Seismic Properties of Partially Molten Rocks: Constraints from Experiments and Observations on Natural Rocks

Abstract:

The increasing resolution of seismic tomography studies as well as techniques that are sensitive to sharp, localized velocity changes indicate that melt is present in diverse tectonic settings in the upper mantle. Unresolved questions are the quantification of seismic velocities and attenuation to melt fraction, and the dynamic stability of this melt. Properties of partially molten rocks are generally sensitive to melt geometry. The melt geometry can be observed in experiments, but the experimental observations have to be extrapolated to mantle grain sizes. High resolution images of experimentally observed melt geometries (Garapic et al., 2013) indicate that the broad attenuation peak observed by forced torsional oscillation experiments (Jackson et al., 2004, Faul et al., 2004) are due to melt ‘squirt’. Characterization of the inferred melt geometry in rocks recovered from mid-ocean ridges and peridotite massifs suggest that the experimental observations are applicable to the upper mantle. The data from the earlier seismic property experiments will be reprocessed with improved apparatus calibration and fitting strategies to provide a ‘global’ fit to the melt-bearing data as a function of grain size and melt fraction. This should allow quantification of the amount of melt imaged in seismic studies and therefore enable more accurate geodynamical models for example of arcs and backarcs, for which the highest resolution tomographic images exist. It should also provide more quantitative constraints on the amount of melt required to generate receiver functions and SS-bounces that suggest the presence of melt in the asthenosphere.