savu basin

Structural and stratigraphic evolution of the Savu Basin, Indonesia

JAMES W. D. RIGG* & ROBERT HALL

SE Asia Research Group, Department of Earth Sciences, Royal Holloway University of London,

Egham, Surrey TW20 0EX, UK

*Corresponding author (e-mail: [email protected])

Abstract: The Savu Basin is located in the Sunda–Banda fore-arc at the position of change fromoceanic subduction to continent–arc collision. It narrows eastward and is bounded to the west bythe island of Sumba that obliquely crosses the fore-arc. New seismic data and published geologicalobservations are used to interpret Australia–Sundaland convergence history. We suggest the basinis underlain by continental crust and was close to sea level in the Early Miocene. Normal faulting inthe Middle Miocene and rapid subsidence to several kilometres was driven by subduction rollback.Arc-derived volcaniclastic turbidites were transported ESE, parallel to the Sumba Ridge, and thenNE. The ridge was elevated as the Australian continental margin arrived at the Banda Trench,causing debris flows and turbidites to flow northwards into the basin which is little deformedexcept for tilting and slumping. South of the ridge fore-arc sediments and Australian sedimentarycover were incorporated in a large accretionary complex formed as continental crust was thrustbeneath the fore-arc. This is bounded to the north by the Savu and Roti Thrusts and to the southby a trough connecting the Java Trench and Timor Trough which formed by south-directed thrust-ing and loading.

The Savu Basin is situated in the fore-arc of theSunda–Banda Arc (Fig. 1) at the margin of theEurasian plate. It is an area of particular interestsince it is immediately north of the position wherethere is a change from subduction of Indian Oceancrust at the Java Trench in the west to collisionbetween the Australian continental margin and theBanda fore-arc to the east.

The Savu Basin has an unusual asymmetricaltriangular shape and narrows eastwards fromc. 200 km at its maximum width to c. 20 km northof Timor and westwards to c. 50 km north ofSumba. The basin is bounded to the west by theisland of Sumba and a submarine ridge thatcrosses the fore-arc obliquely in a NW–SE direc-tion. To the north is the active volcanic arc includ-ing the island of Flores which passes east into anextinct sector of the Banda arc between Alor andWetar. To the south are the smaller islands of Savuand Roti, and to the east the much larger island ofTimor. Timor has been the subject of manystudies, notably concerned with collision of theAustralian continental margin and the Banda vol-canic arc. There is now an unusually short distancebetween the collision complex on Timor and theinactive volcanic islands of Alor and Wetar to thenorth. There has been considerable controversyabout the significance of the Timor Trough whichis significantly shallower than the Java Trench,and in particular whether it is a trench or fore-deep.A sinuous bathymetric trough south of Savu and

Roti and the Savu Basin connects the Timor Troughto the Java Trench.

We have studied a recently acquired 2D seismicdata set from the Savu Basin. Although there are nowells in the basin it is possible to correlate the stra-tigraphy offshore with that on land. The islands ofSumba, Savu, Roti and Timor all emerged in thePliocene or Pleistocene. The stratigraphy anddeformation of Sumba have been well documented(Effendi & Apandi 1980; Fortuin et al. 1994) anddetailed studies of Savu and Roti (Harris et al.2009) also provide valuable information that aid ininterpreting the offshore data. In addition, marinegeophysical investigations combined with gravityand tomographic modelling have recently been usedto interpret the deep crustal structure along a regionaltransect crossing the collision zone and the SavuBasin (Shulgin et al. 2009). The combination ofinformation from these studies with the new seismicdata set provides the basis for interpreting thedevelopment of the fore-arc basin and the collisioncomplex south of the Savu Basin from oceanicsubduction to the earliest stages of arc–continentcollision in the context of a new model of Bandasubduction rollback (Spakman & Hall 2010).

Seismic stratigraphy

The seismic data consists of 32 2D seismic lines thelongest of which is 535 km, covering the southern

From: Hall, R., Cottam, M. A. & Wilson M. E. J. (eds) The SE Asian Gateway: History and Tectonicsof the Australia–Asia Collision. Geological Society, London, Special Publications, 355, 225–240.DOI: 10.1144/SP355.11 0305-8719/11/$15.00 # The Geological Society of London 2011.

and western parts of the Savu Basin, with a total areaof 50 000 km2. Two surveys were combined for thisstudy. The first was acquired in 2002 and comprises2 740 km of long-offset 2D data to a depth of 12 sTWT (two-way travel time). The second wasacquired in 2007 and comprises 3 000 km of long-offset 2D data to a depth of 8 s TWT (Toothill &Lamb 2009). The stratigraphic column (Fig. 2)shows the strata present at the eastern end ofSumba (Fortuin et al. 1994) and correlation withthe seismic sequences identified in this study.

Unit 1

Unit 1 is the deepest unit and is the sequence below apackage of bright reflectors that can be mappedthroughout the entire area as the lowest continuousfeature identifiable in the dataset (Fig. 3). Belowthis feature there is locally some reflectivity andthe sequence can be split into two parts. The upperpart contains some localized reflectors which aresub-parallel, and very bright, within a sequencethat has almost no reflectivity. In places there is noclear boundary between the upper and lower parts

of Unit 1 whereas elsewhere there is a sharp bound-ary, and the lower half of Unit 1 is characterized bymoderately bright, laterally discontinuous reflectiv-ity which indicates bedding and gives an overallmottled appearance to the sequence. Bedding canbe traced for distances of up to 20 km. The faultsthat cut the top horizon cannot be traced into thesequence below.

There are commonly between three and fiveparallel reflectors at the top of Unit 1 with a max-imum thickness of 0.4 s TWT. They are offset byextensional faults, by a maximum of 0.3 s TWT,and are locally rotated. The tops of fault blocksvary in depth from 2 s TWT close to Sumba toa maximum of 7.4 s TWT in the deepest parts ofthe basin.

