# Exploitation de la cohérence locale des données sismiques pour l'imagerie du sous-sol

Abstract : My research activities deal with seismic imaging, mainly in the context of oil and gas exploration. The objective is to derive quantitative images of the sub-surface such as the P-wave velocity model, from seismic measurements recorded at the surface. In that context, I endeavored to look at seismic data in a different manner. Seismic input gathers are function of the source and receiver positions and the recorded time. Individual traces are usually processed separately, for example in the case of Kirchhoff migration to derive the velocity contrasts at the correct position in the depth domain. For alternative methods, such as reverse-time migration, all traces within a single shot are simultaneously back-propagated from the surface. I propose to select a group of few adjacent traces localized around some time or depth values and to reconsider different imaging processing tasks from this angle. The local aspect is justified in the context of oil and gas exploration with the increasingly dense acquisition surveys, but also with the notion of Fresnel zone. The local coherency of seismic data is tackled through two approaches. Firstly, locally coherent events are simply described in 2D by the position of the central trace, a position within the trace and a slope indicating the local coherency. I demonstrate the feasibility to invert for the low frequency components of the velocity models by picking locally coherent events in the depth migrated section. I also show the equivalence between this approach and slope travel time tomography defined in the unmigrated time data. Secondly, I take into account the signature of the data, in particular the band-limited aspect, with the use of curvelets. More applications become then possible: migration, velocity analysis, sensitivity of the migrated section with respect to the velocity model, and other tasks related to the pre-processing steps such as denoising, multiple prediction, ... In each case, I analyze the advantages and limitations of the notion of locallly coherent events. It appears that the choice of the decomposition scheme, e.g. through curvelets, is crucial for subsequent processing. If for example one needs to compress seismic data, then curvelets are not useful. If on the contrary the objective is to derive how a given migrated section depends on the velocity model used for migration, then curvelets are powerful. More precisely, curvelets are useful to suppress coherent noise, with a flexible decomposition scheme. They almost diagonalize the demigration/migration operator providing the sensitivity of a migrated section with respect to the velocity model. Finally, curvelets are beneficial to exploit the multi-scale aspect of seismic data, e.g. to remove aliased data. I also show the possibility to develop new decomposition and reconstruction schemes such as circlets''. As research perspectives, my objective is to go beyond the high frequency approximation for which curvelets are naturally defined. I propose in particular a new formulation for the full waveform inversion, referred as the Differential Full WaveForm Inversion. This method aims at retrieving the sub-surface properties with a local optimization scheme, without the need to derive a precise initial velocity model closed to the solution. More work is needed. This method relies on the local coherency of data and supposes that the signal is well sampled in time and space. Potential applications deal with oil and gas exploration, geotechnics, global and regional seismology.
Keywords :

Cited literature [128 references]

https://tel.archives-ouvertes.fr/tel-00535531
Contributor : Herve Chauris <>
Submitted on : Thursday, November 11, 2010 - 12:42:03 PM
Last modification on : Thursday, April 9, 2020 - 5:08:15 PM
Document(s) archivé(s) le : Saturday, February 12, 2011 - 2:35:42 AM

### Identifiers

• HAL Id : tel-00535531, version 1

### Citation

Hervé Chauris. Exploitation de la cohérence locale des données sismiques pour l'imagerie du sous-sol. Sciences de la Terre. Université Pierre et Marie Curie - Paris VI, 2010. ⟨tel-00535531⟩

Record views