 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
 |
|
||||||||||||||||||
 |
 |
 |
 |
 |
 |
 |
|
||||||||||||
 |
 |
 |
|
||||||||||||||||
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
|||||
 |
 |
 |
 |
 |
 |
 |
 |
|
|||||||||||
 |
NSF-IFREE-MARGINS
Workshop Workshop Report Jim Gill, Simon Klemperer, Bob Stern, Yoshi Tamura, Doug Wiens |
| |
|
Download
this document as a PDF
file (320 KB) or read it online below. |
|
| A workshop on the Izu-Bonin-Mariana (IBM) subduction system was held in Honolulu HI during September 8-12, 2002, under the auspices of the MARGINS Program of NSF. It was co-sponsored by the Japanese Institute for Frontier Research on Earth Evolution (IFREE). Both the USA and Japan have selected the IBM for focused research during the next five to ten years, creating opportunity for joint research activities. Convenors in alphabetical order were J. Gill (Santa Cruz), S. Klemperer (Stanford), R. Stern (Dallas), Y. Tamura (IFREE), and D. Wiens (St. Louis). About 100 scientists attended. About 2/3 were from the USA and most of the rest were from Japan. Many had not worked extensively in the region before, and about 25% were graduate students and postdocs from both countries. There were about twenty invited talks, listed below, about equal time for discussion, and about 50 poster presentations. Abstracts and key visuals from the invited talks are available at this web site. The IBM arc has been selected as the “oceanic cold-subduction” end-member example in which Subduction Factory topics can be addressed effectively with least influence of the upper plate. It is one of two integrated study sites for such projects; Central America is the other. The workshop featured recent research results related to IBM, some of which were sponsored by the MARGINS Subduction Factory initiative. Similarities and differences between the two study sites also were presented, as were comparisons with other oceanic arcs. The meeting started with overviews of the geochemical and geophysical context of general subduction by C. Hawkesworth (Bristol) and M. Gurnis (CIT), and of the IBM arc in particular by R. Stern (UTD) and B. Taylor (UH), respectively. M. Arima (Yokohama) summarized the IBM arc components exposed in the Tanzawa arc-arc collision complex. S. Peacock (ASU) discussed the thermal and flow structure of IBM, and G. Abers (BU) described general slab seismicity and seismological constraints on the dehydration and phase transformations in the downgoing crust. D. Wiens (WUSL) and M. Fouch (ASU) described the seismic tomography and anisotropy of oceanic arcs in general as background to work in the Marianas in 2003. S. Klemperer (Stanford) and K. Suyehiro (JAMSTEC) summarized the crustal structure of the Mariana and Izu sectors, respectively. The second day focused on inputs and outputs to the IBM arc. T. Plank (BU) presented results from ODP Leg 195 for slab inputs, and R. Hickey-Vargas (FIU) summarized mantle inputs from the perspective of back arc basin basalts. P. Fryer (UH), T. Elliott (Bristol), and O. Ishizuka (GSJ) summarized outputs from the forearc, volcanic front, and backarc, respectively. Y. Tamura focused on the abundant felsic volcanic and plutonic outputs of Izu, and J. Ishibashi (Kyushu) summarized IBM’s hydrothermal fluids and deposits, and their geobiology. M. Reagan (Iowa) summarized the history of IBM’s magmatic outputs, and K. Fujioka (JAMSTEC) provided a tectonic overview and estimate of crustal production rates. The third day included reviews by M. Hirschmann (Minnesota) of decompression and flux melting in arcs, and by B. Bourdon (IPG-Paris) of the timescale of such processes from the perspective of U-series disequilibria. P. Kelemen (WHOI) discussed processes generating continental crust in oceanic arcs, especially the Aleutians. S. Schwartz (UCSC) discussed the shallow seismogenic zone and some of the unique observations of that zone in the Izu and Mariana arcs, G. Hirth (WHOI) discussed the effects of water on the material properties of the mantle wedge, and Y. Tatsumi (JAMSTEC) gave an overview of the multiple roles of subduction in crustal and mantle evolution. Summaries were also given of NSF, IFREE, and GEOMAR plans for work on subduction zone processes in IBM and Central America, and the related RIDGE2000 integrated study site in the Lau Basin. Plans for IODP drilling and submersible programs in IBM were presented. The day closed with brief descriptions of about a dozen funded field projects in the IBM system. B. Progress towards Realizing
the Objectives Results were presented from, or in preparation for, all six of the MARGINS Subduction Factory projects that have been funded in the Mariana portion of the IBM arc since 1999. They included >5000 km of multi-channel seismic profiling (Taylor et al.), a 50-OBS wide-angle seismic experiment for crustal structure (Klemperer et al.), passive seismic tomography (Wiens et al.), melt inclusions in phenocrysts (Plank et al.), age dating and geochemical evolution of islands (Reagan et al.), and the southern seamount province (Stern et al.). In addition, results were presented from ODP Leg 195, which drilled sediments and basaltic basement in the incoming Pacific Plate off both the Izu and Mariana arcs. The workshop demonstrated that progress is being made in the IBM focus site towards providing good answers to the questions that motivate the Subduction Factory Initiative: 1) How do forcing functions such as convergence rate and upper plate thickness regulate production of magma and fluid from the Subduction Factory? 2) How does the volatile cycle (H2O and CO2) impact chemical, physical and biological processes from trench to deep mantle? 3) What is the mass balance of chemical species and material across the Subduction Factory, and how does this balance affect continental growth and evolution? In addition, one of the ancillary questions asked in the Science Plan: “How, why and where are new subduction zones started?” seems to be best addressed at this site, and good progress is being made to understand this process as well. Examples of the progress are provided below. 