S31C-0555
The Sunda-Banda Arc Transition: New Insights From Marine Wide-Angle Seismic Data
End of 2006, RV SONNE cruise SO190 SINDBAD (Seismic and Geoacoustic Investigations along the Sunda- Banda Arc Transition) went south of the Indonesian archipelago to acquire various geophysical datasets between 112 °E and 122 °E. The main goal of the project is to investigate the modifications of the lower plate (variability in the plate roughness, transition from oceanic to continental lower plate) and their effects on the tectonics of the upper plate (development of an outer high and forearc basin, accretionary and erosive processes). The tectonic style changes in neighboring margin segments from an oceanic plate-island arc subduction along the eastern Sunda margin to a continental plate-island arc collision along the Banda margin. Moreover, the character of the incoming oceanic plate varies from the rough topography in the area where the Roo Rise is subducting off eastern Java, to the smooth oceanic seafloor of the Argo- Abyssal Plain subducting off Bali, Lombok, and Sumbawa. In order to cover the entire variations of the lower plate, seven seismic refraction profiles were conducted along four major north-south oriented corridors of the margin, at 113 °E, 116 °E, 119 °E, and 121 °E, as well as three profiles running perpendicular to the major corridors. A total of 239 ocean bottom hydrophone and seismometer deployments were successfully recovered. Shooting was conducted along 1020 nm of seismic profiles using a G-gun cluster of 64 l. Here, we present velocity models obtained by applying a tomographic approach which jointly inverts for refracted and reflected phases. Additional geometry and velocity information for the uppermost layers, obtained by prestack depth migration of multichannel seismic reflection data (see poster of Mueller et al. in this session), is incorporated into our models and held fixed during the iterations. http://www.ifm- geomar.de/index.php?id=sindbad
S31C-0556
The Sunda-Banda Arc Transition: New Insights from Marine Multichannel Seismic Data
After the Indian Ocean Mw 9.3 earthquake and tsunami on December 26, 2004, intensive research activities focussed on the Sunda Arc subduction system offshore Sumatra. For this area a broad database is now available interpreted in terms of plate segmentation and outer arc high evolution. In contrast, the highly active easternmost part of this subduction system, as indicated by the south of Java Mw 7.7 earthquake and tsunami on July 17, 2006, has remained almost unexplored until recently. During RV SONNE cruise SO190 from October until December 2006 almost 5000 km of marine geophysical profiles have been acquired at the eastern Sunda Arc and the transition to the Banda Arc. The SINDBAD project (Seismic and Geoacoustic Investigations along the Sunda-Banda Arc Transition) comprises 30-fold multichannel reflection seismics with a 3-km streamer, wide-angle OBH/OBS refraction seismics for deep velocity control (see poster of Planert et al. in this session), swath bathymetry, sediment echosounder, gravimetric and geomagnetic measurements. We present data and interpretations of several 250-380 km long, prestack depth-migrated seismic sections, perpendicular to the deformation front, based on velocity models from focussing analysis and inversion of OBH/OBS refraction data. We focus on the variability of the lower plate and the tectonic response of the overriding plate in terms of outer arc high formation and evolution, forearc basin development, accretion and erosion processes at the base of the overriding plate. The subducting Indo-Australian Plate is characterized by three segments: i) the Roo Rise with rough topography offshore eastern Java ii) the Argo Abyssal Plain with smooth oceanic crust offshore Bali, Lombok, and Sumbawa, and iii) the Scott Plateau with continental crust colliding with the Banda island arc. The forearc responds to differences in the incoming oceanic plate with the absence of a pronounced forearc basin offshore eastern Java and with development of the 4000 m deep forearc Lombok Basin offshore Bali, Lombok, and Sumbawa. The eastern termination of the Lombok Basin is formed by Sumba Island, which shows evidence for recent uplift, probably associated with the collision of the island arc with the continental Scott Plateau. The Sumba area represents the transition from subduction to collision. Our seismic profiles image the bending of the oceanic crust seaward of the trench and associated normal faulting. Landward of the trench, they image the subducting slab beneath the outer arc high, where the former bending-related normal faults appear to be reactivated as reverse faults introducing vertical displacements in the subducting slab. The accretionary prism and the outer arc high are characterized by an ocean-verging system of imbricate thrust sheets with major thrust faults connecting seafloor and detachment. Compression results in shortening and steepening of the imbricated thrust sheets building up the outer arc high. Tilted piggy-back basins and downlaps of tilted sediments in the southern Lombok forearc basin indicate ongoing uplift of the entire outer arc high, abrupt displacements, and recent tectonic activity. http://www.bgr.bund.de
S31C-0557
