T43C-01
Seismic structure and crustal evolution in the Izu-Bonin arc - results from active source seismic studies along the volcanic front and the rear arc -
JAMSTEC has conduced intensive active source seismic surveys, which have been widely covered in the Izu- Bonin arc, since 2004. All of those data have been acquired with equal data acquisition parameters (e.g., 200L air-gun, 5km-spacing OBSs, 204 ch streamer cable, a profile length of 500km). This makes it possible to quantitatively compare structural variations obtained along those profiles. In this study, we show seismological constraints of evolution of the arc crust on the basis of the seismic image along the present-day volcanic front: i.e., 1) crust of continental composition having Vp of 6 – 6.8 km/s has been predominantly generated beneath basaltic volcanic centers along the volcanic front, 2) the bulk compositions of the crust does not changed during evolution from the thin Bonin arc crust to the thick Izu arc crust, but represent more mafic (basaltic) than an average composition of typical continental crusts, 3) a process to return mafic to ultramafic lower crustal components to the mantle is required for an arc crust to evolve into a continental crust. In addition, we newly discovered a seismological evidence of a paleo-Izu arc (presumably Oligocene arc) based on recently acquired seismic data in the rear arc. Along arc structural variation in the rear arc is similar to that we found in the present- day volcanic front; i.e., a volume of the 6-6.8 km/s crustal component (continental composition) shows 50-80 km scale variation, which corresponds to the distribution of the basalt volcanoes. Another important finding is that the along arc structural variation in the rear arc is not correlated with the rear arc seamount chains. A remarkable difference between the two arcs is observed in the thinner parts of the 6-6.8 km/s crustal component. A volume ratio of a cruatal component having over 7 km/s (mafic to ultra-mafic composition) between the basalt volcanoes in the present-day arc shows significantly larger than that in the Oligocene arc. A process to predominantly increase such a high velocity crustal component between the basalt volcanoes may be required during the evolution process from the Oligocene arc to the present-day arc.
T43C-02 INVITED
Crustal growth deduced from seismic structure of the Mariana arc-backarc system
We obtained a seismic velocity model of the Mariana arc-backarc system (MABS) based on active-source seismic profiling under the umbrella of the MARGINS program: US-JAPAN collaborative research multi-scale seismic imaging of the Mariana subduction factory. A 2D profile across the Mariana arc (MA), Mariana Trough (MT), West Mariana Ridge (WMR), and the Parece Vela Basin (PVB) was complemented by a 3D refracion study across the MA and a 2D profile along the arc, The major structural characteristics are (1) crustal thickness variations across the MABS (~20 km in the MA, ~17 km in the WMR, ~6 km in the MT and PVB); (2) of the presence of at least some middle crust with velocity of 6 km/s beneath arc regions (MA and WMR); (3) uneven distribution of lower-velocity lower crusts (6.7–6.9 km/s, LVLC) beneath the volcanic front and adjacent to the MT; (4) slow velocities of less than 8 km/s in the upper mantle, under the arc regions; (5) deep reflectors in the mantle with slow velocity of < 8km/s beneath the arc regions, and (6) a high-velocity lower crust (7.2–7.4 km/s) at the boundary regions between the MA and MT. (7) Crustal thickness beneath the MT axis is a few kilometers thicker than that of the surrounding trough region and the clear reflector distributes beneath the axis. Although the bulk composition of the MA and WMR is still basaltic, simple petrologic modelling of our 2D profiles based on an anatexis model suggests that the volume of lower crustal restites and olivine cumulates after extraction of mid-crustal andesite should be significantly larger than observed, suggesting that part of the lower crust, especially cumulates, is seismically part of the mantle. The LVLC, which may indicate advanced crustal growth, located adjacent to the MT suggests that the backarc opening promoted crustal growth and differentiation. The distribution of the low mantle velocities and the deep reflectors, especially beneath the extinct WMR, suggest that their origin could be explained by transfer of lower crustal residues to the seismically defined mantle across the Moho. In addition, the high-velocity lower crust beneath the arc-backarc transition zone is likely composed of mafic/ultramafic materials created by extensive partial melting of mantle peridotites or last stage of the arc magmatism rather than serpentinized peridotites. The crustal thickening and the deep reflectors beneath the MT axis suggests underplating of basaltic magmas and existence of the low velocity region in the upper mantle.
