The trade-off between volumetric and topographic structure for seismic traveltimes: 660 km topography and mantle structure

SUMMARY Synthetic body-wave traveltime inversion experiments were performed to evaluate the effect of unmodelled 660km discontinuity topography on the inference of aspherical, volumetric mantle structure. Synthetic residuals were computed for S waves refracted through a 660 km discontinuity with topography. We employed a spherical harmonic parametrization for lateral variations of the topography of the 660 km discontinuity and the mantle velocity structure. the radial dependence of the velocity structure in the mantle was parametrized in two ways: smooth Chebyshev polynomial functions and uniform shells. the results of both para-metrizations show that significant smearing of the input topographic signal appears in the models of volumetric mantle structure inferred from the synthetic data. Solving for higher order radial structure with the smooth Chebyshev functions reduces the smearing, but resolution is limited to the half-wavelength of the radial basis functions. More of the input synthetic residual variance is absorbed into the solution for volumetric structure by solving for higher order radial structure with smooth polynomials or with thinner shells directly below 660km depth. However, fundamental differences between the kernels for volumetric and topographic structure restrict recovery of the input signal to approximately 80 per cent. This work points to the possible value of relating volumetric and topographic structure when inverting seismic observations. This may be addressed by incorporating geodynamic modelling and mineral physics results into the modelling of seismic observations.