Firstarrival Tomography of Seismic OBS Data and Prestack Depth Migration of MCS Data from the Sumatra Continental Margin
Prestack depth migration is applied to multi-channel seismic streamer data acquired by RV Sonne cruise 186 near Simeulue island offshore Sumatra. The sea floor depth varies between 50 m near Simeulue and 5 km in the oceanic trench. The sediment is up to 4 km thick in the trench. The oceanic crust is imaged over a distance of 90 km landwards from the trench. It is subducting at an angle of 5-6 degrees on average. The P wave velocities of the sedimentary layers are determined by focusing analysis. These velocities are used to construct an initial model for seismic first arrival tomography, which is applied to wide-angle Ocean Bottom Hydrophone and Seismometer data recorded along the same profile as the MCS data. The inversion of more than 9000 traveltimes from 24 stations with source-receiver offsets smaller than 70 km was performed on a 1000 x 100 grid, which covers 250 km along the profile and 30 km in depth. The inversion converged after 14 iterations with decreasing regularization parameter to a root-mean-square traveltime residuum of 0.060 s, which is of the size of the experimental error. The velocity model shows high P-wave velocities of 6 km/s in the northeast, where the crust is of continental origin. The obtained 2D velocity model is used as input for the prestack depth migration. Seismic tomography was also applied to two addtional OBH/OBS experiments in the same area. The P-wave velocity profiles show a velocity increase of 0.7/s for the first 8 km depth below sea floor with a sea floor depth of 2 km. Detailed bathymetric maps of the survey areas are presented.
S31C-0558
Wide angle seismic refraction imaging of the Northern Sumatra Subduction Zone
Two deep penetrating wide angle seismic refraction profiles were acquired during the SUMATRA-OBS cruise offshore Sumatra, to image the subsurface in the zone of maximum co-seismic slip for the great Sumatra- Andaman earthquake of December 2004. Here we present the 2-D velocity image of the subsurface obtained by traveltime tomographic inversion of the northern profile. This line was shot twice using airguns onboard R/V Marion Dufresne (~8260 cu inch) and the one onboard Western-Geco/Schlumberger vessel the Searcher (~10000 cu inch) with much denser shot spacing. The velocity image spanning a ~520 km long profile almost at orthogonal orientation to the trench, is obtained by the inversion of about ~30000 traveltime picks made on the 56 ocean bottom seismometers (OBSs) that were deployed along the profile. The velocity model depicts the rather thin (~4.5-6.0 km) Indo-Australian oceanic plate in the region, subducting underneath the Sunda plate at an angle of about 8-10°. The subducting slab is overlain by thick volume of the evolved accretionary complex for about 150 km, followed with the system of Aceh basin in forearc region and the volcanic arc domain further east. The high velocity structure just east underneath of the Aceh basin appears to be of continental origin and might be a result of the complex tectonic history of the region. On the eastern end of the line the continental crust with thickness of about 12-15 km appears to be thinned by the pull-apart extension activity of the various branches of Sumatra fault. The experiment was conducted in conjunction with the acquisition of coincident seismic reflection lines by the Western-Geco/Schlumberger's "Searcher" seismic vessel. Future work aims at obtaining a depth migrated subsurface image of the coincident seismic reflection profile WG2 using this velocity model, to further constrain the geometry and structure of this highly active subduction zone.
S31C-0559
Seismic image of the segment boundary of the Sumatra Dec. 2004 and March 2005 megathrust earthquakes
Two mega-thrust events occurred off northern Sumatra on December 26th 2004 (Mw=9.1-9.3) that produced a devastating tsunami, and on March 28th 2005 (Mw=8.6). Seismological investigations, GPS measurements, as well as in-situ and remote observation of vertical motion on the fore-arc islands show both, an abrupt southern termination of the large December 2004 rupture and a sharp northern termination of the rupture zone of the March 2005 mega-thrust. With newly acquired marine geophysical data (reflection and refraction seismics, magnetics, gravity and bathymetry) for the first time the boundary has been imaged. From wide-angle/refraction seismic data it becomes clear that there is an abrupt arc parallel depth change of 3 kilometres within 40 kilometres in the oceanic crust SW of Simeulue Island. The change in depth obtained in wide angle reflections corresponds with a change in depth and also the reflectivity of the oceanic crust reflection in MCS data. We interpret that the abrupt depth change originates from a ramp or tear in the subducted oceanic crust. The discontinuity in the oceanic crust likely trends NNE and is east of a continuation of an extinct fracture zone on the subducting Indo-Australian plate, indicating a pervasive lower plate control on margin structure, particularly its segmentation. The tear may be the reason for rupture propagation termination of the great December 26th 2004 and March 28th 2005 Sumatra Andaman earthquakes. At the ramp or tear, strain is decoupled resulting in a tectonic boundary for earthquake propagation.
