Search

Coastal tide gauges do not only play a central role in the study of climate-related sea level changes but also in tsunami warning systems. Over the past five years, ten GPS-controlled tide gauge systems have been installed by the German Research Centre for Geosciences (GFZ) in Indonesia to assist the development of the Indonesian Tsunami Early Warning System (InaTEWS). These stations are mainly installed at the Indonesian coastline facing the Indian Ocean. The tide gauge systems deliver information about the instantaneous sea level, vertical control information through GPS, and meteorological observations. A tidal analysis at the station's computer allows the detection of rapid changes in the local sea level ("sea level events"/SLE), thus indicating, for example, the arrival time of tsunamis. The technical implementation, communication issues, the operation and the sea level event detection algorithm, and some results from recent earthquakes and tsunamis are described in this paper.

GPS (Global Positioning System) technology is widely used for positioning applications. Many of them have high requirements with respect to precision, reliability or fast product delivery, but usually not all at the same time as it is the case for early warning applications. The tasks for the GPS-based components within the GITEWS project (German Indonesian Tsunami Early Warning System, Rudloff et al., 2009) are to support the determination of sea levels (measured onshore and offshore) and to detect co-seismic land mass displacements with the lowest possible latency (design goal: first reliable results after 5 min). The completed system was designed to fulfil these tasks in near real-time, rather than for scientific research requirements. The obtained data products (movements of GPS antennas) are supporting the warning process in different ways. The measurements from GPS instruments on buoys allow the earliest possible detection or confirmation of tsunami waves on the ocean. Onshore GPS measurements are made collocated with tide gauges or seismological stations and give information about co-seismic land mass movements as recorded, e.g., during the great Sumatra-Andaman earthquake of 2004 (Subarya et al., 2006). This information is important to separate tsunami-caused sea height movements from apparent sea height changes at tide gauge locations (sensor station movement) and also as additional information about earthquakes' mechanisms, as this is an essential information to predict a tsunami (Sobolev et al., 2007). This article gives an end-to-end overview of the GITEWS GPS-component system, from the GPS sensors (GPS receiver with GPS antenna and auxiliary systems, either onshore or offshore) to the early warning centre displays. We describe how the GPS sensors have been installed, how they are operated and the methods used to collect, transfer and process the GPS data in near real-time. This includes the sensor system design, the communication system layout with real-time data streaming, the data processing strategy and the final products of the GPS-based early warning system components.

Since the nineteenth century, several local topocentric geodetic datums have been used for surveying and mapping in Indonesia. In 1975 the Indonesian Datum 1974, which is a national topocentric datum, was introduced and then replaced by the National Geodetic Datum 1995 which is a static geocentric datum realized using GPS observations. In recent years it has been realized that, due to on-going active tectonics in the Indonesian region, the National Geodetic Datum 1995 is inadequate for surveying and mapping in some regions of Indonesia, and also for some current and emerging applications. Initial studies suggested that a semi-dynamic geocentric datum is suitable for Indonesia. The adopted new datum uses the ITRF2008 reference frame with a reference epoch of 1 January 2012. It incorporates several major deformation blocks of Indonesia, and several micro blocks to model specific deformation events such as large earthquakes. Realization of this new semi-dynamic datum will be primarily based on existing data coming from many previous national GPS campaigns that have been conducted by the Geospatial Agency of Indonesia, and from existing GPS CORS stations across Indonesia.

Data collected at approximately 60 Global Positioning System (GPS) sites in southeast Asia show the crustal deformation caused by the 26 December 2004 Sumatra-Andaman earthquake at an unprecedented large scale. Small but significant co-seismic jumps are clearly detected more than 3,000 km from the earthquake epicentre. The nearest sites, still more than 400 km away, show displacements of 10 cm or more. Here we show that the rupture plane for this earthquake must have been at least 1,000 km long and that non-homogeneous slip is required to fit the large displacement gradients revealed by the GPS measurements. Our kinematic analysis of the GPS recordings indicates that the centroid of released deformation is located at least 200 km north of the seismological epicentre. It also provides evidence that the rupture propagated northward sufficiently fast for stations in northern Thailand to have reached their final positions less than 10 min after the earthquake, hence ruling out the hypothesis of a silent slow aseismic rupture.