Interpretation

The oldest rocks reported from SE Sumba areCretaceous marine siltstones and sandstones whichinclude volcaniclastic interbeds (Fortuin et al.1997) and represent submarine fan deposits (vonder Borch et al. 1983). They are unconformably

Fig. 1. Geographical features of the Savu Basin and surrounding area. (a) DEM of satellite gravity-derived bathymetrycombined with SRTM topography (Sandwell & Smith 2009). (b) Contoured bathymetry from Gebco (2003) withcontours at 200, 100, 2000, 4000 and 6000 m.

J. W. D. RIGG & R. HALL226

overlain by shallow marine to non-marine Palaeo-gene sandstones, limestones and volcanic agglomer-ates (von der Borch et al. 1983). We interpret thelower part of Unit 1 to represent the deeper marineCretaceous sequence and the upper part to be thePalaeogene in age. The seismic character of Unit 1is consistent with the descriptions of these rockson land. The bright package at the top of Unit 1 issuggested to be Eocene–Oligocene or possiblyLower Miocene Nummulites limestones (von derBorch et al. 1983; Fortuin et al. 1992, 1997; vander Werff et al. 1994).

The field relationships described by Fortuin et al.(1992, 1994) from Sumba with tilted fault blockscapped by discontinuous carbonates, all overlainunconformably by Neogene strata, are well matchedby the observations from the seismic data. Fortuinet al. (1994) suggest these relationships correspondto breakup of a carbonate platform and rapid subsi-dence in the early Middle Miocene. Fortuin et al.(1994) interpret a Late Burdigalian unconformitywith a thin sequence of conglomerates that contain

older shallow water limestones passing up intochalks and marls. We suggest these probably corre-spond to the top of Unit 1.

Unit 2

This unit is up to 1.6 s TWT in thickness and fills thedepressions created by the extensional faulting ofUnit 1, onlapping the fault surfaces (Fig. 3). Reflec-tors are most clear where the unit is thinner and aretypically bright, subparallel and discontinuous,forming a complex bedding pattern, whereas thethicker parts of Unit 2 are characterized by a moretransparent, less distinctive seismic character. Insome places a series of three or four very bright par-allel reflectors are visible in the middle of the unitand stand out against the more transparent areas.The seismic character of Unit 2 suggest a facieschange which could be from thin bedded lithologiessuch as carbonates to a more uniform lithology.Overall the amplitude of the reflections increasestowards the top of this unit.

Fig. 2. Seismic stratigraphy of the Savu Basin and correlation with the stratigraphy of east Sumba after Fortuin et al.(1992). The colours shown next to the seismic section for the four units are used on subsequent seismic profiles.The vertical scale of seismic sections is two-way travel time (TWT) in seconds.

EVOLUTION OF THE SAVU BASIN 227

The top of this package is truncated by a promi-nent unconformity; this surface has extremelystrong reflectivity and can be mapped throughoutthe basin (Fig. 3). The unconformity cuts acrossbedding in the underlying unit at a low angle. It isnow an irregular surface but this appears to be theresult of younger contractional deformation.

Interpretation

Fortuin et al. (1992, 1994) record a significantchange above the thin Late Burdigalian sequenceto volcaniclastic turbidites, which we suggest corre-sponds to Unit 2. All these rocks are poorly dateddue to reworking and absence of microfossils.Dates from the volcaniclastic turbidites (Fortuinet al. 1994) are from NN5 (14.8–13.5 Ma) toNN11 (8.3–5.5 Ma). Fortuin et al. (1997) suggestthat volcanic input waned during the Tortonian(11.5–7 Ma) and revived for a short time duringthe Messinian. Deposits commonly contain abun-dant fragments of pumice, lava and orientated plagi-oclase, along with broken crystals of volcanic zonedplagioclase, anorthite (40–60%), volcanic quartz,augite, orthopyroxene, amphibole, hematite andminor amounts of hornblende, all of which points

to an association with a typical island arc (Fortuinet al. 1994).

The changes in thickness of this volcanogenicsequence on land between central and east Sumbaresembles variation seen on the seismic lines.Thickness is up to 1.6 s TWT, and locally wellbedded parts of the sequence pass laterally intounreflective packages consistent with rapid depo-sition of volcaniclastic material as debris flowsand turbidites. Fortuin et al. (1997) suggested thatformation of a basin slope, facilitated by NE–SWorientated faults, led to subsidence below the car-bonate compensation depth in east Sumba duringthe Middle Miocene and the Late Miocene incentral Sumba. Dissolution of carbonate, but thepresence of calcareous nannofossils in some sam-ples, indicate depths of 4 to 5 km (Fortuin et al.1994, 1997). Fortuin et al. (1997) suggest thesedeposits were derived from the south and formpart of a 100 km fan that progrades northward.

Unit 3

This unit is a particularly distinctive at the top of theseismic lines crossing the southern part of the SavuBasin. It is characterized by a series of flat sheet-like

Fig. 3. Uninterpreted and interpreted seismic line from the northern edge of the Sumba Ridge illustrating therelationships between Units 1, 2 and 3. Inset map shows location of seismic lines used in this study and the location ofthis line in red.


deposits, picked out by alternations of very brightand lower amplitude reflectors (Fig. 3). The reflec-tors become less bright towards the top of thesection. The unit typically has a thickness of about1 s TWT (maximum thickness of 1.8 s TWT) andisochron maps suggest it was derived from thesouth because it thickens north into the deep SavuBasin from the present-day NW–SE trending sub-marine Sumba Ridge, and then becomes thinnerstill further north in the basin. Unit 3 appears todownlap onto the underlying unconformity to thenorth of the Sumba Ridge (Fig. 4). On the SumbaRidge Units 3 and 2 are broadly conformable, andon the few undeformed sections south of theSumba Ridge Unit 3 onlaps the unconformity.

