GeowgicaL Soci£ty of MalayJUz - circum-Pacific Council for Energy a/lu MineraL I!':0urc~ Tectonic Fl'flITUwork and Energy Ruourcu of tbe W(Jtern Margin of the PacifIC Btum November 27 December 2, 1992, Kuo.Ia Lumpur, MalayJUz Lithosphere structure and dynamics of the Banda Arc collision zone, Eastern Indonesia ADRIAN RICHARDSON Geological Research in Southeast Asia, Department of Geology Royal Holloway and Bedford New College, Egham, Surrey, TW20 O~X, England Present Address: Shellinternationale Petroleum MaatschapplJ B.V. Postbus 162, 2501, AN Den Haag, Oostduinlaan 75, Nederland Abstract: Timor and adjacent islands in Eastern Indonesia are the site of a currentl~ active col~ision between the Australian continental shelf and the volcanic islands of the Banda Arc. PreVIously pubhshed sedimentological and micro-paleontological data from the islands and the subduction trench date the start of this collision at 2.5 to 3 Ma. Previously published work on seismicity in the Banda Arc shows the geometry of the subduction zone to be fairly simple and continuous down to below 650 km depth. In this paper, the geometry of both the top and bottom surfaces of the subd~cting sla~ is mapp~d to a hl:gh~r degree of accuracy than before. The edge of Australia, which was preVIously a rifted paSSIve margm, IS shown to be markedly thickened in the collision zone itself and has been subducted to a depth of about 200 km. In addition, it appears that the subduction zone has major lateral and vertical discontinuities at depth. It is proposed that these discontinuities reflect slab separation during a previous micro­ continental collisional event some 10 to 7 million years ago and that this collision was related to the transfer of continental lithosphere from the Indo-Australian plate to the South East Asian plate. INTRODUCTION into six groups (Audley-Charles, 1968; Barber, 1981; Charlton et al., 1991). The "para-autochthon" forms One of the more fundamental geological the core of the island and consists of problems being addressed by current research is to unmetamorphosed to high grade metamorphosed discover how collisional mountain ranges are sedimentary material which has been interpreted constructed. Understanding is hampered by both as being underplated Late Palaeozoic to Early the geographical remoteness and structural Mesozoic Australian-derived rocks. The complexity of some of the areas under study. "allochthon" is made up of ultrabasic rocks with Another approach to the problem is to study a some metamorphosed sedimentary cover. This unit young, active collision zone which has not yet has a "tectonic" relationship with the "para­ developed to the same degree of complication as a autochthon" and was traditionally regarded as being mature mountain range. The Banda Axc in Eastern derived from the Asian plate forearc region, (Barber Indonesia is, arguably, the youngest mountain belt and Audley-Charles, 1976). However, in the world. A collision is presently underway Sopaheluwakan (1990) has good evidence that the between the northern edge of the Australian unit was metamorphosed during a period of intra­ continent and the volcanic islands of the Banda Axc oceanic thrusting near a spreading centre and was (Fig. 1). The collision has progressed to a stage subsequently emplaced against the "para­ such that a "collision complex" has been built up autochthon" in a collisional event around 38 million parallel to and just south of the volcanic arc. The years ago. The rocks of the "autochthon" consist of higher parts of the collision complex have risen unmetamorphosed Mesozoic to Recent sediments above sea level to form the islands of Sumba, Timor, that have been accreted to the southern side of the Tanimbar and a string of smaller islands. developing orogen. The remaining three groups Timor is the largest island of the chain and it are mud diapir deposits, raised reef terraces and is here that geological research has been historically Recent alluvium. focused since the late 1800s. The age of onset of this collision has been derived by many workers SEISMOLOGICAL DATA using a variety of types of evidence at 2 to 3 million years ago in the region of Timor, (Johnston and Earthquakes are one of the main sources of Bowin, 1981; De Smet et al., 1990; Harris, 1992). information on the large-scale structure and The rocks of Timor have traditionally been split dynamics of tectonically active areas and it is with Geol. Soc. MalaYJia, Bulletin 33, July 1993; pp. 105-118 --'- o J' (J) S' " 10' II' » o 12" :0 5> z ~200metres ::n 13" () »I :0 14' 55 o res z '" Australia 17' IS' '" 75 mm/ylj 20' 2J L-~-::-~~7.-~~~~~:-:-::-:-:-:--:~~~~~~~ ---J 1J' ........