calais,C. W. Stevens, lowthrusting north of theTimor Trough, Geophys. riourn. Roy. and C. Subarya,Accretion of the southernBanda Arc to the Astron.Soc., 85, 365-381, 1986. Australianplate margin determined by globalpositioning sys- McCaffrey,R., P. Molnar,S. W. Roeker,and Y. S. Joyodiwiryo, tem measurements,Tectonics, 15,288-295, 1996. Microearthquakeseismicity and fault plane solutions related to Hamilton, W., Tectonicsof the Indonesianregion, US Govt. arc-continentcollision in the easternSunda arc, Indonesia, Printing Office, 1974. dournalof GeophysicalResearch, 90, 4511-4528, 1985. Harris, R. A., Temporaldistribution of strainin the activeBanda McKenzie, D., and D. Fairhead,Estimates of the effectiveelastic orogen:A Reconciliationof Rival Hypotheses,dournal of thicknessof thecontinental lithosphere from Bouguer and free SoutheastAsian Earth Sciences,6, 373-386, 1991. air gravityanomalies, dournal of GeophysicalResearch, 102, Johnston,C. R., and C. O. Bowin, CrustalReactions Resulting 27,523-27,552, 1997. from the Mid-Pliocene to Recent Continent-island arc Collision McNutt, M. K., M. Diament,and M. G. Kogan,Variations of in the Timor Region,Australian Bureau of Mineral Resources elasticplate thickness at continentalthrust belts, dournal of Geologyand Geophysicsdournal, 6, 223-243, 1981. GeophysicalResearch, 93, 8825-8838, 1988. TANDON ET AL. 357 van der Meulen, M. J., S. J. H. Buiter,J. E. Meulenkamp,and M. Tandon, K, J. M. Lorenzo, and G. O'Brien, Effective elastic J. R Wortel, An early Plioceneuplift of the centralApenninic thicknessof the northernAustralian continental lithosphere foredeepand its geodynamicsignificance, Tectonics, 19, 300- subductingbeneath the Banda Orogen (Indonesia):Inelastic 313, 2000. failureat the startof the continentalsubduction, Tectonophys- Molnar, P., and D. Gray, Subductionof continentallithosphere: ics, 329, 39-60, 2000. someconstraints and uncertainties,Geology, 7, 58-62, 1979. Turcotte,D. L., and G. Schubert,Geodynamics: Application of National GeophysicalData Center,ETOPO-5 bathymetry/topog- ContinuumPhysics to Geological Problems, John Wiley & raphy data,National Oceanicand AtmosphericAdministration, Sons, New York, 1982. US Dept. Commerce,Boulder, Colorado, 1988. Uyeda, S., and H. Kanamori,Backarc opening and the mode of PoudjomDjomani, Y. H., J. D. Fairhead,and W. L. Griffin, The subduction,dournal of GeophysicalResearch, 84, 1049-1061, flexural rigidity of Fennoscandia:reflection of the tectono- 1979. thermal age of the lithosphericmantle, Earth and Planetary Veevers, J. J., Sedimentarysequences of the Timor Trough, Science Letters, 174, 139-154, 1999. Timor, and the Sahulshelf, in: Initial reportsof the Deep Sea Richardson,A., (preprint) Constraintson continent-arccollision Drilling Project, Edited by J. J. Veevers,J. R. Heirtzler, and et in easternIndonesia, 2001. al., 27, 567-569, 1974. Royden, L., and G. D. Karner, Flexure of the continentallitho- Veevers,J. J., J. R. Heirtzler,and et al., Initial reportsof theDeep spherebeneath Apennine and Carpathianforedeep basins, Na- Sea Drilling Project,27, 1060 pp, 1974. ture, 309, 142-144, 1984. Veevers,J. J., D. A. Falvey, and S. Robins, Timor Trough and Royden,L., E. Patacca,and P. Scandone,Segmentation and con- Australia:Facies show topographicwave migrated80 km dur- figurationof subductedlithosphere in Italy: an importantcon- ing the past3 my., Tectonophysics,45, 217-227, 1978. trol on thrust-beltand foredeepbasin