Onlap between the individual layers of this unitcan be seen throughout although there is no apparentpattern. A number of slumped packages are presentwithin Unit 3, some of which are very well imagedby marked contrasts in reflector character. They arecommonly between 0.1 and 0.2 s TWT in thickness,at different levels in Unit 3, and are localized fea-tures that can often only be correlated within a25 km radius. The slumped packages are character-ized by a more transparent seismic character thanthe overlying and underlying well bedded sequenceswith high amplitude reflectors. There are also anumber of more discrete slumped units within the

top part of this unit, which are on average 0.5–1 sTWT in thickness. These are slumps with a welldefined base, which detached at a bedding plane,and at their distal end the basal thrust ramps upagainst underlying sediment. Their lack of internalseismic character makes them stand out from thesheet-like deposits that make up most of Unit 3.This unit is also cut by many small displacementhigh angle normal faults, with spacing of 8–10 km.

Interpretation

Unit 3 is interpreted as Tortonian to Pliocene, withforaminiferal chalks and marls containing varyingamounts of hemipelagic nannofossil oozes and vol-canogenic muds (von der Borch et al. 1983; Fortuinet al. 1992, 1997). There are also some thicklybedded volcanic mass flow deposits. The increasein foraminiferal chalks up-section is probably thereason why reflectors become less bright towardsthe top of the unit.

We suggest the apparent downlap north of theSumba Ridge is the result of uplift of the ridge rotat-ing the former onlapping sequence, supported by therelatively steep dip of beds (up to 5–68) which forthese fine grained sediments is unlikely to relate toan original depositional slope. The onlap betweenthe layers within the correlative unit on Sumba

Fig. 4. Uninterpreted and interpreted seismic line from north of Savu showing apparent northward downlap and tiltingof sediments of Units 2 and 3 away from the elevated Sumba Ridge.


was observed by Fortuin et al. (1997) and is sug-gested to be the result of shifting fan lobes.

The increasing abundance of slumps in the upperparts of the unit suggest that uplift of the SumbaRidge occurred during deposition of Unit 3. Theslumped units with downslope ramps resemble fea-tures described by Bull et al. (2009) interpreted asdue to pre-existing weaknesses in the mechanicalproperties of the basal shear surface or stresses gen-erated as material moves downslope.

Unit 4

The boundary between Units 3 and 4 is often unclear,but in places onlap between these sequences canbe recognized. The strata of Unit 4 are characterizedby a weakly reflective seismic background withinwhich numerous erratically distributed and ran-domly orientated high amplitude reflectors can beseen. Unit 4 is universally present at the base ofthe slope north of the Sumba Ridge. Isochronmaps indicate that material was derived from thesouth and movement into the basin was facilitatedby extensional faulting at the top of the slope,which can be seen clearly throughout Unit 3.

Four phases of slump infill can be identified(Fig. 5). These are separated by minor unconformi-ties and backstep towards their source as the basinwas progressively infilled. Large blocks are oftenpresent within these deposits which in places retainsome of the bedded character of the original deposit.

In the northern part of the basin, deposits arebetter bedded with flat, parallel and bright reflectorswhich are less prominent at greater depths. Theseinterfinger with the slumps. The well beddedsequences could represent turbidite deposits at thedistal ends of debris flows but could also representmaterial carried into the deeper parts of the basinfrom a different source.

An important point is that basin infill varies sig-nificantly from east to west. In the SE, close toTimor and Roti, Unit 3 is thin (up to 0.25 s TWT)and Unit 4 is thick (about 1 s TWT). Close toSumba the thicknesses of Units 3 and 4 are reversed.The sediment slides associated with Unit 4 increasein importance towards the east.

Interpretation

These deposits formed by simultaneous uplift anddownslope transport of debris into the northern partof the Savu Basin during the early Pliocene (vander Werff et al. 1994). Ongoing tilting is fundamentalto the generation of the 4000 km2 of superimposedslumps seen on Sumba (Fortuin et al. 1992; Roep& Fortuin 1996) and on offshore seismic lines.These deposits are associated with acoustic voids,which is a direct result of water expulsion, causedby a considerable overburden being deposited dur-ing a short time period (van Weering et al. 1989a).

The thinning of Unit 3 to the east in the SavuBasin reflects a greater distance from the sediment

Fig. 5. Uninterpreted and interpreted seismic line crossing the Savu Basin showing the units identified in the deeperpart of the basin. On this section Unit 2 is thin and Unit 3 is missing in parts of the section. In contrast Unit 4 ismuch thicker and can be subdivided into several sub-units.


source which was the Sumba Ridge. In contrast, thegreater thickness of Unit 4 in the east is likely to belinked to increased uplift and subsequent slumpingfrom Timor. The well bedded sequences in themost northern parts of the basin may have beenderived from the Banda Volcanic Arc to the north,or represent finer material carried further into thebasin as part of the slumped packages from therising Sumba Ridge. There is some interfingeringof these bedded deposits with Unit 4 and thisfavours a Banda Arc origin. There are a number ofslumped horizons at the very top of the seismicsection closest to the island of Flores and thesealmost certainly relate to downslope depositionfrom the Banda Arc.

Structures

The principal structures mapped from the seismicdataset are shown on Figure 6. These are describedfrom north to south.

Savu Basin

Extensional faulting is clearly visible on the north-ern edge of the Sumba Ridge, and there are numer-ous normal faults further north into the Savu Basin.Stretching of the Savu Basin, facilitated by rotationon domino style fault blocks, created a series of fulland half graben. The main normal faults are com-monly 15 km apart, and although the extensiondirection cannot be precisely determined becauseof the wide spacing between seismic lines, it isbroadly north–south. The faults have not beeninverted. Displacement on these high angle faultshas contributed to the subsidence of the top ofUnit 1 from depths of 2 s on the ridge to 7 s TWTin the deepest parts of the basin. There has beentilting of Units 2 to 4 which in part reflects upliftof the Sumba Ridge but also greater subsidence inthe northern part of the basin.

