# The deep Earth

While we came upon evidence and convergent agreement on a number of deep-earth features such as major seismic discontinuities, phase transitions or deep slabs, we still struggle to find multi-disciplinary comfort on other ends: The termination of earthquake activity at 660 km depth, water in the mantle, or the nature of the reversing geomagnetic field are just a few sources of long-standing disputes in the geophysical community. Do plumes carry a significant amount of heat to the surface? How could we assess deep-mantle stratification due to chemical reservoirs seismically? What does seismic anisotropy tell us exactly about deep-mantle flow? What is the true scale length and amplitude of seismic heterogeneities? How does the interconnection between velocities, temperature, and chemical composition vary with depth and temperature for multiphase minerals? With seismic waves being the primary courier of information about the whole Earth, we can only further constrain these elusive problems by improvements in the actual data quality and coverage itself as well as innovative, new ways of seismic modeling, processing, and interpretation. Several intriguing developments are under way at both levels, such as general broadband digitalization, USArray, floating and ocean-bottom recording devices on the acquiring side, but also powerful seismic-wave propagation algorithms and improved inversion techniques accounting for finite frequencies and full waveforms on the digestive side. Our work primarily deals with the latter, and we approach the deep Earth from three seismic angles: high-frequency waveform modeling of strong 3D heterogeneities, tomographic imaging of volumetric structures, and considerations of core-mantle boundary topography.

## High-frequency 3D waveform modeling

(T. Nissen-Meyer, S. Hempel, E. Vanacore, C. Thomas, S. Rost, M. Thorne)The
axisymmetric spectral-element method AXISEM relies on
spherical symmetry. However, in a narrow spatial scale, it is feasible
to insert lateral heterogeneities which replicate 2.5D structures. We
apply this to ultra-low velocity zones (ULVZ), i.e. small-scale,
strongly heterogeneous structures above the core-mantle boundary and
other proposed deep-earth structures. To resolve these features, one
needs to achieve high seismic frequencies (0.2-1 Hz), for which the
collapsed-dimension approach of AXISEM is very well positioned. We
vary internal ULVZ structure and analyze its effect on the waveforms
of diverse phases such as PcP, ScP or SPdKS. This work is done in
close collaboration with groups in Munster, Utah, Leeds, and Tempe.

## Core-diffraction measurements

(K. Hosseini, T. Nissen-Meyer, K. Sigloch)Waves that graze around the core-mantle boundary cover a wider region within the lowermost mantle than any other seismic phase. Givent the central importance this region takes for mantle and core dynamics, it is highly desirable to consider as many of these waves as possible, both in the context of tomography and waveform modeling. Additionally, diffraction effects are due to frequency-dependent wave effects that cannot be appropriately modeled using asymptotic theory, such that full solutions to the wave equation are necessary. Further, these Pdiff and Sdiff phases can be recorded up to rather high frequencies (up to 0.3Hz), and therefore constrain multiple scales. We embark on an ambitious data analysis and processing project, assembling high-quality Pdiff measurements from around the globe at a vast range of frequency bands, and process these to obtain robust cross-correlation traveltimes. A database of 20,000 arrivals is already collected, and represents a crucial component for waveform tomography of the lowermost mantle.

## Supervirtual core-mantle-boundary interferometry

(P. Bharadwaj, T. Nissen-Meyer, G. Schuster, M. Mai)As mentioned above, core-diffracted waves can add valuable constraints on the dynamics of core and mantle. However, they are subject to considerable scattering and with increasing distance loose significant amounts of energy, especially at higher frequencies. Thus, at large distances, they drown within ambient noise and are unfortunately not usable in single trace analyis. To remedy this issue, we apply so-called supervirtual inteferometry to Pdiff arrivals and show that this stacking interferometric technique can significantly boost signal-to-noise ratios especially at large epicentral distances. Thus, we can enlargen the usable dataset of diffracted waves.

## Core-mantle boundary topography

(A. Colombi, T. Nissen-Meyer, L. Boschi)In global tomography, one assumes several spherical layers to be known, and inverts mainly for volumetric wavespeed perturbations with respect to a reference model. It is however known that several of these discontinuities contain significant topography, mostly related to hot upwellings or cold downwellings. We compute and analyze their effect on waveforms and sensitivity kernels and include these effects as additional inversion parameters. Specifically, we focus on the core-mantle boundary (CMB) which is crucial to both mantle dynamics and the magnetic field generated in the fluid outer core. We set up a synthetic experiment to investigate and quantify the resolving power of tomographic inversions for deep-earth structure and CMB undulations: We vary various critical parameters in fully 3D forward simulations such as core-mantle boundary topography, 3D velocity structure, source-receiver configuration. We thus obtain a database as a guide to classify the validity and imagability of a certain configuration, most notably related to core-mantle boundary topography, which is often ignored in the classical approach to seismic tomography using only volumetric properties of the Earth. We include the effect of weak boundary perturbations with respect to the reference model in the synthetic experiment and inversion. The knowledge gained from the synthetic experiment will be used as a guide to identify and parameterize joint inversions of actual data for core-mantle-boundary topography and volumetric structures.