T53A-1393
Sumisu Volcano, Izu-Bonin Arc, Japan: Site of a Rhyolitic Caldera-Forming Eruption From a Small Open-Ocean Island
Sumisu submarine caldera, an 8 x 10 km Crater Lake type structure along the front of the Izu-Bonin arc, was the site of a 30-60 ka eruption that introduced 50-70 km3 of rhyolite tephra into open-ocean environment. A manned submersible, two ROVs, a DeepTow camera system, and dredge samples (augmented by newly acquired single-channel seismic profiles and SeaBeam mapping) were used to study the caldera and surrounding areas. Caldera-wall studies show that pre-caldera eruptions built a complex of overlapping submarine dacitic and basaltic cones that formed an edifice 40 x 25 km in diameter; its summit grew above sea level to form an island about 200 m high. The caldera-forming eruption likely began on this island and sent a Plinian column high into the air. We interpret that prodigious rates of tephra fallback overwhelmed the Sumisu area, forming huge rafts of floating pumice and forcing large volumes of hot, proximal fallback debris beneath sea level-- generating gravity flows of quenched pumice that traveled downslope along the sea floor. The caldera rim, currently 100-400 m below sea level, is mantled by thin and discontinuous deposits of syn-caldera pumice showing no evidence of heat retention. Ocean currents apparently eroded most proximal pumice away (especially during periods of late Pleistocene sea-level lowering), and the caldera rim is locally mantled by lag lithics. ODP Leg 126 cores in the Sumisu Rift, atop the Rift's east margin, and on the forearc slope, show the sea floor south and southeast to be underlain by non-welded pumice deposits 10-20 m thick. Geochemical data point to Sumisu as the source of these distal deposits, rather than the nearby arc-front centers of Minami Sumisu or Torishima-indicating that the Sumisu pumice gravity flows traveled for distances of at least 70 km. Post-caldera edifice instability resulted in the collapse of a 15 degree sector of the east caldera rim and the formation of spectacular thin-skin slumps that ring much of the edifice.
T53A-1394
Anatahan Activity and Monitoring, 2005
Anatahan volcano began erupting in 2003 and continued with a second eruptive phase in 2004. In January 2005 the volcano began a sequence of eruptions and unrest that continues as of September 2005. The activity has been characterized by punctuated episodes of very steamy strombolian activity and vigorous ash emission. Some of the ash emissions have reached 50,000-foot elevations, with VOG and ash occasionally reaching the Philippines and southernmost Japan, over 1000 miles away. Vigorous ash emission has been almost continuous since June 2005. A M4.8 long-period earthquake (LP) occurred in mid-August, one of the largest LPs recorded on the planet in the last quarter-century. Real-time monitoring consisting of a few telemetered short-period seismometers and acoustic sensors has been severely hampered by ashfall on the small island. Monitoring efforts have been focused on the aircraft/ash hazard, with the goal of providing the FAA and airline industry with rapid notice of seismic signatures that may indicate ash columns rising to the altitude of airline traffic, or nominally above 20,000-30,000 ft.
T53A-1395
Volatile (H2O, CO2, Cl, S) contents of magmas in the Izu-Bonin back and rear arc
In back arc environments mantle melting may be promoted by mantle upwelling and decompression as well as by addition of volatile-rich slab components that lower the mantle solidus. Distinguishing between these two modes has important ramifications for models of subduction zone dynamics. We report results of an ongoing study of volatile contents of primitive submarine lavas from the Izu-Bonin rear arc region (Latitude 30°30' to 32°30' N). A diversity of lava types is evident in this region, and we have studied seven representative samples. From east to west these include samples from the active rift zone that lies immediately behind the volcanic front, backarc knolls adjacent to the rift, and seamount chains that extend west into the Shinkoku basin. Few samples are glassy and so we have concentrated on olivine-hosted melt inclusions, measuring volatile and major element contents by SIMS and EMPA, and will determine trace elements by LA-ICP-MS. In most cases melt inclusions contain appreciable CO2, which suggests that minimal water has been lost from these melts via low pressure degassing. Magma from each portion of the rift have distinctive water contents. Inclusions from the active rift zone have water contents that are typically < 1 wt.%, consistent with a greater degree of decompression and upwelling. Back arc knolls have highly variable water contents, from < 0.5 wt.% up to > 3 wt.%. Western and east seamounts have water contents typically between 1 and 2.5 wt.%. The range of water contents we observe is similar to that from the Mariana trough further south. Overall our results are consistent with existing models that suggest a role for slab-derived fluids across the arc and rear arc region, but suggest that decompression melting may also be locally important. Further trace element studies on these samples are underway.
T53A-1396
The magmatic plumbing of the submarine Hachijo NW volcanic chain, Hachijojima, Japan: long distance lateral magma transport?
Recent geophysical observations on basaltic composite volcanoes in Izu-Bonin arc reveal the process of long distance lateral magma transport within the shallow <ETH> middle crust. Such intrusion events sometimes caused flank fissure eruption and also triggered a formation of collapsed caldera (Miyakejima 2000). To clarify a long-distance magma transport system of the basaltic composite volcano in volcanic arc from geological and petrological aspects, we investigated a submarine volcanic chain (Hachijo NW chain) nested Hachijo Nishiyama volcano, a frontal composite volcano in the northern Izu arc. The volcanic chain extends 15 km from Hachijo-Nishiyama volcano and is composed of ridges and many small cones with basal diameters generally less than 2km. Dredge sampling recovered basaltic lavas and spatters. A diving survey using a ROV (Hyper Dolphin) revealed pillow lava flows on steep slopes and accumulation of spatters and agglutinates near the eruption center. Basalts from the Hachijo NW chain generally have more primitive composition (up to nearly 7% of MgO) compared to the basaltic rocks from the Nishiyama. The bulk magma composition of Hachijo NW chain is controlled by fractionation of clinopyroxene, olivine and plagioclase while plagioclase accumulation was indicated by aluminum-rich character of the Nishiyama volcano and its subaerial satellite cones. Trace element ratios unaffected by melting or crystal fractionation (e.g., Nb/Zr) are not significantly different between the Nishiyama and the Hachijo NW chain. This implies that the sources of magma for these volcanic systems are basically identical. However, ratios affected by melting process are significantly different between the two. Hachijo NW chain shows lower LREE/HREE and Zr/Y, implying difference in degree of partial melting of the source. Other possible processes for producing these differences in trace element characteristics include crustal assimilation. These results obtained so far appear to indicate that the feeding system of the Hachijo NW chain is independent from that of the Nishiyama volcano, even though there are some transitional lavas implying interaction between the two systems. Magma erupted from the Nishiyama volcano seems to have plagioclase accumulation after crystal fractionation of plagioclase and mafic minerals possibly in a shallow crustal magma chamber. Magmas erupted in the Hachijo NW chain, however, experienced much less crystal fractionation in the crust during ascent. They erupted without plagioclase accumulation. The observed difference in their magmas precludes the possibility that the Hachijo NW chain magma was separated and transported from the magma system of Nishiyama volcano before the Nishiyama magma was affected by crystal fractionation and plagioclase accumulation.
