Multiscale imaging of complex structures from multifold wide-aperture seismic data by frequency-domain full-waveform tomography: application to a thrust belt

Abstract

S U M M A R YAn application of full-waveform tomography to dense onshore wide-aperture seismic data recorded in a complex geological setting (thrust belt) is presented.The waveform modelling and tomography are implemented in the frequency domain. The modelling part is solved with a finite-difference method applied to the visco-acoustic wave equation. The inversion is based on a local gradient method. Only the P-wave velocity is involved in the inversion. The inversion is applied iteratively to discrete frequency components by proceeding from low to high frequencies. This defines a multiscale imaging in the sense that high wavenumbers are progressively incorporated in images. The linearized waveform tomography requires an accurate starting velocity model that has been developed by firstarrival traveltime tomography.After specific pre-processing of the data, 16 frequency components ranging between 5.4 and 20 Hz were inverted. Ten iterations were computed per frequency component leading to 160 tomographic models. The waveform tomography has successfully imaged southwestdipping structures previously identified from other geophysical data as being associated with high-resistivity bodies. The relevance of the tomographic images is locally demonstrated by comparison of a velocity-depth function extracted from the waveform tomography models with a coincident vertical seismic profiling (VSP) log available on the profile. Moreover, comparison between observed and synthetic seismograms computed in the (starting) traveltime and waveform tomography models demonstrates unambiguously that the waveform tomography successfully predicts for wide-angle reflections from southwest-dipping geological structures.This study demonstrates that the combination of first-arrival traveltime and frequencydomain full-waveform tomographies applied to dense wide-aperture seismic data is a promising approach to quantitative imaging of complex geological structures. Indeed, wide-aperture acquisition geometries offer the opportunity to develop an accurate background velocity model for the subsequent waveform tomography. This is critical, because the building of the macromodel remains an open question when only near-vertical reflection data are considered.Key words: finite-difference methods, thrust belt, traveltime and full waveform inversions, wide-aperture seismic data. I N T RO D U C T I O NSeismic imaging of complex structures characterized by strong lateral variations in the velocity field remains a challenge. Usually, the seismic imaging problem is applied to conventional surface-seismic multichannel reflection geometries which involve only near-vertical wave propagations (namely, short-angle reflections). In such geometries, the seismic imaging problem can be subdivided into two distinct tasks. Chronologically, the first one is the determination of a smooth background velocity model which describes the large-scale velocity distribution. This macromodel building is generally developed by traveltime tomography or migration-based velocity analys...

A parallel evolution strategy for an earth imaging problem in geophysics

“…For many years, FATT has proven to be stable in generating smooth velocity models of the subsurface. Some examples of application of FWI to real data using a starting model built by FATT are described in [32,37]. Similarly, the stereotomography is regarded as one of the most promising methods for building a smooth velocity model.…”

Simultaneous inversion for velocity and attenuation by waveform tomography

“…This Fréchet matrix can be worked out as the following (Pratt et al, 1998;Ravaut et al, 2004). Taking the first-order derivative to Eq.…”

“…As for the step length k α , it is calculated by a parabolic line-search method (Ravaut et al, 2004, Sourbier et al, 2009. In this method, three step lengths, 0, 1 α and 2 α , are selected first.…”

Acceleration strategies for elastic full waveform inversion workflows in 2D and 3D

“…1 is typically an irregular surface, and hence, finding a global minimum can be challenging if not impossible. In this context, multi-scale techniques [4,9,32,35,40,48], combined with gradient preconditioning and regularization methods, have been developed to sequentially incorporate higher wavenumber information into the inverted model. This results in a reduction of local minima effects by means of either selecting increasing individual frequency samples (in frequency domain) or broadening a low-pass filter by increasing its corner frequency (in time domain).…”