Towards full Waveform Box Tomography in the Deep Mantle: the Case when both Sources and Stations are Outside the Target Region

Abstract : The seismic wavefield can now be accurately computed in earth models of arbitrary complexity using numerical methods, i.e. the Spectral Element Method (SEM), which holds great promise for full waveform seismic tomography. However, the computational cost is one of the limiting factors of this approach. One way of addressing this consists in limiting the size of the region one wishes to image. So far, proposed approaches have either restricted the dataset to events and stations located within the region, or have considered hybrid methods in which the wavefield is calculated using an approximate, faster solver outside of the region, and a 3D solver within the region. We have proposed a methodology (“Box tomography”), that allows us to couple full 3D solvers outside and inside the target region. We first replace the global wavefield by secondary sources on the border of the target region. We then extrapolate the wavefield to stations outside of it by convolving it with precomputed Green’s functions from the box boundaries to the stations (Masson and Romanowicz, 2017,GJI). This methodology is quite general and does not depend on the particular numerical solver. Recently, Clouzet et al. (2018,GJI) presented its first application to the development of a radially anisotropic shear velocity model of the upper mantle in north America. However, while sources were both inside or outside the target region, stations were all located inside the region. The wavefield outside of the box was computed only once in a reference 3D global model, while iterations of the model and wavefield computations were restricted to the target region, using a “regional” solver, in this case RegSEM (Cupillard et al., 2012,GJI). Here, we present the extension of this approach to the case when both events and stations are outside of the target region. A “global” solver is used for the propagation of the wavefield in a reference 3D model outside of the target region and for the computation of the Green's functions from the boundaries to the external receiver. With this methodology, one can compute seismograms to higher frequency at reasonable computational cost, and hence image the target region at a finer scale. We show preliminary results of application of this method to modelling the deep structure under the Yellowstone hotspot.
Document type :
Conference papers
Complete list of metadatas
Contributor : Yder Masson <>
Submitted on : Tuesday, October 9, 2018 - 1:42:57 PM
Last modification on : Wednesday, June 26, 2019 - 2:26:05 PM


  • HAL Id : hal-01891223, version 1



Sevan Adourian, Yder J. Masson, Barbara Romanowicz. Towards full Waveform Box Tomography in the Deep Mantle: the Case when both Sources and Stations are Outside the Target Region. AGU Fall Meeting, Dec 2018, Washington, D.C, United States. ⟨hal-01891223⟩



Record views