T53A-1397
Compositionally Diverse Magmatism in Torishima Volcano, Izu-Bonin Arc: Implications for the Genesis of Silicic Magma in the Intra-Oceanic Arc
Torishima, located 600km south of Tokyo, is an active Quaternary volcano of the Izu-Bonin arc. The total volume of this volcano is 466 cubic km, making it the second largest volcano within the arc. Due to the inaccessibility of this isolated island, the precise geological and petrological studies of this volcano have only begun recently. The results revealed that this volcano had compositionally diverse magmatism in its volcanic history that ranges from basalt to rhyolite. Volcanic history of this island can be mainly separated into three stages; stratovolcano stage, syn-caldera stage, and post-caldera stage. The stratovolcano stage basaltic lava and dike constructed main body of the island. Then simultaneous to the caldera formation, voluminous andesite to dacite pyroclastic eruptions occurred. In addition, abundant rhyolitic pumice was collected from the submarine slope of the volcano. The post-caldera eruptions are characterized by the large subaerial effusive lava flows of basalt to basaltic andesite magmas. The syn-caldera stage andesite and dacite occur as pumice and volcanic glass. These andesite and dacite are characterized by their high MgO (<5.2 wt.%) and Ni (<48 ppm) contents and their disequilibrium phenocrysts assemblages. From petrological and geochemical evidences, these andesite and dacite were produced by the magma mixing of two different end-member magmas; an unevolved basaltic magma of high-Fo ol, high-Mg cpx, and high-An pl phenocrysts, and a rhyolitic magma that have anhydrous phenocryst assemblage of low-Mg cpx, low-An pl, opx, and titanomagnetite. The basaltic end-member magma has similar geochemical and petrological characteristics as to the stratovolcano stage basalt, but has more unevolved composition. The rhyolitic end-member magma has comparable composition and petrography as to the submarine rhyolitic pumice. Tamura and Tatsumi (2002) proposed that a dehydration melting of the hydrous tonalitic middle crust as a possible origin for the rhyolitic magma in the intra-oceanic arc. If so, the rhyolitic end-member magma of Torishima may represent the partial melt of the tonalitic middle crust. To test this hypothesis, dehydration melting experiments of tonalite (SiO2=61wt.%) was done at 3 kb, 900-1000C. The partial melts produced in these experiments have equivalent compositions and residual phase assemblage as to the rhyolitic end-member magma. Other previously proposed processes to produce rhyolitic magma in the intra-oceanic arc, such as partial melting of the mafic lower crust or fractional crystallization from the basaltic magma, are insufficient to explain the major and trace element compositions and anhydrous phenocryst assemblage of the silicic end-member magma. The silicic magmatism of the Torishima volcano is most likely to have generated from the partial melting of the tonalitic middle crust. The basaltic magma played an important role not only as a heat source for the partial melting but also to produce compositionally diverse silicic magmas through magma mixing processes.
T53A-1398
Across-arc geochemical variation of basalt in the Izi-Oshima to Niijima cross chan, norhtern Izu arc
The foundations of the Izu-Bonin (Ogasawara)-Mariana (IBM) arc date back ~45 Ma and have been subjected to subduction of the Pacific Plate along the eastern rim of the Philippine Sea Plate. The present IBM arc has been active since ~17Ma and volcanic islands are located both along and across the arc. In the northern Izu Arc five cross-chain islands are aligned obliquely across the arc. In order of increasing distance from the arc these are Izu-Oshima (IO), Toshima (TS), Udonejima (UD), Niijima (NJ), and Kozushima (KZ). Basalt lavas occur on IO, TS, UD and NJ and are all dated (or estimated) as late Pleistocene to recent, and are thus well suited for examination of the present across-arc geochemical variation. We have analyzed twenty basalt lavas (SiO2 = 49-51 wt.%, Mg\# = 39-57) from these islands for major and trace elements, and demonstrate a number of interesting across arc geochemical variations. Slab fluid signals, as shown by [Pb/Ce]n, [Sb/Gd]n and [Ba/Th]n are highest in the volcanic front at IO (6.6, 10 and 9.8) and decrease towards the rear arc (1.5-1.9, 1.6-18 and 1.8-2.1), consistent with greatest input from the slab fluid signal in the volcanic front basalts. [La/Yb]n is low in the volcanic front IO (0.6 on average). It increases in TS (3.1), whereas decrease through UD (2.6) to NJ (1.6). [Nb/Yb]n and [Nb/Zr]n behave almost same perhaps reflecting different degree of partial melting. However, these ratios do not exactly correlate each other indicating different slab sediment contribution. Source heterogeneity is unlikely because [Nb/Ta]n and [Zr/Hf]n remain broadly constant over the cross-chain. Overall, slab fluid contribution is maximum in the volcanic front and decreases abruptly in the rear-arc. Apparent degree of partial melting is lower in the rear-arc and contribution of slab sediments may also play a role in subtle chemical variation found in rear-arc basalts. Interpretations based on trace elements will be tested by analyzing Sr-Nd-(Hf)-Pb isotopes.