1. How do forcing functions such as convergence rate and upper plate thickness regulate production of magma and fluid from the Subduction Factory? A major difficulty in understanding subduction processes has been inadequate numerical models for temperature and mantle flow in subduction zones. Three new thermal models (Conder et al., Peacock and van Keken, and Kelemen et al.) were presented at this meeting that incorporate temperature-dependent viscosity. These models showed higher slab surface temperatures than previous models, and in addition suggested a possible mechanism for a component of decompression melting beneath the arc. Workshop participants also learned about the first-order subdivision of the IBM forearc into a deforming southern (Mariana) part and a relatively undeformed northern (Izu) part. It seems clear that the abundance of serpentinite diapirs and forearc vents are controlled by this deformation. The lithospheric structure of the upper plate of the Izu segment has been determined by Japanese scientists and a parallel study in the Marianas funded by NSF has begun. A long active source profile across the Mariana arc with 100 ocean bottom seismographs will be completed by Japanese scientists in early 2003. Forcing functions related to mantle flow and possible sequential melting are being investigated by Japanese and US investigators by studies linking magmatic outputs along flow lines from the back-arc basin spreading ridge along cross-chains and into the magmatic front. Waveform inversion results for the structure of the Mariana Trough backarc summarized by D. Wiens suggest higher upper mantle seismic velocities and possibly lower temperatures than in the Lau Basin backarc. No high resolution tomography has been completed in IBM yet but this is one of the goals of the funded passive OBS deployment to commence in 2003. We still do not understand what is causing the first-order variation in magmatic compositions along the arc, from moderately enriched (medium-K calc-alkaline) in the south to extremely enriched (shoshonitic) in the center to ultra-depleted (low-K tholeiitic) in the north. Part of the answer comes from the different types of sediments that are being subducted, as T. Plank showed at the workshop, but this does not explain the observation that Izu arc lavas are higher degree melts than those of the Mariana arc. An important model for explaining why arc melts are so depleted is that they are products of sequential melting, first beneath the back arc, then beneath the volcanic front. This model is difficult to reconcile with the fact that lavas from the Mariana Arc, which are associated with an actively spreading back-arc basin, are less depleted than Izu Arc lavas, which are not. The slab component also seems to differ along strike, with “fluid” and “melt” signatures being combined at the Marianas volcanic front but partitioned between volcanic front and reararc, respectively, in Izu. Workshop participants also saw new GPS results (T. Kato et al.) for the Mariana Arc which radically alter our understanding of how the Mariana Trough is opening, and it is clear from this that GPS studies of plate motion are essential. 2. How does the volatile cycle (H2O and CO2) impact chemical, physical and biological processes from trench to deep mantle? Recent improvements in microanalytical techniques (ion probe, laser ablation ICP-MS, FTIR) have resulted in tremendous advances in understanding element fluxes, especially water, through arc systems, and the IBM system has been the site of many of these advances. Good data sets for water through the Mariana magmatic system are being assembled, and comparable data sets for glass inclusions in Izu ejecta are needed. Initial results suggest higher water contents in Mariana than Izu magmas of similar level of differentiation. The flux of carbon dioxide through the IBM arc magmatic system is controversial, with no good data yet because hydrous magmas lose this gas even at relatively high pressure. Obtaining robust estimates of CO2 flux through convergent margins is a global challenge. Studies of subaerial fumaroles and submarine hydrothermal and forearc vents are needed as part of experiments designed to monitor the fluxes. Estimating sulphur dioxide fluxes is tractable but remains to be widely accomplished. Monitoring SO2 fluxes can be done to advantage using satellite remote sensing, so that involving NASA and NASDA (Japan space agency) in this effort is highly desirable. Measuring the flux of other volatiles (halogens, N, methane, rare gases) and related isotopic compositions is also needed, and these measurements would offer important insights into CO2 flux. Workshop participants saw good examples of how these measurements were progressing in Central America (T. Fisher, D. Hilton, etc.), and this approach could be adapted to IBM. IBM has an additional flux from hydrothermal systems associated with backarc basin spreading system as well as submarine calderas, and techniques developed as part of the RIDGE program could be readily adapted. Japanese scientists are very active in measuring volatile fluxes both at sea and on land and opportunities for interaction exist. Work on how these fluxes affect the biosphere is just getting underway and US investigators may learn a lot from the Japanese ‘Archean Park’ project just getting underway in southern IBM. This project is led by T. Urabe (University of Tokyo) with co-investigators from a number of Japanese institutions including JAMSTEC; US investigators should be encouraged to collaborate. 3. What is the mass balance of chemical species and material across the Subduction Factory, and how does this balance affect continental growth and evolution? Results from ODP Leg 185 were presented by T. Plank, who showed that some of the variation in incompatible elements along the IBM arc system can be related simply to variations in the composition of subducted sediments. IBM is a very good arc system to identify input controls, because the sediments being subducted in the north and the south are distinct. Whether or not this model can explain isotopic variations along the arc remains to be resolved. The magnitude of magmatic fluxes, and whether they vary systematically along the arc, remain important unresolved questions. Because crust formation in arcs is largely vertical, this flux cannot be related to a measurable quantity such as spreading rate. Consequently, it is difficult to measure the magmatic flux directly. Techniques for estimating magmatic flux indirectly (i.e., via SO4 or some other volatile flux) await development. The thickness of the crust, which integrates the arc magmatic flux over the life of the arc, can be measured directly, and good progress is being made in IBM. Japanese scientists have already measured crustal seismic velocities across the Izu arc and interpreted this for crustal structure. The field experiment needed to generate a comparable profile along the Mariana arc has been conducted by S. Klemperer et al. and preliminary