Mantle transition zone topography and low-velocity zone locations in the Yellowstone Region

Earth’s mantle transition zone (MTZ) is a possible global water reservoir and may be responsible for long-term (~100 Ma) ocean-mass regulation. Estimates of water capacities in MTZ minerals are ~1 wt%, far greater than that of rocks of the surrounding mantle. When water-rich material is displaced from the MTZ, partial melting occurs, generating a sharp reduction in seismic velocities detectable with seismic receiver functions (RFs). We estimated RFs for the MTZ beneath the Yellowstone region using earthquakes recorded by ~200 stations of the Earthscope Transportable Array. We found many LVZs both above and below the MTZ, consistent with water release upon phase transformation of hydrated MTZ rock into upper- and lower-mantle mineral assemblages with low water capacities. The locations of LVZs are consistent with mid-mantle flow induced by descent of a Farallon-slab fragment and ascent of the deeply-rooted Yellowstone plume as imaged by seismic tomography.

Methods
Here, we provide the topography of the mantle transition zone (MTZ) and low-velocity zones (LVZs) above and below the MTZ estimated for the Yellowstone region. Seismic data used in this study were recorded on the EarthScope Transportable Array and obtained from the Incorporated Research Institutions for Seismology (IRIS) Data Management Center (http://ds.iris.edu/ds/nodes/dmc/). Topography of the MTZ and LVZs were identified using stacked common conversion point (CCP) (Dueker and Sheehan, 1998; Bianchi et al., 2010) multiple-taper correlation (MTC) RFs (Park and Levin, 2000; Park and Levin, 2016). The topography of the 410- and 660-km discontinuity were calculated with target depths of 410 and 660 km, respectively. LVZs above and below the MTZ were identified with RFs targeted to 410 and 760 km depth, respectively. Target depths (i.e., Ps delay times) were calculated via the US-SL-2014 tomography model (Schmandt and Lin, 2014), obtained from the IRIS Earth Model Collaboration (http://ds.iris.edu/ds/products/emc/). All structures were identified with a cutoff frequency fc of 0.65 Hz. Stacking uncertainty was calculated with the jackknife resampling method; we consider feature robust if the RF value is greater than one standard deviation.