T53A-1399
Co-existing wet and dry basaltic magmas at Torishima volcano, 100 km south of Sumisu caldera, Izu-Bonin arc; implications for arc magma genesis and crustal evolution
Basalts erupted from Sumisu caldera, Izu-Bonin arc, include both highly depleted, wet basalts and moderately depleted, dry basalts, which were defined as low-Zr basalt and high-Zr basalt, respectively (Tamura {\it et al.,} 2005). Higher degrees of partial melting of the mantle source are closely related to higher water contents of the basalts. Tamura {\it et al.} (2005) suggested that within a single volcanic complex, the water content in the mantle wedge is heterogeneous, resulting in different degrees of partial melting of this source. This was the first documented example of co-existing wet and dry basalts in a single volcano from the Izu-Bonin arc, and it remains unclear if this feature is unique to, or even common within, the Izu-Bonin arc. Here, we propose that Torishima Island is a further example of both wet and dry basalts occurring in a single arc volcano, and suggest that this may be a common feature of the Izu-Bonin arc. Torishima (translated into English as bird-island) is only a small island (3 km in diameter), however it is the subaerial exposure of a large submarine volcano (the 466 km3 Torishima volcano), built from depths of > 1000 m below sea level. The 1902 eruption of Torishima killed all 125 inhabitants, and the island has been uninhabited since 1965, when the weather station was evacuated due to an earthquake swarm in 1965 and 1966. Torishima volcano is situated approximately 100 km south of Sumisu caldera, with no arc volcanoes between the two. Furthermore, ODP Site 788, located on the volcanic front between the two volcanoes, found no proximal lithologies such as lava flows or agglutinates that indicate any submarine volcanic centres at least back to Pliocene times. Thus, a comparative study of Sumisu and Torishima volcanoes is important to determine the temporal dimension of along-strike variations in mantle composition and crust production rates, and to clarify the role that hot fingers (Tamura {\it et al.,} 2002) might play in the production of arc magmas and crustal evolution. Torishima Island is comprised of three groups of basaltic lavas, distinguished by small, but clear gaps in SiO2 content (< 50 wt. %, ~51-53 wt. %, and ~54.5 wt. % SiO2). Using FeO*/MgO as proxy for differentiation, the basalts form two trends distinguished by their different K2O and Zr contents and Sr/Zr ratios at the same FeO*/MgO. Torishima Island basalts with SiO2 content < 50 wt %, and basaltic rocks with > 50 wt % SiO2 are comparable to the low-Zr and high-Zr basalts of Sumisu caldera (Tamura {\it et al.,} 2005), respectively. All Torishima volcano basalts are strongly depleted in light rare earth elements (LREE), compared with middle and heavy REE. Importantly, low-Zr basalts at Torishima Island are more LREE depleted than high-Zr basalts, a feature also observed at Sumisu caldera. Our results for Torishima Island basalts suggest that wet and dry basalts co-existing within a single volcanic complex could be a common feature of Izu-Bonin arc volcanoes . A mantle diapir model (Tamura {\it et al.,} 2005) could explain the petrogenesis of both high-Zr and low-Zr basalts at both Torishima Island and Sumisu Caldera.
T53A-1400
Pb isotope systematics of the Izu-Bonin-Mariana arc: revelations from double spike Pb measurements
The Izu-Bonin-Mariana arc volcanoes include a range of isotope compositions that encompass the global diversity recognised in intra-oceanic arcs. Lead isotope ratios of the IBM arc almost span the range of MORB compositions but have previously been recognised as being deflected towards subducted material, and hence promoted as a tracer for sediment input to the volcanic arc. However, interpretation of the composition and proportion of subducted material has been hampered by imprecise Pb isotope analysis. In this study we present new high-precision double spike Pb isotope data from the length of the Izu-Bonin-Mariana arc. The northern section of the Izu-Bonin arc forms tight Pb isotope trends toward the local Pacific Plate sediment which has high Δ 7/4 and 8/4 and low $^{206}$Pb/$^{204}$Pb (~18.6). Further south, between 30$^{o}$-25$^{o}$N, the Pb compositions change dramatically and progressively vector to a low Δ 7/4 and 8/4 and high $^{206}$Pb/$^{204}$Pb (~19.8) composition similar to HiMu. The 25$^{o}$-21$^{o}$N section has a similar HiMu trend, but with higher Δ 7/4 and 8/4 for a given $^{206}$Pb/$^{204}$Pb. The Mariana section, 21$^{o}$-16$^{o}$N, has little evidence for the high Δ 7/4 sediment component recognised in previous studies but indicates the addition of a HiMu-pelagic sediment mix composition. Generally, the Pb isotopes of individual volcanoes are found to have trends coincident with the section of the arc in which they are located. This suggests that the process that generates the heterogeneity within the arc also operates on a local scale. The progressive change to higher $^{206}$Pb/$^{204}$Pb towards the south of the IBM arc can not be linked to a variation in the mantle wedge, which remains at low $^{206}$Pb/$^{204}$Pb (17.8-18.2) Indian Ocean-like compositions. To generate the southward trend requires the subducted material providing Pb for the arc to change sympathetically. Measurements of the isotope composition of the main components within the slab (pelagic and volcaniclastic sediments, seamounts and ocean crust) indicate a viable mixing models can be constructed to represent the southward variation. On the scale of each section of the arc a close correspondence is recognised between the bulk composition of the subducting plate and the overriding arc.