Jakarta is the capital city of Indonesia with a population of about 10 million people, inhabiting an area of about 25 × 25 km. It has been reported for sometime that locations in Jakarta are subsiding at different rates. Up to the present, there has been no comprehensive information about the characteristics and pattern of land subsidence in the Jakarta area. Usually land subsidence in Jakarta is measured using extensometers and ground water level observations, or estimated using geological and hydrological parameters. To give a better picture about land subsidence, geodetic-based monitoring systems utilizing leveling and GPS surveys have also been implemented. The land subsidence characteristics of Jakarta and its surrounding area areinvestigated using data from three repeated leveling surveys performed in1982, 1991, and 1997, and two repeated GPS surveys conducted in 1997and 1999. Leveling surveys detected subsidence up to about 80 cm duringthe period of 1982–1991, and up to about 160 cm during the 1991–1997period; while GPS surveys observed subsidence up to about 20 cm duringthe period of 1997–1999. Comparison with the hydrological data shows thatland subsidence in Jakarta is strongly related to excessive groundwater extraction.

An analysis combining historical triangulation and recent Global Positioning System (GPS) survey measurements in west and north Sumatra, Indonesia, reveals a detailed slip history along the central part of the Sumatran fault. The arc-parallel components of the combined velocity field are consistent with slip rates inferred from GPS data, ranging from 23 to 24 mm/yr. Between 1.0°S and 1.3°N the Sumatran fault appears to be characterized by deep locking depths, of the order of 20 km, and the occurrence of large (Mw ≈ 7) earthquakes. The long-term (1883–1993) strains show simple right-lateral shear, with rates similar to GPS-measured, 1989–1993 strain rates. Coseismic deformation due to the 1892 Tapanuli and 1926 Padang Panjang earthquakes, estimated from triangulation measurements taken before and after the events, indicates that the main shocks were significantly larger than previously reported. The 1892 earthquake had a likely magnitude of Mw ≈ 7.6, while the 1926 events appear to be comparable in size to the subsequent (M ∼ 7) 1943 events and an order of magnitude higher than previously reported.

Global Positioning System (GPS) measurements along the subduction zone of northern Sumatra (2°S to 3°N) reveal that the strain associated with the oblique convergence of the Australian plate with Eurasia is almost fully partitioned between trench-normal contraction within the forearc and trench-parallel shear strain within a few tens of kilometers of the Sumatran fault Kinematic analyses of interplate earthquake slip vectors provide slip rates on the Sumatran fault within a few millimeters per year of GPS and geologic rates, giving us more confidence in the use of slip vectors for inferring slip partitioning elsewhere. The inferred slip rate on the Sumatran fault is ∼1/3 less than the full margin parallel component of plate motion. An across-forearc rotation in the slip vectors suggests that the missing arc-parallel shear occurs seaward of the geodetic network, between the forearc islands and the trench. Simple finite element models are used to explore the conditions under which the change in the principal strain rate directions between the forearc and the arc region can occur. Modeling suggests that neither a preexisting strike-slip fault nor a zone of thermally induced lithospheric weakness in the overriding plate is needed for strain partitioning to occur. In general, forearc slivers form over the region of interplate coupling and are driven along strike by the basal shear. A volcanic arc can help the partitioning process by localizing the margin-parallel shear strain in the upper plate if its crust and mantle are weaker than its surroundings. Interplate slip vectors and geodetic results from Sumatra together suggest that the highest coupling on the plate boundary occurs beneath and seaward of the forearc islands, consistent with inferences about the rupture zones of great nineteenth century earthquakes there. The Sumatra example suggests that geodetic measurements of interseismic, margin-parallel shear strain at oblique convergent margins can be used to map the landward extent of the relatively high basal stress beneath the overriding plate if one can correct for strain localization caused by weak upper plate strike-slip faults.

