Electrical resistivity structure in the mantle across the Mariana subduction system

We have conducted a magnetotelluric (MT) survey across the central Mariana area (around 18°N) to provide a comprehensive electrical resistivity image of the mantle beneath the Mariana subduction zone, fore-arc, arc, and back-arc system, at a depths down to the transition zone. Our transect will address issues of hydration of the mantle wedge and subsequent melting, the origin of arc magmas, and melting processes beneath the slow back-arc spreading ridge axis. Electric and magnetic time series measured with seafloor instruments were cleaned and processed into MT response functions (apparent resistivity and phase), using a bounded influence algorithm with remote reference (Chave and Thomson, 2003, 2004). Topographic effects on the apparent resistivity and phase data were corrected, based on the equation of Nolasco et al. (1998), using correction tensors which were modeled with a three-dimensional forward code (Baba and Seama, 2002). The nonlinear conjugate gradient inversion algorithm, which seeks regularized solutions (Rodi and Mackie, 2001; Baba et al., 2006), was used to find an optimal two-dimensional electrical resistivity structure. This inversion algorithm allows us not only to seek a minimum structure model but also to investigate the effects of electrical anisotropy on the data. In the inversions, the TE mode apparent resistivity data (in which the electric current flow is parallel to the strike of structure) were not used because they are susceptible to resistivity anomalies by off-profile structures, and were seen to be consistently poorly fit. Sensitivity tests were applied to key features found in the obtained structure through inversions of synthetic data. The resultant isotropic electrical resistivity structure shows the following features: (1) a thick resistive Pacific plate lithosphere (roughly 100km in thickness), (2) a high resistivity region roughly in the location of the descending slab (this feature while consistent with the data is not a required feature); (3) a low resistivity region starting at a depth of about 60-70km within the mantle wedge above the slab; (4) a high resistivity region beneath the back-arc spreading ridge axis. The low resistivity feature in the mantle wedge can be attributed to melting initiated by the release of water. We have run tests on the depth to the top of this conductor in an attempt to identify the primary phases responsible for water release. The high resistivity beneath the back-arc spreading center suggests that melt delivery to the ridge crest occurs through segmented (three-dimensional) pathways, reflecting the slow spreading rate of the system. We investigated whether electrical anisotropy is required in our models, but there is no requirement for significant anisotropy either in the mantle wedge or beneath the back-arc basin.