results were presented at this workshop. Further complementary work is being planned by JAMSTEC scientists. There was extensive discussion about the significance of the 6.2 km/sec P-wave velocity layer observed in the Izu cross-section, and the likelihood that tonalites exposed in the Tanzawa mountains in the Izu collision zone represent exposures of that layer. This felsic middle crust is similar in some aspects to the composition of continental crust but dissimilar in trace element concentrations and isotopes, being more depleted in both compatible and incompatible elements. The crustal structure and composition of the two intra-oceanic arc systems for which there are high quality crustal velocity profiles, Izu and the Aleutians, are so different as to require more studies to determine the significance of this variability. Workshop participants were intrigued by the likelihood that arc lower crust delaminates and falls back into the mantle, as has been suggested for the Sierra Nevada of California, but understanding this process remains an important challenge for studies of crustal growth. 4. How, why and where are new subduction zones started? Good progress is being made here. The workshop was excited to see geodynamic models developed by C. Hall and M. Gurnis (CIT) which reproduces the most important features of the IBM subduction initiation process. Uncertainties about the paleogeography of the West Philippine Basin continue to complicate our understanding of how subduction began. R. Hickey-Vargas demonstrated that lavas from the West Philippine Basin are similar to those of the Indian Ocean. M. Reagan showed new geochronologic results for proto-arc igneous activity, along with compositional data for these lavas, but results remain too scattered to resolve genetic models. Another important objective of the workshop was to encourage collaboration between US and Japanese scientists working on related problems. Several such joint opportunities were discussed. 1) IODP Drilling in IBM. Several specific ideas were addressed; see below. 2) Comparative field studies. Despite sharing a common history, IBM sectors differ in important ways. For example, slab dip steepens southward, there is backarc rifting in the north but spreading in the south, and “fluid” and “sediment” components are combined in arc magmas in the south but separated across-arc in the north, at least since 15 Ma ago. Along-strike comparisons offer powerful opportunities to evaluate differences in forcing functions, volatile behavior, and mass balances, but require joint planning and sometimes joint deployment of assets. Several joint field programs have already been planned and funded, including a joint passive land-OBS deployment aboard the Japanese ship Kaiyo to commence in the Marianas in 2003 (Wiens and Suyehiro/Shiobara). The US (Klemperer) and Japanese (Suyehiro/Kodaira) active source experiments in the Mariana have been coordinated and should yield complementary information. Further joint field programs were discussed. 3) Numerical model of subduction zones. A subduction modeling workshop, funded by MARGINS and held in Michigan in October, should rapidly increase progress in this area. The Earth Simulator at IFREE, currently the most powerful computer available to the scientific community, is available to run three-dimensional kinematic models of subduction at high spatial and temporal resolution. Participants felt that models are more limited conceptually than computationally at the moment, but joint efforts in this field could change this quickly. C. Topical Summaries The meeting ended with a full day of small group discussions of specific topics to identify what still needs to be done and how to do it, in light of the oral and poster presentations and related interaction of the preceding days. The topics were chosen by participants and are not mutually exclusive. Brief summaries of these discussions follow. In order to maintain the character of discussion, no effort is made to integrate separate sessions. Indeed, themes that recur in multiple contexts reflect wide consensus. Where applicable, these summaries also draw from poster and oral presentations, and plenary discussion, earlier in the workshop. 1. Fluids and melts from the slab. Many presentations summarized efforts to assess the composition of, and differences between, fluids released by dehydration versus partial melts of subducted sediment and basalt. This information is essential to address volatile cycling through subduction zones. Thermal models presented at the meeting for even the cold circumstances of IBM subduction allow for prolonged dehydration and even melting beneath the 300 km-wide subduction system. Participants called for further field, experimental, and analytical studies of metamorphic rocks exhumed from subduction zones, including serpentinites. Relating P-T history to mineral and rock chemistry is essential when using these rocks to understand effects of dehydration on both fluids and restites. Experimental studies of element partitioning into fluids and melts remain sorely needed. Work to date is exploratory and little consensus has been reached apart from order of magnitude differences in behavior between classes of elements. Analyses of melt inclusions in arc volcanics, especially rapidly quenched crystals in scoria, seem the most tractable source of information about volatile contents in melts because arcs and even backarcs are too shallow and volatiles contents are too high for dense glass to form on submarine lava. Experimental study of the roles of volatiles and dehydration on the material properties of mafic and ultramafic rocks is needed to interpret the evidence of seismic anisotropy and embrittlement earthquakes in arcs. 2. Imaging, Modeling, and Experimenting
