1 Results of Prior NSF Support NSF Grant EAR-9614970, The Recent Structural Morphologic and Dynamic Evolution of the Eastern Tibetan Plateau: A Multi-Disciplinary Study of the Process in Continental Deformation, $1,275,622 from 5/1/97 to 5/31/01, awarded to B.C. Burchfiel, L.H. Royden, K.X. Whipple and R.W. King, M.I.T. This study has produced a new interpretation for the mode of continental deformation in the region of eastern Tibet. We conducted geological field observations, established a GPS network of ~40 stations and a broadband array of 25 seismometers,, identified and obtained preliminary age dates on a low-relief, preuplift surface that caps much of eastern Tibet, and developed a quantitative model for large scale continental deformation to place our results in a geodynamic framework. Geodynamic modeling suggests that many features of Tibet result from continental deformation where the lower crust is very weak and undergoing regional scale flow. Our interpretation is that lower crust along the eastern plateau margin is thickened by evacuation of lower crust from beneath the central plateau, with little shortening of the upper crust. This is reflected in the uplift history of the eastern plateau and incision of Tibet by major rivers. We have obtained the first ages for uplift of the eastern plateau (8-10 Ma). Many publications have resulted from this study; a selected - to fit on one page - subset is below (* indicates MIT student author). Northrup*, C. J., Royden, L. H., and Burchfiel, B. C., l995, Motion of the Pacific plate relative to Eurasia and its potential relation to Cenozoic extension along the eastern margin of Eurasia: Geology, v. 23, p. 673-768. Burchfiel, B. C., Chen., Z., Liu, Y., Royden, L. H., 1995, Tectonics of the Longmen Shan and adjacent regions: Inter. Geol. Rev., v. 37, 661-736. Royden, L.H., 1996, Coupling and Decoupling of Crust and Mantle in Convergent Orogens: Implications for Strain Partitioning in the Crust, J. Geophys. Res., v. 101, 17, 679-17,705. King, R. W., Shen*, F., Burchfiel, B. C., Chen, Z., Li, Y., Liu, Y., Royden, L. H., Wang*, E., Zhang, X., and Zhao, J., 1997, Geodetic measurement of crustal motion in southwest China: Geology, v. 25, 179-182. Royden, L. H., Burchfiel, B. C., King, R. W., Chen, Z., Shen*, F., and Liu, Y., 1997, Surface deformation and lower crustal flow in eastern Tibet: Science, v. 276, 788-790. Wang*, E., and Burchfiel, B.C., 1997, Interpretation of Cenozoic tectonics in the right-lateral accommodation zone between the Ailao Shan Shear zone and the Eastern Himalayan syntaxis: Inter. Geol. Rev., v. 39, 191-219. Wang*, E., Burchfiel, B.C., Royden, L.H., Chen, L., Chen, J., and Li, W., 1998, Late Cenozoic XianshuiheXiaojiang and Red River fault systems of southwestern Sichuan and central Yunnan, China: Geol. Soc. Am. Spec. Pap. 327, 108 pp. Whipple, K.X., Kirby*, E., and Brocklehurst*, S. H., l999, Geomorphic limits to climate-induced increases in topogrpahic relief: Nature, v. 410, 39-43. Wang*, E., and Burchfiel, B.C., 2000, Late Cenozoic to Recent deformation in SW Sichuan and adjacent Yunnan, China, and its role in formation of the SE part of the Tibetan plateau: Geol. Soc. Am. Bull.. v. 112, 413-423. Kirby*, E., Whipple, K. X., Burchfiel, B. C., Tang, W., Berger, G., Sun, Z., and Chen, Z., 2000, Quaternary deformation along the eastern margin of the Tibetan Plateau: Tectonic and topographic evolution of the Min Shan, Sichuan Provice, China: Geol. Soc. Amer. Bulletin. v. 112, 375-393. Chen, Z., Burchfiel, B. C., Liu, Y., King, R. W., Royden, L. H., Tang, W., Wang*, E., Zhao, J., and Zhang, X., 2000, GPS measurements from eastern Tibet and their implcations for India/Eurasia intracontinental deformation: Jour. Geophys. Res., v. 105, p 16,215-16,227. Clark*, M.K. and Royden, L.H., 2000, Topographic Ooze: Building the eastern margin of Tibet by lower crustal flow, Geology, v. 28, 703-706. Shen* F., Royden, L. H., and Burchfiel, B. C., 2001, Large scale crustal deformation of the Tibetan plateau: Jour. Geophys. Res.v. 106, p. 6793-6816. Kirby*, E., Reiners, P., Krol, M., Whipple, K.X., Farley, K., Liu, Y., Tang, W. and Chen, B., Later Cenozoic uplift and landscape evolution along the eastern margin of the Tibetan plateau: Inferences from 40Ar/39Ar and (UTh)/He Thermochronology, Tectonics, submitted. Burchfiel, B. C., and Wang*, E., 2002, Northwest-trending, middle Cenozoic, left-lateral fautls in southern Yunnan, China, and their tectonic significance: Jour. Structural Geol., v. 25/5, p. 781-792 Clark*, M. K., Schoenbohm*, L. M., Royden, L. H., Whipple, K. X., Burchfiel, B. C., Zhang, X., Tang, W., Wang, E., and Chen, L., 2002, Surface uplift, tectonics and erosion of the Eastern Tibet from large-scale drainage patterns: Tectonics, p. 1-14. Clark*, M., Bush, J., and Royden, L., submitted, Crustally-derived dynamic topography produced near strength heterogeneities in the continental crust: An application of the Hele Shaw Cell for modeling weak lower crustal flow, JGR. Clark*, M.K., House, M.A., Royden, L.H., Burchfiel, B.C., Whipple, K.X., Zhong, X. and Tang, W., submitted, Late Cenozoic Uplift of Southern Tibet, Science. Schoenbaum*, L.M., Whipple, K.X., Burchfiel, B.C., and Chen, L., in press, Geomorphic constraints on surface uplift, exhumation, and plateau growth in the Red River region, Yunnan Province, China, Tectonics. 2 Introduction Within the major oceanic subduction systems, the interaction of dense subducting lithosphere with the surrounding mantle controls the geometry of the descending slab, the motion of the trench with respect to the adjacent plates, and the directions and rates of global plate motions. Because complex feedback systems between plate motions and subduction zones exist on a global scale, such plate boundaries must be understood by studies of mantle convection at a global (or nearly global) scale. Within the Mediterranean region, where subduction boundaries are typically less than a thousand kilometers in length (Fig. 1a), subduction probably exerts little to no effect on the motions of the major tectonic plates, implying that feedback system between subduction and global plate motions is negligible. Thus subduction processes within the Mediterranean can be studied at a regional scale. This proposal is for study of the active Hellenic subduction system, located in the east-central Mediterranean region and sandwiched between the slowly converging African and Eurasian plates. Because the Hellenic subduction system exhibits both spatial and temporal variability in subduction rate, it offers an excellent opportunity to quantify the relationships among subduction rate, trench retreat, density of subducted lithosphere and deformation of the over-riding lithosphere. It also offers an excellent opportunity to study the dynamic interaction between subducting lithosphere and the surrounding mantle. Subduction Systems of the Mediterranean Region The Hellenic system is one of several Miocene to Recent subduction systems within the Mediterranean region that exhibit thrust faulting and subduction along arcuate belts with low topographic relief (Figs. 1 and 2, McKenzie, 1972, 1978; Le Pichon, 1982; other examples include the Carpathian and Apennine subduction systems; Royden 1983; Lonergan and White, 1977; Malinverno and Ryan, 1986, Jolivet and Faccenna, 2000). These subduction systems can be distinguished from thrust belts like the European Alps by the presence of widespread extension within the upper plate lithosphere and by the migration, or “retreat”, of the trench away from the rigid part of the over-riding plate (Fig. 3). They also share a number of distinguishing geological features, such as little involvement of crystalline basement in the thrust complex, low topographic relief and anomalously deep foreland basins. Subduction along these young retreating boundaries has been generally oblique to the direction of Europe-Africa convergence, so that, unlike the European Alps, the relationship of these subduction systems to the overall convergence between Africa and Eurasia is not immediately obvious. Royden (1993ab) has suggested an evolutionary sequence whereby these subduction systems evolve naturally from continental collision along irregularly shaped continental boundaries. She concluded that the primary driving mechanism for retreating subduction is gravity acting on negatively buoyant subducted lithosphere. Similar conclusions have been reached by other authors (e.g. Le Pichon, 1982, Malinverno and Ryan, 1986, Lonergan and White, 1977; Faccenna et al., 2001), although the importance of slab pull in driving deformation is controversial (e.g. Dewey and Sengor, 1979; Le Pichon and Angelier, 1979; Mercier et al., 1979, Platt and Vissers, 1989; Jackson, 1994; Meijer and Wortel, 1997; Hatzfeld et al., 1997). A direct consequence of this hypothesis is that subduction at retreating subduction boundaries will occur only when the subducting plate has a sufficiently high density to drive the subduction (e.g. Faccenna et al, 2001). Subduction ends when large volumes of buoyant material enter the subduction system. Within this conceptual framework, differences in the evolution and duration of retreating subduction systems are largely attributed to the size and configuration of the deep water regions, underlain by dense lithosphere, that are available for subduction. Within the Mediterranean region there is a close correspondence between the density (water depth) of lithosphere that enters these subduction boundaries and the occurrence of active subduction (Fig. 1; Royden, 1993ab). Nevertheless, the length and time scales over which subduction rate, geometry and slab density are related and the details of slab-mantle interactions in these systems remain poorly understood. This proposal is aimed at addressing these general issues through a detailed analysis of the young Hellenic subduction system. 3 Active Tectonics of the Hellenic System Over the past decades Greece has been the focus of intense study by the earth science community. The classic papers of McKenzie (1972, 1978), Dewey and Sengor (1979), Angelier et al. (1982) and Le Pichon (1982) were published more than two decades ago. Since that time much new data has come to light but our understanding of the regional tectonics and driving forces for deformation has changed little. Although much excellent work has been done in recent years, most of the observational studies have been undertaken piecemeal or are aimed at local problems and the results are dispersed among a plethora of publications. Arguably, the most significant advance in Hellenic geology has come in the refinement of the rates of active deformation as determined by GPS measurements throughout Greece and adjacent regions (e.g. Clarke et al., 1998; McClusky et al., 2000). Precise dating and quantitative determinations of paleotemperatures and pressures associated with metamorphic events has led to a more precise description of metamorphic and extensional events and their spatial and temporal distribution. Indeed, the large volume of relatively new data that is available from Greece now makes it possible to formulate highly specific hypotheses about the dynamics of the system, and to test these hypotheses quantitatively through a targeted and tightly coordinated plan of research. This would not be possible in an area where the geology is not well known. Along these lines, we have drawn mainly on published data in constructing the following description of the tectonic setting and the development of the Hellenic system. However, many of the correlations that we highlight in this discussion are our own and come about only through the broad perspective of dynamic and kinematic processes that involve the entire Hellenic system. Many of these correlations have not, to our knowledge, been discussed elsewhere in the literature. This is particularly the case for the correlations between subduction rate, subduction zone geometry and the character of subducted lithosphere, in space and in time, which form the underpinnings for our proposed study. Active Tectonics of the Subduction System The Hellenic thrust belt involves faulting and folding of sedimentary rocks along a length of approximately 1000 km, from the southern Mediterranean northward through the Ionian Sea to the southeastern Adriatic Sea (Figs. 1 and 4). (We confine most of the discussion of the Hellenic Arc system to Greece because data from Albania are difficult to obtain and not required for this study.) We distinguish northern and southern segments of the Hellenic thrust belt that are dextrally discontinuous along the west-southwest trending Kephalonia Transform (e.g. Dewey and Sengor, 1979; Finetti, 1982; Kahle and Muller, 1993; Kahle et al., 1995). GPS data show that the rate of convergence across the southern segment of the belt is ~40 mm/yr (Fig. 5, Clarke et al., 1998; McClusky et al., 2000). Convergence across the northern segment of the Hellenides is ~5 mm/yr relative to Africa and ~10 mm/yr relative to southern peninsular Italy. Onland, the transition from fast to slow subduction rates occurs across a broad zone that is approximately continuous with the Kephalonia Transform (e.g. Goldsworthy et al., 2003; Armijo et al., 1996; Roberts and Jackson, 1991). The Kephalonia Transform coincides with a dramatic change in the lithosphere entering the Hellenic subduction system (Fig. 1b). Lithosphere north of the Kephalonia Transform is continental with crust ~25 km thick. This crust is overlain by sediments deposited in shallow water since Triassic time (e.g. Jacobshagen et al., 1978). The modern water depths are between 0 and 1 km. This lithosphere has a lower average density than the underlying mantle because its water depth is less than the ~2.5 km depth of typical zero-age oceanic lithosphere. Lithosphere south of the Kephalonia Transform underlies the deep water Ionian Sea and is thought to be oceanic in character [Makris, 1985; de Voogd et al., 1992]. Its age is unknown but is probably early Mesozoic and ~8 km of sediments overlie the basement. Water depths in the Ionian Sea are between 3 and 4 km, with maximum depths of ~5 km along parts of the Hellenic trench. This lithosphere has a higher average density than the underlying mantle because its water depth is greater than the ~2.5 km depth of typical zero-age oceanic lithosphere. 