T53A-1401
Decompression Basaltic Volcanism in the Lzu Arc Extensional Zone
Current understanding of the sources of Izu arc volcanism is largely based upon studies of the Quaternary volcanoes of the volcanic front and the 12-3 Ma backarc western seamount chains. Th/La ratios increase, and Nd and Hf isotope ratios decrease across the arc from the volcanic front to the backarc, suggesting more subducted sediment in the backarc source. Evidence of slab-derived fluid decreases across the arc. These contrasts are both temporal and spatial. The greatest amount of sediment implied for the Izu backarc is less than the least for the Mariana volcanic front if the initial mantle were the same. The Izu volcanic front source has the least sediment of all NW Pacific arcs. Basaltic volcanism within the ~100 km wide extensional zone between the volcanic front and western seamount chains has been widespread along the full length of the arc since 3 Ma. Its mantle source is less affected by subducted sediment than in the rear-arc seamounts, and is less affected by slab-derived fluids than the volcanic front. This is particularly true of the active (<1 Ma) rifts a mere 25 km west of the volcanic front. Such rifts are characterized by lower Pb and Sr isotope, Th/La, and Hf/Sm ratios, and higher Nd and Hf isotope ratios, than in either the volcanic front or backarc seamount chains. Along-strike and across-strike chemical variations within the extensional zone are minor and apparently unrelated to extrapolations of the seamount chains. Decompression melting minimally affected by subduction seems to have dominated uniformly within the zone since the inception of rifting at ~3 Ma.
T53A-1402
The petrological study on the dredged volcanic rocks from forearc seamount along Ogasawara-Mariana forearc
Hahajima Seamount is one of forearc seamount located at about 20 km west of Ogasawara Trench, and about 110 km southeast of Ogasawara Islands.Serpentinized ultramafic rocks, mafic rocks, metamorphic rocks, and volcanic rocks were dredged from this large seamount. The dredged volcanic rocks during KH03-3 and KT04-28 can be subdivided into two types: boninitic series and tholeiitic series. The boninitic rocks include boninite, boninitic andesite-dacite and rhyolite. Some boninitic andesites show typical adakitic or bajaitic chemical characteristics, TiO2=0.3; FeO*/MgO=0.5; Sr 500-600 ppm; Y 7-8 ppm. The boninitic rocks are similar to the nearby Chichijima Island type volcanic rocks. The tholeiitic series include aphyric basalt and phyric basaltic andesite, bulk compositions of these rocks are similar to anther neaby Hahajiam Island type island arc tholeiitic rocks :TiO2=0.5; SiO2 = 50-55, and a few basalts show chemical characteristics similarly to MORBs. There is no petrological evidence to show any relation between these two types of magma in Hahajiam Seamount. To explain the association of boninitic and tholeiitic volcanic rocks in the Hahajima Seamount, it is anticipated that there had been two distinct magma sources at different time interval in this area. At an early stage of subduction, may be the mantle wedge was so much hot that the subducted oceanic crust and wadge mantle were partly melted at a shallower depth, and thus adakitic and boninitic magmas were derived in this special geo-tectonic environment,respectively. After the formation of boninitic magma stage, the subducted oceanic crust became cold, the tholeiitic magma was derived at a deeper depth where low temperature mantle wedge produce common island arc magmatism. The dredged volcanic rocks from other forearc seamount in Mariana during KH05-1-5 will be also discussed.
T53A-1403
Seismic Stratigraphy of the Mariana Forearc Sedimentary Basin
A grid of seismic reflection profiles across the Mariana forearc between 14N-18N reveals a sedimentary basin between the Oligocene-Miocene frontal arc and the Eocene outer forearc highs. We identify and correlate several seismic stratigraphic units and use them to constrain the local and regional tectonics, which vary significantly from north to south. Four major sediment packages are distinguished in the southern forearc basin. The oldest unit, U-4, is conformable to arcward-tilted, rotated fault blocks formed during early extension, possibly associated with early Oligocene rifting prior to Parece Vela Basin spreading. Onlap relationships between the oldest sedimentary units indicate that deposition occurred before, during and after block rotation. On one profile, the U-4 sequence is deformed above a blind thrust fault in an otherwise extensional environment. Sediments that comprise the third unit, U-3, thin trenchward and onlap onto U-4. U-2 sediments onlap both sides of the basin and are characterized by nearly uniform thicknesses across the southern section. They currently dip trenchward, but are bypassed and onlapped arcward by thin recent deposits, U-1, on the three southern lines, suggesting recent relative subsidence of the outer forearc. The onset of this subsidence (during deposition of the upper strata of U-2) may have generated slope instability that triggered a large submarine slump off the frontal arc high into the forearc basin ENE of Saipan. The seismic stratigraphic units reveal both pre- and post-slump depositional boundaries including a possible post-slump debris apron around the perimeter of the toe thrust. The central region (near 16N), absent of the large rotated basement fault blocks found in the south, is characterized by high-angle normal faults that offset the seafloor by as much as 200 m. The upper section of U-4 is visible in isolated sections, but the coherency of the oldest layers is lost. Because a clear basement reflection is not resolved in this area, it is uncertain whether the absence of the oldest sediment reflections represents a lack of deposition or the limits of our imaging capabilities. The basin stratigraphy reveals a northward thickening of U-2 and U-3, indicating greater extension and increased sediment supply in the central region during deposition. U-1 is absent suggesting that the large relative subsidence of the outer forearc is restricted to the southern region. The stratigraphy of the northern forearc basin (near 18N) is interrupted by several local basement highs. U-4 and the lower sediments of U-3 are not imaged in this area. The upper strata of U-3 are resolvable in small basins formed between local highs. Above this, U-2 comprises most of the coherent basin fill. Ongoing work seeks to correlate these sequences with dated cores drilled in the area at ODP Leg 60 Sites 458 and 459.