Savu Thrust

The Savu Thrust is actually a zone of thrusting at thenorthern margin of Savu Island (Harris et al. 2009).The faults dip south and at the rear of this zone somehave displacements of more than 2 s TWT (Fig. 7).The most important thrust reaches the seabedwhereas others are blind. The geometry of the mainSavu Thrust is obscured by poor seismic quality inuplifted areas. There are numerous smaller reversefaults both behind and in front of the main thrust,which in some places break through to the surface.Immediately in front of the main thrust there is afootwall syncline in Unit 3 associated with minorthrusts that do not reach the seabed. Folds furthernorth of the main thrust are associated with

inversion of older extensional faults that havebeen reactivated as blind thrusts which cut upsection to Unit 3. Behind the main thrust, faultsand associated folds are progressively steepened,leading to the formation of back thrusts. Faultpropagation folds in the form of hanging wall anti-clines and footwall synclines can be seen abovethe tips of all thrusts. No bends in the fault planescan be seen and there are steep front limbs andsome forelimb thinning (Fig. 7).

The main Savu Thrust is two separate thrusttraces offset by roughly 5 km just east of Savu(Fig. 6). The western thrust can be traced on landinto north Savu (Harris et al. 2009) and the easternthrust is offset to the south and can be traced fromthe eastern side of the island. The displacement isgreatest in the middle of each of these faults, anddiminishes to the east and west.

There are several episodes of deformationassociated with the thrust zone, marked by promi-nent unconformities in deformed Unit 3. The firstmajor phase of movement folded deeper parts ofUnit 3 and folds are onlapped by younger reflectors.There are numerous subsequent subtle onlaps ofreflectors near the top of Unit 3, associated with syn-deformation thickening away from the main fault.

Roti Thrust

This thrust zone (Fig. 8) is located to the NW of Rotiand has a similar style to the Savu Thrust. It islocated offshore 20 km north of Roti, has the sameNNE–SSW orientation as the island, and runs paral-lel to it for its entire 70 km length. The faults dipsouthward and again displacement is greatest inthe centre of the fault zone. The thrust zone is about25 km south of the eastern strand of the Savu Thrust.The thrust zone is wider but the maximum dis-placement on the most important thrusts is lessthan in the Savu thrust zone, at roughly 1 s TWT.The net displacement on both zones appears com-parable (Fig. 6).

Like the Savu Thrust, several phases of defor-mation can be identified by mapping of minorunconformities in the deformed zone with onlapand thickening of packages in front of the mainthrust. A difference between the two thrust zonesis that the Roti thrust is associated with fewer back-thrusts and there are more inverted normal faults infront of the main thrust (Fig. 8).

NE–SW thrusting south of Sumba

On the south side of the Sumba Ridge, in the area SEof Sumba there are numerous northward dippingthrust faults. They are associated with a number ofanticlines which are very well imaged on seismiclines and are 1000–1400 m across. The orientation


Fig. 6. (a) Location of seismic lines in the Savu Basin area. The parts of seismic lines shown in this paper are marked ingreen with the corresponding figure number in square brackets. (b) Summary structure map with the principaltectonic features identified in the area of study. The coloured shaded area north of the Java Trench and Timor Trough is,from west to east, the transition from the accretionary fore-arc complex, which passes east into a deformed zoneincluding fore-arc and Australian sedimentary cover forming the Savu–Roti Ridge, into the arc–continent collisionzone of Timor. The dashed blue line crossing the Savu–Roti Ridge and Timor is the approximate position of theformer Banda Trench. The dotted black line north of Timor is the inferred northern limit of continental basement. Thedashed red line below Sumba is the inferred northern and western limits of the subducted Scott Plateau. See text fordiscussion. The heavy black line marks the position of the section drawn in Figure 10.


of faults and fold axes is difficult to determinebecause of the wide spacing of seismic lines in thisarea but they are clearly not parallel to the SumbaRidge and our best estimate is that they have atrend of about 0308 (Fig. 6).

This zone of thrusting is roughly 60 km acrossand is overlain by an almost undeformed sequenceof sediments which dips northwards and is almost1 s TWT thick. This sequence appears to be theequivalent of Unit 3 in the Savu Basin and sedimentswere probably derived from the NW, originallythinned to the SE, and have now been tilted by therise of the Savu–Roti Ridge. The zone of thrusting,which is now a sedimentary basin, widens towardsthe Lombok basin to the west.

NE–SW thrusting within the Savu–Roti Ridge

Within the Savu–Roti Ridge are numerous thrustfaults which are visible in the upper 1 s TWT ofthe seismic lines. The trend of these faults is very

similar to those seen to the south of Sumba. Thereare two particularly clear zones of deformedmaterial, shown on Figure 6. The southern zone ischaracterized by small, well imaged northward-dipping thrusts, which become progressively rotatedand steepened towards the north. The majority offaults within the northern, slightly broader, zone ofdeformation closer to Sumba dip north, althoughsome smaller antithetic faults can also be seen.Between these two zones some faults and reflectorscan be seen in places but overall the Savu–RotiRidge appears to be a seismically opaque mass ofdeformed material.

Discussion

From the Eocene to the Early Miocene there hadbeen subduction of oceanic lithosphere at the JavaTrench as Australia moved north (Hall 2002). Theregion around Sumba was situated at the SE

Fig. 7. Seismic section showing the Savu Thrust north of Savu Island. The upper part of Unit 3 could correspond toUnit 4 in the deeper parts of the basin.


corner of Sundaland in the Early Miocene, at theeastern end of the Sunda Arc, when Australiancontinental crust of the Sula Spur began to collidewith the North Sulawesi volcanic arc. South of theSula Spur was the Banda embayment, an area ofJurassic–Cretaceous ocean crust within the Austra-lian continental margin with Timor on its south side.At about 15 Ma there was a major change withinitiation of new subduction in the Banda region(Spakman & Hall 2010). The Java Trench becamealigned with southern side of the Sula Spur andoceanic lithosphere of the Banda embaymentbegan to subduct due to its negative buoyancy andthe subduction hinge rolled back to the SE, form-ing the west-plunging lithospheric fold definedtoday by seismicity. Figure 9 shows our interpre-tation of the Savu Basin in the context of subductionin the Banda region. Collision of the volcanic arcwith the Australian continental margin in Timorbegan at about 4 Ma (Audley-Charles 1986; Hall2002).