We model spatial variations in horizontal displacements of 117 geodetic sites measured during annual surveys in 1989-1996 with the Global Positioning System (GPS) as elastic strain across a locked strike-slip fault to infer the contemporary slip rate, locking depth, and location of the Sumatran fault (SF) in northern Sumatra (1 S-3 N). GPS-derived slip rate estimates increase slightly northward from 23 plus or minus 3 mm/yr at 0.8 deg S to 26 plus or minus 2mm/yr at 2.7 N. They agree with geologic estimates north of the Equator, but at 0.5 S they are about 10 mm/yr higher. Strain appears to be distributed asymmetrically about the fault. South of 2 N, about 5 mm/yr of shear is required within the offshore forearc, west of the fault, to achieve a closer agreement of fault locations inferred from GPS velocities with geologically identified traces of the SF. Locking depth estimates are on the order of 10-20 km. The western branch of the major fault bifurcation near 1 N slips at a rate five times higher than the eastern branch. The two main strands of the fault at the northwestern tip of Sumatra (5.5 N) appear to be nearly free of horizontal strain; significant slip must occur away from the two strands, probably further east at two other geologically active branches. The Banda Aceh embayment is extruded to the northwest at a rate of 5 plus or minus 2 mm/yr. Within the estimated velocity uncertainties of several mm/yr, fault-normal deformation along the SF is insignificant. Almost strain free, the northern part of the back-arc basin is part of a rigid Sunda shelf, while the northern forearc is subjected to 8 plus or minus 5 x 10 (exp -8)/yr of extension nearly parallel to the arc.

Global Positioning System (GPS) measurements from 1992 to 1995 indicate that the left-lateral Palu fault in central Sulawesi slips at a rate of 38±8 mm/a with a locking depth between 2 and 8 km. From the measured slip rate and the historic seismicity of the fault, we estimate that the Palu fault currently has stored enough strain to produce a Mw>7 earthquake. The Palu and other nearby faults accommodate rapid clockwise rotation of nearly 4°/Ma of E Sulawesi relative to eastern Sunda. The rotation of east Sulawesi transfers E-W shortening between the Pacific and Eurasian plates to N-S subduction of the Celebes Basin beneath Sulawesi.

SUMMARY The point of convergence of the Eurasian, Philippine and Australian plates is situated adjacent to the island of Sulawesi, Indonesia. The relative plate velocities are estimated by NUVEL1 to be 7 to 9 cm yr {1 . The complex tectonic mechanism of the triple junction has been studied over a two-year period in the course of the GEODYSSEA Southeast Asian Project. The GPS investigations concentrate on measurements of both the Sulawesi (eastern Indonesia) part of the inter-regional GEODYSSEA network and a local subnetwork on Sulawesi. Motions derived using data from the subnetwork con¢rm what the results of the inter-regional GEODYSSEA network have suggested; that is, that current deformation is high, and there are distinct deformation domains in the study area on Sulawesi. The tectonic mechanism of the triple junction has been analysed using a rigid microblock model. The triple junction area can best be interpreted as a headland of the Australian Plate de£ected by its collision with the Philippine Plate, thereby identifying the driving forces of the current deformation. The northern part is dominated by the Sula domain, which shows clockwise rotation. To the south, it is connected to the Australian Plate by an ensemble of microblocks undergoing counterclockwise rotation. In addition to the above, our tectonic model permits the determination of the local in£uence of two large earthquakes (M~7.8, 1996 January 1 and M~7.0, 1996 July 22) on the motion of the station Tomini (north Sulawesi). More observations and a denser GPS network are planned in order to study the behaviour of the Palu-Koro Fault, the main fault on the western limit of the Sula block.

