Full text loading...
-
Elastic Model Low-To-Intermediate Wavenumber Inversion Using Reflection Traveltime and Waveform of Multicomponent Data
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 80th EAGE Conference and Exhibition 2018, Jun 2018, Volume 2018, p.1 - 5
Abstract
If the starting velocity model does not appropriately contain low-to-intermediate components to avoid cycle skipping, full-waveform inversion (FWI) is likely to converge to a local minimum. Generally, these components could be retrieved from diving or/and refracted waves. However, long offsets are required to record these transmitted waves. For better depth penetration, reflected waves should be used to recover the long-wavelength velocity structures. Retrieving background P- and S-wave velocities with multicomponent data is more difficult. In this abstract, we propose a three-stage inversion of the background P- and S-wave velocities using reflectioa traveltime and waveform of the multicomponent seismograms, respectively.
For traveltime inversion, we use a scalar acoustic propagator to extrapolate the normal and adjoint wavefields in the first two stages. This can avoid artefacts in the calculated gradients due to the non-phyiscal mode conversion and save lots of computational resources. To honor the waveforms of the multi-component data, we use an elastic propagator to extrapolate the normal and adjoint vector wavefields. The gradients are preconditioned through P/S mode decomposition to mitigate the artefacts and to improve the two-parameter inversion in the last stage. Numerical example will demonstrate the validaty of this hierarchical inversion approach.