We suggest that the SE corner of Sundalandwhich includes the region around Sumba and theSavu Basin was underlain by continental crust thathad been accreted to Sundaland in the mid

Cretaceous (Hall et al. 2009). This suggestion issupported by crustal thicknesses and densities forthis region presented by Shulgin et al. (2009).Volcanic activity between the Paleocene andEocene marked a brief phase of subduction whichceased during the Oligocene when shallow watercarbonates were deposited on Sumba. We interpretthe oldest rocks, which make up Unit 1 in the SavuBasin, to represent the Late Cretaceous to EarlyMiocene interval. We correlate the distinctivebright reflectors at the top of Unit 1 with the Palaeo-gene or Lower Miocene shallow marine limestonesdescribed on Sumba (von der Borch et al. 1983;Fortuin et al. 1994, 1997; van der Werff et al.1994). This horizon can be traced across the wholeof the Savu Basin implying that the whole regionwas close to sea level in the Early Miocene.

Subsidence and volcanism

Burollet & Salle (1982) suggested subsidence onSumba began in the Early to Middle Miocene.Based on dating by Fortuin et al. (1994) it appearsto have begun at about 15 Ma with breakup of a car-bonate platform leading to rapid subsidence in the

Fig. 8. Seismic section showing the geometry of the Roti Thrust, 25 km NW of Roti Island. The upper part of Unit 3could correspond to Unit 4 in the deeper parts of the basin.


early Middle Miocene (Fig. 9). The shallow waterOligo-Miocene limestones have equivalents in theeast, on Timor, in the Cablac Limestone (seeAudley-Charles 2011), implying that most of whatis now the Banda fore-arc between Sumba andEast Timor was at sea level before the earlyMiddle Miocene. On Sumba, subsidence coincidedwith a significant change to volcaniclastic turbi-dites, marking the beginning of volcanic activity(Fortuin et al. 1992, 1994), that we correlate withUnit 2 offshore. Volcanic input is recorded onSumba from NN5 (14.8–13.5 Ma) to NN11 (8.3–5.5 Ma) with a possible decline in the Tortonian(11.5–7 Ma).

Fortuin et al. (1997) suggest these deposits werederived from the south and form part of a fan that

prograded northwards across what is now eastSumba and the offshore region to the east, whichimplies a short-lived volcanic arc to the south ofSumba. We agree that material was derived fromthe south but there are several problems with thesouthern arc interpretation. It would have beenunusually close to the subduction trench (van derWerff et al. 1994), significantly south of otherMiddle Miocene volcanic activity along the SundaArc from Java to Sumbawa, and would requirean exceptionally steep dip on the subducted slabfrom the trench to the typical 100þ km depth tothe Benioff zone for volcanoes between Javaand the Banda Arc at the present day (Englandet al. 2004). The shift of the position of the arcto its current position in Sumbawa–Flores would

Fig. 9. Cartoon showing evolution of the geometry and tectonic features of the Savu Basin and surrounding region fromthe Middle Miocene to the present-day. Inset maps show reconstructions of the Banda region based on Spakman & Hall(2010). See text for detailed discussion. In the Middle Miocene the formerly shallow marine are subsided as trenchrollback began. The dashed blue line continuing east from the Java Trench on the 3 and 0 Ma maps is the inferredposition of the former Banda Trench. By 3 Ma continental crust of the Australian margin had arrived at the trench, andtoday has been subducted beneath the Sumba Ridge.


require a dramatic reduction in slab dip to itspresent angle.

An alternative is that volcaniclastic input mayhave originated in the volcanic islands to the NWof Sumba and was transported by movement ofmaterial as turbidite flows SE through the LombokBasin and then northward into the Savu Basin. Wesuggest that Sumba, although submerged, was con-sistently a relative bathymetric high in the fore-arcapproximately parallel to the present orientation ofthe Sumba Ridge from about the Middle Mioceneonwards. West Sumba was at very shallow depthsthrough the Neogene (Fortuin et al. 1997) and tothe west of Sumba sediment thicknesses are muchthinner than further west in the Lombok Basin(van Weering et al. 1989b), implying an east–west to NW–SE orientated ‘proto-Sumba Ridge’extending from Sumbawa to East Sumba. NE–SWorientated faults in Sumba, which allowed EastSumba to subside to depths of 4 to 5 km (Fortuinet al. 1994, 1997) before the Tortonian, and wouldhave opened a passage into the Savu Basin fromthe SW (Fig. 9).

Because of the orientation and spacing ofseismic lines it is difficult to be certain of the orien-tation of extensional faults. Fortuin et al. (1994,1997) observed NE–SW faults on Sumba. Offshoreto the north of Sumba, based on those that can becorrelated between seismic lines, faults appear tobe oriented close to north–south. In the SavuBasin extensional faults have a broadly east–westtrend, curving from WNW in the west to ENE inthe east.

The Sumba Ridge is a feature with invertednormal faults on its north and south sides indicatingWNW–ESE trending extensional faults continuedas far east as the present longitude of Savu(Fig. 9). We suggest the WNW-trending SumbaRidge is parallel to a deep basement trend that wasinherited from Australian continental basementwhich accreted to the SE corner of Sundaland inthe mid Cretaceous (Hall et al. 2009). In contrast,north–south faults close to the volcanic arc wereprobably formed by along-arc extension, and weinterpret the east–west to ENE-extensional faultsin the basin to have formed in response to subduc-tion initiation and rollback into the Banda embay-ment which began at about 15 Ma, based onregional arguments (Hall 2002, 2009, 2011;Spakman & Hall 2010). Fleury et al. (2009) reportunpublished K–Ar ages of 16 Ma from Hendaryono(1998) for the oldest volcanic rocks on Flores con-sistent with this age estimate.

Thrusting

The oldest thrusts are those north of the Savu–Rotiridge to the SE of Sumba. These have a ENE–WSW

trend and are now covered by up to 1 s TWT of sed-iment. It is not possible to correlate these sequenceswith the Savu Basin as there are no lines that crossfrom SW to NE of the Sumba Ridge but theseismic character of the thrusted sequences is verysimilar to Unit 3. We interpret these as structuresformed in the accretionary complex north of, butclose to, the former Banda Trench. Thrusts withsimilar strike are found in the Savu–Roti Ridgewithin narrow zones all within a much broaderzone of seismically opaque, apparently highlydeformed material, that forms most of the ridgeand resembles accretionary complexes close tosubduction trenches.

