Traveltime Dispersion in an Isotropic Elastic Mantle: Dominance of the Lower Mantle Signal in Differential-frequency Time Residuals

We study wavefield effects in elastic isotropic 3-D seismic structures derived from the temperature field of a high resolution mantle circulation model. More specifically, we quantify the structural dispersion of traveltime residuals of direct P- and S-waves in a model with realistic length-scales and magnitudes of the variations in seismic velocities and density. 3-D global wave propagation is simulated using a spectral element method, and traveltime residuals are measured in four different frequency bands by cross-correlation of 3-D and 1-D synthetic waveforms. Intrinsic (dissipative) attenuation is deliberately neglected, so that any variation of traveltimes with frequency can be attributed to structural effects. Additional simulations are performed for a model in which 3-D structure is removed in the upper 800 km to isolate the dispersion signal of the lower mantle. One question that we address is whether the structural length-scales inherent to a vigorously convecting mantle give rise to significant body-wave dispersion.

In our synthetic dataset, the difference between long-period and short-period traveltime residuals generally increases with increasing short-period residual. However, we do not find an exact linear dependence, and in case of P-waves even non-monotonic behaviour. At largest short-period residuals, average dispersion is on the order of 2 s for both P- and S-waves and even larger when structure is confined to the lower mantle. Dispersion also appears to be asymmetric; that is, larger for negative than for positive residuals. The standard deviations of both P- and S-wave residuals also increase with increasing period and we discuss possible explanations for this behaviour. Overall, wavefield effects in both models are generally stronger for P-waves than for S-waves at the same frequencies.

We also find that for certain combinations of periods, the difference between the respective residuals is very similar between the "whole mantle" and the "lower mantle" model. Signals from upper mantle structure seem to mostly cancel in these cases, at least for certain length-scales. "Differential-frequency" traveltime residuals could thus represent a new seismological observable that may prove useful for tomographic studies of the lower mantle.