T43C-03
3-D Crustal Structure and Variable Growth of the Mariana Island Arc from Seismic Tomography
A 3-D seismic refraction survey was acquired over the Mariana volcanic arc at 14.5° -18.5° N. First
arrival travel times from this survey, and from a separate 2-D survey acquired approximately perpendicular to the
arc, have been simultaneously inverted for a 3-D P wave velocity model using seismic tomography subject to
smoothness constraints. In the final 3-D tomographic velocity model, the 7.4 km/s isovelocity contour correlates
best with the Moho along the 2-D profile, which was inferred using both first arrivals and wide-angle reflections,
and this contour is used as a proxy for the Moho. The active arc, which initiated only 3-4 Ma ago, has an average
crustal thickness of 18 km. Approximately 40 km to the east, the inactive remnant of the rifted Eocene arc has an
average crustal thickness of 21 km, due primarily to a thicker lower crustal layer with velocities of 6.5-7.0 km/s. P
wave velocities within the upper crust of the active arc are approximately 380 m/s lower than in the remnant
Eocene arc, but are up to 280 m/s higher at a depth of 15 km. These results suggest an evolution of arc structure
with increasing age: we infer closure of fractures and porosity in the upper crust through hydrothermal circulation
and a reduction in the mafic character of the mid-lower crust as a result of intracrustal differentiation. Comparison
of averaged 1-D velocity functions with sonic velocity measurements on samples from the Tanzawa complex in
the Izu collision zone suggests that although tonalitic rocks may exist in the transition from upper to middle crust,
much of the crust is basaltic. Middle crust with velocities of 6.0-6.5 km/s, which comprises 14% of the arc crust, is
best developed beneath the Eocene arc, but varies in thickness between 2 km and 6 km along strike. Crustal
production clearly varies both temporally and spatially. Under the assumption that a 40 km wide corridor along the
modern arc has not been affected by earlier arc magmatism, we estimate that crustal production has been
116 km
T43C-04
Seismic Attenuation Structure of the Mariana Subduction System
Analysis of the anelastic structure of a subduction zone can place first-order constraints on variations in temperature and volatile content. We investigate seismic attenuation across the western Pacific Mariana subduction system using data from the 2003-2004 Mariana Subduction Factory Imaging Experiment. This 11 month experiment consisted of 20 broadband stations deployed on islands and 58 semi-broadband ocean bottom seismographs deployed across the forearc, island arc, and back-arc spreading center. Following Stachnik et al. [2004], we compute the amplitude spectra for P and SH arrivals for a given local earthquake and simultaneously invert for the path-averaged attenuation parameter, t*, for each waveform and the seismic moment and corner frequency for each earthquake. Measurements are taken over a frequency band of 0.05 to ~10 Hz, although poor signal to noise along high-attenuation mantle wedge paths limit observations to < 2-3 Hz in some cases. Tomographic inversion of the ~3000 t* measurements shows low attenuation along raypaths traveling through the subducting Pacific plate. A high attenuation anomaly beneath the island arc extends to ~60-80 km depth and is distinctly separated from the higher attenuation anomaly beneath the backarc spreading center. The westward-trending asymmetric backarc anomaly extends to ~100 km depth. We interpret the arc and trough anomalies as due predominantly to increased temperatures. Further exploitation of attenuation-temperature relationships will elucidate which anomalies are thermally controlled versus which are controlled by other mechanisms, such as the presence of volatiles or partial melt.