4 In the northern Hellenides convergence is confined to a narrow zone located west of the coast of mainland Greece; a bathymetric trench is not present here (Finetti and Morelli, 1972, 1973; Baker et al, 1997). Convergence along the southern Hellenides involves shortening of sedimentary rocks across a wide region of the Mediterranean sea floor, reaching from the Hellenic trench to the east side of the Mediterranean ridge (e.g. Kopf et al., 2003, Fig. 4) These folded and thrust-faulted sedimentary rocks appear to be detached from the underlying oceanic basement. Seismic and gravity data indicate that the basal detachment is ~12-15 km deep at the trench (Hirn et al, 1997; Clément et al., 2000). Deformation outboard of the trench is thin-skinned, involving only a veneer of sedimentary rocks and subduction to depth of the oceanic basement occurs only beneath the Hellenic trench. If we consider the shallow expression of the Hellenic subduction boundary to coincide with the zone where basement rocks are subducted to depth, then the Kephalonia Transform forms a pronounced discontinuity in the Hellenic subduction boundary (e.g. Dewey and Sengor, 1979). In the northern Hellenides the subduction boundary is a narrow northwest-striking zone of convergence located west of the Greek coast. In the southern Hellenides, the subduction boundary coincides with the northwesttrending Hellenic trench for hundreds of kilometers until the subduction boundary bends eastward towards Crete. Between the northern and southern Hellenides, the subduction boundary, or at least its shallow expression, is discontinuous by ~100 km along a right-stepping discontinuity that currently displays right-slip displacements (Sachpazi et al., 2000; Kahle et al., 1995). This discontinuity corresponds closely to the transition from shallow to deep water lithosphere and to the change in convergence rate from ~10 to ~40 mm/yr (Fig. 1b). The Hellenic slab exhibits a Bennioff zone to ~150 km depth (Papazachos et al., 2000 and references contained therein). Seismic tomography reveals a high-velocity slab-like structure dipping 45° to the northeast and north, extending to at least 600 km depth and perhaps into the lower mantle (Fig. 6, Spakman et al., 1993; van der Hilst et al., 1997; Karason, 2000). Gravity data indicate that the slab has an anomalously high density relative to the surrounding mantle. Gravity anomalies over the Aegean Sea are 50-100 mg more positive than expected from point-wise compensation of the Aegean and consistent with a dense slab dipping 45° northeast up to depths of ~250 km (Royden, 1993a; Tsokas and Hansen, 1997). Large, regional-scale isostatic gravity anomalies are not present along the northern portion of the Hellenides (Moretti and Royden, 1988). These observations suggest that the anomalously dense mass associated with the southern portion of the subducted lithosphere is not present, or is greatly reduced, along the northern portion of the subduction boundary. This is consistent with low-density lithosphere currently entering the subduction boundary in the northern Hellenides and high-density lithosphere entering the subduction boundary in the southern Hellenides. Active Tectonics of the Upper Plate: Central Hellenic Shear Zone and Aegean Domain The continental lithosphere above the Hellenic subduction system includes the emergent regions of mainland Greece, the Peloponnesus, Crete, westernmost Turkey and the submerged region of the Aegean Sea. GPS data show that the central and southern Aegean domains, Crete, and some of the Peloponnesus, are currently behaving as an almost rigid block with little internal deformation (Fig. 7, McClusky et al., 2000). Earthquake data show little seismic deformation of this region, with intense seismicity along its boundaries (Fig. 6). GPS velocities show that the northern boundary of the Aegean block is a broad east-northeasttrending zone of shear which we shall refer to as the Central Hellenic Shear Zone. It is approximately continuous with the Kephalonia Transform to the west and with the North Aegean Trough and North Anatolian Fault to the east. The displacement rate across the Central Hellenic Shear Zone is ~24 mm/yr in a N45°E direction, giving oblique right slip across the zone as a whole (Figs. 5 and 7). Estimating the trend of the Central Hellenic Shear Zone at ~N60°E, this resolves into ~22 mm/yr of right-slip parallel to the zone and ~9 mm/yr of extension perpendicular to it. The distribution of active deformation through the Central Hellenic Shear Zone is reasonably well known from GPS, earthquake locations and mapping of active fault breaks (e.g. Goldsworthy et al., 2003 5 and papers cited therein.) Within and adjacent to the Gulf of Corinth, active tectonic structures have been studied extensively but other areas are more poorly known (Freyberg, 1973; Jackson and McKenzie, 1983; Armijo et al., 1996). Earthquake focal mechanisms indicate mainly normal faulting on east-west trending faults with a few events showing right slip on northeast-trending faults or mixtures of the two modes (Fig. 6). Surface ruptures along active faults indicate that extension is largely accommodated on east-west to northwest-trending faults, resulting in water depths of ~900 m in the Central Gulf of Corinth. Extension has been accompanied by rapid uplift on the south side of the gulf where Pliocene-Quaternary sedimentary rocks are present at elevations up to 1200-1600 m (Philipson, 1893, Keraudren and Sorel, 1987). GPS data within and adjacent to the Central Hellenic Shear Zone show two primary zones of deformation, the Gulf of Corinth and the Gulf of Evoikos (Clarke et al., 1997; Goldsworthy et al., 2003). These data indicate little active rotation between the Peloponnesus and mainland Greece, but ~7°/m.y. clockwise rotation of Boetia (the region between the Gulf of Corinth and the Gulf of Evia, Fig. 4). These data, which are locally accurate to ±3 mm/yr, are not sufficient to discriminate finer details within the Central Hellenic Shear Zone. Small rates of displacement (and earthquakes) also occur on active faults north of the Central Hellenic Shear Zone, but their total contribution to Aegean-Eurasia velocities is small (Figs. 5 and 6). Within the Aegean Sea, the northern boundary of the Aegean block is more narrowly defined, corresponding a region of right-slip and extension in the North Aegean Trough (e.g. Taymaz et al., 1991; Roussos and Lyssimachou, 1991). Here Pliocene-Quaternary subsidence has resulted in a deep water basin with water depths of ~1 km as compared to the shallow water in surrounding parts of the northern Aegean. Earthquakes with right-slip focal mechanisms also occur in the north Aegean south of the North Aegean Trough, but GPS displacement rates across this zone are small. The GPS velocity field shows that the Aegean domain is bordered to the east by a broad northwest-trending zone of extension and left-slip in western Turkey, which we refer to as the West Anatolian Shear Zone. The displacement rate across this zone is ~18 mm/yr in a direction N15°E (Figs. 5 and 7). Estimating the trend of this broad deformation zone at ~N50°W, this resolves into ~13 mm/yr of left-slip parallel to the zone and ~11 mm/yr of extension perpendicular the zone. This zone is poorly studied. Little is known in detail about the geologic structures that accommodate active displacement although Quaternary extensional structures have been identified. East of the West Anatolian Shear Zone, much of Anatolia behaves as a rigid block moving eastward at ~20-25 mm/yr relative to Eurasia (Fig. 5). This lithospheric block is bounded to the north by the dextral North Anatolian Fault, a well-defined structure that accommodates almost all of the motion of Anatolia relative to Eurasia. It ends to the east in the collision zone between Arabia and Eurasia and extends westward into the northern Aegean Sea, where it merges with the northern boundary of the Aegean block. Total right-slip displacement on the North Anatolian Fault is estimated at ~85 km (Sengor, 1979). The Young Volcanic Arc Pliocene-Quaternary arc volcanoes form an arcuate chain located approximately 200 km north and east of the Hellenic trench (Fytikas et al., 1984). These volcanic centers lie above lithosphere subducted along the southern segment of the Hellenides (Figs. 4 and 8). No Pliocene-Quaternary arc volcanoes are present above lithosphere subducted along the northern segment of the Hellenic trench, with the possible exception of the northernmost volcano in the chain, which lies within the Central Hellenic Shear Zone. Cenozoic (Mainly Late) Evolution of the Hellenic Arc System Data that constrain the active deformation of a tectonic system give only a snapshot of the system’s current state. Because dynamic processes operate in space and in time, information about how a tectonic system has evolved is needed in the search for governing processes. Here we summarize geological data that constrain when and how the southern Hellenic subduction system became 6 distinguished from the northern Hellenic system and how these changes are related to the character of the subducted lithosphere and the rate of subduction (We summarize the most important relationships and chronology at the end of this section so that the bored reader can skip all but the summary section.) The Hellenic Thrust Belt and Subduction System North- to northeast subduction along the Hellenic belt has been continuous since Jurassic time (e.g. Mercier et al., 1975, Jacobshagen, 1978; Papanikolaou, 1993). From early Tertiary to late Miocene time, subduction resulted in stacking of the external Hellenic thrust sheets by west and south-vergent thrust faulting. Most of these thrust sheets consist of sedimentary cover stripped from its basement while the underlying crust and lithosphere were subducted beneath the Hellenides (Burchfiel, 1980). Thrusting progressed from internal to external zones, presumably paralleling the southward advance of the volcanic arc and subduction boundary (Fig. 8). A roughly delineated boundary separates the southern segment of the external Hellenides, exposed on the Peloponnesus, from the northern segment of the external Hellenides, exposed on mainland Greece. This zone coincides approximately with the Central Hellenic Shear Zone but is more narrowly defined. Some tectonic units, such as the Pindos nappe, are present north and south of this zone and continue across it. Other units of the southern Hellenides can be correlated with similar units in the northern Hellenides, but display differences in paleogeographic, sedimentary and/or metamorphic facies (for example, the Olympus and Mani marbles). Still other units, like the Parnassos unit, exist only on one side of this boundary zone. Hence this geologic boundary has been a persistent feature during the evolution of the Hellenic system. Because the Central Hellenic Shear Zone is approximately coincident with this older transverse zone of displacement, the task of identifying young structures that accommodate recent offset is complicated by older deformations that have occurred along approximately the same zone. Subducted Lithosphere: Oceanic or Continental? Beginning in late Eocene time, northeast-dipping subduction beneath the external Hellenides involved subduction of oceanic lithosphere (the Pindos zone, subducted until late Eocene time); subduction of continental lithosphere overlain by shallow-water carbonates (the Tripolis-GavrovoOlympus zones, subducted in late Eocene to Oligocene time); subduction of continental lithosphere overlain by deep-water carbonates (the Ionian-Mani zones, subducted in Oligocene to earliest Miocene time; the Ionian zone is not to be confused with the modern Ionian Sea); and subduction of continental lithosphere overlain by shallow-water carbonates (the Paxos zone, sometimes confusingly called the preApulian zone, subducted beginning in early Miocene time). (For a review see Jacobshagen et al., 1978.) Thus subduction of continental crust occurred throughout the Hellenides from Oligocene to Late Miocene time. In the northern Hellenides, subduction of Paxos zone continental lithosphere continues today. A widespread blueschist event in occurred Oligocene time, when rocks of the Ambelakia and Arna units were metamorphosed and thrust westward over the continental crust of the Olympus/Mani zone. These high-pressure metamorphic rocks are present along almost the entire portion of the external Hellenides. They were emplaced in late Oligocene to earliest Miocene time and occur above the Olympus/Mani units and below the Tripolis and Pindos units (e.g. Schermer, 1990; Bassias and Triboulet, 1994). Within the southern Hellenides, the Paxos zone continental lithosphere has been completely subducted and dense oceanic lithosphere of the modern Ionian Sea is currently entering the trench. The time at which Ionian lithosphere entered the southern Hellenic subduction zone is not well constrained by geological data, but is late Miocene or younger based on the youngest exposed thrust sheets of the external Hellenides. In the northwestern Peloponnesus, along the west coast of mainland Greece and in the Ionian islands to the west of mainland Greece, upper Miocene (Messinian) sedimentary rocks are overthrust by the outer thrust sheets of the Hellenides. These upper Miocene rocks comprise the uppermost part of the thick sedimentary sequence of the Paxos zone, reflecting deposition in shallow water conditions beginning with platform carbonate deposition in Triassic time. This implies that shallow 7 water continental lithosphere was subducted beneath the southern Hellenides as late as 6-8 Ma, constraining the oceanic lithosphere of the Ionian Sea to have entered the southern trench no earlier than ~8 Ma (or perhaps slightly earlier if overthrusting of these occurred within the thrust belt after they were stripped from their continental basement.) Continuity of Thrust Belt The Hellenic thrust belt was continuous from northern to southern Hellenides from Eocene through Late Miocene time as indicated by the largely continuous frontal thrusts of the Pindos zone (late Eocene) and thrust faults of late Miocene (Messinian) age (Fig. 4). The time at which the modern discontinuity developed in the Hellenic subduction boundary is not well dated but sediments deposited in east-west trending depocenters unconformably overlie the late Miocene thrust faults on islands within the Kephalonia Transform zone (e.g. Zakinthos). Marine seismic reflection data reveal areas of Pliocene normal faulting in the western part of the Kephalonia Transform region, with rapid subsidence from shallow water to more than 1 km water depth (Monopolis and Bruneton, 1982; Kamberis et al., 1996). Thus structures that disrupt the