- ___ 1)5' - · I~ 106' lOT lOS' 109' 11 0' Ill ' 11 2" Il l" 114" liS' 1] 6' 11 7' liS' 11 9' 120' 121' 122' 123" 12·V 125 ' 126' 121' 128' 129' 130' 13 1' Dr 133" 134 ' Figure 1. Location map of Eastern Indonesia. liTHOSPHERE STRUCTURE AND DYNAMICS OF THE BANDA ARC COLLISION ZONE, EAST INDONESIA 107 these data that this paper is largely concerned. INTERPRETATION This part of Indonesia is one of the most seismically active areas in the world. The International The "Top of Envelope" plot (Fig. 2) is interpreted Seismological Centre (ISC) currently records about as showing the geometry of the active part of the 5 earthquakes per week within the area of the map Wadati-Benioff zone since the top of a subduction in Figure 1. A number of papers have already been slab is relatively cold and brittle. This is in keeping published about the seismicity of the Banda Arc with many other works, for instance those referred area (Fitch and Molnar, 1970; Cardwell and Isacks, to above. The shallow portions « 200 km depth) to 1978; Hamilton, 1979; McCaffrey et al., 1985; Eva the west and to the east of Timor are seismically et al., 1988; McCaffrey, 1988, 1989; Ritsema et al., very active and define a fairly smooth surface. The 1989). All of these works regard the geometry of high activity in the eastern region is largely due to the Wadati-Benioff zone to be a simple concave stresses within the plate near the region of greatest bowl shape to the east and a steeply dipping single curvature. Most events for which fault plane surface to the west penetrating to below 600 km solutions have been calculated in this region are depth; see, for instance, Figure 5 of Cardwell and north-east directed thrusts whose orientation is Isacks (1978). The Wadati-Benioff zone contours in parallel to the direction of motion of the Australian all the above works are only rough approximations plate. In contrast, there is little shallow seismic through the data at depth intervals of 100 km. activity under eastern Timor. The probable The data set for this study is an extract from explanation for this observation is that subduction the ISC catalogue from 1 January 1964 to 31 August under eastern Timor has now ceased. Eastern 1988 and the USGS Preliminary Determination of Timor is the location of the highest mountains and Epicenters Bulletin from 1 September 1988 to 31 has the smallest arc-orogen cross-strike width and December 1991. Also included are earthquakes is therefore the most developed part of the collision recorded by temporary regional networks zone. Inactivity of this segment of the subduction (McCaffrey et al., 1985), giving a total of about zone is in accord with the work of Johnston and 8500 events. Undefined depth events that had Bowin (1981) on sediments in the trench to the been assigned a default depth of 33 km were south of eastern Timor which indicates that discarded but all other events down to the smallest subduction in this region has been inactive for the recorded magnitudes were retained. last 600 000 years. Continuing deformation in this The aim of this work was to create a three­ segment ofthe collision zone is being accommodated dimensional model of the structure of the by a high rate of uplift (Johnston and Bowin, 1981; lithospheric components of the Banda Arc region. Milsom and Audley-Charles 1986; De Smet et al. Locations of earthquake hypocentres were initially 1989; De Smet et al.; 1990) and back-arc thrusting plotted on maps. The plotted earthquakes were north of Alor and Wetar Islands (Nabelek and then grouped into horizontal slices 10 km in McCaffrey, 1983; Silver et al., 1983; McCaffrey and thickness. The surface curvature of the earth was Nabelek, 1984, 1986; McCaffrey, 1989). ignored. The depth error for any given The deeper portion of the subduction zone is teleseismicallY located earthquake is of the order of more difficult to contour confidently because of the 10-20 km (Dziewonski et al. 1981). The sample relative scarcity of events. Nevertheless, it is clear group is large enough, however, for the statistical that to the west of Timor, the seismic zone descends errors to have no impact on the overall distribution vertically to below 650 km depth. To the west of of events. With the aid of maps and sections, the Timor and to the west of Sumba, there appear to be outer surfaces of the envelope of active seismicity large gaps in the vertical portion of the seismic were contoured by hand with a contour interval of zone. 10 km. The resulting contour maps are rather There is still great debate about the actual complicated since, in many regions, the contoured mechanism of deep earthquakes (Davies, 1980; Goto surfaces are almost vertical and the contours lie on et al., 1985, 1987; Frohlich, 1987). Nevertheless, it top of each other. The results are therefore appears from hypocentre plots that earthquakes at displayed in isometric view (Figs.