S31C-0560
The Subducting Investigator Fracture Zone Offshore Siberut as Imaged by Seismic Tomography.
At convergent plate boundaries the structure of the subducting oceanic plate is one of the key factors which govern the development of the whole subduction zone. It effects the size and position of the seismogenic zone as well as the structure of the accretionary prism, the structural style of the outer high and the sedimentary thickness in the fore-arc basins. Further on, variations of the oceanic crust perpendicular to the direction of subduction can result into a segmentation of the seismogenic zone. We are presenting results of a wide-angle seismic experiment which was conducted in August 2006 on board of the German research vessel Sonne. The profile has a total length of 200 km and is situated parallel to the deep sea trench offshore the island of Siberut. In this area the Investigator Fracture Zone is subducted at an oblique angle of 70 degrees below the Indonesian island arc. Seaward of the deformation front a MCS-profile shows that the topography of the oceanic basement has a variability of up to 2 km. To investigate the effects of this variable topography onto the subduction zone, we applied a refraction tomography on the available wide-angle data. The results give insights into the distribution of the seismic velocity within the accretionary prism and the subducting oceanic crust. An additionally inversion of wide-angle reflections obtained from the downgoing oceanic crust shows that the topography of the Investigator Fracture Zone is still a remarkable feature in the depth of 14 km below seafloor.
S31C-0561
The 2004 Sumatra Earthquake Mw 9.3: Seismological and Geophysical Investigations in the Andaman-Nicobar Islands
The December 26, 2004 Sumatra-Andaman earthquake (MW 9.3) is the fourth largest event (M=9.0) in the world during the last 100 years. It occurred by thrust faulting on the interplate thrust of the subducting Indian plate and overriding Burmese platelet. The main shock rupture, ~1300 km long and 200 km wide, propagated from north of Sumatra to Andaman Nicobar Islands; the slow rupture generated a tsunami which killed about 300,000 people. The epicenter of the earthquake is located at 3.90°N and 94.26°E with a focal depth at 28 km (USGS). The past significant earthquakes in this zone are the 31 December, 1881 (M 7.9), 26 June, 1941 (M 7.7), 20 January, 1982 (M 6.3) and 14 September, 2002 (M 6.0).
S31C-0562
Estimations of Fault Parameters of the 2004 Sumatra Earthquake
The rupture directivity for the 2004 Sumatra earthquake is analyzed by the differences of phase-delay time of Rayleigh-wave between the main shock and its nearby earthquakes at periods of 140-160 sec. A long source- process time (~463.0 sec) and large rupture length (~1164.0 km) are derived through the rupture directivity analysis. The source-process time for this earthquake is larger than those for the 1960 Chile and 1964 Alaska earthquakes. This might be due to the long rupture occurring during the faulting of the earthquake. The estimated rise time, 92.0 sec, for the 2004 Sumatra earthquake is approximately 20% of the whole source duration and is larger than those for the 1960 Chile and the 1964 Alaska earthquakes. This probably reflects a fundamental difference of frictional properties between these earthquakes. When taking the rise time into account, the estimated rupture velocity is approximately 3.1 km/sec, which is higher than previous studies from hydroacoustic data and regional seismic network. In this study, additional evidences from surface-wave phase-delay time analysis would confirm the basic features of rupture for the 2004 Sumatra earthquake. This result can also provide constraints on the study of source rupture for this earthquake.
S31C-0563
Seismic Evidence for a Mantle Megathrust Producing the Giant Sumatra-Andaman Earthquake
The great (M = 9.3) Sumatra-Andaman earthquake of 26 December 2004 was the third largest subduction event in the last 50 years. The rupture initiated at 30-40 km depth northwest of Simeulue Island and propagated as far as the northern Andaman Islands, breaking a total area of about 1300 x 150 km2. The earthquake was caused by a sudden slip -up to 30 m - releasing stress accumulated for many hundreds of years on an interface (megathrust) along which the Indo-Australian plate subducts beneath Sunda and Burma plates. Using high- resolution deep seismic reflection imaging, we show that the subducting oceanic crust and Moho are sliced by at least six NE dipping thrust faults that may have their roots in the oceanic mantle, requiring the presence of a megathrust in the mantle. Near the subduction front, these deep-rooted faults continue in the overlying sediments and are associated with seafloor scarps and steep slopes, suggesting that these faults are active and extend to the seafloor. The presence of steeply dipping thrust earthquakes near the subduction front, some of which are in the down-going plate, further corroborates these observations. Beneath the forearc basin, aftershocks lie below the imaged basalt-sediment interface, confirming the presence of active thrust in the down-going plate. We also observed a reflective zone just above the plate interface, which might be a slice of unerplatted oceanic crust. Taken together, these results imply that very strong coupling, appropriate for brittle failure of mantle rocks, accounts for the initiation of such an unusually powerful tsunamigenic event. A megathrust in the mantle would increase the width of seismogenic zone both towards up dip and down dip. Perhaps the 2004 Sumatran event should be considered an example of a novel class of exceptionally large and infrequent megathrust earthquakes ("Mantle Megaquakes": Mw: 9-10), typical of subduction zones and great collision ranges such as the Himalayas, rupturing deep mantle interfaces with a mechanical strength much greater than that of thrusts in the crust.