late Miocene thrust front were active by Pliocene time. Paleosubduction: Rates, Magnitude and Timing Throughout Cenozoic time, the down-going lithosphere beneath the Hellenides has been part of the African plate. Thus the rate of subduction beneath the Hellenides must be equal to the rate of trench migration toward Africa and we can equate subduction rate with the rate of trench migration and/or the rate of slab roll-back. Paleosubduction rates and magnitudes are difficult to determine precisely. Faccenna et al. (2003) estimate a subduction rate slightly in excess of 10 mm/yr for Oligocene to Recent. Several other independent estimates of subduction rate can also be made: (1) A series of volcanic arcs exist within the Aegean domain (e.g. Fytikas et al., 1984; Papanikalaou, 1993; and references contained therein). These arc progress from north to south, advancing ~300 km from Oligocene to recent time (Fig. 8), and until Late Miocene time were erupted above both northern and southern segments of the Hellenic subduction system. The rate of advance of the arc gives an estimate of the rate of subduction, although this is a lower bound because previously erupted arcs may have been moved southward due to later extension within the Aegean domain. The late Miocene volcanic arc is ~150 km north and east of the modern volcanic arc and ~150 km south of the Oligocene arc. Comparing the position of the Oligocene volcanic arc to that of the Late Miocene arc, yields ~150 km of subduction from Oligocene to Late Miocene time. Dividing by a time interval of ~20 m.y., yields an average subduction rate of ~8 mm/yr from 25-6 Ma. This applies to the northern and southern Hellenides and is similar to the modern convergence rate across the northern Hellenides. Taking the difference in position between the late Miocene and the modern volcanic arcs (~150 km) and dividing by 6 m.y. we obtain an average subduction rate of 25 mm/yr for the southern Hellenides in Pliocene-Quaternary time. This is slower than the modern convergence rate of 40 mm/yr across the southern Hellenides, but shows that subduction rates in the southern Hellenides increased markedly in late Miocene to Pliocene time. Alternatively, dividing the 150 km distance between the late Miocene and modern arcs by the modern subduction rate of 40 mm/yr yields ~4 m.y, indicating that the subduction rate increased beneath the southern Hellenides at or before 4 Ma. (2) The magnitudes and rates of subduction along the Hellenic trench can also be estimated from overthrusting of the relatively autochthonous basement beneath the Hellenides. In the eastern thrust belt, windows of the Olympus and Mani marbles give and the age of emplacement of the overlying thrust sheets of ~25-30 Ma. Rocks in the westernmost part of the Hellenides that correlate with the Olympus and Mani marbles (in the Gavrovo and Ionian zones) were overthrust at ~6-8 Ma. The distance between these internal basement windows and the external thrust faults, measured perpendicular to the trend of the Hellenides, is ~150-200 km. Dividing by the intervening time of 17-25 m.y. yields an average shortening rate of 6-12 mm/yr. This is a minimum rate for subduction because the relatively autochthonous basement exposed in the eastern Peloponnesus and at Olympus has probably been detached and transported toward 8 the foreland, but these results are at least consistent with the estimates of subduction rate determined from the same period. (3) If the modern discontinuity in the Hellenic subduction boundary across the Kephalonia Transform corresponds to a difference in subduction between the northern and southern Hellenides, then the subduction zone in the southern Hellenides has advanced 100 km further towards the foreland than the subduction zone in the northern Hellenides. Because thrust sheets emplaced in the external Hellenides from late Eocene to late Miocene time are continuous from the northern to southern Hellenides, the additional subduction beneath the Southern Hellenides was probably post late Miocene. Dividing 100 km by 6-8 m.y., yields a difference in average subduction rate between the northern and southern Hellenides of 12-17 mm/yr. Adding to this current rate of 5-10 mm/yr in the northern Hellenides, we arrive at an average subduction rate for the southern Hellenides of 17-27 mm/yr for Pliocene-Quaternary time. This is similar to the rate of subduction estimated for the southern Hellenides from advance of the volcanic arc. Late Cenozoic Evolution of the Hellenic Upper Plate Constraints on the relative rates of displacement beneath the northern and southern Hellenides can be obtained by examining the deformation of upper plate lithosphere along the Central Hellenic Shear Zone (Fig. 4). Because this zone separates the northern Hellenides from the southern Hellenides, the timing and magnitude of shear along this boundary should correlate with the relative rates of trench retreat. We correlate the initiation of this shear zone with the incipient disruption of the Hellenic subduction boundary into northern and southern segments. In this section we summarize briefly what is currently known about displacement within this broad shear zone, the timing of initiation, and the tectonic features that predate the formation of the Central Hellenic Shear Zone. Since at least Eocene time, extension within the upper plate of the Hellenic subduction system has occurred along dominantly arc-parallel structures, producing a series of back-arc basins with extension generally progressing from more internal to more external zones (Fig. 4, Mercier et al., 1989; Papanikolaou, 1993). (We use the term “back-arc extension” loosely, to indicate regional-scale extension of upper plate lithosphere. Extension may occur within and in front of the volcanic arc, but occurs largely behind the arc of the thrust belt and/or accretionary prism.) Extensional structures are preserved as regional-scale low-angle detachment faults that separate high-grade rocks in their footwalls from unmetamorphosed and highly extended hanging wall rocks. Such structures include the Early Miocene Strymon detachment system in northern Greece, the Middle Miocene detachment fault at Mt. Olympus, and the Middle to Late Miocene shear zones of the Central Aegean (Lister et al., 1984; Schermer, 1993; Dinter and Royden, 1993; Forster and Lister, 1999). In Late Miocene (Tortonian) time, extension resulted in normal faulting and subsidence of the Cretan Basin, forming an arc-parallel zone of subsidence north of the island of Crete (e.g. Jongsma et al., 1977). Subsidence of the Cretan Basin occurred in the hanging wall(s) of gently north to northeast dipping detachment faults exposed on the islands of Kythera (Papanikolaou and Danamos, 1991) and Crete (Fassoulas et al., 1994; Jolivet et al., 1996; Ring et al., 2001). Our recent mapping reveals that this zone of extensional faulting also extends northwestwards along the eastern Peloponnesus to the Gulf of Corinth (Figs. 4 and 9). The fault system is a low-angle extensional feature dipping 15° to 35° northeast and is similar to many of the extensional detachment systems mapped in the Basin and Range Province (Fig. 10, Skourtsos, 2002). This Late Miocene extensional complex contains the youngest arc-parallel extensional structures of regional extent within the Hellenic system. It is approximately coeval with the last thrusting and subduction events prior to disruption of the Hellenic subduction system by the Kephalonia Transform and Central Hellenic Shear Zone. No younger regional-scale arc-parallel extensional structures exist in the Aegean domain. Thus a fundamental change in the tectonics of the Aegean region occurred in late Miocene or early Pliocene time. This is reflected by termination of arc-parallel extension on the Parnon detachment system and initiation of arc-disruption by faulting along the Central Hellenic Shear Zone. 9 In Late Miocene or Pliocene time, younger east-west trending faults formed within the Central Hellenic Shear Zone and its offshore continuations to the east and west. Many of these cross-cut and disrupt the Hellenic thrust system (Figs. 4, 9 and 11). Within the Central Hellenic Shear Zone, the Gulf of Corinth has been extensively studied (e.g. Freyberg, 1973; Collier and Dart, 1991; Briole et al., 2000; Ori, 1989; Poulimenos, 1993; Leeder et al., 2002), as has the late Miocene Megara basin exposed near Corinth (e.g. Bentham et al., 1991). Other cross-cutting structures, dominantly extensional and with large displacement are fairly poorly known. These include the northwest-trending Beotikos-Kifissos detachment fault, which is no longer active but has a hanging wall basin containing probable PlioceneQuaternary, possibly latest Miocene, sediments. However, the initiation age, cessation age and magnitude of extension across this detachment are not well known. To the northeast faults bounding the Gulf of Evia and the northeastern (Aegean) margin of Evia are active (Roberts and Jackson, 1991). The age of initiation of extension in the Gulf of Evia is not known but seismic data indicate that the crust has been thinned to ~18 km beneath the gulf (Makris, 1977), suggesting significant extension over a period of several million years. East of the Central Hellenic Shear Zone, the dominantly dextral North Aegean Trough and the North Anatolian Fault also have a predominantly Pliocene age of initiation (Armijo, 1999, 2000; Taymaz et al., 1998; Barka and Kadinsky-Cade, 1988; Westaway, 1994) . GPS data show that the modern deformation zone along the North Anatolian Fault continues westward through the Central Hellenic Shear Zone and into the Kephalonia Transform (Kahle et al., 1995, 2000; McClusky et al., 2000). However, the nearly continuous exposure of internal Hellenic thrust sheets (especially the front of the Pindos nappe) indicate that a hypothesized 100 km of dextral displacement cannot be accommodated by a localized right slip within the Central Hellenic Shear Zone. Instead, structures that are present within the Central Hellenic Shear Zone are dominantly extensional and eastwest to west-northwest-trending. Reconciliation of the continuous nature of older thrust structures within the Central Hellenic Shear Zone with significant amounts of dextral shear across the zone remains an important problem. Collisional Events in Eastern Turkey In eastern Turkey the continental lithospheres of Arabia and Eurasia collided at ~12 Ma (e.g. Sengor et al., 1985). Here collisional and post-collisional convergence have resulted in significant crustal thickening and the development of high topography throughout eastern Turkey (Fig. 1a). These events probably contribute to the driving forces for the westward motion of Anatolia and perhaps for southwestward motion of the Aegean region (for example, McKenzie, 1972). In proposing to study the relationship of Hellenic subduction to deformation and motion of the greater Aegean region, we do not assume that the collisional tectonics of eastern Turkey have no effect on motion or deformation in the Aegean region. Indeed, it is likely that the events in eastern Turkey have some influence on the current motion of the Aegean block. However, GPS data show that the Aegean domain is moving nearly as fast relative to Anatolia (~18 mm/yr) as Anatolia is moving relative to Eurasia (~22 mm/yr). The Aegean domain also moves faster relative to Eurasia (~30 mm/yr) and in a different direction than Anatolia. This indicates that there are other important driving forces for the modern motion of the Aegean domain. This proposal is aimed at understanding the extent to which the evolution and modern setting of the Aegean domain, particularly along its western margin, can be understood in light of coeval events along the Hellenic trench and in the underlying subduction zone. The possible role of continental collision in eastern Turkey, and the westward extent of its influence, is something that we will continue to reassess throughout the proposed study. Summary of Key Points Key points, on which we base our current interpretation and working hypothesis for the Hellenic subduction system, can be abstracted from the previous sections and include: The modern subduction rates and the density of the lithosphere currently entering the Hellenic subduction zone change abruptly across the Kephalonia Transform, a dextral discontinuity in the subduction boundary with an apparent displacement of ~100 km. North of the Kephalonia Transform 10 lithosphere entering the subduction boundary is continental and subduction rates are 5-10 mm/yr. South of the Kephalonia Transform lithosphere entering the trench is oceanic and subduction rates are 40 mm/yr. The Kephalonia Transform also correlates with the apparent density of subducted lithosphere to several hundred kilometers depth, with a denser slab occurring beneath the southern Hellenides. The Kephalonia Transform continues eastward to a broad zone of dextral shear in central Greece, the Central Hellenic Shear Zone. This shear zone currently exhibits extensional (at ~11 mm/yr) and right slip (at ~22 mm/yr) components of displacement. It continues eastward to the North Anatolian Fault, which has approximately 100 km of total right slip at a modern rate of ~25 mm/yr. The modern volcanic arc exists only above lithosphere subducted beneath the southern Hellenides. Late Cenozoic Development of the Hellenic System: The average rate of subduction beneath the Hellenides appears to have been more-or-less constant from 25-8 Ma, at approximately 5-10 mm/yr. Lithosphere entering the subduction zone over this time period was continental, with shallow water sea floor entering the subduction boundary since 15-20 Ma. Over this time period the Hellenic thrust front appears to have been a continuous from northern to southern Hellenides and the Kephalonia Transform and the Central Hellenic Shear Zone did not yet exist. Subduction of continental lithosphere at 5-10 mm/yr continues unchanged to present time. But, beginning sometime during the interval 4-8 Ma, the subduction rate in the southern Hellenides increased rapidly to its current rate of 40 mm/yr. This is approximately the time of entry of oceanic lithosphere into the Hellenic trench, although this can only be constrained to be post 6-8 Ma. East-west trending faults within the Kephalonia Transform were active by Pliocene time while east-west trending faults of the Central Hellenic Shear Zone first became active in latest Miocene to Pliocene time. This is the same time as initiation of subsidence of the North Aegean