Long-period surface waves and mantle boundary undulations

Abstract

SUMMARY We address two questions with synthetic experiments. (1) Is there sufficient signal in long-period surface wave data to provide useful constraints on topography on the internal boundaries in the upper mantle and around the Transition Zone? (2) What is the magnitude and nature of the bias caused by undulations on internal discontinuities on inversions for volumetric structures that do not also include boundary structures? The synthetic experiments rest on models of boundary topography on three upper mantle and Transition Zone discontinuities at 220, 400 and 660 km depth. The 660 km topographic model is that of Shearer & Masters (1992). We consider the effect of two different 400 km models; one is identical to Shearer and Master's model (Model A) and the other is their model sign flipped (Model B). The 220 km model is ad hoc, but has been motivated by the work of Revenaugh & Jordan (1989, 1991b) who argue that the 220 km discontinuity is not a global feature at all, but has a continental affinity. For simplicity, our 220 km model is global, but the jump in material properties across the discontinuity has been reduced relative PREM by a factor of two and is based on a continent-ocean model which is depressed up to 25 km below continents and ascends 10 km under oceans. These resulting boundary models have a significant effect on long-period fundamental spheroidal mode structure coefficients, with a magnitude on average of 15-30per cent relative to the volumetric models M84A + LO2.56 between harmonic degrees 20≤l≤75. Inversion for volumetric structures of the structure coefficients computed with the boundary models alone, devoid of volumetric structure, shows that simple mantle volumetric structures can be found to fit the boundary structure coefficients nearly perfectly. This demonstrates the essential trade-off between boundary and volumetric mantle structures using surface wave data alone. Destructive interference between the effects of the 660km boundary and Model B of the 400 km boundary requires a somewhat oscillatory volumetric mantle model to fit the structure coefficients, but the resulting volumetric model is no more oscillatory than current models of the upper mantle. Both boundary Models A and B produce on average a 30 per cent volumetric bias relative to M84A + LO2.56 from the base of the crust to 1000 km depth. The relative nature of the 400 km and 660 km undulations strongly affects the radial distribution of the volumetric bias, but has less of an effect on the average size of the bias.