T32A-01
Seismic Structure of the Subducted Cocos Plate
The Meso-American Subduction Experiment (MASE) was designed to determine the critical parameters to necessary to simulate the subduction process in Central Mexico . A preliminary analysis of the data shows a 200km section of the slab that is subhorizontal and to within the resolution of the receiver functions it underplates the continental crust with no intervening asthenosphere. This is an interesting situation because the short-term (GPS) and long-term (geologic) strain measurements show almost no compressive strain in this region. This would imply that the crust is decoupled from the subducting slab. Near the coast, the receiver functions show that the slab cuts through the crust at an approximately a 15-degree angle, and under the Trans-Mexican Volcanic Belt the slab becomes detached from the crust, but its geometry at depth is not yet determined from the receiver functions, but a well-developed mantle wedge is apparent from the attenuation of regional earthquakes.
T32A-02
Seismic structure of the Rivera subduction zone - the MARS experiment
The subduction zone of western Mexico is a unique region on Earth where microplate capture and overriding plate disruption are occurring today. The small Rivera plate is subducting beneath western most Mexico primarily beneath Jalisco state while to the east it is the Cocos plate that is subducting. Above the Rivera plate the Jalisco block of Mexico is bounded by the north trending Colima Rift and the northwest trending Tepic-Chapala Rift and may form a microplate in its own right. Magmatism is present throughout the region and is unusual for a subduction zone in that geochemical analyses indicate an ocean island basalt component to some of the lavas. Also, Colima volcano is offset trenchward from other volcanoes in the Mexican Volcanic Belt. Little is known of the subducting Rivera plate geometry due to the paucity of seismicity within the plate yet the geometry of the Rivera and Cocos plates at depth are likely critical for understanding the tectonic evolution of western Mexico. The MARS (MApping the Rivera Subduction zone) project consists of the deployment of 50 broadband seismometers covering the Jalisco block from the coast to the Tepic-Chapala rift in the north and about 150 km to the west of the Colima rift. The instruments were deployed in January, 2006 and will be removed in June, 2007. The goal of the project is to seismically image the subducting Rivera and Cocos plates at depth as well as the mantle wedge above the plates. A number of different analyses of MARS data are underway including teleseismic tomography, receiver function analysis, and shear wave splitting analysis. The preliminary tomography results clearly show both subducting plates with a sharp change in dip to the east of the Colima rift probably indicating a tear between the two plates along a trend more eastward than the trend of the rift. The images also show extremely slow shallow mantle velocities beneath the Tepic-Chapala rift but not beneath the Colima rift. Receiver functions indicate an average Moho depth of roughly 40 km in the continental interior but are very complicated near the coast. Finally, shear wave splitting observations indicate a generally north fast direction with a reduction in magnitude of splits beneath the Colima rift.
T32A-03 INVITED
Numerical modeling of volcanic arc development
We have created a new coupled geochemical-petrological-thermomechanical numerical model of subduction associated with volcanic arc development. The model includes spontaneous slab bending, subducted crust dehydration, aqueous fluid transport, mantle wedge melting and melt extraction resulting in crustal growth. Two major volcanic arc settings are modeled so far: active continental margins, and intraoceanic subduction. In case of Pacific-type continental margin two fundamentally different regimes of melt productivity are observed in numerical experiments which are in line with natural observations: (1) During continuous convergence with coupled plates highest amounts of melts are formed immediately after the initiation of subduction and then decrease rapidly with time due to the steepening of the slab inclination angle precluding formation of partially molten mantle wedge plumes; (2) During subduction associated with slab delamination and trench retreat resulting in the formation of a pronounced back arc basin with a spreading center in the middle melt production increases with time due to shallowing/stabilization of slab inclination associated with upward asthenospheric mantle flow toward the extension region facilitating propagation of hydrous partially molten plumes from the slab. In case of spontaneous nucleation of retreating oceanic subduction two scenarios of tecono-magmatic evolution are distinguished: (1) decay and, ultimately, the cessation of subduction and related magmatic activity, (2) increase in subduction rate (to up to ~12 cm/yr) and stabilization of subduction and magmatic arc growth. In the first case the duration of subduction correlates positively with the intensity of melt extraction: the period of continued subduction increases from 15,4 Myrs to 47,6 Myrs with the increase of melt extraction threshold from 1% to 9%. In scenario (1) the magmatic arc crust includes large amounts of rocks formed by melting of subducted crust atop the thermally relaxing slab. In contrast, in case of stable self-sustaining subduction, magmatic rocks produced by partial melting of hydrated mantle wedge clearly dominate the crust. In several numerical experiments an intra-arc extension is observed during subduction. This process results in splitting of previously formed magmatic arc crust by a newly formed spreading center. In all conducted numerical experiments the loci of magmatic activity and intensity of crustal growth is strongly dependent on the dynamics of hydrous partially molten upwellings (cold plumes) rising from the slab. The material forming these plumes can be homogenous (composed of hydrated mantle) or heterogeneous (composed of both hydrated mantle and subducted crustal rocks). In case of heterogeneous plume growth material mix chaotically resulting in attenuation and duplication of the original layering on scales of 1-1000 m. Comparison of numerical results with geological observations from the Horoman ultramafic complex in Japan suggests that mixing and differentiation processes related to development of partially molten plumes above slabs may be responsible for strongly layered lithologically mixed (marble cake) structure of asthenospheric mantle wedges.
