 |
| Titel |
Importance of a 3D forward modeling tool for surface wave analysis methods |
| VerfasserIn |
Damien Pageot, Mathieu Le Feuvre, Donatienne Leparoux, Philippe Côte, Capdeville Yann |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250131406
|
| Publikation (Nr.) |
 EGU/EGU2016-11812.pdf |
|
|
|
|
|
| Zusammenfassung |
| Since a few years, seismic surface waves analysis methods (SWM) have been widely
developed and tested in the context of subsurface characterization and have demonstrated
their effectiveness for sounding and monitoring purposes, e.g., high-resolution tomography of
the principal geological units of California or real time monitoring of the Piton de la
Fournaise volcano. Historically, these methods are mostly developed under the assumption of
semi-infinite 1D layered medium without topography. The forward modeling is generally
based on Thomson-Haskell matrix based modeling algorithm and the inversion is driven by
Monte-Carlo sampling.
Given their efficiency, SWM have been transfered to several scale of which civil engineering
structures in order to, e.g., determine the so-called V s30 parameter or assess other critical
constructional parameters in pavement engineering. However, at this scale, many structures
may often exhibit 3D surface variations which drastically limit the efficiency of SWM
application. Indeed, even in the case of an homogeneous structure, 3D geometry can bias the
dispersion diagram of Rayleigh waves up to obtain discontinuous phase velocity curves
which drastically impact the 1D mean velocity model obtained from dispersion
inversion.
Taking advantages of high-performance computing center accessibility and wave propagation
modeling algorithm development, it is now possible to consider the use of a 3D elastic
forward modeling algorithm instead of Thomson-Haskell method in the SWM inversion
process. We use a parallelized 3D elastic modeling code based on the spectral element
method which allows to obtain accurate synthetic data with very low numerical dispersion
and a reasonable numerical cost.
In this study, we choose dike embankments as an illustrative example. We first show that their
longitudinal geometry may have a significant effect on dispersion diagrams of Rayleigh
waves. Then, we demonstrate the necessity of 3D elastic modeling as a forward problem for
the inversion of dispersion curves. |
|
|
|
|
|
|