S31C-0564
Designing a Geologically Satisfying Model of the 2004 Sumatra-Andaman Earthquake
Deformation models are simplified versions of actual geologic systems. A suitable model design reflects a balance of problem domain complexity and computational simplicity. Designing a valid conceptual model is a critical, and often overlooked, step in the modeling process and is the focus of this study of the M9+ Sumatra- Andaman earthquake of 26 December 2004. This earthquake is the largest recorded by GPS measurements, which can be used to verify the results of our model. The GPS measurements are accurate to within millimeters, so it is imperative that the model be as geologically accurate as possible. In building this model we employed the use of seismic and tomography data, geologic maps, and structure and cross-section images. The first iteration of our 3-D finite element model (FEM) was constructed using a single cross-section generated through the southern part of Sumatra. The results of this study indicated that observed GPS measurements corresponded well to the predicted GPS measurements of the FEM to the south. However, the northern observed and predicted measurements were not as consistently correlated. This could partly be due to the simplification of propagating a single cross-section along the strike of the Sunda trench. This second iteration of the FEM will merge both the southern cross-section and a northern cross-section taken through the Andaman Islands including the back-arc spreading center to the east. Merging the cross-sections will account for the changing geology and structure of the island arc from south to north, and ultimately generate more reliable deformation predictions.
S31C-0565
Evidence for Late Holocene Paleo Earthquakes on the Aceh Portion of the Sumatran Subduction Zone, Northwest Sumatra
We investigated coastal lowland environments of the northwestern coast of Sumatra to determine whether there is stratigraphic evidence for rapid subsidence accompanying prehistoric subduction zone earthquakes on the Aceh portion of the Sumatran subduction zone. Characterizing late Holocene vertical land-level changes will help identify paleo subduction zone earthquakes and will help constrain the width of the slip patch during these earthquakes. Our preliminary results suggest that, at least at two times in the past ca. 4000 years, regional coseismic subsidence events have occurred along the northwestern Aceh coast and are recorded in stratigraphy from coastal wetland environments. The stratigraphic signature of two paleo earthquakes each includes a regionally-extensive buried tidal wetland soil. Although there are tsunami deposits above each of these regionally extensive buried soils, these tsunami deposits are not as thick or as extensive as the 2004 tsunami deposit that presently blankets the coastal lowland of northwest Sumatra. We also document that the 2004 Andaman-Aceh earthquake caused more than a meter of subsidence along portions of the northwestern Sumatra coast. Although our work is preliminary, we note two significant conclusions to date. First, paleo earthquakes of the Aceh portion of the Sumatran subduction zone are well recorded in coastal wetland stratigraphy as regionally extensive buried soils preserved because of coseismic subsidence. Second, unlike coastal areas of the southwest Pacific and eastern Indian oceans that are removed from active subduction margins, the Aceh coastal lowland stratigraphy records a multiple-thousand year record of relative sea level change, probably because in the particular setting of northwestern coastal Sumatra, local vertical tectonic displacement is the dominant process affecting relative sea level history in the last ca. 5000 years. This history of vertical tectonic displacements records a time-series of long term seismic behavior of the Sunda megathrust along the northwest coast of Sumatra.