years of GPS measurements across the Palu-Koro fault (Sulawesi, Indonesia) state left lateral strike-slip (3.4 cm/yr) with a small normal component (0.4 cm/yr). The measurements on intermediate points of the transect show that the fault is locked, at a depth estimated around 8-16 km by a very simple model. The best measured baseline linking the two most separate sites (Watatu and Toboli) shows very interestingly a constant rate of 2.6 cm/yr from 1992 to 1995, and an increased rate of 6.3 cm/yr since then. This increase is due to an earthquake on January 1, 1996 in the Minahasa trench, presumably affecting station Toboli's position by 2 cm of co-seismic and 2 cm of post-seismic displacement, which is added to its long term velocity. We interpret this phenomenon as additional stress loading on the Palu fault, driven by the slip during the earthquake 100 km to the North-East.

Deformation above the Sumatra subduction zone, revealed by Global Positioning System (GPS) geodetic surveys, shows nearly complete coupling of the forearc to the subducting plate south of 0.5°S and half as much to the north. The abrupt change in plate coupling coincides with the boundary between the rupture zones of the 1833 and 1861 (Mw>8) thrust earthquakes. The rupture boundary appears as an abrupt change in strain accumulation well into the interseismic cycle, suggesting that seismic segmentation is controlled by properties of the plate interface that persist through more than one earthquake cycle. Structural evidence indicates that differences in basal shear stress may be related to elevated pore pressure in the north.

Global Positioning System geodetic measurements at thirteen locations in Indonesia and four in Australia reveal that the Australian continent has accreted the Banda island arc to its margin. Small relative velocities of five sites on west Java, south Kalimantan, Bali, and south Sulawesi define a rigid Sunda shelf that moves relative to northern Australia in a manner consistent with pole locations from NUVEL-1 Australia-Eurasia but at a rate that is about 7% slower. Block-like northward motion of the southern Banda arc toward the Sunda shelf at nearly the same rate as Australia suggests that the Timor trough is now inactive as a thrust. Little of the convergence of Australia with Eurasia is accommodated by strain within the Banda arc structure. Most of the convergence appears to occur as northward translation of the rigid arc with shortening on the Flores and Wetar thrusts and possibly on faults within the back arc basin.

Global Positioning System (GPS) measurements made in 1991, 1992 and 1993 provide preliminary estimates of slip distribution between the Australian and Pacific plates in Irian Jaya, Indonesia. We interpret the GPS results with constraints from earthquake mechanisms and slip vectors, recent marine surveys, and geology. Three GPS sites in southeastern Irian Jaya show motions that are within 10 mm/yr of the expected motion of Australia. A coast-to-coast N-S baseline along 140.5°E crosses all known onland regions of active deformation but reveals no more than 15 mm/yr of shortening and 20 mm/yr of left-lateral shear in the 27-month period. The remaining 40 mm/yr of expected convergence between the Pacific and Australian plates probably occurs at the New Guinea trough. GPS sites on the island of Biak, at 136°E, and at Sorong, near the western tip of Bird's Head (at 131°E), both move 90–100 mm/yr in a WSW direction relative to Irian Jaya, but less than 15 mm/yr relative to each other. These sites are on either side of the Sorong fault and demonstrate that it is not presently the major boundary between the Australian and Pacific plates. Instead the plate boundary is now south of the Sorong and Biak sites. Earthquakes suggest possible structures that accomodate motion between Bird's Head and Australia but the relative importance of them remains uncertain.