on Mantle Wedge Processes In order to better understand mantle motions, efforts should be made to map seismic anisotropy, and use this to infer fabrics and thus convection patterns in the mantle wedge. 3D patterns of mantle anisotropy are needed which perhaps can be obtained from integrated studies of shear wave splitting and surface wave dispersion. Further experiments are needed to reveal how shear wave splitting fabrics can be interpreted to reveal mantle fabrics, given a plausible range of water contents. Eventually the IFREE ‘Earth Simulator’ can be used for developing more realistic models of flow in the mantle wedge, and it was recommended that the US and Japanese modeling communities find ways to develop more complex models of mantle flow, which hopefully can be tested with carefully crafted field programs. With respect to understanding fluid motions and distributions, experiments are needed to understand the significance of S-wave velocity anomalies observed in the mantle wedge. Seismic tomography reveals areas of the mantle where seismic velocities are relatively low, but the extent to which this reflects temperature, fluids, or melts is difficult to determine. Patterns of melt and fluid distribution may be distinct but the details are controversial. Fluids released from the slab are probably controlled by essentially 2-D variations of temperature, pressure, composition of the mantle, and kinetics, but the magmatic output is distinctly 3 dimensional. How is the former, sheet-like ‘reverse rain’ converted into the latter, clustering of volcanoes? Is there any way to accomplish this other than Rayleigh-Taylor instability in the melt region? What are the implications of the several thousand-year timescale inferred from 226Ra-disequilibrium for mantle viscosity and melt ascent mechanisms? Related to this, what is the significance of the single positive correlation between H2O content and percent melting (‘Stolper and Newman’ trend)? Does the same correlation apply to arc magmatic systems as to back-arc ones, and if so, what does this tell us about the different ways that these melts evolve? One of the most important – and tractable – issues is how hydrous depleted tholeiites of the Izu arc are. These lavas show trace element characteristics that are consistent with a large fluid flux, such as high Ba/La, but phenocryst melt inclusions need to be analyzed for water contents and trace element compositions in order to resolve this question. Understanding the thermal structure of the mantle wedge is critical for a wide range of geodynamic and geochemical models. Reliable heat flow measurements, ideally transects, are needed in both the Marianas and Izu. This work will not be easy because of the difficulty of inserting probes into volcaniclastic sediments as well as potential problems with thermal transients in the water column. We also need a better understanding of how the mantle wedge immediately above the slab accepts the water released from the slab. Is it injected through fractures and, if so, is there seismic evidence for this? Alternatively, porous flow of fluids from the slab into overlying peridotites should result in extensive serpentinization and chloritization. Can this be identified or discounted using geochemical and geophysical techniques? We need to refine and test models of sequential melting beneath backarcs and then arcs. Is this what is responsible for the large extents of melting inferred from trace element systematics of volcanic front lavas? Finally, what is the distribution of Indian Ocean versus Pacific type mantle beneath the IBM arc system, and how can that distribution be used to distinguish wedge from slab fluxes? 3. Time scale of dehydration and melting processes Several oral and poster presentations dealt with the implications of U-series disequilibria in volcanic rocks for processes at IBM and elsewhere. The Marianas volcanic front provides one of the classic “apparent U-Th” isochrons suggesting the addition of sediment melt >350 Ka before volcanism, slab dehydration ~30 Ka beforehand, and a final fluid addition a few thousand years beforehand. However, evidence was presented for volcanoes behind the Marianas magmatic front that this sequence of events is not necessarily the case. 226Ra and 231Pa excesses were seen as stronger time constraints than U-Th disequilibria, indicating very recent fluid addition that triggered flux melting, and slow ingrowth during advection of the mantle wedge, respectively. Various participants suggested that 226Ra excesses may date from the last slab dehydration, the time of mixing of “fluid” and “melt” that originated from the slab at different times and places, or the final breakdown of hydrous phases in down-dragged peridotite in the mantle wedge. These topics address all three of the Subduction Factory themes. The time scale of advection of the mantle wedge is related to forcing functions. At least U and Ra are transported by slab-derived fluids, so they track and date the fluid flux. And all the nuclides are sensitive to the mass balance between mantle, fluid, and slab melt components. Participants called for more experiments to better determine solid/fluid partition coefficients for U, Th, Ba, Nb, and REE at variable fO2, CO3, and halogen contents, especially for garnet and amphibole. They also noted uncertainty in the stability of garnet in depleted peridotite, and the modal mineralogy of dehydrating amphibolite to eclogite beneath volcanic arcs to backarcs. Further study of IBM volcanics, especially from the backarc, are needed to invert rock chemical analyses to constrain the composition of slab fluids or melts or both. Two-dimensional U-series melting models need to be refined by combining decompression and flux melting in ways that are consistent with geodynamical models of mantle wedge convection. The rapid ascent indicated by correlations between 226Ra disequilibria and trace element slab signatures seems to require channelized melt flow without time for chemical equilibrium. This constraint is strongest at the Mariana volcanic front, and there was debate whether it also applies elsewhere in IBM. It was noted that U-series nuclides retain evidence of processes deep in the wedge while major elements re-equilibrate at shallower pressure. There was debate whether melt ascent mechanisms are different in arcs compared to ridges and ocean islands, or whether one simply has independent evidence of speed only in arcs such that similarly rapid ascent applies elsewhere too. Crustal differentiation processes in IBM and similar arcs must take in no more than a few thousand years to transform primitive magmas into andesites and dacites (differentiation time). Indeed, the processes may be so fast that even 226Ra disequilibria (half-time 1600 years) may be insensitive to them. Consequently, participants urged attention to even shorter-lived nuclides, and on diffusion and crystal size distribution studies to constrain crustal level timescales and processes. This is especially important if, as proposed at the meeting, tholeiitic melts differentiate rapidly and erupt whereas calcalkaline ones freeze and are later defrosted (remelted) to yield the felsic rocks characteristic of continental crust. Differentiation processes are so fast, and eruption volumes so small, that magma chambers large enough to be imaged seismically may not even exist beneath arc volcanoes not characterized by rhyolite. Practitioners pledged to provide full information about their numerical models, perhaps extending to sharing code, and to more thoroughly test the extent to which relatively long-lived U-series disequilibria are affected by shallow level processes. 