Diffraction-angle filtering of gradient for acoustic full-waveform inversion

Geophysics ◽  
2020 ◽  
pp. 1-87
Author(s):  
Ju-Won Oh ◽  
Jiubing Cheng ◽  
Dong-Joo Min

Seismic full-waveform inversion (FWI) estimates the subsurface velocity structures by reducing data misfit between observed and modeled data. Simultaneous matching of transmitted and reflected waves in seismic FWI causes different updates of different wavenumber components of a given model depending on the diffraction angle between incident and diffracted rays. Motivated by the inverse scattering imaging condition and elastic full-waveform inversion, we propose applying a diffraction-angle filtering technique in acoustic FWI, which enables us to separate transmission and reflection energy in the partial derivative wavefields. The diffraction-angle filtering is applied to the virtual source, which is the model parameter perturbation acting as a source for the partial derivative wavefields. The diffraction-angle filtering consists of two diffraction-angle filters (DAF), DAF-I and DAF-II. DAF-I, which is derived from the particle acceleration of the incidence wavefields, suppresses energies at either small or large diffraction angles by simply changing the sign of the weighting factor. DAF-I is exactly identical to the conventional inverse scattering approach. DAF-II, which is derived from the artificial shear strain of the incident P-wave, additionally suppresses energies at intermediate diffraction angles. With this mechanism, we can design various types of diffraction-angle filtering to control the updates of wavenumber components of the misfit gradient with respect to the P-wave velocity. For the synthetic Marmousi-II data and real ocean-bottom seismic data from the North Sea, we demonstrate that the diffraction-angle filtering enables us to control low-, intermediate- and high-wavenumber components of the gradient direction.

Geophysics ◽  
2021 ◽  
pp. 1-52
Author(s):  
Yuzhu Liu ◽  
Xinquan Huang ◽  
Jizhong Yang ◽  
Xueyi Liu ◽  
Bin Li ◽  
...  

Thin sand-mud-coal interbedded layers and multiples caused by shallow water pose great challenges to conventional 3D multi-channel seismic techniques used to detect the deeply buried reservoirs in the Qiuyue field. In 2017, a dense ocean-bottom seismometer (OBS) acquisition program acquired a four-component dataset in East China Sea. To delineate the deep reservoir structures in the Qiuyue field, we applied a full-waveform inversion (FWI) workflow to this dense four-component OBS dataset. After preprocessing, including receiver geometry correction, moveout correction, component rotation, and energy transformation from 3D to 2D, a preconditioned first-arrival traveltime tomography based on an improved scattering integral algorithm is applied to construct an initial P-wave velocity model. To eliminate the influence of the wavelet estimation process, a convolutional-wavefield-based objective function for the preprocessed hydrophone component is used during acoustic FWI. By inverting the waveforms associated with early arrivals, a relatively high-resolution underground P-wave velocity model is obtained, with updates at 2.0 km and 4.7 km depth. Initial S-wave velocity and density models are then constructed based on their prior relationships to the P-wave velocity, accompanied by a reciprocal source-independent elastic full-waveform inversion to refine both velocity models. Compared to a traditional workflow, guided by stacking velocity analysis or migration velocity analysis, and using only the pressure component or other single-component, the workflow presented in this study represents a good approach for inverting the four-component OBS dataset to characterize sub-seafloor velocity structures.


Geophysics ◽  
2019 ◽  
Vol 84 (5) ◽  
pp. B311-B324 ◽  
Author(s):  
Laura Gassner ◽  
Tobias Gerach ◽  
Thomas Hertweck ◽  
Thomas Bohlen

Evidence for gas-hydrate occurrence in the Western Black Sea is found from seismic measurements revealing bottom-simulating reflectors (BSRs) of varying distinctness. From an ocean-bottom seismic data set, low-resolution traveltime-tomography models of P-wave velocity [Formula: see text] are constructed. They serve as input for acoustic full-waveform inversion (FWI), which we apply to derive high-resolution parameter models aiding the interpretation of the seismic data for potential hydrate and gas deposits. Synthetic tests indicate the applicability of the FWI approach to robustly reconstruct [Formula: see text] models with a typical hydrate and gas signature. Models of S-wave velocity [Formula: see text] containing a hydrate signature can only be reconstructed when the parameter distribution of [Formula: see text] is already well-known. When we add noise to the modeled data to simulate field-data conditions, it prevents the reconstruction of [Formula: see text] completely, justifying the application of an acoustic approach. We invert for [Formula: see text] models from field data of two parallel profiles of 14 km length with a distance of 1 km. Results indicate a characteristic velocity trend for hydrate and gas occurrence at BSR depth in the first of the analyzed profiles. We find no indications for gas accumulations below the BSR on the second profile and only weak indications for hydrate. These differences in the [Formula: see text] signature are consistent with the reflectivity behavior of the migrated seismic streamer data of both profiles in which a zone of high-reflectivity amplitudes is coincident with the potential gas zone derived from the FWI result. Calculating saturation estimates for the potential hydrate and gas zones yields values of up to 30% and 1.2%, respectively.


Geophysics ◽  
2018 ◽  
Vol 83 (2) ◽  
pp. R129-R140 ◽  
Author(s):  
Ju-Won Oh ◽  
Mahesh Kalita ◽  
Tariq Alkhalifah

We have developed an efficient elastic full-waveform inversion (FWI) based on the P-wave excitation amplitude (maximum energy arrival) approximation in the source wavefields. Because, based on the P-wave excitation approximation (ExA), the gradient direction is approximated by the crosscorrelation of source and receiver wavefields at only excitation time, it estimates the gradient direction faster than its conventional counterpart. In addition to this computational speedup, the P-wave ExA automatically ignores SP and SS correlations in the approximated gradient direction. In elastic FWI for ocean bottom cable (OBC) data, the descent direction for the S-wave velocity is often degraded by undesired long-wavelength features from the SS correlation. For this reason, the P-wave excitation approach increases the convergence rate of multiparameter FWI compared with the conventional approach. The modified 2D Marmousi model with OBC acquisition is used to verify the differences between the conventional method and ExA. Finally, the feasibility of the proposed method is demonstrated on a real OBC data from the North Sea.