Convergence across the Java Trench has been estimated for the first time, from annual Global Positioning System (GPS) measurements commencing in 1989. The directions of motion of Christmas and Cocos Island are within 1 deg of that predicted by the No-Net Rotation (NNR) NUVEL-1 plate motion model for the Australian plate although their rates are 25% and 37% less than predcited, respectively. The motion of West Java differs significantly from the NNR NUVEL-1 prediction for the Eurasian plate with a 1 deg difference in direction and a 40% increase in rate. We infer that either West Java moves with a distinct Southeast Asian plate or this region experiences plate margin deformation. The convergence of Christmas Island with respect to West Java is 67 +/- mm/yr in a direction N11 deg E +/- 4 deg which is orthogonal to the trench. The magnitude of convergence agrees well with rescaled NUVEL-1 relative plate model which predicts a value of 71 mm/yr between Australia and Eurasia. The direction of motion matches the direction inferred from earthquake slip vectors at the trench but may be more northerly than the N20 deg E +/- 3 deg predicted by NUVEL-1. On June 2, 1994, almost a year after the last GPS survey, an M(sub W) = 7.5 earthquake with slip vector direction N5 deg occurred south of central Java.

[1] We use survey mode and continuous GPS data from 1991 to 2007 to examine fault segmentation in the earthquake cycle at the Sumatra megathrust, site of the 26 December 2004 Mw 9.1 Sumatra-Andaman, the 28 March 2005 Mw 8.7 Nias-Simeulue, and the 12 September 2007 Mw 8.4 Mentawai earthquakes. These data, including new observations from 2006 and 2007, allow us to observe the final few years of one earthquake cycle and the beginning of the next. Our analysis reveals that the megathrust is segmented, a characteristic that may persist through multiple earthquake cycles. The Nias-Simeulue earthquake ruptured approximately the same region that broke in 1861, a 300 km long segment abutting the Sumatra-Andaman rupture zone. Farther southeast, the Mentawai segment of the megathrust (0.5°S–5°S), which produced M > 8 earthquakes in 1797 and 1833, is fully locked in the interseismic period but is flanked by two freely slipping regions, the Batu Islands in the NW and Enggano in the SE. The 12 September 2007 Mentawai earthquake sequence ruptured only the southern one third of the 1833 rupture zone. We model postseismic deformation from the Sumatra-Andaman and Nias-Simeulue earthquakes and find that afterslip was concentrated updip and downdip, respectively, from the main shocks. Comparing the velocity fields before and after 2001, we find the subduction zone underneath the Batu Islands and Enggano, which, prior to the earthquakes, was partially to fully coupled, appears now to be slipping freely. Thus, while the segmentation of the subduction zone is preserved, interseismic coupling on the subduction fault may vary with time.

A unique GPS velocity field that spans the entire Southeast Asia region is presented. It is based on 10 years (1994–2004) of GPS data at more than 100 sites in Indonesia, Malaysia, Thailand, Myanmar, the Philippines, and Vietnam. The majority of the horizontal velocity vectors have a demonstrated global accuracy of ?1 mm/yr (at 95% confidence level). The results have been used to (better) characterize the Sundaland block boundaries and to derive a new geokinematic model for the region. The rotation pole of the undeformed core of the Sundaland block is located at 49.0°N–94.2°E, with a clockwise rotation rate of 0.34°/Myr. With respect to both geodetically and geophysically defined Eurasia plate models, Sundaland moves eastward at a velocity of 6 ± 1 to 10 ± 1 mm/yr from south to north, respectively. Contrary to previous studies, Sundaland is shown to move independently with respect to South China, the eastern part of Java, the island of Sulawesi, and the northern tip of Borneo. The Red River fault in South China and Vietnam is still active and accommodates a strike?slip motion of ?2 mm/yr. Although Sundaland internal deformation is general very small (less than 7 nanostrain/yr), important accumulation of elastic deformation occurs along its boundaries with fast?moving neighboring plates. In particular in northern Sumatra and Malaysia, inland?pointing trench?perpendicular residual velocities were detected prior to the megathrust earthquake of 26 December 2004. Earlier studies in Sumatra already showed this but underestimated the extent of the deformation zone, which reaches more than 600 km away from the trench. This study shows that only a regional Southeast Asia network spanning thousands of kilometers can provide a reference frame solid enough to analyze intraplate and interplate deformation in detail.