4. Volatile fluxes and cycles Understanding how both major (H2O, CO2, SO2) and minor (N2, CH4, noble gases) abundance volatile elements move through subduction zones is one of the three objectives of the Subduction Factory science plan. The IBM arc system is particularly well-suited for understanding volatile cycling because (a) inputs of the incoming plate (especially sediments) are distributed regularly and have been well-sampled by ODP coring, and (b) there are four distinct regions across the arc system where outputs can be sampled: forearc serpentinite diapirs, arc magmatic front, arc cross-chains, and back-arc basins. The input part of the volatile budget is relatively well-understood through the ODP program, although the volatile characteristics of the basaltic basement is poorly known. Volatile estimates based upon ophiolites or composite sections of basement may not be appropriate due to differences in age and alteration history. Another important uncertainty is whether or not a significant fraction of water is stored in subducted serpentinized mantle in IBM and contributes to the volatile input. In contrast, less is known about the output flux of volatiles. Outputs from the forearc, principally from serpentinite seamounts, are almost completely unconstrained. The magmatic arc flux is becoming known only for certain sections of the arc, and then only for two major volatiles (H2O and SO2). For example, preliminary studies were reported for volatiles in glass inclusions in phenocrysts from three volcanoes along the southern part of the IBM magmatic front, but none from the northern part. There are some data for rare gases, principally He, but only from the southern magmatic front. Conversely, there are some data for SO2 emissions from active volcanoes in the north (including one of the largest SO2 outputs on earth, currently at Miyakejima), but not the south. In order to use what little data are available, better estimates of magma production rates and hydrothermal activity are needed in order to quantify volatile fluxes. Temporal changes in IBM magma production rates and volatile fluxes are completely unconstrained. Furthermore, almost nothing is known about volatile fluxes from backarc cross-chains, although this flux may be subordinate to the magmatic front and back-arc basin. In the arc and back-arc, the low CO2 contents of IBM submarine glasses indicates loss of volatiles through degassing, complicating estimates of primitive volatile abundances. We have a good understanding of water fluxes associated with back-arc basin igneous activity in the Mariana Trough, and have some rare gas data for these rocks. New GPS data suggest that we need to reconsider how the Mariana Trough is opening, and this will require changes in our estimates of melt production and volatile budget. This group strongly recommended a program devoted to completing a first-order understanding of the IBM volatile budget. This will require new analyses of all four magmatic outputs, using both remote sensing and direct sampling to determine arc magmatic outputs, and analyses of both glasses and melt inclusions from the arc front, arc cross-chains and the back-arc basins. We need to investigate water column chemistry for volatile emissions from the fore-arc, arc, and back-arc basin. This effort should be co-ordinated between USA and Japanese scientists. These results should be integrated into existing volatile databases (including OIB and MORB samples) in order to construct a first-order balance for volatile cycling between internal and exterior terrestrial reservoirs. We also need better estimates of the stability of various phases that retain and release volatiles in the downgoing slab, and to examine whether there is a significant amount of serpentinite in the downgoing slab. This is a tractable problem that can be addressed by an OBS crustal refraction study designed to measure Poisson’s ratio along with determination of epicenter depths of outer rise earthquakes. 5. Nature and distribution of primitive melts in the mantle wedge The conditions of melting provide tests of how forcing functions regulate melt production, and whether continental crust is a primary or secondary product of mantle melting in subduction zones. Presentations summarized the high temperature (1300-1400oC) and shallow pressure (~1 GPa) at which primitive arc and backarc melts have been shown experimentally to be in equilibrium with mantle minerals. They also emphasized the positive correlation between apparent percent melting and amount of slab component in several arcs including the Marianas, and that the same amount of slab component will produce more melt at higher temperature. The resulting melts range from silica-undersaturated through tholeiitic basalts, to high-magnesium basaltic andesites and andesites, and eventually boninite depending on pressure, volatiles, and melt fraction. The P-T conditions of equilibrium are thought to record the final stage of melt ascent, and to include a decompression component. Participants noted the absence of primitive magmas of any kind along the modern IBM volcanic front, and called for further exploration plus re-examination of existing collections to find such rocks. The backarc shows greater promise. Integration of refined mapping of the hydrous peridotite solidus with seismic profiles beneath volcanoes should result in better knowledge of how melts are distributed and segregate. 