evolution,Geology, 15, Warris, B. J., Plate tectonics and the evolution of the Timor Sea, 714-717, 1987. northwestAustralia, The Australian Petroleum Exploration As- Sandwell,D. T., and W. H. F. Smith, Marine gravity anomaly sociation, 13-18, 1973. from Geosat and ERS 1 satellite altimetry, dournal of Geo- Watts,A. B., S. H. Lamb, J. D. Fairhead,and J. F. Dewey, Litho- physical Research, 102, 10039-10054, 1997. spheric flexure and bending of the central Andes, Earth and Simandjuntak,T. O, and A. J. Barber, Contrastingtectonic styles Planetary ScienceLetters, 134, 9-21, 1995. in the Neogene orogenicbelts of Indonesia,in: Tectonicevolu- Watts, A. B., The strengthof the earth's crust,Marine Technol- tion of southeastAsia, edited by R. Hall and D. Blundell, Geo- ogy Societydournal, 17, 5-17, 1983. logical SocietySpecial Publication, 106, 185-201, 1996. Wessel, P., Analytical Solutions for 3D flexural deformation Simons,F. J., M. T. Zuber, and J. Korenaga,Isostatic response of semi-infinite elasticplates, Geophysicaldournal International, the Australianlithosphere: Estimation of effectiveelastic thick- 124, 907-918, 1996. nessand anisotropyusing multitaper spectralanalysis, dournal Wessel,P., and W. H. F. Smith,New versionof genericmapping of GeophysicalResearch, 105, 19,163-19,184, 2000. tools, Eos. Trans. AGU Electron. Suppl., Aug. 15, 1995. Smith,W. H. F., andD. T. Sandwell,Global s•aflor topography (Availableat http://www.agu.org/eos-elec/9514e.html) from satellitealtimetry and shipdepth soundings, Science, 277, Widiyantoro, S., and R. D. van der Hilst, Structureand evolution 1956-1962, 1997 of lithosphericslab beneaththe Sundaarc, Indonesia,Science, Snyder,D. B., H. Prasetyo,D. J. Blundell, C. J. Pigram, A. J. 271, 1566-1570, 1996. Barber,A. Richardson,and S. Tjokosaproetro,A dual doubly Wong a Ton, S. Y. M., and M. J. R. Wortel, Slab detachmentin vergent orogen in the Banda Arc continent-arccollision zone as continentalcollision zones: An analysisof controllingparame- observedon deepseismic reflection profiles, Tectonics, 15, 34- ters, GeophysicalResearch Letters, 24, 2095-2098, 1997. 53, 1996. Sopaheluwaken,J., H. Helmers,S. Tjokrosapoetro,and E. Surya Nila, Medium pressuremetamorphism with inverted thermal gradientassociated with ophiolitenappe emplacement, Nether- Juan M. Lorenzo, Departmentof Geology and Geophysics, landsdournal of Sea Research,24, 333-343, 1989. Louisiana State University, Baton Rouge, LA 70803, USA. Stagg,H. M. J., Tectonic elementsof the north west shelf, Austra- (j uan• eo1. lsu. edu ) lia, scale 1:2500000,Australian Geological Survey Organiza- GeoffreyW. O'Brien, AustralianGeological Survey Organiza- tion, Canberra, 1993. tion, GPO Box 378, Canberra, Australia, 2601. Stewart,J., and A. B. Watts, Gravity anomaliesand spatialvaria- (gobrien•selenite.agso.gov.au) tion of flexural rigidity at mountainranges, dournal of Geo- Kush Tandon, Departmentof Geology and Energy Institute, physical Research, 102, 5327-5352, 1997. University of Louisiana, Lafayette, LA 70503, USA. Storti, F., F. Salvani, and K. McClay, Synchronousand velocity- (kxt2777•louisiana.edu kushtandon•yahoo.com) partitionedthrusting and thrustpolarity reversal in experimen- Sri Widiyantoro,Department of Geophysicsand Meteorology, tally produced,doubly-vergent thrust wedges: Implications for Bandung Institute of Technology, Bandung, Indonesia. naturalorogens, Tectonics, 19, 378-396, 2000. (sriwid•geoph.itb.ac.id)