Basin and right-slip along the North Anatolian Fault. Together, the Kephalonia Transform, the Central Hellenic Shear Zone and the North Aegean Basin constitute a system of Pliocene-Quaternary faults that disrupt arc-parallel thrust and extensional features of the older Hellenic system. Their initiation was coeval with the cessation of arc-parallel extension in the south-central Aegean domain. Interpretation and Working Hypothesis To our knowledge, the correlations between events that we have outlined in this proposal are largely new and not discussed in the literature. They indicate a fundamental reorganization of the Hellenic subduction system between latest Miocene and Pliocene time. Reorganization involved the subduction boundary, the subducted slab and the lithosphere of the over-riding plate. It was approximately coincident, in space and time, with entry of Ionian oceanic lithosphere into the Hellenic trench. Our current interpretation of these data is that the entry of dense oceanic lithosphere into the southern segment of the Hellenic subduction boundary is largely responsible for the post-Miocene evolution of the Hellenic system. Our working hypothesis is that the negative buoyancy of this dense oceanic slab has brought about an increase in the rate of subduction and trench migration in the southern Hellenides beginning in latest Miocene or early Pliocene time. This has caused an additional 100 km of subduction and trench roll back in the southern Hellenides, as compared to the northern Hellenides, producing ~100 km of dextral offset along the Kephalonia Transform, the Central Hellenic Shear Zone, and the North Aegean Trough. If this interpretation is correct, then the Hellenic system provides us with an unparalleled opportunity to study the dynamic interactions that occur within subduction systems. We propose to test this interpretation, and its dynamic implications, through a focused set of interdisciplinary studies. Before proceeding to our proposed research plan, we present two preliminary examples how one might connect kinematics and dynamics within the Hellenic system. Both examples are highly speculative and have not been tested or refined. Nevertheless, they are useful preliminaries to a discussion of the proposed project and demonstrate how various components of our multi-component research plan will be needed to understand the evolution and physical processes operating within this dynamic system. Preliminary Kinematic Model for the Central Hellenic Shear Zone 11 One component of our working hypothesis is that right shear across the Central Hellenic shear zone accommodates differential subduction rates between the northern and southern Hellenic subduction zones. Structures that might accommodate right-shear across the Central Hellenic shear zone are not obvious, a feature that has long prevented tectonicists from extending the North Anatolian Fault Zone as a narrow feature through mainland Greece. Since the advent of GPS measurements, it is clear that active displacement of ~25 mm/yr does traverse the Central Hellenic Shear Zone. Goldsworthy et al. (2003) have suggested that this occurs through a combination of extensional faulting and block-rotation within the shear zone. Similarly, if ~100 km of Pliocene-Quaternary displacement is to be accommodated on the dominantly extensional faults of the Central Hellenic Shear Zone, then rotational features are probably of great importance (e.g. McKenzie and Jackson, 1983) However, it is also clear that many currently active faults were not active in Pliocene time, and that there are large Pliocene extensional structures that are not active. Hence the location of faulting has changed with time. An important part of our proposed study will be to document how faulting and rotation of crustal fragments in the Central Hellenic Shear Zone have evolved through time. Fig. 12 contains a highly speculative reconstruction showing how 100 km of overall right-shear might have been accommodated by known extensional structures within the Central Hellenic Shear Zone. We begin with the present configuration of the Hellenides, including the modern subduction boundaries, the Pindos (late Eocene) thrust front, and two zones of late Miocene thrusting. We subdivide the Central Hellenic shear zone into four major blocks separated from one another by known zones of PlioceneQuaternary extension, including the well studied extensional system in the Gulf of Corinth and the poorly known extensional structures of the Beotikos-Kifissos normal fault system and normal faults in the Gulf of Evia. Based on our hypothesis that the 100 km dextral discontinuity in the subduction boundary is due to 100 km of additional trench roll-back of the southern Hellenic trench, we restore the southern trench segment, relative to the northern trench segment, by moving it 100 km northeast (without rotation). This aligns the northern and southern subduction boundaries as we hypothesize they might have been in late Miocene time. Holding mainland Greece fixed, the Central Hellenic Shear Zone is reconstructed by allowing extension across the accommodating fault zones to increase to the southeast, producing clockwise rotation of crustal fragments within the shear zone. The Ionian islands (Zakinthos and Kephalonia) move with the adjacent crustal blocks as indicated by the color scheme in Fig. 12, except that they close up against the west coast of Greece. The result is ~100 km of westward motion in the southern Peleponnesus and progressively lesser amounts of translation for regions north and east. This reconstruction produces clockwise rotation of crustal blocks within the Central Hellenic shear zone, with ~30° clockwise rotation of the Peloponnesus and progressively lesser amounts in crustal blocks to the east. We note that GPS data show little rotation of the Peloponnesus relative to mainland Greece. However, paleomagnetic data indicate approximately 30° of clockwise rotation of the southern Peloponnesus since about 5 Ma (and more in the Ionian islands, Kissel and Laj, 1988; Kissel et al., 1989, Duermeijeret al., 1999, 2000). This is consistent with the 30° clockwise bend in the overall trend of the Pindos and late Miocene thrust fronts as they extend south from mainland Greece into the Peloponnesus. This reconstruction is highly speculative and almost certainly incorrect in many details, but some features of the reconstruction are encouraging. First, it is possible to accommodate 100 km of right shear through the Central Hellenic Shear Zone by predominantly extensional deformation. Second, clockwise rotation of fragments brings the Pindos and late Miocene thrust fronts into parallelism across the Central Hellenic Shear Zone and produces thrust faults that maintain an approximately constant distance from the trench for a distance of ~500 km along strike. Third, and most important, it produces a hypothesis that can be tested in the field through geologic mapping, dating and paleomagnetic studies. Preliminary Geodynamic Model 12 An integral part of our proposed work is theoretical investigation of how subducted slabs interact with the surrounding mantle. For slabs with a large component of trench retreat and slab roll-back, this is a fundamentally 3D problem because flow occurs around the subducting lithosphere and stress on the slab is probably a function of the along-strike length of the subduction zone. This is perhaps one reason why few self-consistent trench models have been incorporated into studies of mantle dynamics. The few works that do include a trench whose motion is computed from the mantle flow, and not specified a priori, include numerical (for example Zhong and Gurnis, 1995; Becker et al., 1999; Tan and Gurnis; 2003;) and analogue models (Faccenna et al., 2001; Funiciello et al., 2003; Bellahsen et al., 2003). This section contains a preliminary example of how slab density may control subduction geometry and rate. Other models are described in the section on Proposed Study. This example addresses the case of no overall plate convergence so that subduction rate and upper plate extension rate are equal (Fig. 3, right side); the subducted lithosphere sinks through the surrounding mantle due to its negative buoyancy. We treat the subducting slab as a thin sheet with zero flexural rigidity (Fig. 13). Slab length does not change in a down-dip direction but stress is transmitted longitudinally along the slab. Buoyancy is proportional to the density difference between the slab and surrounding mantle, integrated through the slab. Buoyancy is also linearly proportional to water depth above the slab prior to its entry into the subduction zone, thus water depth can be used as a proxy for buoyancy. Water depths of 0 km indicate buoyant lithosphere, 2.5 km depths (mid-ocean ridge depth) indicate neutrally buoyant lithosphere, and 6.5 km depths indicate very negatively buoyant lithosphere. Slab buoyancy is allowed to vary as a function of position to simulate oceanic and continental regions. Forces that act on the slab are (1) vertical forces proportional to the buoyancy of slab, (2) forces normal to the slab due to viscous flow in the surrounding mantle, and (2) shear forces along the slab due to viscous flow in the surrounding mantle. The crux of the model is how to specify the viscous stresses that act on the slab, which will be the subject of much of our proposed theoretical work. Here we use a simple Stokes relation for motion of a rigid body through a viscous fluid of viscosity µ. Thus shear stress is proportional to µ/λ times the slab-parallel velocity of the plate in a lower mantle reference frame, where λ is some length scale that characterizes the dimensions of viscous flow in the surrounding mantle. Normal stress is proportional to 2µ/λ times the slab-normal velocity of the plate in the same reference frame. For a known slab geometry and buoyancy distribution, the velocity of the slab is computed by balancing the horizontal and vertical forces on the slab: - !(Ccos !) + qs cos ! + qn sin != 0 !s !(Csin !) - qs sin ! + qn cos ! + Fg = 0 !s (1) where C is the internal compressional force that acts longitudinally along the slab, qs and qn are the viscous shear and normal stresses due to flow in the surrounding mantle, θ is the dip of the subducted slab, s is position along the slab and Fg is the downward force related to the buoyancy contrast between slab and mantle. These equations can be combined to eliminate C, and qs and qn can be replaced by terms proportional to components of slab velocity. Fig. 14 shows steady-state geometries and rates of subduction (trench roll-back) for three uniformly buoyant lithospheric slabs. Rates of subduction vary from 10 to 70 mm/yr, increasing with the pre-subduction water depth of the slab. The angle of subduction increases with increasing slab buoyancy, so that the densest slab is subducted at the shallowest angle. The latter is not realistic. It is a function of how viscous stresses are applied to the shallow slab and we have found that slab dip is highly sensitive to the normal stress applied to the shallowest parts of the slab. Slab geometry and rate change when lithosphere of variable density enters the subduction zone. When buoyant continental material of large extent enters the trench, the rate of subduction drops from 70 mm/yr to <10 mm/yr in 5-10 m.y (Fig. 15). Slab angle steepens to nearly 90° and continental material is subducted to depths of ~300 km. If the continental block is only 200 km wide (approximately the largest 13 block that can be subducted) slab behavior is nearly identical to that of the large continent for 10 m.y. but, as dense oceanic lithosphere enters the trench behind the continental block, the rate of subduction increases to its initial value over 20-30 m.y. Fig. 16 shows the same model applied to a simulation of the Hellenic subduction system. We begin with steady-state subduction of low density lithosphere at 10 mm/yr. At 8 Ma, the water depth of lithosphere entering the trench is increased by 2.5 km to simulate entry of oceanic lithosphere into the southern Hellenic trench in late Miocene time. By 0 Ma the rate of subduction has increased to about ~40 mm/yr. If we compare this to the case where there is to change in the buoyancy of subducted material at 8 Ma (simulating the northern Hellenides), model results show approximately 100 km of additional subduction. Model results also show that entry of the dense lithosphere into the subduction system only affects the subducted slab to depths of ~200 km. At greater depth the model slab geometries are identical from northern to southern Hellenides. These preliminary results are approximately consistent with the time at which oceanic lithosphere entered the southern Hellenic trench, with the inferred 8 m.y. time interval for the increase of the southern Hellenic subduction rate from 10 to 40 mm/yr, and with the apparent 100 km dextral displacement of the subduction boundary between northern and southern Hellenides. Although much remains to be done to improve this preliminary geodynamic model of subduction, these first results are encouraging and help to guide our proposed plan of research. Proposed Study We propose a multidisciplinary study of the Hellenic subduction system aimed at better understanding the dynamics of the subducting lithosphere, its interaction with the surrounding mantle, and its connection to surface deformation and tectonics. Better documentation of the magnitudes, rates and timing of the various events within the young Hellenic system will enable us to better constrain these dynamic interactions and the length and time scales over which they occur. The proposed study is an international collaborative effort conducted jointly with colleagues from the University of Athens (Professor Dimitrios Papanikalaou), Athens Technical University (Professors Charis Biliris and Dimitrios Paradisos), and the University of Rome (Professor Claudio Faccenna and Dr. Francesca Funicello). Each of these scientists has had extensive past collaboration with at least one of the PI’s. The proposed work has three basic components, In brief, they include (1) Field and laboratory studies directed largely at constraining young and active near-surface kinematics through the Central Hellenic Shear Zone (and adjacent regions), including structural and stratigraphic mapping, paleomagnetic studies of crustal rotation, isotopic dating and analysis, and GPS. (2) Passive seismologic studies aimed at imaging the geometry and character of the shallow (<200 km) subducted slab beneath the northern and southern Hellenides. (3) Theoretical studies (modeling and analog) of the kinematics and dynamics of subduction systems like the Hellenic arc. Field Geological Studies The answers to many of the questions that arise as part of this proposal can be obtained only through detailed field studies. Establishing the age of initiation, temporal evolution and total displacement through the shear zone is crucial to understanding when and how the southern Hellenic subduction system became differentiated from the northern Hellenic system. Within a program of structural and stratigraphic mapping, closely coordinated with paleomagnetic studies, isotopic dating and GPS measurements, the questions that need to be answered include: How much right-lateral shear has occurred across the Central Hellenic shear zone, how is it accommodated, today and in the past, and how does it connect westward to the Kephalonia Transform and eastwards to the North Aegean Trough and the North Anatolian Fault? If overall shear is accommodated largely by extensional structures, how many extensional structures exist? Is displacement on these structures consistent with the GPS velocity field? 