Global multiple-frequency S-wave tomography

I am working on: high-resolution tomographic imaging of the Earth's mantle, using S-wave travel times measured in different frequency bands.

(1) << Better understanding the Earth's interior with seismic tomography >>

A large contribution to our knowledge of the structure and evolution of our planet has been obtained from the analysis of seismic waves triggered by earthquakes. As these waves are sensitive to some properties of the media through which they propagate, they contain information on the Earth's structure. Global seismic tomography aims at extracting this information from seismograms. The dynamics of our planet largely occurs in the mantle, which represents 84 per cent of the total volume of the Earth, and extends from the base of the crust (a shallow layer that does not exceed 70 km thickness), to the Core Mantle Boundary (CMB), at 2889 km depth. A detailed knowledge of the geometry and amplitude of seismic anomalies within the mantle is crucial for better constraining the physical parameters and the forms of mantle convection. This requires high resolution 3-D seismic images of the entire mantle, that can be compared directly with geodynamical and geochemical findings. The aim of my PhD/Post-doc research has been to obtain a high-resolution 3-D shear-wave tomographic model of the Earth's (lower) mantle, that could contribute to a better understanding of mantle dynamics (cf. FIGURE 1).


FIGURE 1: This picture shows, in 3-D, the shear-wave velocity anomalies in the whole-mantle as imaged by the recent technique of multiple-frequency tomography. Blue/red colours correspond to fast/slow velocity regions, with respect to the 1-D reference model IASP91. This first model was built during Zaroli's PhD thesis (2010). It has further been improved and published in Zaroli et al. (2013). [click to enlarge]


(2) << Multiple-frequency tomography: towards higher-resolution seismic imaging >>

The "resolution" of global body wave 3-D seismic tomographic models has significantly improved in the last 25 years, thanks to growing international seismic networks, improved computational facilities, and development of new seismological tools which extract more information from seismograms. Until recently, ray theory (RT) formed the backbone of global seismic tomography, mainly because of its simplicity and short computing time, and has largely contributed to better resolve the 3-D seismic structure of the deep Earth (e.g. Grand et al. 1997; Albarede & van der Hilst, 1999; Fukao et al. 2001; Romanowicz 2003). RT is based on the approximation that the travel time of a body wave only depends upon the 3-D seismic structure along the (infinitesimally narrow) geometrical ray path. This assumes that seismic waves have an infinite-frequency, or a zero wavelength. In reality, seismic waves have wavelengths ranging from 10 to 1,000 km, or even more. RT is then only applicable if the heterogeneities are much larger than the Fresnel zone. Therefore, it breaks down when used for imaging small scale heterogeneities, because diffraction effects make travel time (and amplitude) anomalies dependent on Earth structure in a 3-D region around the geometrical ray path. Since wave diffraction phenomena are not taken into account in RT based seismic tomography, it seems that progress towards "higher-resolution" imaging of small-scale features present in the mantle requires a movement away from RT. Recently, "finite-frequency" approaches (in contrast to the "infinite-frequency" RT) have emerged in seismic tomography (e.g. Marquering et al. 1998; Dahlen et al. 2000; Zhao et al. 2000; Komatitsch et. al 2002; Calvet & Chevrot 2005; Tromp et al. 2005; Nissen-Meyer & Dahlen 2007), in order to take wave diffraction effects into account, and make the imaging of smaller objects possible. In "finite-frequency" tomography, body wave travel time (and amplitude) anomalies are "frequency-dependent", and the geometrical ray paths are replaced by volumetric sensitivity Frechet kernels (cf. FIGURE 2). Therefore, measuring travel times at several periods increases the amount of independent informations, since at each period the seismic waveform is influenced by a different weighted average of the Earth's structure, through the corresponding 3-D sensitivity kernel. One can exploit this frequency-dependency, by simultaneously inverting body wave travel times from different frequency bands, which is called "multiple-frequency tomography".

FIGURE 2: Example of analytical S phase Frechet kernels at 34 s period. Black dashed line: geometrical ray path. Source depth: 0 km. Epicentral distance : 64.5 degrees (Zaroli, 2010, PhD thesis; Zaroli et al, 2013). [click to enlarge]

Some results

FIGURE 3: Zaroli et al, 2013, Solid Earth [click to enlarge]