T32A-04 INVITED
Across-arc Volatile Variations in the Michoacan-Guanajuato Volcanic Field, Mexico: Investigating Slab Devolatilization and the Mantle Source for Basaltic arc Volcanoes
Volatile abundances in mafic arc magmas provide information on slab devolatilization processes, and thus give insight into fluid recycling in subduction zones. In order to investigate both magmatic and mantle volatile (H2O, CO2, S, Cl) variations in a subduction zone, we have analyzed olivine-hosted melt inclusions from nine basaltic centers located at varying distances from the trench in the Michoacan-Guanajuato Volcanic Field (MGVF), Mexico. Additionally, we have used trace element data to assess the variability in mantle source compositions across the arc. By sampling mainly primitive basaltic centers (many contain Fo88-91 olivine), we can compare magmatic volatile contents of near-primary magmas across the arc. Our data show that magmatic H2O is high (3.0- 5.0 wt%) from the volcanic front to 160 km behind the front. Other volatiles (CO2, S, and Cl) also show high concentrations across the arc. We used these data to estimate mantle volatile contents based on trace element partial melting models (e.g., Kelley et al., 2006). The degree of mantle partial melting beneath the MGVF varies from 13-24% for our analyzed cones. Calculated mantle volatile contents (0.5-1.0 wt% H2O, 100-200 ppm Cl, and 100-650 ppm CO2) are highly elevated compared to depleted MORB mantle values. Additionally, trace element data for each of the cones show that the mantle beneath the MGVF is heterogeneous. Most cones show the relative depletions in Nb and Ta typical of subduction zone melts. However, two samples show enriched Nb-Ta values, suggesting these melts originated from a more OIB-like source. Consistently high mantle H2O across the arc, combined with high Cl and CO2, suggest that flux of water and other volatiles is not limited to the region beneath the volcanic front. Instead, these data support several dehydration models, including continuous dehydration of the slab to great depths, down-dragging of hydrated mantle above the slab, and trenchward migration of the volcanic arc over time. However, better geophysical modeling of the thermal structure of the downgoing slab coupled with age constraints on volcanism in the region are necessary to further assess models for fluid fluxing beneath the MGVF.
T32A-05
Exploring grain deformation processes of underthurst sediments beneath the Ecuadorian margin based on multichannel seismic data
In this work we quantify the physical properties variations of underthrust sediments along the first tens of km of subduction of the non-accretionary sourthern Ecuadorian margin using high-quality multichannel seismic data. Two pre-stack depth-migrated profiles (SIS-72 and SIS-18) have revealed the presence of three zones along the subduction channel (SC) characterized by contrasting velocity and velocity-derived physical properties. These variations agree with expected based on the different stages of deformation of granular media subjected to progressively increasing confining pressures observed in laboratory experiments. The three zones likely correspond to transformational steps of underthrust sediments governed by different physical processes that control its mechanical behaviour at increasing confining pressures. The seismic images show that the sediments feeding the SC are series of stratified layers related to turbiditic sequences that fill the trench in SIS-72, and chaotic seismic facies suggesting coarser mass wasting deposits in SIS-18. Within Zone I (0-9 km from the trench and 5-6 km depth in line SIS-72, 0-5 km and 4.5-5 km depth in line SIS-18), progressively increasing velocity and decreasing velocity-derived porosity indicate continuous compaction of unconsolidated (fluid-supported) incoming material, accompanied by effective fluid drainage along the décollement and/or across the highly porous accretionary wedge. Comparison with experimental results in granular media indicate that the dominant processes at this range of pressures (<30 MPa) are grain rolling, particle rotation and frictional slip at grain contacts, with no o little deformation. Within Zone II (9-25 km and 6-7 km depth in SIS-72, 5-12 km and 5-5.5 km depth in SIS-18), the underthrust material is under the critical porosity, becoming frame-supported. Velocity and porosity remain constant regardless the increasing confining pressure, suggesting undrained conditions due to a low permeability within either the underthurst sediments or the décollement, and resulting in overpressured fluids within the SC. Laboratory experiments show that, at pressures of 30-100 MPa, the dominant process is Hertzian-like, elastic deformation of individual grains with increasing grain damage. Within Zone III (25-30 km and 7-7.3 km depth in SIS-72, 12-16 km in and 5.5-6.5 km depth SIS-18) velocity increases and porosity falls rapidly, probably indicating sudden sediment compaction and subsequent release of over-pressured fluids. Experimental results agree with this behaviour showing pervasive grain cracking (cataclasis), acceleration of compactive strain rates and, eventually, a sudden collapse of the system. The highest pressures at which these transformations occur in SIS-72 with respect to SIS-18 could be due to the larger grain diameter and broader size distribution of underthurst sediments in SIS-18.