14 How much rotation of crustal fragments has taken place within the Central Hellenic Shear Zone and what role did it play in accommodating right-lateral shear? When did the present day pattern of deformation begin? How has it evolved through time? What is the deeper crustal response to the shallow crustal deformation? How is it related to the underlying slab geometry? Pliocene-Quaternary Structures in the Central Hellenic Shear Zone Published work and our preliminary field studies indicate that most of the young and active structures in the Central Hellenic Shear Zone are normal faults and associated hanging wall sedimentary basins. These normal faults trend northwest (in the east) to east-west (in the west) and are mainly confined to the Central Hellenic Shear Zone, although a few structures extend north and south of the shear zone proper. Within the Central Hellenic Shear Zone, the major graben structures bounding the Gulf of Corinth have been studied extensively (e.g. Briole et al., 2000). Other important extensional structures within the Central Hellenic shear zone have not been well studied. These include the northeast-dipping Beotikos-Kifissos normal fault and active extensional faults that bound the Gulf of Evia (Figs. 4 and 911). The Gulf of Evia and the Evia Basin, located west and east of the island of Evia, are bounded by east-dipping normal faults. Some of these faults are exposed on land on the west of the gulf, but east of Evia the structures are mostly under water (Jackson, and McKenzie, l988; Armijo et al., 1996). We propose to map selected regions along the Beotikos-Kifissos and Evoikos fault systems to constrain the timing and magnitude of extension. Hanging wall strata are poorly dated, but are generally mapped as Pliocene-Quaternary. Some hanging wall units contain lignite and floral fossils that should yield reasonable age constraints on these young normal fault systems. We also propose isotopic dating of volcanic detritus in the hanging wall basins. Our preliminary mapping has shown that in addition to the major graben structures, numerous smaller extensional faults and related basins form a complex interrelated pattern of extensional deformation within the Central Hellenic Shear Zone. These structures have not generally been recognized as extensional and are not active. However, they appear to have accommodated some of the earliest extension within the Central Hellenic Shear Zone. For example, in the northern Peloponnesus, near Feneos, we have identified an extensional system that in map pattern displays curved fault traces (Fig. 17). The fault patterns in this area are complex and successive generations of faults intersect one another. The faults appear to be listric in cross section and have hanging wall sediment accumulations. Hanging wall units are not aligned and the detached blocks have rotated around vertical and horizontal axes and relative to one another. The presence of syn-tectonic and post-tectonic sediments is important as they appear to be Pliocene-Quaternary in age and constrain the age of this extensional faulting. We have identified similar, but as yet unmapped, structures in several areas in the northern Peloponnesus and in areas immediately north of the Gulf of Corinth. These young structures are not large but are relatively easy to recognize because they are well expressed in the topography as small valleys and basins of deposition bounded by arcuate mountain fronts. We propose systematic mapping of such structures within the Central Hellenic Shear Zone, beginning with the Feneos region in the northern Peloponnesus and extensional structures north of the Gulf of Corinth. These faults are important because they will constrain the timing of early deformation within the Central Hellenic Shear Zone. These structural studies will be augmented by paleomagnetic measurements of block rotations and isotopic dating of hanging wall and overlap sedimentary assemblages. Interaction of Central Hellenic Shear Zone Structures with Older Hellenic Structures The young faults of the Central Hellenic Shear Zone cut older structures that are part of a regional system of arc-parallel thrust and extensional faults. These arc-parallel faults formed as the Hellenic thrust belt and associated extension migrated to the southwest until late Miocene time (Fig. 4). Our current hypothesis is that initiation of the Central Hellenic shear zone marks a major change in the tectonic regime of the Hellenic orogen as arc-parallel features became inactive and were cut by faults of the 15 Central Hellenic Shear Zone. To investigate this hypothesis we propose to map the arc-parallel extensional faults that immediately predate the Central Hellenic Shear Zone. Along the eastern side of the Peloponessus we have identified a previously unknown low-angle arc-parallel extensional fault system of regional extent, referred to as the Parnon detachment system, which parallels the older thrust units (Fig. 10). Normal faults within this detachment system displace the older thrust sheets and bring higher structural units down onto lower structural units, thinning or eliminating units within the original thrust complex. Within the upper plate of the Parnon detachment system high-angle faults can be shown to terminate downward at the basal low-angle fault contact. This detachment system is typical of the low-angle faults that are widespread throughout the Hellenic orogen (Fig. 4). We propose detailed mapping to constrain the age of the latest movements along the Parnon detachment system. Preliminary age control indicates that the Parnon detachment system may be as young as late Miocene as it appears to be the continuation of the dated late Miocene detachment on the island of Kythera (Papanikolaou and Danamos, l991). Particularly important will be mapping in the area where northwest trending faults of the Parnon detachment system and similar faults are intersected by younger east-west trending faults of the Central Hellenic shear zone. This occurs in the northeastern Peloponnesus where young east-west trending faults cut and in places reactivated older extensional structures of the Parnon detachment system. For example, the marine Megara basin is exposed in the northeastern Peloponnesus. Our mapping indicates that it may be a hanging wall basin to the Parnon detachment system. It contains lignite-bearing deposits that are tilted ~20° to the northeast and cut by east-west trending normal faults. These data suggest that the Parnon detachment was active here until late Miocene time (6-10 Ma). We propose to map key portions of the northern part of the Parnon detachment system and its hanging wall deposits to better constrain the geometry, direction of motion and age of faulting. Where the Parnon detachment system exposes lowgrade Mani and higher grade Arna metamorphic rocks in the footwall, we will collect samples for U/Th/He radiometric dating to constrain the time of exhumation. A detailed knowledge of older structures in the vicinity of the Central Hellenic shear zone will also be helpful in constraining the younger displacements within and across the shear zone. The older system of thrust faults and superposed regional detachment faults can be recognized in mainland Greece and in the Peloponnesus. However, the trend of the external belt appears to be rotated ~30° clockwise across the Central Hellenic shear zone (Figs. 4 and 12). We propose to test the continuity of these and other major structures across the Central Hellenic shear zone through more detailed field correlation of structure and stratigraphy than is now available, and to test the interpretation of clockwise rotation through mapping and paleomagnetic sampling. These structural and stratigraphic observations will be integrated with isotopic (and paleontologic) dating of hanging wall basin fill, and with paleomagnetic measurements to provide a temporal record of extension and rotation within the Central Hellenic Shear Zone. These results will be compared to estimates of overall displacement and rotation from regional lines of evidence and with seismic images of the subducted lithosphere to yield a regional picture of deformation from the surface into the mantle. Paleomagnetic Studies Paleomagnetic studies in the early 1980s were among the first to recognize that rotation occurred during the deformation in central Greece and the Peloponnesus from at least mid-Tertiary time to the present. Kissel and coworkers demonstrated that a large region west of the Aegean, extending from the Peloponnesus northward through Greece to Albania, experienced two apparent pulses of clockwise rotation around a pole located in Albania, with 25° of rotation in the Middle Miocene, and an additional 25° since early Pliocene time (Kissel et al., 2003; see also Sperenza et al. 1995; Duermeijeret al., 1999, 2000 and references contained therein). This zone of rotation extends across the Kefalonia Transform, the Central Hellenic Shear Zone, and other large-scale tectonic structures, showing no obvious relationship to them. Paleomagnetic data from Kissel et al. (2003) from within the Central Hellenic Sheer Zone show 16 that superimposed on this large-scale clockwise rotation is an additional 25° of clockwise rotation of Evia, apparently of post-Miocene age (and in approximately the same region that GPS data show undergoing rapid clockwise rotations today). This additional rotation was not found further east on the island of Skyros in the westernAegean. The authors interpreted this to be a reflection of rotation of faultbounded blocks within the Central Hellenic Shear Zone related to slip along the surrounding faults. These data are in partial agreement with our preliminary kinematic model of this region (Fig. 12) but there is still much that remains to be understood. First, Kissel et al.’s (2003) rotational data were from only two blocks and it is possible that differential rotations occur within subregions of the Central Hellenic Shear Zone. A recent study by Mattei et al. (in press) that sampled young sedimentary rocks south of the Gulf of Corinth found evidence for both clockwise and anticlockwise rotation in the western part of the Central Hellenic Shear Zone, although they the uncertainty on these rotations is extremely large. Second, Kissel et al. (2003) were not able to date precisely the observed rotations at sites in the Central Hellenic Shear Zone. Finally, the origin and nature of the large scale rotation of this part of Greece and its connection to the past and present tectonic structures remains largely a mystery, even after 20 years of paleomagnetic work. However, it seems significant that the second (post-Miocene) phase of rotation roughly coincides with the transition from subduction of continental to oceanic lithosphere in the Southern Hellenides. We are thus encouraged to pursue more extensive paleomagnetic work in the Central Hellenic Shear Zone and surrounding regions. We propose to conduct densely distributed sampling of all the crustal blocks (in so far as we can identify them) in the Central Hellenic Shear Zone so as to reconstruct a continuous record of the magnitude and orientation of progressive rotations (and the related geological structures) throughout the region. Our approach will be to conduct combined field mapping, paleomagnetic and geochronological work to obtain precise time constraints on these rotations. One of our aims will be to determine how closely linked in time are the onset of faulting in the Central Hellenic Shear Zone, the onset of the second clockwise rotational phase noted by Kissel et al. (2003), and the entry of dense oceanic lithosphere into the southern Hellenic trench. A close correspondence in timing would clearly implicate a connection between these processes. GPS Measurements To determine present-day crustal motions in southern Greece, we will build on the alreadyextensive GPS observations in this region (Fig. 18). Most of the high-precision data have been collected by our collaborators at the National Technical University of Athens working with scientists at MIT, ETHZurich, and a consortium of universities in the UK. The continuous GPS stations of the Hellenic Arc network and many of the survey-mode stations of the Peloponnesus and Aegean have velocity uncertainties at the level 1-2 mm/yr, as shown in Figs. 5 and 7 and can resolve strain and rotations at a sufficient to test our hypotheses. Current estimates of velocities from the 66 stations in the Central Greece Network [Clarke et al., 1997] have uncertainties of 3-4 mm/yr but may be improved with additional measurements made by the UK consortium through 2000 (expected release date for the newer data is highly uncertain and no earlier than 2005). In order to tie the UK network to the MIT analysis and to refine estimates of strain in the Central Hellenic Shear Zone, we propose to make additional measurements at 10 stations within the Central Hellenic Shear Zone in years 1, 3 and 5 of our study. Our MIT colleagues R. Reilinger and S. McClusky have been funded to carry out additional measurements in the southern Peloponnesus and Crete in 2004 and 2007. These new measurements will be south of the Central Hellenic Shear Zone but will be useful in constraining the detailed velocity field of the Peloponnesus relative to northern Greece. Thus we anticipate that the very modest GPS effort proposed