T53A-1404
Imaging the Subducting Pacific Plate Beneath the Mariana Forearc
Three pairs of multi-channel seismic (MCS) profiles near 15N, 16.5N and 18N image the subducting Pacific Plate from the outer trench rise to beneath the Mariana forearc. The subducting oceanic crust and superposed seamounts are deformed by plate flexure and cut by the resulting normal faults. Pacific Plate Moho is imaged 2.1 seconds TWTT below basement on the northern profile. The Mariana Trench deepens from about 7.5 km near 18N to 9 km at 14.5N. Beneath the outer forearc we observe a prominent, low frequency reflection from the top of the down-going plate. The slab reflection is strongest and most continuous on the southern profile where it can be traced 70 km arc-ward from the trench axis. The subducting slab is also imaged on 8 shorter crossing profiles. Teleseismic earthquake data poorly constrain the location of the slab in the outer forearc because of large depth uncertainties. To better determine the shallow slab geometry we depth converted four MCS profiles post-migration. For the range of inferred basement velocities of 5-6.5 km/s the slab dips an average of 7-10 degrees and reaches depths of 18-22 km in the south, 70 km from the trench axis. Serpentinite seamounts on the lower slope of the outer forearc obstruct the slab reflection. A subducting seamount is imaged as an out-of-plane reflection between the trench axis and the southeast flank of Big Blue serpentinite seamount. In areas where water depths are less than 4 km, the slab reflection is also obscured by the seafloor multiple. We used multiple suppression techniques such as Radon Velocity Filter, dip filters, and bottom mutes to enhance the slab reflection. These MCS sections image, for the first time, the geometry of the subducting Pacific Plate beneath the outer Mariana forearc.
T53A-1405
Seismicity of the Shallow Thrust Zone and Celestial and Big Blue Serpentinite Seamounts in the Northern Mariana Arc From Ocean Bottom Seismographs
The Northern Mariana forearc is the only arc with large serpentinite seamounts and is also unusual in that it lacks large (Mw > 7) shallow thrust zone earthquakes. During June, 2003 to May, 2004, three ocean bottom seismometers (OBS) were located in triangular arrays around both Big Blue and Celestial seamounts. These ocean bottom seismographs were used along with approximately forty other OBSs and twenty land stations deployed throughout the rest of the Mariana arc to study seismicity associated with the seamounts as well as the seismicity of the shallow thrust zone and subducting plate beneath the seamounts. All six of the OBSs at the seamounts operated successfully during the first 50 days of the deployment, providing high resolution earthquake locations during this time period. The results indicate a lack of significant seismicity within or directly beneath the seamounts, and suggest that the seamounts are located above a relatively aseismic portion of the subduction zone. A few well-located events in the vicinity of the seamounts suggest that the shallow thrust zone is located at a depth of 15-25 km beneath the seamounts. Patches of extremely high seismicity in the shallow thrust zone are located to the west of both seamounts at depths of 30-40 km. The patches are approximately 20 km in width and suggest extremely heterogeneous faulting properties along the shallow thrust zone. Another zone of earthquakes at depths of about 60 km initiate immediately to the west of the seamounts and extend towards the west, representing faulting within the down-going plate. This zone of seismicity appears to be continuous with the lower zone of the Mariana double seismic zone. We suggest that the relatively aseismic character of the outer forearc of the Mariana islands results from large-scale serpentinization of the outermost mantle wedge, as serpentinite is characterized by stable sliding rather then stick-slip behavior. In addition, the lack of large subduction zone thrust earthquakes results from the extreme heterogeneity of the shallow thrust interface, as revealed by the patches of high seismicity and aseismicity at depths of 30-40 km.
T53A-1406
Earthquake Swarms Detected by the Mariana SUBFAC Seismograph Deployment: Submarine Volcanic Activity and Tectonic Deformation of the Mariana Microplate
The 48 ocean bottom seismographs (OBS) and 20 land seismographs deployed from June 2003 to May, 2004 for the Mariana Subduction Factory Imaging Experiment recorded four significant swarms of seismicity in the Central Mariana Arc. Earthquakes were detected using the Antelope database and located using the dbgenloc program. A large swarm of 128 events occurred between East and West Diamente Seamounts from mid-July to mid-August of 2003. The best-located earthquakes form an east-west trending feature on the western slope of East Diamente seamount extending to mid-Diamente seamount. Active hydrothermal activity was observed on East Diamente during an April, 2004 NOAA cruise. The location of these earthquakes and their tight temporal grouping indicates a possible submarine eruption or magma movement associated with East Diamente submarine volcano. Another possibly volcanogenic swarm consisting of 44 events was located on the northwest slope of West Sarigan seamount from December, 2003 to January, 2004. West Sarigan is a submarine volcanic cone and the location of this swarm ~ 10 km northwest of the summit suggests possible volcanic activity. Two other swarms are located away from regions of known volcanism and probably reflect recurrent seismicity associated with deformation of the Mariana arc. A strong swarm composed of 25 events occurred in the Mariana backarc in a region of relatively featureless bathymetry between Sarigan Island and the backarc spreading center during August, 2003. The swarm earthquakes define an E-W lineation that is not due to earthquake mis-location. An August, 2005 Mw 5.4 earthquake at this location has a CMT solution indicating N-S compression on E-W trending faults. The final swarm, during October, 2003, is located on the forearc rise to the east of Sarigan and consists of 45 events. The earthquake locations suggest a southwest-Northeast trending fault. A Mw 5.2 event occurred at this location in 1990; its CMT solution shows northeast-southwest extension. These final two swarms indicate a high rate and complex pattern of internal deformation of the Mariana microplate between the Mariana trench and the backarc spreading center.
T53A-1407
Long Term Seismic Observation in Mariana by OBSs : Double Seismic Zone and Upper Mantle Structure
In order to obtain the deep arc structural image of Mariana, a large-scale seismic observation by using 58 long-term ocean bottom seismometers (LTOBS) had been performed from June 2003 until April 2004, which is a part of the MARGINS program funded by the NSF. Prior to this observation, a pilot long-term seismic array observation was conducted in the same area by using 10 LTOBSs from Oct. 2001 until Feb. 2003. At that time, 8 LTOBSs were recovered but one had no data. Recently, 2 LTOBSs, had troubles in the releasing, were recovered by the manned submersible (Shinkai 6500, Jamstec) for the research of the malfunction in July 2005. By using all 9 LTOBS's data, those are about 11 months long, hypocenter determination was performed and more than 3000 local events were found. Even with the 1D velocity structure based on the iasp91 model, double seismic zones and a systematic shift of epicenters between the PDE and this study were observed. To investigate the detail of hypocenter distribution and the 3D velocity structure, the DD inversion (tomoDD: Zhang and Thurber, 2003) was applied for this data set with the 1D structure initial model except for the crust, which has been surveyed by using a dense airgun-OBS system (Takahashi et al., 2003). The result of relocated hypocenters shows clear double seismic zones until about 200 km depth, a high activity area around the fore-arc serpentine sea-mount, the Big Blue, and a lined focuses along the current ridge axis in the back-arc basin, and the result of the tomography shows a image of subducting slab and a low-Vs region below the same sea-mount mentioned. The wedge mantle structure was not clearly resolved due to the inadequate source-receiver coverage, which will be done in the recent experiment.
