The most explosive, deadly volcanic eruptions occur in subduction zones. Their occurrence and explosivity is intimately tied to the transport of water and other volatiles to great depth, and its release into hot mantle or along the plate boundary. Yet, understanding how volatiles released from subducting plates produce melts within the mantle wedge of subduction zones remains a challenge. A series of recent seismic and petrologic projects in subduction zones and results from the laboratory have dramatically increased the available information on these processes, and preliminary analyses of these data have begun to place bounds on the extent and depths of melting, style of melt transport, and variations between sites. This proposal plans to take the next step, the integration of geophysical observations, geochemical measurements, and rheological models, to address fundamental problems regarding melting process in the mantle. We will compare recent results from the Nicaragua-Costa Rica subduction zone, which exhibits strong along-strike variations in geophysical and geochemical observations, with those from the Marianas where volcanoes sample across-strike from the arc front to active back-arc spreading center.
Broadband imaging in Central America and the Marianas indicates a region of flowing hot mantle from seismic attenuation and P-wave velocities, mantle fabric from shear-wave splitting measurements, and the geometry of melt transport from Vp/Vs anomalies. These results also show clear along-strike variations in most parameters, interpretable as variations in slab hydration, melt transport, and mantle wedge H2O content. In both regions, extensive sampling and analysis of volcanic products has provided for the first time direct measurement of magmatic water content, which appears to vary in tandem with classic indicators of slab fluid/sediment in Central America, and with distance from the arc front in the Marianas. The integration of these observations will be made possible by advances in our understanding of mantle rheology in the presence of water and melt, including both experimental and theoretical breakthroughs, for example on the behavior of melt under shearing conditions and on the physical controls on grain size, and on quantitative parameterizations of melt fabric. Our approach will be to integrate all datasets in a consistent manner, then iteratively test theoretical predictions against them. Results should distinguish models of melt generation and transport, for example of vertical porous flow, inclined flow, or ascent of cold diapirs. Once calibrated, the relationships needed to make these comparisons will be made available to interpret structure in other subduction zones and regions of melt production globally. The integration of disparate observations is a key step in making them understandable to a wide audience, and this integrated multidisciplinary approach will be tested in undergraduate classrooms and graduate seminars. The calibration of relationships between seismic observables and thermodynamic melting parameters should be broadly applicable and serve to enhance infrastructure for research.