here will be highly leveraged and will tie the independently measured UK network to the reference frame of the greater Aegean region and the existing large regional GPS data set at MIT. 17 We will perform the analysis of the GPS data at MIT using the GAMIT/GLOBK software [King and Bock, 2003; Herring, 2003] as described in McClusky et al. [2000], combining the new observations with the full set of data acquired for the eastern Mediterranean since 1989. Isotopic Studies We propose a program of radiogenic isotopic and U-Pb and U-Th-He studies to support our field mapping within the Central Hellenic Shear Zone and adjacent regions. Dating the youngest movement on the extensional faults of the south-central Aegean domain, such as the Parnon detachment system of the eastern Peloponnesus, is necessary to establish the time at which arc-parallel extensional features were still active within the Aegean extensional domain. Dating of the oldest motions on the east-west to northwest trending extensional faults of the Central Hellenic Shear zone is important to establish the time at which arc-disruption began. It is conceivable that the two sets of structures overlap in time, recording a gradual transition from arc-parallel extension to arc-disruption and the termination of extension in the south-central Aegean domain. Biostratigraphic dates are available for a few parts of the Late Miocene to Present sedimentary sections of Central Greece, particularly where young marine sediments have been uplifted and exposed in areas adjacent to the Aegean Sea, the Gulf of Corinth, and on islands within the Ionian and Aegean Seas. These data are sufficient to suggest a late Miocene age for the Parnon detachment system (dated in one place) and Pliocene-Quaternary ages for faults within the Central Hellenic Shear Zone. Unfortunately, much of the sections are non marine and tight age control is largely lacking, except for the young (<500 kyr) marine terraces along the south side of the Gulf of Corinth and a few other scattered ages. Within the Central Hellenic Shear Zone, the important Beotikos-Kifissos detachment system is largely undated, as are the older displacements on the active Gulf of Evia. Dating of the onset and cessation of extension in these areas, as well as the older, inactive normal faults near the Gulf of Corinth, will be a high priority for our proposed study of the Central Hellenic Shear Zone. In so far as possible, we aim to date pre-, syn- and post-kinematic deposits associate with faulting in this area, so that in conjunction with our structural mapping it will be possible to reconstruct the magnitude and locus of extension within the shear zone and how it has evolved with time. We propose two primary dating techniques to achieve this objective. First, footwall rocks for the Parnon detachment system and for some of the extensional faults in the Central Hellenic shear zone expose the low grade metamorphic rocks of the Mani, Arna and base Tripolis (Tyros beds) units. (The Arna is higher grade than the Mani or Tyros units, but the metamorphism is Oligocene and not related to extensional unroofing.) Paleotemperature estimates in the structurally deepest Mani unit are ~250°C, while the structurally highest Tyros beds show low greenschist facies metamorphism (Figs. 10 and 11). We propose to date the unroofing ages of these footwall rocks using U/Th/He geochronolgy. We have budgeted funds to do these analyses either at MIT or Cal Tech. The MIT lab is running but given the back log we could easily switch to Cal Tech if necessary. Second, we propose to date hanging wall basin sediments, and overlap assemblages, using U/Pb geochronology on volcanic detritus within the basin fill. Although thick ash beds are relatively rare, the presence of an active arc approximately 100 km to the east has provided a volcanic contribution to these sedimentary sections. Thus we propose to generate a detailed chronology by the dating of thin beds of volcanic ash using zircons. Bowring has shown that precise dates may be obtained on Miocene and even younger rocks and we will get around the problem of having to find fresh igneous minerals for Ar-Ar geochronology. Radiogenic isotopic studies will be to examine temporal changes in arc magmatism. We will use a full suite of Nd, Sr, and Pb isotopes to evaluate whether there is a contribution from subducted continental lithosphere and whether we can detect when it became involved in arc magmas. Proposed Seismic Studies A two-stage, passive seismic investigation is proposed to image the subducted Mediterranean slab in the 0-250 km depth range, to the north and south of the Central Hellenic Shear Zone. Previous seismic 18 analyses in the region have imaged the general outline of the slab, but the available data coverage and employed methodologies have not yielded the geometrical resolution necessary for the geodynamic modeling purposes of this study. Furthermore, most previous studies have focused on the southern portion of the Hellenic subduction system, leaving the northern part poorly constrained In this experiment, appropriate array design and application of new high-resolution teleseismic imaging techniques will provide improved resolution of the slab beneath northern and southern Hellenides and a means of detecting the transition from subducted continental to subducted oceanic lithosphere. Other goals include providing constraints on devolatilization and hydrous reactions in the slab and mantle wedge, and the coupling between the slab, the overlying accretionary prism and the Aegean lithosphere. These new geometrical constraints will be directly applicable to geodynamic simulations of the Hellenic subduction system. In the next section we provide a brief overview of past seismological work in the study area and then proceed to the specific objectives and methods of the proposed experiment. Previous seismological results For more than 30 years the eastern Mediterranean has been the focus of relentless seismological scrutiny. A variety of seismic approaches have been employed to image the subsurface structure of the Hellenic subduction system. The studies range from local investigations of the upper crust through seismic reflection and refraction profiles to large-scale imaging of the subducted slab from global tomography. We can divide seismic images of the Hellenic system into three distinct depth domains: shallow (0-30 km), intermediate (30-180 km) and deep (>180 km). In the shallow domain, marine reflection and refraction profiles west of the trench support the idea that the crust beneath the Ionian Sea is oceanic in origin, with an average thickness of 8 km, overlain by up to 8-10 km of sediment [Makris, 1985; de Voogd et al., 1992; Kopf et al., 2003]. East of the trench, a series of local marine and land-based seismic experiments in the Ionian Islands show the subduction décollement as a shallow-dipping discontinuity at ~12 km depth [Hirn et al, 1997; Clément et al., 2000; Sachpazi et al., 2000]. Landward to the east, seismic signal generated in the shallow domain is dominated by the structure of the continental crust and Moho topography. Travel-time tomography and receiver functions show that the continental Moho reaches depths in excess of 40 km beneath the Peloponnesus and mainland Greece [Makris, 1985; Papazachos and Nolet, 1997; Tiberi et al., 2000; Marone et al., 2003]. The crust thins to the northeast, reaching thicknesses of ~30 km beneath the Gulf of Evia and the central Aegean Sea. The seismic signal in the intermediate depth domain is dominated by interactions with the cold subducted lithosphere, which exhibits higher than average wavespeeds. Independent tomographic results using local and teleseismic events image the slab as a tabular anomaly dipping shallowly to the northeast beneath the Peloponnesus, at an angle of ~10-12° [Papazachos and Nolet, 1997; Tiberi et al., 2000]. The 20 km slab thickness obtained by Papazachos and Nolet (1997) is surprisingly thin for a lithospheric plate, but their results may not be well constrained for sub-horizontal geometries. Both studies detect a sudden increase in subduction angle at ~70 km depth beneath the Gulf of Corinth. Below that point, the slab dips ~30° to a depth of 150-170 km below the western Aegean Sea. The distribution of earthquake foci outlines a diffuse Wadati-Benioff zone extending to depths <180 km (Papazachos et al., 2000), with an increase in subduction angle at ~90-100 km beneath the northeast Peloponnesus. This kink in the slab is deeper than that imaged tomographically, but this is reasonable because the Wadati-Benioff zone does not necessarily represent the top of the slab. Anisotropy in the mantle beneath the Hellenic subduction system has been investigated through SKS splitting [Schmid et al, 2003] showing small splitting times and poorly constrained polarization directions. Tomographic inversion of Pn waves by Hearn [1999] shows a clear preferential alignment of the fast axis parallel to the arc. 19 Below 180 km depth, the geometry of the Hellenic subduction system is exclusively constrained by regional and global tomography [Spakman et al., 1993; van der Hilst et al., 1997]. Models show the subducted slab as a fast wavespeed anomaly that dips steeply to the northwest and extends to a depths of ~1000 km (Fig. 6). Objectives of the Proposed Seismic Experiment To test geodynamic models of the Hellenic subduction system, robust constraints are required for the geometry and internal structure of the subducted slab and for slab interactions with the overlying mantle wedge. Although the southern slab segment has been clearly detected in previous seismic investigations, its finer structure at intermediate depth remains poorly constrained. The proposed seismic experiment is designed to improve resolution in this depth range and address the following questions: a) what are the position and geometry of the subduction décollement (i.e., top of the slab)? b) what is the thickness of the subducted slab? c) how are earthquake foci distributed with respect to the slab (i.e., at the décollement, in the subducted crust or mantle?) d) is there evidence for devolatilization of the slab and association hydration of the mantle wedge? e) what morphological features distinguish the southern Hellenic subduction system, where oceanic crust is being subducted, from the northern Hellenic subduction system, where continental crust enters the trench? f) is there a change in signature of the subducted crust with depth associated with the inferred transition from oceanic to continental composition? g) why is there a sudden increase in subduction angle near 70 km depth? Is this related to eclogitization or to the transition from oceanic to continental subducted crust? We propose deployment of two dense arrays of three-component, broadband seismometers to the north and south of the Central Hellenic Shear Zone (Figs. 19 and 20). Our rationale for two lines is to image and compare the distinct morphologies of the slab in the two regimes of the Hellenic subduction system. The total duration of the experiment will be 3 years, with deployments of 1.5 years each for the southern and northern lines. Each line will comprise 40 stations with average station spacing of 7.5 km. Two-Dimensional GRT Inversion The array configuration is primarily devised for the 2D Generalized Radon Transform (GRT) Inversion, a teleseismic imaging/migration approach recently developed at the University of British Columbia [Bostock and Rondenay, 1999; Bostock et al., 2001]. This method is designed to image complex discontinuities (rapid transitions in material properties) beneath dense broadband seismic arrays using the scattered wavefield comprised in the teleseismic P-coda. In contrast to many previous migration approaches that yield images of subsurface properties analogous to scattering potential, this method directly inverts for P-wave and S-wave velocity perturbations and density. The 2D GRT inversion is a powerful tool to image structure at intermediate depth, in subduction zone settings and stable continental areas (Fig. 21). For example, images obtained in the Cascadia subduction zone clearly delineate the geometry of the subducted oceanic crust, a change in subduction angle at ~40 km depth, and a disruption near that inflection point that has been interpreted as devolitization of the crust and hydration/serpentinization of the mantle wedge [Rondenay et al., 2001; Bostock et al., 2002]. The example shown in Fig. 21 clearly demonstrates that this method will image currently unresolved structure of the Hellenic subduction system. The expected results will address many of our objectives, including the geometry of the décollement, the distribution of earthquakes in the slab (obtained by relocating foci and mapping them onto the imaged sections with additional constraints from tomography as described below), hydrous reactions in the crust and mantle wedge, morphological differences between north and south subduction regimes, and possible down-dip transition from an oceanic to a continental slab in the southern 20 subduction system (to be determined from changes in crustal/slab thickness and signature). Our capacity to differentiate between dehydration of subducted crust and the transition from oceanic to continental subducted crust will rely on establishing morphological differences between the northern and southern regimes (where continental and oceanic crust, respectively, are entering the subduction zone) and on the characteristic signatures associated with eclogitization of subducted crust and hydration of the mantle wedge (determined from GRT images and additional geophysical observables discussed below). The resolution afforded by 2D GRT inversion has been thoroughly assessed on theoretical grounds [Bostock et al., 2001], through synthetic tests [see Fig. 21; Shragge et al., 2001], and as a function of experimental geometry [Rondenay et al., 2002]. These results ensure that the method will accurately resolve structure such as subducted crust down to a depth of at least 200 km, with a maximum dip resolution ranging from 85° near the surface, to 45° at 200 km depth (dependent on event distribution and array aperture), and a volume resolution of ~5 km (based on the high cut-off frequency of 0.3 Hz chosen to avoid scattering from surface topography). Note that 7.5 km station spacing and 1.5 year duration for each array are needed for comprehensive sampling of scatterers in the model, so as to yield accurate material property perturbations [Rondenay