T53A-1408
Crustal Growth of the Izu - Ogasawara (Bonin) - Mariana Oceanic Island arc Inferred from Seismic Velocity Structures
It is said that ancient earth consists of the basaltic materials without granitic ones, despite of current large continents has thick granitic materials. Because the oceanic island arc has andesitic middle crust despite of the initial arc has no andesitic materials, the arc is regarded as one of steps of crustal evolution to a continent. Suyehiro et al. (1996), Takahashi et al. (2004) and Crawford et al. (2003) found thick andesitic middle crust with velocity of 6 km/s in northern Izu, Mariana, and Tonga arcs, respectively. However, the central and the eastern Aleutian arcs have little or no andesitic middle crust (Holbrook et al., 1999). To understand the nature of the crustal growth, we have to clarify not only the andesitic middle crust but also the lower crust and the upper mantle, because repeated crustal differentiation may occur within the middle and lower crusts. In 2005, JAMSTEC performed an active seismic experiment across the southern Ogasawara (Bonin) arc using an airgun array and 110 ocean bottom seismographs. A seismic main line runs perpendicularly across the Eocene arc of initial oceanic island arc (the Ogasawara ridge), the current active arc with non-mature and the old Miocene arc. From the preliminary results, the three arcs have similar crustal thickness. However, these crustal components seem to be different each other. The Eocene arc has relative thin middle crust and thick lower crust. The current active arc has thin middle crust and thick lower crust with high velocity of over 7 km/s. The Miocene arc has thick middle crust similar to the northern Izu arc. The Ogasawara trough, which is regarded as a part of Oligocene arc just before the backarc opening, has a crustal thickness of 15-20 km. Moreover, according to reflection imaging recorded by OBSs, we clarified the large fault between the current active arc and the Ogasawara trough, which cuts entire of the crust. This fault suggests that a process of the initial stage of the backarc opening might follow the simple shear mechanism. Using above three velocity models within this arc, we discuss crustal transformation accompanied to the crustal growth to the upper mantle.
T53A-1409
Crustal Structure of the Southern Kyushu-Palau Ridge, the Other Half of the Proto Izu-Bonin-Mariana Island Arc
The Kyushu-Palau Ridge (KPR) is a bathymetric high extending north-south direction at the center of the Philippine Sea and considered as a remnant of a proto Izu-Bonin-Mariana island arc separated by the backarc spreading of the Shikoku and Parece Vela (Oki-No-Tori-Shima) Basin. We conducted a wide-angle and multi-channel seismic experiment to investigate variation in crustal structures along the KPR. The experiment consisted of four seismic lines that were selected to represent the variations in seafloor topography of the southern KPR. All of the profiles cross the KPR perpendicularly and their profile lengths range from 175 to 375 km. The controlled seismic source was a tuned array of 36 airguns with a total volume of 8,040 inch$^3$. In the wide-angle seismic survey, we shot the airgun array at an interval of 200 m (90 s) for each line. We used 200 ocean bottom seismographs (OBS) at an interval of 5 km as receivers. The travel time and amplitude data obtained by the OBSs were modeled by a tomographic inversion and two-dimensional ray tracing. The maximum crustal thickness beneath the KPR varies from 14 to 20 km according to the profile and the crust is significantly thicker than those of the both sides of each profile, that is, the oceanic crusts of the West Philippine Basin to the west and of the Parece Vela Basin to the east. The thickest crust among the four profiles is found in the region where the KPR connects with the Oki-No-Tori-Shima island. The thicker crust beneath the KPR is mainly due to a fat lower crust. The thick (> 5 km) middle crust with P wave velocity of 6.0-6.3 km/s that characterizes the northern Izu-Bonin island-arc crust does not exist so clearly in our profiles. This may relate to the fact that the southern KPR is deeper in seawater and does not have a mature island arc crust compared with the northern KPR where existence of a 6 km/s layer has been reported previously.
T53A-1410
Modeling the role of back-arc spreading in controlling 3-D circulation and temperature patterns in subduction zones
Subduction of oceanic lithosphere provides a dominant driving force for mantle dynamics and plate tectonics, and strongly modulates the thermal evolution of the mantle. Magma generation in arc environments is related to slab temperatures, slab dehydration/wedge hydration processes and circulation patterns in the mantle wedge. A series of laboratory experiments is used to model three-dimensional aspects of flow in subduction zones, and the consequent temperature variations in the slab and overlying mantle wedge. The experiments utilize a tank of glucose syrup to simulate the mantle and a Phenolic plate to represent subducting oceanic lithosphere. Different modes of plate sinking are produced using hydraulic pistons. The effects of longitudinal, rollback and slab-steepening components of slab motions are considered, along with different thicknesses of the over-riding lithosphere. Models look specifically at how distinct modes of back-arc spreading alter subduction zone temperatures and flow in the mantle wedge. Results show remarkably different temperature and circulation patterns when spreading is produced by rollback of the trench-slab-arc relative to a stationary overriding back-arc plate versus spreading due to motion of the overriding plate away from a fixed trench location. For rollback-induced spreading, flow trajectories in the wedge are shallow (e.g., limited upwelling), both the sub-arc and back-arc regions are supplied by material flowing around the receding slab. Flow lines in the sub-arc wedge are strongly trench-parallel. In these cases, strong lateral variations in slab surface temperature (SST) are recorded (hot at plate center, cool at plate edge). When the trench is fixed in space and spreading is produced by motion of the overriding plate, strong vertical flow velocities are recorded in the wedge, both the shallow sub-arc and back-arc regions are supplied by flow from under the overriding plate producing strong vertical shear. In these cases SSTs are nearly uniform across the plate. Results have implications for geochemical and seismic models of 3-D flow in subduction zones influenced by back-arc spreading, such as the Marianas.