[1] The 2004 catastrophic Indian Ocean tsunami has strongly emphasized the need for reliable tsunami early warning systems. Another giant tsunamigenic earthquake may occur west of Sumatra, close to the large city of Padang. We demonstrate that the presence of islands between the trench and the Sumatran coast makes earthquake-induced tsunamis especially sensitive to slip distribution on the rupture plane as wave heights at Padang may differ by more than a factor of 5 for earthquakes having the same seismic moment (magnitude) and rupture zone geometry but different slip distribution. Hence reliable prediction of tsunami wave heights for Padang cannot be provided using traditional, earthquake-magnitude-based methods. We show, however, that such a prediction can be issued within 10 minutes of an earthquake by incorporating special types of near-field GPS arrays (‘‘GPS-Shield’’). These arrays measure both vertical and horizontal displacements and can resolve higher order features of the slip distribution on the fault than the seismic moment if placed above the rupture zone or are less than 100 km away of the rupture zone. Stations in the arrays are located as close as possible to the trench and are aligned perpendicular to the trench, i.e., parallel to the expected gradient of surface coseismic displacement. In the case of Sumatra and Java, the GPS-Shield arrays should be placed at Mentawai Islands, located between the trench and Sumatra and directly at the Sumatra and Java western coasts. We demonstrate that the ‘‘GPS-Shield’’ can also be applied to northern Chile, where giant earthquakes may also occur in the near future. Moreover, this concept may be applied globally to many other tsunamigenic active margins where the land is located above or close to seismogenic zones.

The island of Sulawesi, eastern Indonesia, is located within the triple junction of the Australian, Philippine, and Sunda plates and accommodates the convergence of continental fragments with the Sunda margin. We quantify the kinematics of Sulawesi by modeling GPS velocities and earthquake slip vectors as a combination of rigid block rotations and elastic deformation around faults. We find that the deformation can be reasonably described by a small number of rapidly rotating crustal blocks. Relative to the Sunda Plate, the southwestern part of Sulawesi (Makassar Block) rotates anticlockwise at ?1.4°/Myr. The northeastern part of Sulawesi, the Bangai?Sula domain, comprises three blocks: the central North Sula Block moves toward the NNW and rotates clockwise at ?2.5°/Myr, the northeastern Manado Block rotates clockwise at ?3°/Myr about a nearby axis, and East Sulawesi is pinched between the North Sula and Makassar blocks. Along the boundary between the Makassar Block and the Sunda Plate, GPS measurements suggest that the trench accommodates ?15 mm/yr of slip within the Makassar Strait with current elastic strain accumulation. The tectonic boundary between North Sula and Manado blocks is the Gorontalo Fault, moving right laterally at about 11 mm/yr and accumulating elastic strain. The 42 mm/yr relative motion between North Sula and Makassar blocks is accommodated on the Palu?Koro left?lateral strike?slip fault zone. The data also indicate a pull?apart structure in Palu area, where the fault shows a transtensive motion and may have a complex geometry involving several active strands. Sulawesi provides a primary example of how collision can be accommodated by crustal block rotation instead of mountain building.

Nature, v. 440, n. 7080, p. 46-51, 2006. http://dx.doi.org/10.1038/nature04522; International audience; The Sumatra–Andaman earthquake of 26 December 2004 is the first giant earthquake (moment magnitude Mw > 9.0) to have occurred since the advent of modern space-based geodesy and broadband seismology. It therefore provides an unprecedented opportunity to investigate the characteristics of one of these enormous and rare events. Here we report estimates of the ground displacement associated with this event, using near-field Global Positioning System (GPS) surveys in northwestern Sumatra combined with in situ and remote observations of the vertical motion of coral reefs. These data show that the earthquake was generated by rupture of the Sunda subduction megathrust over a distance of >1,500 kilometres and a width of