et al., 2002]. We need adequate event distribution to illuminate slab structure from updip (WNW) and downdip directions. A plot of world seismicity over 1.5 years (01/2000-06/2002) for events of magnitude ≥5.7 and teleseismic epicentral distances 30°<∆<100° shows that this condition will be met (Fig. 20). We must also validate the assumption that 2D imaging is appropriate for the Hellenic subduction system. The applicability of the 2D assumption can be tested from the forecasted event distribution (Fig. 20) in the context of slab geometry (dashed line in Fig. 19). For the Hellenic subduction system, the structure must extend linearly for a minimum of ~100 km on either side of the imaging plane. Results from previous tomographic studies [e.g., Papazachos and Nolet, 1997] and placement of the proposed array across a relatively straight portion of the crescent-shaped subduction complex validate this second condition. Note that the imaging plane does not have to be exactly parallel to the array; data may be projected onto any vertical plane oblique to the array, allowing us to optimize model geometry for the 2D assumption. The validity of images obtained in the proposed study will be thoroughly assessed by generating synthetic datasets for these models. For this purpose, we will use the finite-difference scheme of Hicks [2003], which allows for arbitrary source and receiver positions – a geometrical condition that prevails in all teleseismic experiments. Additional Complementary Analyses To better constrain the structure of the Hellenic subduction system and the possible occurrence of hydrous reactions or partial melt, data from our temporary dense arrays will be augmented with data from permanent stations of the Greek National Seismic Network (Fig. 19) and other temporary arrays [e.g., Papazachos and Nolet, 1997; Papazachos et al., 2000]. Travel-time tomography using local and teleseismic events will be used in conjunction with the 2D GRT inversion to obtain new images of long wavelength (i.e., volumetric) P- and S-wavespeed anomalies in the Hellenic subduction system, with 2D GRT results projected as constraints onto the model space of the tomography. This will reduce smearing of anomalies in the tomographic model and yield a more robust assessment of the thickness of the subducted slab at intermediate depth. Tomographic results will also be used to obtain estimates of the Vp/Vs ratio (or δlnVs/δlnVp) and bulk sound velocity throughout the Hellenic subduction system. These seismic observables will help constrain the local influence of temperature, fluids and partial melt on mantle wedge rheology [e.g., Karato, 1993, 2001; Christensen, 1996; Wiens and Smith, 2001]. Seismic attenuation, another important rheological indicator, may be inverted for using P and S spectra from local events [see, e.g., Tsumura et al., 2000]. Earthquake foci located near the imaged profiles will be relocated using these new tomographic models and mapped onto the slab structure imaged by 2D GRT inversion. These results will yield improved constraints on the distribution of earthquake foci with respect to the subduction décollement and the subducted Moho. Shear-wave splitting will be performed on teleseismic and, perhaps, regional 21 events. Teleseismic SKS splitting measurements will be used to verify whether, with an increase in data coverage and number of stations, more consistent estimates of the anisotropic parameters than those of Schmid et al [2003] can be obtained for the mantle underlying the Hellenic subduction system. If possible, splitting measurements will be conducted on the waveforms of large local events, although these seldom occur in the western Hellenic subduction system. Indeed, there have only been four events with magnitude >4.0 and focal depth >60 km over the past 3 years (2000-2003). As a consequence, local events may only loosely constrain anisotropy of the mantle wedge. Modeling of Subduction Zone Dynamics There has been much work on the interplay of surface plate motions and mantle flow (see recent reviews by King, 2001, and Christensen, 2001), but most existing models are restricted to 2D and prescribe, rather than calculate, the trench positions through time. Exceptions include numerical studies by Zhong, Gurnis and coworkers (e.g. Zhong and Gurnis, 1995, and Gurnis et al., 2000) and laboratory work by Faccenna and coworkers (see below). Here we propose to develop and investigate a variety of subduction models with self-consistent trench motions, where slab migration and trench retreat arise as a natural result of slab/mantle rheology and density, rather than being specified a priori. This problem is fundamentally three dimensional because once the end of the slab has descended to the base of the upper mantle, flow is constricted around the end of the slab and slab roll-back occurs only through lateral flow of mantle material around the unconfined edges of the slab. Thus we propose a three pronged investigative approach (purely viscous, coherent slab, and analog), using 2D and 3D analyses, to provide maximum insight into this difficult problem in 3D fluid flow. Within the context of the Hellenic system, constraints from geology and geodesy (past and present-day kinematics) and tomography (density structure at depth) will enable us to evaluate the role of slab dynamics in plate boundary evolution. Our geodynamic studies will proceed hand-in-hand with field studies as it is our experience that theoretical results are invaluable for guiding observational efforts as well as for interpreting the results of the observations. Coherent Slab Models (led by PI Royden) The preliminary geodynamic model presented earlier in this proposal is perhaps the most simplistic slab model that can be constructed. It has the advantage of simplicity and of being semianalytical, but is lacking a well-grounded exploration of the viscous stresses on the subducted slab. We propose to expand this approach to incorporate explicitly the effects of viscous flow within the surrounding mantle. The intrinsic behavior of the slab, and the way in which stresses are transmitted along the slab, are satisfactorily managed by treating the slab as a thin, flexurally competent sheet. Equation 1 is easily modified to incorporate flexural behavior of the slab, and we currently have working models for a range of slab rheologies, including time and depth-dependent slab properties. Variable slab buoyancy and variable densities within of the upper plate lithosphere and “fore-arc” region are also easily computed and incorporated into our existing models. The crucial issue is determining how viscous stresses are applied to the subducted slab by the surrounding mantle. Proper analysis of these stresses involves computation of fluid flow within the mantle as a function of slab geometry and velocity, which in turn are controlled by slab density and mantle viscosity. This problem is complex because toroidal flow around the subducting slab is the dominant mode of mantle flow when a slab rolls back through the surrounding mantle. (This may be different from the dynamics of subduction when the convergence rate and subduction rate are comparable, so that the slab may descend through the mantle without a significant slab-normal component of motion). We propose to solve the Stokes equation in 3D for mantle surrounding a subducting slab with prescribed geometry and velocity conditions at the slab surface (using the method of Successive OverRelaxation on a finite-difference grid; for this computation finite element techniques are not needed). We will develop a self-consistent, steady-state model of subduction in 3D. In particular, for a specified slab width, density, rheology and side boundary conditions there is a unique slab geometry such that the 22 buoyancy and viscous forces acting on the slab sum to zero and the slab is in steady-state. We propose to use an iterative approach to find this unique geometry, so that although the mantle flow field will be instantaneous, we will use successive numerical approximations of slab geometry to converge on the final, consistent, steady-state slab geometry. This in itself will be an important outcome because the controls on 3D slab geometry in subduction boundaries are currently not well understood. Questions that we hope to address include: What is the 3D steady-state geometry of subducted lithosphere? What is the flow field in the mantle? What are the viscous stresses acting on the slab? How do these results depend on slab width, density and flexural properties? While the problem of slab roll-back is inherently three-dimensional, solution of the Stokes equation in 3D is probably not feasible at every time-step during non steady-state evolution. But, because non-steady-state behavior is crucial to understanding the Hellenic, and many other, subduction systems, we will attempt to extract some simple “rules” from the 3D analysis that will enable us to make meaningful computations in 2D. We anticipate that viscous stress will be a function of velocity, slab width, distance from slab edge, depth, viscous layering in the mantle, etc. With the answers to these and other questions, we hope that we will be able to make useful, reasonably accurate models of subduction, trench migration and slab geometry with 2D numerical studies. Our goal is to understand how slab density, and its variability, affect the subduction process, how flow is distributed within the mantle, and the length and time scales over which slab density affects subduction dynamics. These studies will be informed by, and conducted in parallel with, purely viscous numerical models and 3d analog models, as described below. Purely Viscous Flow Models (led by PI Becker) Alongside the coherent-slab approach, we propose to study subduction dynamics and trench motion with purely viscous numerical models. The overall goals are similar to those described for the coherent slab modeling. The proposed work involves adapting existing fluid models for the Hellenic subduction zone to study the effect of lithospheric density variations (oceanic/continental material) flow constrainment, 3-D background currents, and viscosity stratification on back-arc motion, slab shapes and margin evolution. We build upon the numerical modeling experience of Becker, who has previously worked with Faccenna's group in Italy. Our studies to date compare 2D viscous computations with Faccenna's 3D analog models (Becker et al., 1999) and use the results to interpret the tectonic evolution of the Central Mediterranean subduction zone (Faccenna et al., 2001). Currently Becker is involved in a follow-up study of trench migration for simplified, but dynamically consistent, 2D subduction models (Enns et al., in prep.). By way of example, Fig. 22 shows dense oceanic lithosphere sinking within a mantle whose (Newtonian) viscosity increases by a factor of 50 at 660 km (cf. Hager et al., 1985). The trench is allowed to move freely. Two sets of mechanical side-boundary conditions illustrate the importance of flow constrainment, which is only poorly explored by 2D models. Our results show that trench migration and slab morphology vary dramatically depending on how the slab and the lower mantle interact. We intend to build on these general results and adapt this model to our study region in the Hellenides, with the inclusion of density variations in the subducting lithosphere. We plan to analyze similarities and differences between these purely viscous models and the coherent-slab models described above, in order to determine how the existence of a competent slab influences the dynamics of mantle flow, and to what extent the rheology of the slab itself plays an important role. Laboratory models of subduction clearly show that 3D currents around the slab are important for the shape of streamlines in the mantle, affecting slab arcuity, morphology, and episodicity in trench motion (e.g. Funiciello et al., 2003; Kincaid and Griffith, 2003). Thus we will strive to explore 3D viscous models numerically. Toward this end, Becker has begun collaboration with F. Funiciello (U Roma TRE), C. Piromallo (INGV, Rome), and Faccenna. This study focuses on the large scale temporal 23 evolution of mantle structure in the whole Mediterranean region as it results from the complex collision process between the African and European plates. For 3D flow, we use the parallelized finite element code CITCOM (Moresi and Solomatov, 1999; Zhong et al., 2000) obtained from the Caltech geoframework.org site. We have begun study of isolated viscous, thermal slabs for evolving, prescribed surface plate motions (cf. Tan et al., 2002). We plan to construct a regional, high resolution model for the Hellenic setting and incorporate the new constraints that will become available during the study, with close collaboration with the observational team. The methods that will be developed by Becker and coworkers will be useful for both the Hellenic implementation, and the parallel, complementary effort on the large scale structure of the region with the Italian group. Tasks before us include incorporation of a tracer advection scheme for the parallel version of CITCOM (for compositional advection) and inclusion of a regional finite element flow model in global circulation calculations. The latter will be of importance to the wider community of geodynamicists. The problem of embedding regional, high resolution computations within a consistent treatment of global, 3D, earth like flow also arises in other subjects like the study of regional seismic anisotropy (cf. CSEDI proposal by D. Blackman, UCSD). We intend to explore prescribing velocity boundary conditions from a Hager and O'Connell (1981) semi-analytical calculation to a regional CITCOM mesh, and use the global version of CITCOM with mesh refinement in the Hellenic arc. Analog Modeling Analog, or laboratory, modeling is a useful tool for gaining insight into complex 3D process, including subduction and plate interactions. Faccenna and co-workers in Rome have been conducting analog experiments of subduction systems for several years, with particular application to the Mediterranean subduction systems. These experiments are usually performed inside a large tank, using a visco-elastic material like silicone putty (rhodorsil gomme) to simulate the lithosphere and honey or glucose syrup to simulate a viscous mantle. The viscosity and density of the simulated slab and mantle can be varied over a large range to simulate the depth and/or temperature dependent rheology in the mantle and