T53A-1411
Seismic Anisotropy and Mantle Flow Across the Mariana Subduction System
We investigate seismic anisotropy across the Mariana subduction system with local and teleseismic shear wave splitting analysis. Data from 48 ocean bottom seismographs and 20 land stations deployed by the Mariana SUBFAC imaging experiment provides an unparalleled opportunity to study anisotropy and mantle flow in an island arc-backarc system. We employ both the minimum eigenvalue and cross-correlation methods to solve for the optimal fast polarization direction and time lag. We also use a hierarchical cluster analysis method to choose the most stable window length (Teanby et al., 2004) over which to compute the best fit splitting parameters. Subsequent visual inspection of hodograms ensures a robust solution. Observed fast directions are highly dependent on event location and depth, suggesting that the anisotropy pattern of the arc is highly heterogeneous. The northern part of the arc system shows approximately arc-parallel fast directions for intermediate depth events and a northwesterly rotation with increasing event depth. Deep events may be highly influenced by anisotropy in the slab, but intermediate depth events should provide unbiased indication of the anisotropy in the mantle wedge. Fast directions surrounding the spreading center also show ~N-S orientation, whereas raypaths sampling west of the spreading center show NW-SE and E-W fast directions, or approximately parallel to the absolute plate motion (APM) of the subducting Pacific Plate. Fast directions at mid-latitude island arc stations are also sub-parallel to APM. The southern part of the arc shows small splitting magnitudes and exhibits a variety of fast directions, with variable orientations south of Guam, roughly APM-parallel near Guam, and ~NNW north of Guam. Application of spatial averaging techniques (Audoine et al., 2004) suggest the overall pattern of fast directions is predominantly along-strike beneath the arc, changing to APM sub-parallel to the west of the backarc spreading center. The overall pattern of arc-parallel fast directions in the arc and APM parallel fast directions beyond the backarc spreading center could be due to either the predominance of along-strike flow near the slab and slab induced counterflow in the far backarc, or to a different slip system of olivine near the water rich slab (Jung and Karato, 2001).
T53A-1412
3-D Seismic Tomography of the Mariana Mantle Wedge From the 2003-2004 Passive Component of the Mariana Subduction Factory Imaging Experiment
High-resolution 3D P-wave velocity structure of the Mariana island-arc and back-arc is determined from body-wave arrival times recorded by 48 ocean bottom seismographs and 20 land based seismometers deployed from June 2003-May 2004. We use ~10,000 P-wave arrival times from ~500 of the largest local events that occurred during the recording period to invert for the seismic velocity structure. Outgoing teleseismic arrival times reported by the PDE bulletins are also used to further constrain the structure and earthquake locations. Additionally, incoming teleseismic arrival times from large teleseismic events are determined using multi-channel cross correlation, providing travel time data that is insensitive to local earthquake locations. An iterative procedure was used during the inversion, in which earthquakes are relocated and rays are retraced through the structure between each iteration via a bending method. Preliminary results reveal some interesting features that will provide constraints on geodynamic processes of the Mariana subduction factory. The slab anomaly is resolved down to depths of 600 km. A slow velocity anomaly of ~8% is present within the upper 150 km beneath the Mariana back-arc spreading center. Slow velocity anomalies are also found within the upper 100 km beneath the active volcanic arc and in the uppermost fore-arc mantle. Slow upper mantle velocities in the uppermost fore-arc could be indicative of widespread serpentinization of the mantle. The deeper portion of the forearc (50-100 km depth) shows fast velocities, indicating cold, unserpentinized mantle.
T53A-1413
Rayleigh Wave Phase Velocity Structure of the Mariana Mantle Wedge From a Combined Array of Land Stations and OBSs
We study the shear velocity structure of the upper mantle in the Mariana mantle wedge using Rayleigh wave phase velocities. We obtain interstation and average phase velocity curves from vertical component Rayleigh waves recorded at 10 broadband land stations and 47 ocean bottom seismographs deployed from June 2003 to May 2004. The ocean bottom seismographs show Rayleigh waves with excellent signal to noise ratios at periods from 15-100 seconds. To calculate the phase velocities we cross correlate the phase-matched filters for each station pair. The use of the phase-matched filter is advantageous because it allows for the isolation of the fundamental Rayleigh mode without the contamination of multiples and increases the signal-to-Noise ratio. We use array analysis to determine the optimum back-azimuth correction for each event to correct for wavefront distortion from structure along the teleseismic paths outside the array. We then apply the back-azimuth correction to each station pair to determine the interstation phase velocity. Individual phase velocity measurements for pairs of stations are then used in a tomographic inversion to determine the phase velocity structure of the backarc at various periods. We use the average phase velocity across the array from events with a variety of back-azimuths to examine azimuthal anisotropy in the region. Preliminary results resolve slower velocities in the backarc and faster velocities in the forearc, as expected. We also investigate the use of noise correlation methods to determine Rayleigh wave phase and group velocities. This method may offer great advantages for short term deployments of broadband sensors since Rayleigh wave Green's Functions can be determined from noise correlation of only several months of data. Preliminary analysis shows that Rayleigh wave Green's Functions can be observed from noise correlation for many OBS and land station pairs.
