I work to improve resolution of the structure of the deep Earth and integrate my results with the work of geodynamics, mineral physics, and geochemistry to better understand the Earth’s evolution. My tools are observational seismology and waveform modelling. I investigate structures in the mantle and core across a range of spatial scales often using single stations and seismic arrays of various sizes.

Mantle dynamics
Inner core growth
Contamination of deep Earth data
Alaskan subduction zone structure

I recently presented to the Earthscope Alaska group on imaging of the mantle and inner core using Alaskan data.

Watch me present on my methods and their applications to sharpening resolution of the outer core here:

Watch me present on my recent work on the inner core and the structure of the Alaskan subduction zone here:

Note that the some of the details of this work have since been updated.

The mantle:
Mantle convection constantly introduces heterogeneity into the mantle through subduction. Small, kilometre-scale heterogeneity distributed throughout the mantle is responsible for much of the energy that we observe in the high-frequency seismic wavefield. The distribution of this heterogeneity is likely controlled by the patterns of mantle convection. By observing scattered seismic energy and mapping the systematic distribution of the causative heterogeneities we can begin to track the movement of the deep Earth through time.


The scattering of seismic body wave energy by discrete volumes with sharply contrasting elastic parameters (bulk and shear moduli and density) is responsible for the long codas of energy that often follow the arrivals of major phases. The direct wave makes investigation of the coda waves challenging. However, for some ray geometries (such as PKPdf, the wave that samples the Earth’s mantle, outer and inner cores), scattered seismic waves can arrive as precursors the direct wave, thus can be more easily identified.

Seismic arrays, collections of closely spaced seismometers, allow resolution of the direction of the incoming energy. By identifying scattered seismic energy with seismic arrays I am able to precisely determine the direction, and thus the location in the Earth of the heterogeneity responsible for the scattering. With this approach I have mapped volumetric heterogeneity throughout the mantle under South Africa (Frost et al., 2013). Through analysis of the frequency content of the scattered energy I resolved the size of the heterogeneity to be around 3-7 km, and determined that heterogeneity within 300 km of the Core-Mantle Boundary (CMB) is more concentrated around the edges of large, convective mantle structure (the LLSVPs) (Frost et al., 2017a). I further mapped scattering heterogeneities within the mantle from the CMB to the surface, finding that heterogeneities are more common close to mantle features likely associated with convection (Frost et al., 2018).

The core:
The iron core occupies only 15% of the volume of the Earth. yet supplies a significant proportion of the heat budget, and is the source of the magnetic field. Interactions across the CMB may affect the composition of the Earth’s lower mantle, and modulate the pattern of the magnetic field. Meanwhile, the inner core affects the pattern of convection in the outer core, and thus likely the magnetic field. The growth of the inner core is likely recorded in its crystal structure, which may manifest as seismically resolvable velocity variations.


Through travel time anomalies on the body wave PKPdf, the seismic velocity in the inner core (IC) has been observed to depend on the direction along which is it is sampled, known as anisotropy. This is supported by measurements of core-sensitive free oscillations of the Earth that show anomalous splitting. The fast axis of anisotropy most is most likely aligned with the rotation axis, and observations suggest that the magnitude of the anisotropy shows hemispherical differences, with the western hemisphere of the IC being more anisotropic than the eastern hemisphere. However, sampling of Earth’s interior is limited by the global distribution of sources and receivers, which is particularly poor along paths close to the fast axis of anisotropy, thus observed patterns may not be representative of the IC.

In my work I aim to resolve the details of the velocity structure of the IC, and thus uncover its history so that we can understand its role in whole Earth evolution. I have applied array methods to study the low amplitude core wave PKPdf2 (or P′P′) at angles close to the fast axis of anisotropy and found that the upper outer core need not be as strongly anisotropy as was previously proposed (Frost and Romanowicz, 2017). Meanwhile, specific paths show stronger travel time anomalies, suggesting an origin outside of the IC (Frost et al., 2020).

New data from recently seismic deployments in Alaska and Antarctica have allowed me to better understand the pattern of anisotropy, especially along paths through the deep IC. I find that anisotropy continues to the core’s centre, and the fast axis remains aligned with the rotation axis, in contrast to several recent studies (Frost and Romanowicz, 2019). Furthermore, anisotropy increases further into the IC, but this pattern appears to depend not on depth, but on axial distance from the rotation axis (Frost et al., 2021).

Through mapping the structure of the core we will be better able to understand its growth history, and thus its influence on the rest of the Earth. I combine geodynamic models of core growth with mineral physics calculations of crystal texturing to resolve the pattern of core growth and its age. Our results suggest that the inner core slowly translated laterally in the plane of the equator from under Brazil towards Indonesia by about 0.3 radii during its 1.0 billion year lifetime (Frost et al., 2021).

Myself and several collaborators have been funded to extend this work to investigate the strain required to deform Iron in the inner core, the strength of flows in a growing inner core, and more detailed patterns of the inner core’s seismic structure, which will all build towards a holistic model of the inner core’s evolution.

The contamination of seismic images by mantle heterogeneity:
The seismic signals that penetrate into the deep Earth may contain interference from complexities in the shallow Earth. In particular, subduction introduces strong velocity heterogeneity into the mantle. Many seismic sources are in regions of subduction, and sometimes seismic stations may overlie these heterogeneities. I aim to understand how these plates act as a lens distorting the seismic waves and affecting our view of the deep Earth, especially the inner and outer cores core. This will help to better bring the life and history of the inner core into focus.


Adapting array methodology to lower frequency data, I have investigated the influence of mantle structure on the wave SmKS, which has been used to study structure in the upper outer core (OC). Arrays allow me to detect changes in the direction of the incoming wave, in addition to travel time anomalies, all of which help me to map the 3D structure of the mantle.

The recently deployed Alaskan element of Earthscope’s transportable array provides an opportunity to study the effects of upper mantle structure on the inner core wave PKPdf. The effect of the Alaskan slab can be seen in travel time anomalies that are usually attributed to the anisotropic character of the inner core. In an NSF funded project “Resolving the influence of mantle heterogeneity on estimates of inner core anisotropy”, I employed array methods to measure the slowness, back-azimuth, and travel time anomalies of PKPdf, ab, and bc across Alaska (Frost et al., 2020; Frost and Romanowicz, 2021). We use our recent tomographic model of Alaska (Roecker, Frost, and Romanowicz, 2018, AGU) to predict the effect of the Alaskan upper mantle on PKPdf wave direction and find a very similar pattern. We use then consider the finite frequency effects of the Alaskan upper mantle: we use waveform modelling to predict the effects of the Alaskan slab on the PKPdf wave and match our observations. This suggests that the PKPdf travel times observed in Alaska are at least partially contaminated by the upper mantle structure. This helps me to both improve the resolution of inner core anisotropy, and understand the geometry of the subducted slab at depth.

High resolution imaging of subduction under Alaska
Motivated by the observation that PKP travel times in Alaska can be predicted using tomographic models and by the desire to remove mantle contamination and use these data to study the inner core, myself and colleagues are working to construct a new tomographic model of the subducting Pacific plate and the surrounding mantle using Full Waveform Inversion. Full Waveform Inversion (FWI) seeks to explain all wiggles in the seismogram, not just the travel time of the first arrival, massively increasing the amount of information used.

The Pacific plate subducting under Alaska.

By exploring the mantle beneath Alaska we will learn about the tectonic processes operating in the northern hemisphere for the past 200 million years, and our model may help to better explain volcanism and seismicity in Alaska.

%d bloggers like this: