Sp receiver functions generated at melting onset: insight from geodynamic models and synthetic receiver functions

Recent studies of converted seismic phases in regions of mantle upwelling and melting have identified Sp phases originating from a depth similar to the expected depth of melting onset. Using 1D melt migration models together with 1D synthetic receiver functions, we have investigated the conditions required to generate the observed converted phases. The magnitude of the synthetic melting onset phases increases with increasing upwelling velocity, increasing shear viscosity, and decreasing water content. The water content of erupted lavas at Hawaii and Galapagos suggest a hydrated mantle source. For the standard relationship of seismic velocity on melt fraction used here, if water is homogeneously distributed within the mantle source, the resulting seismic velocity gradients would not produce the observed converted phases. One way to explain the observed phases is that wet and dry components in the mantle source are heterogeneously distributed and melt independently. A relatively dry plume that entrains water as it passes through a hydrated transition zone could result in favorable conditions for generation the observed phases. Alternatively, experiment and theory show a sharp reduction in relaxed elastic modulus at the onset of melting, attributed to decreasing diffusion length scales upon the formation of an interconnected melt network. Thus, a stronger dependence of seismic velocity on low melt fractions at melting onset may account for the observed converted phases.