The Sumba Ridge has clearly been elevatedduring the collision process. The uplift of the ridgepostdates the thrusts south of Sumba which are over-lain by up to 1 s TWT of sediment but predates theSavu and Roti Thrusts. This uplift clearly postdatesUnit 3 which now dips northwards from the ridge.Unit 4 in the Savu Basin includes numerous slumpedsequences that have moved northwards from theridge. There is no observable inversion of normalfaults, or thrusting associated with the uplift ofSumba Island and the Sumba Ridge which appearsrather to have deformed as a broad upwarp. Wesuggest this marks the first arrival of the Australianmargin at the Banda Trench in the Sumba region,probably at about 2–3 Ma, which postdated col-lision of the volcanic arc and Australian continentalcrust in East Timor. In the Savu Basin there areslumps in Unit 4 which are directed basinwardfrom Sumba and Timor but it is not possible toidentify relative timing from the seismic data set.East Timor uplift began at about 4 Ma (Audley-Charles 1986; de Smet et al. 1990). On Sumbathere was uplift from depths of more than 5 km toemergence of up to 1 km above sea level since4 Ma (Pirazzoli et al. 1993; Fortuin et al. 1997).

The Savu and Roti Thrusts are much youngerthan all other structures and are currently active.Strictly speaking, these are zones of deformationthat include multiple thrusts rather than singlefaults. Both zones include thrusts that emerge atthe sea floor and blind thrusts that are associatedwith deformation of the sequence above andfolding of the seabed. The two main thrust fault seg-ments associated with the Savu Thrust identified onthe seismic lines can be correlated onshore withfault segments identified by Harris et al. (2009)which generate topography on the island itself.According to Harris et al. (2009) there are up toseven forward propagating limbs, some of whichbreak through, closely resembling what is seen onthe seismic lines. Harris et al. (2009) suggest thatmost structural features are generally orientedENE–WSW, sub-parallel to the structural grain ofthe Scott Plateau and Australian continental margin


to the east. This is true close to Savu but over alarger area the Savu Thrust zone is closer to east–west. North of Roti the second thrust zone is verysimilar to that north of Savu, but has a differentNNE–SSW orientation, broadly parallel to theisland. This too is a zone of active thrusting. TheSavu and Roti Thrust are separated by about25 km and do not link up. Both die out to the eastand west. Roosmawati & Harris (2009) show thatsignificant and rapid uplift from water depths ofmore than 2 km to emergence of the islands ofSavu and Roti began at about 2 Ma and they arethe first parts of the accretionary wedge west ofTimor to emerge. This is supported by the seismicevidence which shows that the area south of thetwo thrust zones is generally seismically opaquewith localized zones in which north-vergentthrusts can be recognized. We assume that thisaccretionary zone includes material from both theBanda fore-arc and the deep Australian continentalmargin, and this is also suggested by observationson land in Roti and Savu (Harris et al. 2009;Roosmawati & Harris 2009).

The seismic lines show clearly that the Savu andRoti Thrusts are young and not lithosphere-scalefeatures as they die out rapidly to the east andwest. There are no other thrust zones north ofthem in the Savu Basin, which rules out the continu-ous major thrust zone that is often traced from EastTimor to the south of Sumba (e.g. Fortuin et al.1997; Audley-Charles 2004; Shulgin et al. 2009).In the Savu Basin the only contractional defor-mation seen on the seismic lines are slumps inUnit 4, and there are no significant features on theseabed which is essentially flat or gently sloping.It appears that the accretionary complex thatformed north of the Banda Trench during oceanicsubduction has been overridden and/or incorpor-ated in the wedge of deformed material south ofthe Savu and Roti Thrusts, except in the extremeSE where it is seen beneath flat lying sedimentssouth of Sumba.

The Benioff zone thrust of the Banda subduc-tion zone has no surface expression and is beneaththe Savu–Roti Ridge. Most of the convergencebetween the Australian continent and the Banda vol-canic arc is concentrated in a zone of contractionaldeformation about 120 km wide within which areSavu and the Savu–Roti Ridge, and further eastthe island of Roti. The southern limit of this zoneis a trough where there is south-directed thrustingwhich can be traced west to the Java Trench andeast to join the Timor Trough and is entirelywithin the Australian margin. East of Roti the con-tractional zone becomes wider and more substantialwith elevations on Timor of more than 3 km and thezone of thrusting has overridden the former fore-arcto the north and reduced the distance to the former

volcanic arc. This is the reason for NE-narrowingof the Savu Basin.

The shape of the Australian continental marginwas the cause of the deformation history and thecomplex shape of the collision zone. In the regionsouth of Savu and Timor there were rectilinearsteps in the continental–ocean boundary similar tothose seen today at the Scott and Exmouth Plateausof the NW Australian shelf. The first volcanic arc–continent collision was north of East Timor anddeformation propagated west with time. Harris(1991) and Roosmawati & Harris (2009) suggestedthat this sector of the collision can be understood interms of the present convergent rate between SEAsia and the Australian plate, which is broadlytrue. However, in addition to the shape of the conti-nental margin and the plate convergent vector it isalso important to recognize that there was a com-ponent of convergence due to the SE-rollback ofthe Banda Trench into the Banda embayment(Spakman & Hall 2010). After collision in EastTimor most oceanic crust had been subducted asfar as the west edge of the Scott Plateau, whereasto the east a significant area of oceanic crustremained which has been subducted since 2 Ma.

