A long-standing question in the study of Earth’s deep interior is the origin of seismic mantle heterogeneity. The challenge is to efficiently mine the wealth of information available in complex seismic waveforms and to separate the potential contributions of thermal anomalies and compositional variations. High expectations to gain new insight currently lie within the application of high-performance computing to geophysical problems. Modern supercomputers allow, for example, the simulation of global mantle flow at Earth-like convective vigor or seismic wave propagation through complex three-dimensional structures. The sophisticated computational tools incorporate a variety of physical phenomena and result in synthetic datasets that show a complexity comparable to real observations. However, it is so far not clear how to combine the results from the various disciplines in a consistent manner to obtain a better understanding of deep Earth structure from the expensive large-scale numerical simulations. In particular, it is important to understand how to build conceptual models of Earth’s mantle based on geodynamic considerations that can be quantitatively assessed and used to test specific hypotheses. One specific goal is to generate seismic heterogeneity from dynamic flow calculations that can be used in global wave propagation simulations so that synthetic seismograms can be directly compared to seismic data without the need to perform inversions.

In the multi-disciplinary study presented here, a new method is developed to theoretically predict and assess seismic mantle heterogeneity. Forward modeling of global mantle flow is combined with information from mineral physics and seismology. Temperatures inside the mantle are obtained by generating a new class of mantle circulation models at very high numerical resolution. The global average grid spacing of ~25 km (around 80 million finite elements) allows for the simulation of flow at Rayleigh numbers on the order of 10^9 and to resolve a thermal boundary layer thickness of around 100 km. To assess the predicted present day temperature fields, the geodynamic flow calculations are post-processed with published thermodynamically self-consistent models of mantle mineralogy for a pyrolite composition to convert thermal structure into elastic parameters. Quantitative predictions of the magnitudes of seismic velocity and density variations are thereby possible due to the appropriately high numerical resolution necessary to obtain temperature variations that are consistent with the mineralogical conversion. The resulting structures are compared to tomographic models based on a variety of statistical measures taking into account the limited resolving power of the seismic data. In a final step, the geodynamic models are investigated with respect to the influence of strong convective mass transport on the stability of Earth’s rotation axis. This additional and independent analysis provides information on whether strongly bottom heated isochemical mantle circulation can be reconciled with paleomagnetic estimates of true polar wander. One specific question that can be addressed with this approach is the origin of two large regions of strongly reduced seismic velocities in the lowermost mantle. Several seismological observations are interpreted as being caused by compositional variations. However, a large number of recent geodynamical, mineralogical and also seismological studies argue for a strong thermal gradient across the core-mantle boundary that might provide an alternative explanation through the resulting large temperature variations. Here, the forward modeling approach is used to test the assumption whether the presence of a strong thermal gradient in isochemical whole mantle flow is compatible with a variety of geophysical observations.

The results show that the temperature variations deduced from the new high-resolution mantle circulation models are capable of explaining gross statis