T53A-1414
Asymmetric Seafloor Spreading along the Central Mariana Trough: Insights from Sidescan Sonar and Multibeam Bathymetry Data
The Mariana Trough is one of active and slow-spreading (2~4 cm/yr) backarc basin. The basin exhibits remarkable bow-shape and extremely asymmetry geometry. Spreading axes of the trough are distributed near the Mariana volcanic arc, eastern part of the basin rather than the centerline. To elucidate a seafloor spreading styles of the Mariana Trough, the KR03-12 cruise of R/V {\it Kairei} was carried out along the spreading axes of the central Mariana Trough October to November in 2003. The main tool of the cruise was WADATSUMI sidescan sonar system operating at a frequency of 100 kHz for high-resolution observation of the seafloor. Detailed observation of the sidescan sonar imagery exhibits that two direction of faults and fissures, and related hummocky ridges within axial valley are developed on axial valley floor. One group trends parallel to the axial valley wall, and another group trends ~15° oblique to the valley wall. The obliquely trending group is obviously underlain by the parallel-trending one, and developed only on the eastern half of the axial valley. On the other hand, multibeam bathymetry data obtained during our cruise exhibit the highest backscattering strength at the eastern off-axis of the spreading axis, indicating the recent volcanic activities. Both of the newest tectonism and volcanism found on the eastern off-axis area indicate that the most active part of the ridge has moved eastward from the centerline of the axial valley. The overview of bathymetry around spreading axes seems very symmetric although some features such as inclination of axial valley walls and spacing of larger faults exhibit asymmetric character. Based on our observation, we propose that the _gridge jump event_h has occurred after MOR-like normal and symmetric seafloor spreading. Based on the distance between the centerline of the axial valley to the most active volcanic zone (detected by multibeam bathymetry data), the distance of the latest jumping event is estimated as 0~5.5 km. Magnetic polarity data [Deschamps et al., 2005] indicate that the jump has occurred at least twice since 0.78 Ma.
T53A-1415
Gravity anomaly across the Yap Trench, Sorol Trough, and southernmost Parece Vela Basin and its implications for the flexural deformation of the lithosphere and regional isostasy
In June 2005, R/V Hakuho-maru (KH05-01-Leg 3) conducted a geological and geophysical survey of the southern tip of the Parece Vela Basin (PVB). The survey also profiled the Yap trench, the Yap arc and back-arc region, and Sorol Trough and collected multibeam bathymetry, gravity and magnetic data. In addition, one multichannel seismic reflection profiling across the Yap trench and two dredge rock samplings in the southwestern PVB were carried out. The shipboard free-air gravity field was measured by ZLS Dynamic Gravity Meter D-004 with calibration ties performed at Ocean Research Institute, University of Tokyo and at Apra Harbor in Guam. The shipboard gravity anomaly data show clear match with those derived from satellite altimetry. Also included in our analysis is the shipboard gravity data previously collected by R/V Onnuri. The Yap trench is unique in that it has a short trench-arc distance (approx. 50 km). This proximity has long been interpreted as feature resulting from a collision of over-thickened Caroline Ridge with the trench. In recent years, however, a new hypothesis has been put forward that such feature can be explained by initiation or rejuvenation of subduction, and that the style of subduction changes between north and south of the Sorol Trough. Our survey also revealed peculiar hook-shaped structures in the southernmost PVB and other evidences for large-scale, complex rotational deformation on the seafloor, whose origin remains unclear at this stage. To better understand the nature of these structures and features across Yap trench, Sorol Trough and in southernmost PVB, we examine the regional isostasy using the recently collected bathymetric and gravity data. The density information is deduced from studies conducted at other subduction systems, including Izu-Bonin Mariana trench, and from our own seismic experiment. Preliminary analysis shows that much of the features may be maintained by the flexural rigidity of the lithosphere, especially near and along the trench. To assess the significance of flexural strength of the crust, we apply various flexural models and see if they can explain the observed bathymetry and gravity. The results of this study may provide new clues that will help us to understand the overall tectonic framework of the region at the boundary between two plates, the Philippine Sea and Caroline plates, and their past interaction.
T53A-1416
Tectonic Evolution of the Southern tip of the Parece Vela Basin
The southern tip of the Parece Vela Basin was mapped using state-of-the-art instruments for the first time. The basin is known as an extinct backarc basin behind the Mariana arc-trench system and has developed from ~26 to 12 Ma. The backarc spreading consists of two stages: early east-west spreading and later NE-SW spreading accompanied by several oceanic core complexes. The remnant spreading center, the Parece Vela Rift, seems to connect the Yap Trench at its southern end (~$12°N) and is not traceable in the southern tip of the basin (9~$11°N) west of the Yap Trench. The evolution of the area seems to be linked to the collision of the Caroline Ridge to the Yap Trench, however no systematic mapping had been done before and the tectonics of the area remained enigmatic. New mapping/seismic reflection/dredging results reveal the complex structure of the area, which cannot be seen in northern part of the basin. Relatively continuous N-S fabrics are found in the northern part of the studied area and these fabrics develops within a V-shaped triangle zone. The short NW-SE abyssal hills offset by the NE-SW fracture zones are recognized in the very narrow area just east of the V-shaped area of N-S fabrics. These fabrics indicate the southward propagation of the N-S trending ridge and following NE-SW opening as same as seen in the northern part of the basin, although the eastern wing of the basin was lost. The western part of the area is completely different from the other part of the basin. The most prominent morphology is en echelon, curved deeps near the Kyushu-Palau Ridge. Two deeps are crescent-shaped and curve towards northward. The northern deep is ~6100 m and the abyssal hills seem approximately perpendicular to the deep. The southwestern extension of the northern deep is a narrow curved rift trending $030° and the rift develops within a topographic high. The southern deep is characterized with voluminous dome, which consists of branched topographic highs. The morphological pattern with curved deeps is very much like those of the Pito Deep in the Easter Microplate and of the Endeavor Deep in the Juan Fernandez Microplate. It is likely that the rotational deformation associated with continuous rift propagation and with some finite broad transform zone is related to the origin of the deeps. The area may be the remnant old lithosphere created before the Parece Vela Basin formation and indicate the robust magmatism in the past.