T43C-05 INVITED
Mantle Structure, Melting and Flow in the Nicaragua-Costa Rica and Izu Bonin-Marianas Subduction Zones
Recent broadband seismometer experiments in the Nicaragua-Costa Rica and Izu Bonin-Marianas subduction zones (the NSF MARGINS Subduction Factory focus sites) have led to enhanced imaging of mantle wedge, slab and upper plate structure. Here we relate results from the TUCAN Experiment (Nicaragua-Costa Rica) to mantle temperature, hydration, flow, melting and melt transport, and compare them to findings from the MARIANA array. Velocity and attenuation tomography based on TUCAN data reveal a high velocity, low attenuation subducting slab, a shallow wedge corner with intermediate attenuation, and a slower, more highly attenuating mantle wedge beneath the arc and back-arc. However, velocity and attenuation structures also contain strong regional variations, and these results correlate with arc geochemical data that suggest higher concentrations of slab- derived fluids and greater extents of melting beneath Nicaragua. Beneath Nicaragua, the shallow slab is slower and more attenuating than the slab beneath Costa Rica, consistent with greater slab hydration. In addition, the mantle wedge at depths of 60-100 km is more highly attenuating in Nicaragua than in Costa Rica. Comparison of shear attenuation values with mantle temperatures inferred from arc magmas suggests that much of the difference in attenuation between the Nicaraguan and Costa Rican wedges can be explained by greater water concentrations beneath Nicaragua. A particularly intriguing finding is a column of high Vp/Vs (P-wave velocity/S- wave velocity) that rises from the slab interface directly beneath the arc in Nicaragua. This anomaly could reflect the presence of melt. In three-dimensional models of anisotropy obtained by tomographically inverting shear-wave splitting measurements from local events recorded by the TUCAN array, olivine a-axes are predominantly arc-parallel in the mantle wedge beneath the arc and back-arc at depths of 50 to 150 km (except in northern Nicaragua). The arc-parallel a-axes extend into mantle wedge well beyond the cold, shallow wedge corner where B-type olivine fabric may occur. The observed anisotropy cannot be explained by simple two-dimensional arc-normal corner flow, and instead suggests significant arc-parallel flow. In SKS splitting measurements, fast directions are roughly arc-parallel, and large SKS splitting times indicate that arc-parallel-fast anisotropy is also present beneath the subducting plate. Comparisons of mantle models between the TUCAN and MARIANA experiments are still underway, but a few initial points are apparent. First, as in Nicaragua-Costa Rica, the mantle wedge beneath the arc in the Mariana system is dominated by anisotropy with a fast symmetry axis parallel to the arc. This result suggests the presence of arc-parallel flow in both subduction zone mantle wedges, despite their many tectonic differences (an older, steeper, more tightly curved slab and the presence of active back-arc spreading in the Marianas). Second, while both subduction zones contain a zone of high attenuation in the wedge beneath the arc, the Mariana anomaly is shallower, and the Mariana wedge also contains a second distinct and deeper volume of high attenuation beneath the active spreading center.
T43C-06
Improving Seismic Constraints on Subduction Zone Geometry
Accurate slab geometries are necessary for 3D flow modeling, and for understanding the variations in temperature and melting geometry between different subduction zones. Recent studies have shown that the depth to slab beneath arc volcanoes varies by as much as a factor of two between subduction zones, but these results are based on teleseismic earthquake catalogs with potentially large errors. When available, local seismic arrays provide better constraints. The TUCAN array (Tomography Under Costa Rica and Nicaragua) deployed 48 three component broadband PASSCAL instruments for 18 months with station spacing of 10-50 km across the Central America arc. This dataset provides some of the best control anywhere for ground-truth comparison of teleseismic catalogs in steeply dipping subduction zones. Joint inversion of TUCAN arrival times for velocity and hypocenters illuminate a 10-15 km thick Wadati-Benioff zone (WBZ), with absolute hypocenter uncertainties of 1-5 km. Besides providing accurate hypocenters, the tomographic images provide independent constraints on melting and temperature, through the imaging of low Vp (7.5-7.8 km/s) and highly attenuating (40<Qs<100) mantle wedge, and the imaging of a high Vp/Vs (≥ 1.78) column inferred to be melt. International Seismic Centre (ISC) and EHB (Engdahl et al., 1998) hypocenters show a teleseismic Wadati- Benioff zone (TWBZ) that lies 15 km below that of the TUCAN catalog on average at 80-200 km depth, with similar results for local catalogs based on a 1D velocity model. However, the width of the TWBZ is 30-80 km or 3-5 times that indicated by TUCAN hypocenters; this additional width suggests errors of +/- 10-33 km. Commonly, the top of the subducting slab is assumed to lie at the top of the WBZ seismicity, for example if double seismic zones are expected. Because of the large scatter, the TWBZ is biased too shallow compared to the TUCAN data, vertically by as much as 50 km for the steeply-dipping Nicaragua slab. Relative relocations of hypocenters from ISC arrival times reduce this scatter to be more consistent with TUCAN hypocenters, with a center of seismicity less than 5 km deeper and a WBZ thickness 2-3 times greater. Thus, in regions with no local array, relative relocations of teleseismic catalogs can provide slab geometries consistent with regional earthquakes. We extend the slab inferred from TUCAN using relative relocations, to generate an accurate 3D description of the slab suitable for high-resolution geodynamic modeling.