T32A-06
Structural variations within the Jalisco Block displayed by a magnetotelluric transect
The Jalisco Block has been described as an uplifted domain by several authors; the term block has implicitly involved the whole structure. Alvarez et al. (2006) advanced the idea that the block is actually divided in two domains, which they called the uplifted and the static domains. The uplifted domain lays about 1000 m above the static domain; the latter occupies the coastal area. In addition to the topographic differences, the above authors pointed out at contrasting seismic, magmatic, and aeromagnetic responses between the two domains. In order to clarify the origin of such a difference a magnetotelluric survey was performed involving 13 stations, 10 of them constituting a transect of 200 km across the two domains, while the remaining 3 constituted probes in neighboring areas. The four stations across the coastal region suggest the presence of a dipping conductor, which might be associated with the down going slab of the oceanic plate. The following three stations of the transect show a high conductivity anomaly zone 30 km wide, starting at depths of 8-10 km, which coincides with a region of intense magmatism at the surface; these stations are located in the uplifted domain. Then, the last three stations along the transect reveal an abrupt transition in the subsurface from a conductive zone to a region of high resistivity. This division correlates spatially with the previous finding that the coastal region shows positive magnetic anomalies whereas the landwards portion of the Jalisco Block corresponds to a wide negative anomaly. These results support the idea of two tectonic domains within the Jalisco Block.
T32A-07 INVITED
Adakitic-like volcanism in Southern Mexico and subduction of the Tehuantepec Ridge
The origin of El Chichón volcano is poorly understood, and our attempt in this study is to demonstrate that Tehuantepec Ridge, a major tectonic discontinuity on the Cocos plate, plays a key role in the slab dehydration budget and therefore in partial melting of the mantle beneath El Chichón. Using marine magnetic anomalies we show that the upper mantle beneath TR undergo partial serpentinization, a 5-7 km thick serpentinized root extending along TR and below the oceanic crust. Another key aspect of the magnetic anomaly over southern México is a long-wavelength (~150 km) high amplitude (~500 nT) magnetic anomaly located between the trench and the coast. Using a 2D joint magnetic-gravity forward model, constrained by the subduction P-T structure, slab geometry and seismicity, we find a highly magnetic and low-density source located at 40-130 km depth. We interpret this result as a serpentinized mantle wedge by fluids expelled from the subducting Cocos plate beneath southern Mexico. Such a deep hydrated mantle requires a low temperature wedge (T<600° C) because serpentine is stable below this temperature and also the magnetic properties are preserved for temperature less than the Currie point for magnetite (~580° C). This result explains the lack of volcanism in southern México where the slab depth is ~ 100 km. Using phase diagrams for sediments, basalt and peridotite, and the subduction P-T structure beneath El Chichón we find that sediments and basalt dehydrate ~ 50% at depths corresponding with the location of serpentinized mantle wedge, whereas the serpentinized root beneath TR strongly dehydrates (60-80%) at higher depths (170-180 km) comparable with the slab depth beneath El Chichón. We conclude that this strong deserpentinization pulse of mantle lithosphere beneath TR at great depths triggers arc melting, explaining the unusual location and probably the adakitic signature of El Chichón.
T32A-08 INVITED
The onset of flat subduction in Central Mexico controlled by a low viscosity wedge.
It is generally accepted that dehydration of subducting lithosphere transport fluids into the mantle wedge, causing a decrease in the viscosity. Such decreasing in viscosity can form a well-defined low viscosity wedge (LVW) or a low viscosity channel (LVC) on top of the subducting slab. Using numerical models we study the effect of low viscosity and channels on time-dependent geodynamic models. The results show a dramatic effect of low viscosity on the slab dip evolution through time. A reduction of the wedge viscosity with an order of magnitude produces the slab dip to increase from an initial 30° to 50° after 10 Ma of convergence. Also, the slab overthickening observed in previous published dynamic models due to slab attachment to the overriding plate, is eliminated by introducing a LVW or LVC for models with or without trench rollback. Our study shows two end member models, depending on the maximum depth extent of the LVW. Models with a LVW extending down to 400 km depth show a steep slab geometry (dip > 50°), whereas shallow LVW produces flat slabs, including perfectly flat slabs (dips = 0°). Assuming that slab and sediments dehydration are responsible for lowering the mantle viscosity, the maximum depth extent of the LVW is controlled by the age of the incoming plate and the convergence rate. Thus, significant variations in LVW geometry and viscosity have significant impact on the slab geometry and volcanic arc evolution through time.