Upper mantle meltdistribution from petrologicallyconstrained magnetotellurics

Three parameters: temperature, hydrogen content, and the presence of partial melt, are the dominant controls on the rheology of the convecting upper mantle. As such, they determine the dynamics that control plate tectonics and continental evolution. Since hydrogen depresses the peridotite solidus temperature, these parameters are strongly linked petrologically. We have developed a genetic algorithm code to statistically assess the likelihood that a section of upper mantle contains partial melt. This code uses magnetotelluric observations and petrological constraints on composition and solidus temperatures and allows for uncertainties in the geotherm and the electrical conductivity structure. We have applied this code to the convecting upper mantle beneath (1) a stable continent (the Superior Craton); (2) a hot spot (Tristan da Cunha); (3) stable, old oceanic lithosphere (the northwest Pacific Ocean); and (4) young oceanic lithosphere (adjacent to the East Pacific Rise). Results show that the volume of melt in the convecting upper mantle is heterogeneous. The highest melt proportions are beneath the hot spot while little to no melt is required in the other regions. All regions show low water contents (generally <50 wt ppm in olivine) in the shallow convecting upper mantle, making it unlikely that water causes a large or sharp viscosity contrast between the lithosphere and the convecting mantle. Results differ significantly for different experimental olivine hydrogen conductivity models, highlighting the importance of reconciling these experimental constraints.