T43C-07
Crustal Thickness Along the Central American Volcanic Front
Subduction zone processes alter the upper plate in a number of ways, including accretion, magmatic addition, serpentinization of the mantle wedge and formation of mafic cumulates in the lower crust. All of these changes affect seismic velocities, and characterizing the structure of underlying terranes in Central America establishes a baseline for composition and continental growth. Tomography Under Costa Rica and Nicaragua (TUCAN) is a PASSCAL deployment of broadband seismometers over an 18-month period. The network has two dense cross arc lines and two along arc lines that cross terrane boundaries. Teleseismic P and PP arrivals recorded on the TUCAN network have been used to estimate crustal thickness and Vp/Vs, and to develop receiver function images. Surface reflected mode conversions (Ppms and Psms) enhance resolution. Crustal thickness ranges from 25 to 44 km with formal errors ranging 1.6-9.2 km. The thinnest crust (24.6 +/- 3.5 km) lies directly beneath the arc in Nicaragua, whereas the thickest crust (43.5 +/- 2.5 km) lies in the backarc in Nicaragua and beneath the Costa Rican arc (37.9 +/- 5.2). Changes in crustal thickness and Vp/Vs show two distinct terrane boundaries crossing the arc. Vp/Vs indicate continental crust (Vp/Vs=1.71-1.77) in Nicaragua, with a transition to gabbroic crust (Vp/Vs=1.82-1.88) in Costa Rica where fragments of the Caribbean large Igneous Province have been found. Crustal thickness beneath the arc in Costa Rica yields a crustal growth rate of 16-36 km3/km/Ma, assuming a base crustal thickness of 30-32 km with 6-14 Ma of magmatism. The Moho shows strong velocity contrasts throughout the study area, and is the only interface seen in the backarc, but it is complicated by interferences caused by shallow structure beneath the arc and forearc. Forward modeling indicates that reverberations in sediment layers interfere with the Ps arrival, however surface reflected arrivals (Ppms) require a velocity contrast on the order of 0.5–1.0 km/s at the Moho. Strong velocity contrasts across the Moho indicate that if melt or serpentinization are present in the upper mantle, they have minimal effect on velocities. Initial migration results are also presented.
T43C-08
Results From the TICO-CAVA Land Seismic Refraction Survey
Project TICO-CAVA (Transects to Investigate the Crustal Origin of the Central American Volcanic Arc) is a key part of the MARGINS Subduction Factory initiative to quantify, characterize, and understand the volcanic crust produced in Costa Rica by seismically imaging its volume, extent, seismic properties, and lateral variability. From this information we hope to estimate the major-element composition of the arc crust and draw inferences about the processes controlling volcanic output of the subduction factory in Central America. We will do this with two intimately linked seismic surveys, an onshore explosion refraction survey (2005), and an onshore-offshore airgun survey (2008). Here we present results from Phase I of the project, an onshore explosion seismic refraction survey of the volcanic arc and look forward to phase II, a large onshore-offshore seismic survey in both the Atlantic and Pacific oceans. We acquired seismic refraction/wide-angle reflection data on two lines. Line 1 is 154 km long from the Pacific Ocean to the Caribbean Sea in central Costa Rica, intersecting the main volcanic arc at Volcán Barva. Line 2 is 280 km long and spans the entire length of the active arc in Costa Rica, from north of Volcán OrosÃÆ' near the Nicaraguan border, to south of Volcán IrazÃÆ'º. 748 seismometers were deployed on each transect, resulting in a ~200 m receiver spacing on Line 1 and ~370 m spacing on Line 2. Data quality ranges from fair to excellent. Data on the cross-arc line (Line 1) show upper-crustal refractions that can be tracked from coast to coast and, at the longest offsets, deep reflections that may be Moho reflections. Sediments on the Atlantic coastal plain (2-3 km/s) are about 1.5 km thick; beneath this and beneath the volcanic arc is a low-velocity carapace of volcanic material (3-5 km/s) that is 2-4 km thick. Beneath the low-velocity velocities reach 6.0-6.4 km/s at 5-10 km depth. Deeper structure will become better resolved as analysis continues and the onshore-offshore data are collected. The along-arc line (Line 2) shows evidence for distinctly different crustal structure beneath the Guanacaste volcanos and the Cordillera Central. In addition, low-velocity zones that may represent magma chambers exist beneath Poás volcano and between Irazu and Turrialba volcanoes.