S31C-0566
A Biostratigraphical Record of Repeated Great Earthquakes on the Sunda Subduction Megathrust, Northern Sumatra
The 2004 Andaman-Aceh earthquake and tsunami focused global attention on the northernmost part of the Sumatran subduction zone. Important questions were raised about the past geologic history of the Sunda megathrust offshore of northernmost Sumatra. In response, we have collected stratigraphic data from the northwest coast of Sumatra toward the following goals (1) determining the recurrence interval between great earthquakes; (2) quantifying the relative timing of coseismic and interseismic vertical land level changes; (3) documentation of subsidence precursory to the main coseismic subsidence event; (4) documentation of uplift after the main coseismic subsidence event caused by slow, down dip afterslip on the megathrust; and (5) development of a chronology of tsunami invading the Sumatran coast. In order to achieve these goals, we have begun analysis of the coastal wetlands of northwest Sumatra. Through careful site selection and litho- and biostratigraphical analyses, we aim to resolve both rapid and gradual changes in relative sea level at a resolution of ~10 cm. These data will help characterize earthquake-induced rapid subsidence and gradual sea-level changes precursory to, and subsequent to, coseismic land level changes. Preliminary results suggest two great earthquake events have occurred within the late Holocene, prior to 2004 Andaman-Aceh earthquake. The signatures of two older events, located approximately 2.5 and 3.5 m below mean sea level, are each recognized by the presence of a regionally extensive buried organic soil overlain by a medium to coarse grained quartz sand unit, interpreted as a tsunami deposit. Calcareous foraminifera are present in every lithostratigraphic unit, with noticeable assemblage changes between the medium to coarse grained quartz sand units and the underlying buried soil deposits. The 2004 event is also documented by a buried soil and a surficial coarse grained quartz sand tsunami deposit.
S31C-0567
Discovery Of Pre-2004 Tsunami Deposits On Simeulue Island, Southern Aceh Province, Indonesia
Presumably, megathrust earthquakes and tsunamis occurred repeatedly along the Sunda megathrust before 2004 and 2005. To determine its past tsunami history, we conducted field surveys on Simeulue Island, which straddles the boundary of the 2004 and 2005 ruptures. Shallow excavations exposed paleo-tsunami deposits at two coastal lowland sites in southern Simeulue Island. The paleo-tsunami deposits at the Inor|Naibos and Busong Bay sites appear as laterally continuous sand layers in terrestrial sedimentary successions. Two paleo-tsunami layers exist at Busong Bay; the younger one is bounded by peat layers and the older one separates peat and underlying muddy sediment that was probably deposited in an inter-tidal or sub-tidal environment. Both tsunami deposits are rich in coral clasts. A fresh, uneroded coral boulder from the upper tsunami layer yielded a U/Th age that is consistent with emplacement in 1861, the year of an historical great (M`8.5) earthquake there. The lower tsunami layer may have been emplaced during an earlier uplift event around A.D. 1799, documented by an uplifted coral microatoll at the site. This interpretation is consistent with the sedimentary facies change from mud to peat across the lower tsunami layer, which probably represents an environmental change due to uplift. The single tsunami layer at the Inor|Naibos site demarcates a sedimentary facies change. As at Busong Bay, it is accompanied by coral boulders, which we think are likely to yield U/Th ages consistent with deposition contemporaneous with one of the layers at Busong Bay.
S31C-0568
HYDRODYNAMIC SIMULATIONS OF FAR-FIELD TSUNAMI RISK IN THE INDIAN OCEAN, WITH SPECIAL EMPHASIS ON AFRICA
We evaluate far-field tsunami risk in the Indian Ocean Basin based on hydrodymanic simulations of eight case studies of possible mega earthquakes at the major seismic zones surrounding the basin. They represent worst-case scenarios of seismic rupture along the full extent of seismogenic faults having supported large earthquakes in the historical record. We consider seismic sources located at the extremities of the 2004 Sumatra-Andaman rupture, namely along the Southern coast of Sumatra and in the Andaman-Myanmar province; along the Makran coast of Pakistan and Iran; and also along the Southern coast of Java, where the possibility of a large interplate thrust earthquake cannot be entirely dismissed. Following Ando's [1975] remark concerning the capricious character of segmentation during repeated mega-earthquakes at subduction zones, we consider events even bigger than identified in the historical record, especially in Southern Sumatra. We also envision risk to South Africa from a potential mega-event in the South Sandwich Islands. The results of our hydrodynamic simulations indicate that the distribution of maximum amplitudes in the Indian Ocean Basin is primarily controlled by the classical effect of source directivity, and additionally by refraction and focusing along bathymetric features. As a result, many provinces in the basin could be threatened by higher tsunami amplitudes than in 2004. This pattern is particularly important along the coast of East Africa, from Somalia to and including South Africa, in Madagascar and the Mascarene Islands, especially under a South Sumatra scenario involving an earthquake comparable to, or even possibly larger than, the 1833 event, whose epicentral area is widely believed to be under enhanced seismic risk as a result of stress transfer from the 2004 and 2005 ruptures to the Northwest.