Jakarta is the capital city of Indonesia with a population of about 12 million people, inhabiting an area of about 25-km by 25-km. It has been reported for quite sometime that several locations in Jakarta are subsiding at different rates. Leveling surveys performed in 1982, 1991 and 1997 have detected the subsidence up to about 80 cm during the period of 1982–1991, and up to about 160 cm during the 1991–1997 period; while GPS surveys observed the subsidence up to about 50 cm during the period of 1997–2002. InSAR technique using JERS-1/SAR L-band data estimated a subsidence rate of about 5 to 10cm during the period of 1993 to 1995. Maximum subsidence were found in the northwestern and central eastern parts of Jakarta, while minimum subsidence were found in the southern part. InSAR results show good correspondence with the results from Leveling and GPS Surveys

Terrestrial Reference Frame 2000. We compute poles of rotation for the Australia, Eurasia, and Pacific plates based on our analysis of the global GPS data. We find that regional tectonics is dominated by the interaction of four discrete, rotating blocks spanning significant areas of the Sunda Shelf, the South Banda arc, the Bird’s Head region of New Guinea, and East Sulawesi. The largest, the Sunda Shelf block (SSH), is estimated to be moving 6 ± 3 mm/yr SE relative to Eurasia. The South Banda block (SBB) rotates clockwise relative to both the SSH and Australia plate, resulting in 15 ± 8 mm/yr of motion across the Timor trough and 60 ± 3 mm/yr of shortening across the Flores Sea. Southern New Guinea forms part of the Australia plate from which the Bird’s Head block (BHB) moves rapidly WSW, subducting beneath the Seram trough. The East Sulawesi block rotates clockwise about a nearby axis with respect to the Sunda Shelf, thereby transferring east-west shortening between the Pacific and Eurasia plates into north-south shortening across the North Sulawesi trench. Except for the Sunda Shelf, the crustal blocks are all experiencing significant internal deformation. In this respect, crustal motion in those regions does not fit the microplate tectonics model. INDEX TERMS: 1206 Geodesy and Gravity: Crustal movements—interplate (8155); 1243 Geodesy and Gravity: Space geodetic surveys; 8150 Tectonophysics: Plate boundary—general (3040); 8158 Tectonophysics: Plate motions—present and recent (3040); 9320 Information Related to Geographic Region: Asia; KEYWORDS: crustal motion, Indonesia tectonics, GPS, current plate motions, Southeast Asia

The catastrophic consequences of the 2004 Indian Ocean and the recent (17 July) Java tsunamis demand the development of modern and robust tsunami early warning systems.The greatest challenge of the German Indonesian Tsunami Early Warning System (GITEWS), led by the National Center of Geosciences (GeoForschungsZentrum) in Potsdam, Germany, is to provide early tsunami warnings for the Indian Ocean coast of Indonesia where tsunamis can arrive 20–40 minutes after an earthquake. This article shows that reliable prediction of tsunami wave heights at the Sumatra coast cannot be provided using traditional, earthquake-magnitude-based methods. Instead, a novel, real-time warning technique based on special types of near-field global positioning system (GPS) arrays—a ‘GPS shield’—may be more appropriate for Sumatra and elsewhere around the globe.

[1] Global Positioning System (GPS) measurements made in Sulawesi, Indonesia, from 1992 to 1999 detected coseismic and transient postseismic deformation related to the 1 January 1996, Mw = 7.9 earthquake on the North Sulawesi (Minahassa) trench. These motions are superimposed on the long-term secular motion (40 mm/yr) of the left-lateral Palu fault in central Sulawesi and continued for about 1.5–2 years. Following the earthquake, a string of earthquakes (of magnitude >6) migrated along the Minahassa trench, from west to east. Subsequently, two earthquakes of magnitude >6 occurred on or near the Palu fault migrating toward the south. Modeling the increase in Coulomb stress generated by the successive earthquakes agrees with the hypothesis of interacting events. An unclamping effect, possibly due to fluid migration in the Palu area, is also suggested by the stress computations and the detected (GPS) displacements.