New constraints on the properties of the Yellowstone mantle plume from P and S wave attenuation tomography

We estimate attenuation ( t *) for teleseismic P and S arrivals to seismometers in the Yellowstone Intermountain Seismic Array; tomographically invert these data for upper mantle Qp ‐1 and Qs ‐1 structure; and, with the aid of the upper mantle velocity model of Waite et al. (2006), interpret the results for mantle temperature, partial melt, and water content. Because attenuation analysis is susceptible to contamination by noise, we employ time‐ and frequency‐domain analyses and a careful assessment of the uncertainty associated with each estimate by using the misfit between actual and predicted traces or spectra. The greatest noise source is signal‐generated noise, which affects S wave results more than P wave results. S waves also appear more susceptible to the effects of plume focusing. We find the upper mantle to be highly attenuative, but that above 200–250 km the low‐velocity plume is progressively less attenuative than the adjacent mantle with decreasing depth. We conclude that water dissolved in mantle minerals causes the upper mantle to be highly attenuative, that the Yellowstone plume is only ~50°C warmer than the surrounding mantle below the North America lithosphere, and that melting within this plume begins at depths of 200–250 km. We attribute the lower attenuation in the partially molten plume to the dehydration of the solid matrix as water partitions into the melt. The source of the upper mantle hydration is attributed to subduction, including water flux from a hydrated transition zone and Laramide‐age shallow subduction and direct hydration of North America lithosphere.