6. Subduction initiation We cannot fully understand subduction zones unless we understand how they begin and evolve. The initial location and mechanisms of subduction provide boundary conditions on the subsequent evolution of the system. The IBM arc system began 40-50 Ma and is recognized as perhaps the best site on Earth where the process of spontaneous subduction initiation can be studied. The group identified four key areas to pursue: paleogeographic reconstructions and initiation mechanisms, studies of the "proto-arc" crust exposed in the forearc; constraints on paleo-temperatures and mantle flow; and examination of the temporal evolution of the system. Many of these problems can be pursued with samples and surveys in hand. Constraining viable mechanisms of subduction initiation in the IBM arc system requires both accurate paleogeographic reconstructions and an assessment of permissible geophysical mechanisms. There are several current models of pre-subduction geometries in the region. Much of the uncertainty derives from conflicts between paleomagnetic data and geologic or geochemical constraints. The group identified five important parallel studies to be pursued. A first priority is an accurate reconstruction of the Eocene crustal fragments in the region, using the wealth of recent geologic and geophysical information from the IBM and West Philippine Basin (WPB). High precision age measurements of the principal Eocene units are needed to assess across and along arc trends in age (recent studies have shown that many previous ages are in error because of metamorphism and alteration). This age work should integrate constraints from biostratigraphy and possibly Sr-isotope stratigraphy. Geophysical modeling that identifies permissible mechanisms of subduction initiation (as was demonstrated in new models presented at the meeting) and critical constraints are important. Also needed are models that integrate the tomographically imaged extent of subducted slabs with proposed past plate motions. Finally, there are a few fragments of crust in the WPB that may hold key, pre-50 Ma records, including the far northwestern corner, and the northernmost and southernmost crustal fragments. Geological and geophysical investigations of these crustal fragments should be pursued to constrain their composition, age, vertical histories, sedimentologic histories, and paleomagnetic records. Forearc sequences preserve the magmatic consequences of subduction initiation, and this record can be inverted to constrain mantle flow and thermal regime, along with what was released from the sinking slab. These are the second and third areas of study suggested by the working group, and are an important key to quantifying magmatic fluxes in the system, as the earliest arc volcanism was at a rate much higher than typical of the mature arc. These crustal pieces are also one of the best existing analogs for many large ophiolites. The opportunity now exists to use the magmatic record preserved in the forearc in tandem with the results of experimental petrology to constrain mantle temperatures and flow and thus refine the next generation of geodynamic models for subduction initiation. In addition to the work detailed above, characterization of the protoarc sequences requires expanded studies of the existing lower crustal and upper mantle materials dredged from the forearc (including metamorphic histories, ion probe, P-T, fluid inclusion, etc.) and examining the stratigraphic succession of protoarc magmatism. Vertical sampling of the volcanic succession, dyke swarms, and upper-level gabbros would also allow us to test the hypotheses that forearc crust is not trapped oceanic crust but forms entirely at the time of subduction initiation, and that forearc subduction initiation sequences are analogues for ophiolites. Such vertical sampling can be approached through dredge and submersible studies on the steep western scarps of Guam and Chichi-jima, and by deep drilling on those same islands. The work detailed above would lead to a clear understanding of the mechanisms, conditions, and consequences of subduction initiation in the IBM arc system. Such an understanding is a cornerstone for creating accurate temporal models for the evolution of convergent plate margins. The integrated geologic history of the IBM can provide an important analog for examining the role of intraoceanic island arcs in crustal evolution throughout earth history. 7. Crustal evolution and intermediate/felsic magmas This group focused first on the
tonalites particularly well-known in the northern IBM collision
zone, and possible relations of these exposures to the “6.2
km/s” layer shown on the seismic transect across the IBM
arc at 32°N; then on the restites expected to be formed
during tonalite production; and finally on problems of arc growth
rates. The above studies are required before we can tell whether the measured volume of “7.2 km/s” material in our seismic velocity models is or is not compatible with the volume of less-mafic rocks. We don’t yet know whether these presumed gabbros are related to ongoing rifting in the Izu segment, or whether they are a consequence of normal arc underplating, though this could be tested with additional seismic surveys in areas not actively rifting such as the Bonin segment. We also need estimates of the volume of ultramafic cumulates produced by any given petrologic model. Depending on temperature, these rocks should have compressional wave seismic velocities of about 7.8 km/s, indistinguishable from tectonized mantle harzburgites below the petrologic Moho. If the crustal cumulates are not themselves tectonized (as in the ophiolite section exposed in the trench wall south of Guam), and are sufficiently thick, then a very careful Pn-anisotropy study could in principle distinguish these rock types. Currently, even Pn (uppermost mantle) velocity has not been well constrained below the active IBM arc. When we understand all these layers (felsic, intermediate, mafic and ultramafic), we will be in a better position to evaluate crustal growth variations in time and space. Several studies are needed to help relate measured sections to true growth rates. We need to know the amount of crustal thinning caused by stretching in the Izu segment, whether a factor of two or far less. More thinning would suggest a greater crustal thickness at earlier times, perhaps leading to lower crustal pressures great enough to form garnet and to delaminate the deepest crust. We need to know the thickness of the crust on which the proto-arc was built, to be able to subtract this from our estimates of arc crustal volume: those models in which plumes help initiate subduction also imply thicker initial crust. To model the growth rate with time, we need more drilling of volcaniclastic sequences. To model the growth rate in space will require additional surveys which take account of the different likely length scales on which changes occur. Current seismic surveys can potentially study change at the c. 50 km wavelength, controlled perhaps by “fingers” in the mantle (Tamura et al., 2002) represented by volcano spacing. Comparison of existing and planned surveys of crustal velocity structure in the Izu, Bonin and Mariana segments will allow evaluation of variations at the c. 1000 km wavelength affected by the plate tectonic history, and plate interactions. But much additional work will be required to understand potential variations at intermediate scales (c. 250 km) perhaps controlled by changes in kinematic forcing functions such as variable rates of convergence at the trench or of back-arc spreading, and along-arc stretching. Finally, to understand the physical mechanisms of crustal growth will require detailed studies of the magmatic plumbing of at least one active volcano, followed by active monitoring and time-lapse seismology to image the paths and rates of magma movement. Recent work at Miyakejima in the Izu arc provides an example (Kodaira et al., 2002). 