zones of varying buoyancy in the slab. Because almost any boundary conditions, density and geometry can be imposed on the slab, this method is highly flexible and complementary to numerical modeling efforts. The main issue is to ensure that the scaling from tank experiments to numerical models is carried out correctly, and that the experimental set-up scales appropriately with the earth. Much of Faccenna previous work has focused on the way subduction is initiated (Faccenna et al., 1999), on the evolution of the slab as it descends into the upper mantle (Becker et al., 1999) and on the way the slab interacts with the 660-km discontinuity (Faccenna et al., 2001; Funiciello et al., 2003). The latter experiment shows that the dynamics, geometry and rate of subduction depend on the details of slabmantle interaction. So far, most of our experiments assume a constant density contrast between the subducting slab and the mantle. However, in the context of this proposal we have examined the role of buoyancy in slab dynamics has been tested with a preliminary simulation of a continental plate and a buoyant continental block entering trench (Figs. 23 and 24). These results are qualitatively consistent with the preliminary subduction model shown in Figs. 14 and 15 and provide encouragement that there will be a strong connection between theoretical and analog models of subduction of variable density slabs. Becker and Faccenna have previously collaborated, combining numerical and analog approaches to achieve a better understanding of subduction initiation and evolution and we will build on this history of collaboration. Our proposed study of the Hellenic subduction system, and of the subduction process in general, will proceed alongside the ongoing study of Becker, Faccenna, Funicello and Piromallo, into the temporal evolution of mantle structure within the whole Mediterranean region. We expect strong interactions between the Italian group, Becker and Royden, with an MIT PhD student to be co-advised by Royden and Becker. Management Plan The proposed study is an international collaborative effort conducted jointly between U.S. 24 investigators and European colleagues from Greece and Italy. Our European colleagues will supply funding for their portions of the proposed work, but the bulk of the costs and the overall management of the research plan will lie with the MIT group. The lead PI (Leigh Royden) and two of the CoI's (Clark Burchfiel and Bob King) have much experience in organizing and carrying out large, multidisciplinary studies in conjunction with foreign colleagues. In particular, these investigators have a long history of successful participation in CD-funded research program in eastern Tibet, with Burchfiel as lead PI (see Results of Prior NSF Funding). The study proposed here is of similar scope, duration and funding level to the east Tibetan project, although logistics in Greece will be considerably less difficult. We note that it is relatively unusual to have all (but one, Thorsten Becker) of the US investigators come from the same institution. However, the MIT earth science department is large and diverse, providing an unusually broad range of expertise within a single department. Our experience has been that, by having most of the primary US investigators at MIT, we are able to achieve a high level of interaction and coordination among PI’s due to proximity and daily interactions. Collaborative Workshop/School in Greece, Years 1, 3 and 5 In order to maximize our collaborative efforts and coordinate the scientific and logistical aspects of the proposed work, we plan to hold a two-week workshop in Greece in years 1, 3 and 5. This workshop will function partly as a school to teach our Greek collaborators (and ourselves) about new ideas, techniques and scientific approaches that are relevant to this project and to tectonics studies in general. It is important that these workshops be held in Greece because it will enable many Greek geoscientists to attend all or part of the workshop and will be easily accessible to our Italian colleagues. We particularly aim to include Greek and Italian graduate students, some of whom will be associated with our study, but some of whom will not be. We intend to have all of the US investigators and students associated with the project attend each workshop, so that the workshop will also serve as a venue for the exchange of scientific culture between US, Greek and Italian students as well as the more senior scientists. We believe that these workshops will be most effective if held in a relatively isolated environment where the participants will be away from their usual daily activities and can spend time discussing science outside of the formal workshop hours. We have been offered the use of conference and lodging facilities in a small village outside of Kalamata (in the Peloponnesus) by the vice Mayor of Kalamata. This is within a half-day's drive of Athens and will be easily accessible to all participants. We estimate approximately 20-30 participants for the workshop, the numbers varying depending on whether we are discussing general scientific issues in tectonics (more participants) or the details of our collaborative study (fewer participants). Travel to the workshop and lodging for participants will be funded by each of the collaborative groups, so that we need only to fund travel and lodging for the US investigators. Travel costs will be highly leveraged as many of the participants will be traveling to Greece for field work at the same time. Roles of Principle Investigators and Foreign Collaborators As lead PI, Leigh Royden will be responsible for coordinating all aspects of the proposed work (see above). In addition, she will conduct field geologic studies in conjunction with Clark Burchfiel and Dimitrios Papanikalaou and modeling studies of subduction zone dynamics in conjunction with Thorsten Becker, Claudio Faccenna and Francesco Funicello. She will also take the lead in tectonic analyses and integration of data for the Hellenic region. Royden has had extensive experience with the tectonics of Mediterranean subduction systems, conducted numerous modeling studies of continental deformation and dynamics, and, although not primarily a field geologist, has conducted and supervised field geologic studies in Greece, Tibet and elsewhere. Clark Burchfiel will conduct geological field mapping studies, and together with Leigh Royden and Dimitris Papanikalaou, be responsible for supervising student geologic mapping. He will also be involved in tectonic analysis of the Hellenic system. Burchfiel has a long history of international field mapping and regional tectonic analysis. He has worked in the Mediterranean/Balkan region off and on for over 30 years. 25 Robert King will be responsible for GPS measurements and analysis. He has worked extensively with existing GPS data from the Aegean region. Stephane Rondenay will be responsible for the proposed seismic arrays, processing of seismic data, and integration of regional observations with global tomographic data. He has successfully conducted a similar experiment in the Cascadia subduction boundary. Rondenay is a first-time NSF investigator. Ben Weiss will be responsible for laboratory analysis of paleomagnetic samples. He has much laboratory experience. Jointly with Royden and Burchfiel, he will be responsible for citing and collecting the paleomagnetic samples. Sam Bowring will be responsible for geochronology, collecting samples in coordination with Royden and Burchfiel, and for all U/Pb laboratory analyses. He is the director of the U/Pb facility at MIT and has done extensive high resolution U/Pb dating as well as isotopic characterization of arc magmas. Thorsten Becker will be responsible for dynamic modeling of subduction zones, along with Leigh Royden and Claudio Faccenna. He is a post-doctoral fellow at Scripps Institute of Oceanography and has written papers on the viscous dynamics of subduction zone roll back and trench migration, with particular reference to the Mediterranean subduction systems. Becker, the single PI not at MIT, will be closely integrated into the project through joint theoretical work with Leigh Royden and co-supervision of a PhD student at MIT. Becker is a first-time NSF investigator. Foreign Collaborators Dimitrios Papanikalaou is our principle counterpart in Greece. He is Professor of Geology at the University of Athens and the leading tectonician in Greece. Until September, 2002, he was the Secretary of Civil Protection within the Greek government, prior to which he was director of the Greek National Marine Research Center and head of the Earthquake Planning Office. He is primarily a field and structural geologist. He will involve (and fund) graduate and undergraduate students from the University of Athens in our field research, particularly the structural and stratigraphic mapping and tectonic analysis. Charis Biliris and Dimitrios Paradisos are Professors of Geodesy at the National Technical University in Athens. They are involved in virtually all GPS work in Greece. They have successfully collaborated with King and other MIT scientists on past geodetic studies. Professors Claudio Faccenna and Massimo Mattei and Dr. Francesca Funicello are our Italian collaborators from the University of Rome, Three. Faccenna and Funicello will be responsible for analog models of subduction systems, an extension of their ongoing work in this area. These analog experiments will be conducted with strong coordination with the theoretical studies of Becker and Royden. Mattei will be responsible for conducting paleomagnetic sampling and measurements in collaboration with the US PI’s. Faccenna is a structural geologist and tectonician with extensive knowledge of the Mediterranean region. Funicello has a background in numerical modeling of subduction at the ETH. Mattei is a paleomagnetist with prior experience working in the young sedimentary rocks of central Greece. Multi-Institutional Budget MIT Scripps Year 1 $480,803 $22,751 Year 2 $440,035 $23,786 Year 3 $488,861 $24,873 Year 4 $436,310 $29,979 Year 5 $547,109 $31,376 Total $2,393,178 $132,765 In addition to the funds requested in these proposals, our foreign counterparts will provide resources for the proposed work. Papanikolaou will provide resources to fund himself and Greek graduate students for field and structural mapping. He will also provide funds and personnel for the routine maintenance of the seismic network. Faccenna will provide funding for analog and theoretical modeling for himself and Funicello and will host project visitors in Rome for joint theoretical and analog work. Mattei will provide additional resources for paleomagnetic fieldwork and measurements. Our Greek counterparts will provide conference facilities for our bi-annual workshops (see below). Scientists 26 from each country will fund themselves and their students for travel to and accommodation at these workshops. Education and Human Resources Women are reasonably well represented among U.S. college students studying science, but at every subsequent step along the educational corridor the percentage of women scientists drops. Within US universities, the percentage of earth science faculty who are women is ~12%. At in the earth sciences at MIT, 78% of the undergraduates are female, 42% of the PhD students are female, but only 8% of the faculty are female. Many female scientists cite the choice between career and family as a reason for opting out. The lead PI of this proposal, Leigh Royden, provides an example for the next generation of female scientists that women can succeed as faculty in top research universities without giving up children and family. More than half her former PhD students have been women; many of whom are now faculty in other institutions. A recent survey has shown that MIT PhD students hold more faculty positions in U.S. universities than PhD's from any other department in the country. This speaks to the quality of the doctoral education at MIT and the quality of the students who choose MIT for graduate work. This proposal seeks funding for four PhD students over a 5-year period, which will fund these students from start to finish of their graduate careers. Thus the proposed study will, if funded, contribute to the scientific infrastructure of the U.S. by educating the next generation of research faculty. The proposed study will also feed back into the graduate and undergraduate curriculum at MIT. For example, unfunded collaborations between Leigh Royden and Dimitrios Papanikalaou led to a spring lecture series on the geology of Greece and an MIT-supported two-week field trip to Greece for graduate and undergraduate students in June of 2003. Undergraduates were fully funded by the department while graduate students each contributed $200 toward expenses. Events like this are instrumental in drawing undergraduates into the earth sciences The same unfunded work in Greece has provided for two senior undergraduate thesis projects supervised by Leigh Royden, one completed in June ’03 and one currently underway (both female students). The proposed study is for an internationally collaborative study, involving senior geoscientists and graduate students from the U.S., Greece and Italy. We plan a series of three workshops in Greece throughout our 5-year study period. Although a crucial component of the workshops will be coordination of the research efforts and results, much of the workshop will be devoted to discussion and lectures on new scientific ideas, approaches and techniques. We will involve Greek scientists and students outside of our immediate collaborators so that the workshop will become part school, part research effort. Indeed, one of the best aspects of the MIT student trip to Greece was the interaction of U.S. and Greek students throughout the trip, with a stimulating exchange of scientific and cultural ideas. Many of the US and Greek students are still in email contact as a result of the friendships that developed during the field trip. The Vice Mayor of Kalamata (southern Peloponnesus) has offered us conference facilities in a secluded village on the outskirts of Kalamata because he feels strongly about building a high level of scientific and cultural events within the city and surrounding areas. We anticipate that these workshops will help to disseminate recent scientific advances to an international group of scholars and students, and will make for strong collegial ties between U.S., Italian and Greek scientists. 27 28 29 30 31 32 33 34 35