Cross section

The cross section (Fig. 10) illustrates the key fea-tures of the convergence between the Scott Plateauand the Savu Basin. The section is located alongone of the long north–south seismic lines just eastof Savu and is close to the line of section ofShulgin et al. (2009). We have used the crustalthicknesses and densities of their section which arebased on seismic reflection, refraction, tomographicand gravity data. We suggest the position of the sub-ducted ocean–continent boundary is now north ofthe elevated Savu–Roti Ridge and it was thearrival of the edge of this thickened continentalplateau which has driven the young thrusting atthe Savu and Roti Thrusts. The crust of the Bandafore-arc and the Australian margin have similarthicknesses and densities because both are continen-tal, and both are ultimately Australian. The ScottPlateau was stretched during Late Jurassic riftingbut remained part of the Australian continent. Incontrast, the continental crust beneath the SavuBasin was stretched during Late Jurassic riftingbut then separated from Australia before accretionto the Sundaland margin in the mid Cretaceous(Hall et al. 2009). It was then stretched againduring Middle and Late Miocene subduction roll-back into the Banda embayment (Spakman & Hall2010).

Thick Middle Miocene to Pliocene sediments(Unit 2 and lower Unit 3) of the Savu Basin dipand thin northwards into the basin from the elevated


Sumba Ridge, whereas Pliocene to Recent sedi-ments (upper Unit 3 and Unit 4) thicken northwardsdue to slumping from the Sumba Ridge into thebasin. On the north side of the basin Unit 4 probablythickens southward because of sediment carriedsouth from the volcanic arc. There is no significantdeformation in the basin. The elevated ridge is anaccretionary complex composed partly of Bandafore-arc sediments and partly of Australian marginsedimentary cover. To the south the surface of thiswedge dips south and is formed entirely of detachedAustralian margin sedimentary cover.

Conclusions

The Savu Basin records the Miocene to Recenthistory of convergence between Australia and theSE part of Sundaland. We interpret this region tobe underlain by continental crust that was added tothe Sundaland margin in the mid Cretaceous.Before the Middle Miocene the region includingSumba and the Savu Basin was close to sea leveland subsided rapidly in the late Middle Miocenein response to extension induced by subductionrollback at the Banda Trench as the Java Trenchpropagated east into the Banda Embayment. Theextension is marked by widespread normal fault-ing. A thick succession of volcaniclastic turbiditeswas deposited in the basin and was derived fromthe SW. This material is interpreted as derivedfrom the Sunda Arc to the west of Sumba with flow

being influenced by the relatively shallow SumbaRidge which caused turbidity currents to flow firstSE and then NE into the Savu Basin. The SumbaRidge is likely to be a feature that reflects the deepstructure of the Sunda margin. The western partremained a shallow bathymetric feature during theNeogene although the SE part subsided to depthsof more than 4 km.

The Sumba Ridge was elevated as continentalcrust of the Australian margin arrived at the BandaTrench and was flexed into a broad upwarp thattilted the volcaniclastic turbidite sequence andlater caused debris flows and turbidites to flownorthwards into the basin. Slumps seen on seismiclines came from both Sumba and Timor. Fortuinet al. (1992) noted that slumping affects the areabetween Sumba and Timor more than areas withsimilar or steeper slopes such as those offshoreFlores where there is also abundant seismicity.This may reflect the combination of tectonic stee-pening and the layered turbidites and chalk inter-beds which detached along bedding surfaces as thecollision complex was elevated. Apart from tiltingand slumping the Savu Basin is little deformedand there are no thrusts within it.

South of Sumba a small part of the originalaccretionary complex is preserved, now buriedbeneath almost 1 s of sediment. Most of the pre-collisional accretionary complex has been incorpor-ated in a large accretionary zone which is south ofthe Savu and Roti Thrusts and includes bothBanda fore-arc material and Australian sedimentary

Fig. 10. Cross section along a north–south transect close to Savu Island (see Fig. 6 for location) sub-parallel to thelongest seismic line crossing the Savu Basin which illustrates how the Australian plate and the Banda fore-arc interact inthe Savu region. Faults are shown schematically in red. Thicknesses of seismic units are slightly exaggerated for clarity.Crustal densities and deep structure were inferred using information from a transect by Shulgin et al. (2009).


cover. This deformed complex is bounded to thenorth and south by north- and south-vergent thrusts;the former trench is now deep beneath this complexand the lithospheric faults do not emerge at thesurface. This deformed zone has developed as theAustralian continental crust has been thrust beneaththe Banda fore-arc—we suggest the northern edgeof the Scott Plateau is now beneath the SumbaRidge, contributing to the young thrusting north ofSavu and Roti. These thrusts reflect the shape ofthe pre-collisional Australian margin and havecaused the islands of Savu and Roti to rise fromdepths of more than 2 km since 2 Ma and emergevery recently. Underthrusting of the Sumba Ridgeby the Scott Plateau is probably contributing to thecontinued elevation of Sumba and extensional col-lapse seen on the island. The Timor Trough con-nects to the Java Trench as a bathymetric featurealong the southern zone of south-vergent thrustingbut it is entirely a shallow feature entirely withinthe Australian margin. The trough may haveformed at the site of an earlier deeper bathymetricfeature within the continental margin but itsprimary cause is south-directed thrusting andloading by the collisional complex now about120 km wide south of Savu.

We thank the consortium of oil companies who support theSE Asia Research Group, Steve Toothill and CGG Veritasfor permission to use the seismic data and for helpfuldiscussion, and Chris Elders, Mike Audley-Charles andWim Spakman for help and discussion.

References

Audley-Charles, M. G. 1986. Rates of Neogene andQuaternary tectonic movements in the SouthernBanda Arc based on micropalaeontology. Journal ofthe Geological Society, London, 143, 161–175.

Audley-Charles, M. G. 2004. Ocean trench blocked andobliterated by Banda forearc collision with Australianproximal continental slope. Tectonophysics, 389,65–79.

Audley-Charles, M. G. 2011. Tectonic post-collisionprocesses in Timor. In: Hall, R., Cottam, M. A. &Wilson, M. E. J. (eds) The SE Asian Gateway:History and Tectonics of the Australia–Asia Collision.Geological Society, London, Special Publications,355, 235–260.

Bull, S., Cartwright, J. & Huuse, M. 2009. A review ofkinematic indicators from mass–transport complexesusing 3D seismic data. Marine and PetroleumGeology, 26, 1132–1151.