8. Critical geologic, geochemical, and geophysical data still required. Based on the lectures of the first three days of the IBM Workshop, and discussions in the other breakout sessions, participants were able to identify a large number of data that are still required if we are to understand the IBM subduction factory. If we are to understand subduction initiation, we need to know the Eocene paleogeography and geodynamic setting of the IBM system, including seafloor ages and spreading directions, obtainable in part by MCS surveys, swath mapping, and ultimately drilling legs. At least some additional magnetic surveys are required to fill out our knowledge of the spreading history of the back-arc basins; while the modern kinematic forcing functions (modern rates of back-arc spreading, or along-arc stretching) could easily be measured geodetically with modest additional programs of GPS observation. Some major inputs into the subduction factory are still uncertain, including fluids trapped by hydrothermal alteration of oceanic basalts, in seamounts, and, more speculatively, in the Pacific mantle. Direct sampling of altered seamounts can answer part of this question. Because serpentinization of subducting oceanic mantle is likely to happen between the outer rise and trench due to crustal-scale normal faulting, highly detailed gravity models, or possibly trench-parallel seismic surveys, will be needed in this bathymetrically-challenging environment. Validating theoretical models of fluid release from the down-going slab will require very detailed earthquake hypocentral locations to study spatial and temporal variations in seismicity. Better earthquake locations and mechanisms are also needed to understand the mineral physics of subduction earthquakes, and the nature of asperities in the subduction system. To further our understanding of the mantle wedge, we need many more laboratory and theoretical studies of the relations between seismic observables (P, S velocity, attenuation, anisotropy) and physical properties (temperature, melt, fluids, composition, flow). Studies of the melt distribution in the wedge will probably be best done with OBEM experiments that span from the forearc to the back-arc spreading center, though better seismic tomographic images are also needed. Along-strike active source profiles are needed to determine the composition and extent of the mantle wedge. Theoretical studies of mantle flow and dynamics are needed to provide end-member models to test. Further knowledge of the outputs of the Subduction Factory are needed in some key areas. Geochemists need to carry out a transect of fluid release across IBM from trench to arc; and to study the volatile systematics of available hydrothermal systems. Long-term monitoring of these systems is also needed, not only for the geologic outputs but also to understand the macro-fauna and microbial populations that are both supported by and an influence on Subduction Factory inputs and outputs. Crustal structure has been measured by seismic experiments in just a few places, and new surveys are needed both to examine the composition of the active arc where it is unaffected by modern rifting; and to evaluate the along-strike variability of the arc. One key aspect of the physical state of the arc that is essentially unknown is its thermal structure. Heat-flow measurements in the forearc would help determine stress at the plate boundary; whereas measurements over the active arc would allow study of crustal melting and evaluation of lower-crustal rheology and hence the potential for delamination. Such measurements will be challenging because of the data density required to overcome problems of subsurface water transport, the difficulty of penetrating the volcaniclastic carapace of the active arc, and the largely unknown bottom currents and ocean thermal structure at the seafloor in this area. More optimistically, new regional mapping surveys are allowing us to identify the best areas for studies such as these, and particularly sites such as submerged fault scarps that are appropriate for direct physical sampling of Subduction Factory outputs by dredging and diving. Some areas still need basic bathymetric mapping, such as the back-arc knoll region (27°-30°N), the northern Mariana forearc (20°-24°N), and the West Mariana Ridge; and more and better multi-channel seismic reflection profiles are needed nearly everywhere. Many of these data needs are critically
dependent on the availability of instrument pools, some existing,
some needing further development. Geophysical transects are
limited by the availability of broadband and short-period OBSs
(and the current impracticality of burying broadband sensors),
and of OBEMs; while physical sampling is limited by the availability
of deep submersibles and ROVs, and the continuing need to develop
prod-type drills for shallow borehole sampling. 9. IODP planning Proponents of scientific drilling in the IBM focus area need to generate competitive proposals that address global scientific issues, and convincingly argue that drilled core can, and is required to, solve the problem. Further, proposals will need to demonstrate that the target problem cannot be solved by deep drilling on land. Even though the new drill ships will not be available for another 5 years or more, the lead time required for necessary preparatory work is long, and projects need to be planned now if they are to stand a chance of completion during the MARGINS decade. The breakout group discussed several problems that could be addressed by IODP drilling, and suggested ways or places in which these might be tackled. Our knowledge of the IBM subduction factory would be advanced by deep drilling to constrain models of arc volcanism, perhaps on an extinct (cold) Western Seamount Chain volcano such as Manji seamount. We can study arc crust by drilling to mid-crustal levels in the thin arc crust, to ground-truth the seismically derived