Burollet, P. F. & Salle, C. 1982. Histoire geologiquede l’ile de Sumba (Indonesie). Bulletin de la Societegeologique de France, 24, 573–580.

De Smet, M. E. M., Fortuin, A. R. et al. 1990. Detectionof collision-related vertical movements in the OuterBanda Arc (Timor, Indonesia), using micropaleontolo-gical data. Journal of Southeast Asian Earth Sciences,4, 337–356.

Effendi, A. C. & Apandi, C. 1980. Geological map ofSumba quadrangle, Nusa Tenggara. GeologicalResearch & Development Centre, Ministry of Minesand Energy, Bandung, Indonesia.

England, P., Engdahl, R. & Thatcher, W. 2004. Sys-tematic variation in the depths of slabs beneath arcvolcanoes. Geophysical Journal International, 156,377–408.

Fleury, J.-M., Pubelier, M. & Urreiztieta, M. 2009.Structural expression of forearc crust uplift due to sub-ducting asperity. Lithos, 113, 318–330.

Fortuin, A. R., Roep, T. B., Sumosusastro, P. A., van

Weering, T. C. E. & van der Werff, W. 1992.Slumping and sliding in Miocene and Recent devel-oping arc basins, onshore and offshore Sumba(Indonesia). Marine Geology, 108, 345–363.

Fortuin, A. R., Roep, T. B. & Sumosusastro, P. A. 1994.The Neogene sediments of East Sumba, Indonesia –products of a lost arc? Journal of Southeast AsianEarth Sciences, 9, 67–79.

Fortuin, A. R., van der Werff, W. & Wensink, G. 1997.Neogene basin history and paleomagnetism of a riftedand inverted forearc region, on- and offshore Sumba,Eastern Indonesia. Journal of Asian Earth Sciences,15, 61–88.

GEBCO 2003. General Bathymetric Chart of the Oceans.IHO-UNESCO, Digital Edition 2003, http://www.gebco.net/.

Hall, R. 2002. Cenozoic geological and plate tectonicevolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. Journalof Asian Earth Sciences, 20, 353–434.

Hall, R. 2009. Hydrocarbon basins in SE Asia: under-standing why they are there. Petroleum Geoscience,15, 131–146.

Hall, R. 2011. Australia–SE Asia collision: plate tec-tonics and crustal flow. In: Hall, R., Cottam, M. A.& Wilson, M. E. J. (eds) The SE Asian Gateway:History and Tectonics of the Australia–Asia Collision.Geological Society, London, Special Publications,355, 73–104.

Hall, R., Clements, B. & Smyth, H. R. 2009. Sunda-land: Basement character, structure and plate tectonicdevelopment. In: Proceedings Indonesian PetroleumAssociation, 33rd Annual Convention, IPA09-G-134,1–27.

Harris, R., Vorkink, M. W., Prasetyadi, C., Zobell, E.,Roosmawati, N. & Apthorpe, M. 2009. Transitionfrom subduction to arc-continent collision: Geologicand neotectonic evolution of Savu Island, Indonesia.Geosphere, 5, 152–171.

Harris, R. A. 1991. Temporal distribution of strain in theactive Banda orogen: a reconciliation of rival hypoth-eses. Journal of Southeast Asian Earth Sciences, 6,373–386.

Hendaryono, A. 1998. Etude geologique de l’ile deFlores. PhD thesis, Universite de Savoie, Chambery,France.

Pirazzoli, P. A., Radtke, U. et al. 1993. A onemillion-year-long sequence of marine terraces onSumba island, Indonesia. Marine Geology, 109,221–236.

Roep, T. B. & Fortuin, A. R. 1996. A submarine slidescar and channel filled with slide blocks and


http://www.gebco.net/



megarippled Globigerina sands of possible contouriteorigin from the Pliocene of Sumba, Indonesia. Sedi-mentary Geology, 103, 145–160.

Roosmawati, N. & Harris, R. 2009. Surface uplifthistory of the incipient Banda arc–continentcollision: geology and synorogenic foraminifera ofRote and Savu Islands, Indonesia. Tectonophysics,479, 95–110.

Sandwell, D. T. & Smith, W. H. F. 2009. Global marinegravity from retracked Geosat and ERS-1 altimetry:Ridge Segmentation v. spreading rate. Journal ofGeophysical Research, 114, B01411, doi: 10.1029/2008JB006008.

Shulgin, A., Kopp, H. et al. 2009. Sunda–Banda arctransition: Incipient continent–island arc collision(northwest Australia). Geophysical ReserarchLetters, 36, L10304, doi: 10.1029/2009GL037533.

Spakman, W. & Hall, R. 2010. Surface deformation andslab-mantle interaction during Banda Arc subductionrollback. Nature Geoscience, 3, 562–566.

Toothill, S. & Lamb, D. 2009. Hydrocarbon prospectiv-ity of the Savu Sea basin. In: Proceedings Indonesian

Petroleum Association, 33rd Annual Convention,IPA09-G-013, 1–12.

van der Werff, W., Kusnida, D., Prasetyo, H. & van

Weering, T. C. E. 1994. Origin of the Sumba forearcbasement. Marine and Petroleum Geology, 11,363–374.

van Weering, T. C. E., Kusnida, D., Tjokrosapoetro,S., Lubis, S. & Kridoharto, P. 1989a. Slumping,sliding and the occurrence of acoustic voids in recentand subrecent sediments of the Savu Forearc Basin,Indonesia. Netherlands Journal of Sea Research, 24,415–430.

van Weering, T. C. E., Kusnida, D., Tjokrosapoetro,S., Lubis, S., Kridoharto, P. & Munadi, S. 1989b.The seismic structure of the Lombok and Savuforearc basins, Indonesia. Netherlands Journal of SeaResearch, 24, 251–262.

von der Borch, C. C., Grady, A. E., Hardjoprawiro,S., Prasetyo, H. & Hadiwisastra, S. 1983. Mesozoicand late Tertiary submarine fan sequences and theirtectonic significance, Sumba, Indonesia. SedimentaryGeology, 37, 113–132.


savu basin

Documents