structure. Potentially we can even study the petrologic Moho of the arc by offset drilling along km-high escarpments that exist at various places across the IBM arc system. One example is in the southernmost IBM arc system, near the 11-km deep Challenger Deep, To understand subduction initiation requires knowledge of paleogeography, which may best be obtained by drilling pieces of oceanic crust in the West Philippine Basin that predate the birth of the IBM convergent plate boundary. The subsequent arc history needs better constraints of the sort available from drilling volcaniclastic sections. Tighter constraints are needed on the inputs to the subduction factory, and can be gained, for example, by deepening ODP site #1149. Study of fluid-mantle interaction can be advanced by drilling serpentinite seamounts. These seamounts also study of geo-bio interactions, as would drilling of hydrothermal sites that are important loci of arc metallogenesis. Either of these latter targets, as well as the most active arc volcanoes, would also be candidates for long-term monitoring of active processes. 10. Future off-shore projects There is an interest and need for a wide variety of seagoing projects in the IBM arc. a) Multibeam/Swath/Sidescan/Deep Tow Surveys. Reconnaissance surveys are needed in the Izu-Bonin arc, W. Shikoku Basin, Northern Mariana forearc (20°-25° N), Eastern and Southwest Parece-Vela Basins, and the rifting area in the Northern Mariana/Volcano arc. A nested-studies approach should be taken to study specific targets with deep tow systems, such as arc volcanoes, back-arc spreading segments, forearc fluid vent sites on serpentinite seamounts, and fault scarps. b) Seismic studies. Passive seismic tomographic studies using ocean bottom seismographs are needed to map the upper mantle structure and magma production zones beneath the arc and backarc. Similar deployments are needed to study seismicity patterns possibly associated with serpentinite seamounts, and the relationship of shallow thrust zone seismicity to heat flow and composition. Most of these objectives will be met in the Mariana region by the 2003-2004 US-Japan Passive OBS survey, and a further 2005 Japanese active-passive survey planned in the Izu arc. Future passive OBS work will require the development of buried sensors and possible cabled observatories. There is also a need for active source seismic studies to evaluate crustal structure prior to future drilling on land. MCS work is required to evaluate N-S extension along strike in the Marianas forearc, back-arc basin bounding structures, serpentinite seamount structure, and the possible existence of magma chambers in the arc/backarc. Some of these objectives will be met by the 2002 Ewing and 2003 Kaiyo active source cruises. c) Gravity-Magnetics-EM. Gravity/Magnetic data are needed to constrain the initiation of rifting in the northern Mariana arc. All ships should collect this data routinely and make it available publicly. Seafloor gravity measurements may help study detailed variations in porosity near volcanoes, spreading segments, and serpentinite seamounts. Seafloor EM measurements are needed to help constrain the large-scale distributions of fluids and melt in the arc and backarc. Some seafloor EM data are currently being collected by Japanese groups. d) Heat Flow. A heat flow transect across the entire system is needed to constrain thermal models for the forearc, as well as large scale flow models for the mantle wedge. e) Sea Floor Sampling. Reconnaissance sample collection using dredging is needed to understand forearc faults and backarc- bounding faults. A seafloor-tethered drill can help sample serpentinite seamounts and fault scarps, and could undertake offset drilling of much of the arc infrastructure. An ROV/Submersible drill can be used to study the subsurface structure of serpentinite seamounts. f) ROV/Submersible. Submersible studies are desired at tectonic windows, forearc/cross-arc scarps, hydrothermal systems, on the inner trench slope, and volcanic and serpentinite seamounts. Only by detailed submersible mapping and sampling can these processes be understand at the level now possible at mid-ocean ridges. A major impediment has been the perception that NSF is not interested in sending ALVIN to the Western Pacific, which was last in the region in 1987. US-Japan collaboration in the IBM arc could be greatly improved with development of joint studies of problems of mutual scientific interest, such as serpentinite diapirs and vents, recent arc and back arc basin volcanism and hydrothermal venting, and the crustal structure of the southern Mariana Trough and Challenger Deep region. g) Organizational issues. The Margins community should provide a letter in support of using data from improved satellite gravity measurements now controlled by the U.S. Defense Department to study seafloor tectonics. A list of planned projects and contact information should be provided to facilitate cooperation. 11. Future on-land projects Many of the islands of the IBM chain have not been well characterized, and accurate geological maps are available for only a few. Future work will involve geological studies, sample collection, and visits to the islands to install seismic, EM, GPS, or other monitoring equipment. There was considerable interest in sampling scoria and volatiles from the Mariana and Izu-Bonin volcanic islands. There is also interest in geological mapping of some of the volcanic islands. Juan Camacho from the Commonwealth of the Northern Mariana Islands detailed some of the possible logistical constraints on reaching the islands. It may be possible for geological sampling to be combined with existing projects that must reach the islands for deployment of equipment, such as the 2003-2004 passive seismic survey being carried out by Douglas Wiens. Tobias Fisher (New Mexico), David Hilton (Scripps), Erik Hauri (DTM), Teriyuki Kato (U. Tokyo), Douglas Wiens (Washington Univ.), and Terry Plank (Boston U) all have interest in reaching the northern Mariana islands and agreed to coordinate activities. There is also a lot of interest in further study of the forearc islands, particularly because of their importance for questions of arc initiation. In particular, the characterization of the spatial and temporal evolution of the protoarc sequence should receive high priority. In addition, preliminary site studies are needed at possible on-land drilling targets on Guam and Chichi Jima. The drilling of an in-situ arc ophiolite should be a long-term priority of the on-land studies. |
|