Geophysics ◽  
2016 ◽  
Vol 81 (6) ◽  
pp. R363-R383 ◽  
Author(s):  
Patrick Amestoy ◽  
Romain Brossier ◽  
Alfredo Buttari ◽  
Jean-Yves L’Excellent ◽  
Theo Mary ◽  
...  

Wide-azimuth long-offset ocean bottom cable (OBC)/ocean bottom node surveys provide a suitable framework to perform computationally efficient frequency-domain full-waveform inversion (FWI) with a few discrete frequencies. Frequency-domain seismic modeling is performed efficiently with moderate computational resources for a large number of sources with a sparse multifrontal direct solver (Gauss-elimination techniques for sparse matrices). Approximate solutions of the time-harmonic wave equation are computed using a block low-rank (BLR) approximation, leading to a significant reduction in the operation count and in the volume of communication during the lower upper (LU) factorization as well as offering great potential for reduction in the memory demand. Moreover, the sparsity of the seismic source vectors is exploited to speed up the forward elimination step during the computation of the monochromatic wavefields. The relevance and the computational efficiency of the frequency-domain FWI performed in the viscoacoustic vertical transverse isotropic (VTI) approximation was tested with a real 3D OBC case study from the North Sea. The FWI subsurface models indicate a dramatic resolution improvement relative to the initial model built by reflection traveltime tomography. The amplitude errors introduced in the modeled wavefields by the BLR approximation for different low-rank thresholds have a negligible footprint in the FWI results. With respect to a standard multifrontal sparse direct factorization, and without compromise of the accuracy of the imaging, the BLR approximation can bring a reduction of the LU factor size by a factor of up to three. This reduction is not yet exploited to reduce the effective memory usage (ongoing work). The flop reduction can be larger than a factor of 10 and can bring a factor of time reduction of around three. Moreover, this reduction factor tends to increase with frequency, namely with the matrix size. Frequency-domain viscoacoustic VTI FWI can be viewed as an efficient tool to build an initial model for elastic FWI of 4C OBC data.


Geophysics ◽  
2010 ◽  
Vol 75 (6) ◽  
pp. R109-R119 ◽  
Author(s):  
Timothy J. Sears ◽  
Penny J. Barton ◽  
Satish C. Singh

Elastic full waveform inversion of multichannel seismic data represents a data-driven form of analysis leading to direct quantification of the subsurface elastic parameters in the depth domain. Previous studies have focused on marine streamer data using acoustic or elastic inversion schemes for the inversion of P-wave data. In this paper, P- and S-wave velocities are inverted for using wide-angle multicomponent ocean-bottom cable (OBC) seismic data. Inversion is undertaken using a two-dimensional elastic algorithm operating in the time domain, which allows accurate modeling and inversion of the full elastic wavefield, including P- and mode-converted PS-waves and their respective amplitude variation with offset (AVO) responses. Results are presented from the application of this technique to an OBC seismic data set from the Alba Field, North Sea. After building an initial velocity model and extracting a seismic wavelet, the data are inverted instages. In the first stage, the intermediate wavelength P-wave velocity structure is recovered from the wide-angle data and then the short-scale detail from near-offset data using P-wave data on the [Formula: see text] (vertical geophone) component. In the second stage, intermediate wavelengths of S-wave velocity are inverted for, which exploits the information captured in the P-wave’s elastic AVO response. In the third stage, the earlier models are built on to invert mode-converted PS-wave events on the [Formula: see text] (horizontal geophone) component for S-wave velocity, targeting first shallow and then deeper structure. Inversion of [Formula: see text] alone has been able to delineate the Alba Field in P- and S-wave velocity, with the main field and outlier sands visible on the 2D results. Inversion of PS-wave data has demonstrated the potential of using converted waves to resolve shorter wavelength detail. Even at the low frequencies [Formula: see text] inverted here, improved spatial resolution was obtained by inverting S-wave data compared with P-wave data inversion results.


Geophysics ◽  
2020 ◽  
pp. 1-122
Author(s):  
Nishant Kamath ◽  
Romain Brossier ◽  
Ludovic Métivier ◽  
Arnaud Pladys ◽  
Pengliang Yang

Full waveform inversion (FWI) applications on 3D ocean bottom cable (OBC) data fromthe Valhall oil field in the North Sea have demonstrated the importance of appropriately ac-counting for attenuation. The Valhall field contains unconsolidated shallow sediments anda low velocity anomaly in its center - indicative of gas clouds - which have a significantattenuation imprint on the data. The challenge in which we are interested is to performtime-domain visco-acoustic 3D FWI, which requires more sophisticated tools than in thefrequency domain wherein attenuation can be incorporated in a straightforward manner.The benefit of employing a visco-acoustic, instead of a purely acoustic, modeling engineis illustrated. We show that, in the frequency band employed (2.5 - 7.0 Hz), it is betterto reconstruct velocity only keeping attenuation fixed, because simultaneous inversion ofvelocity and quality factor Q does not provide reliable Q-updates. We design an efficienttime-domain workflow combining a random source decimation algorithm, modeling usingstandard linear solid mechanisms, and wavefield preconditioning. Our results are similarto those obtained from state-of-the-art frequency-domain algorithms, at a lower computa-tional cost compared to conventional checkpointing techniques. We clearly illustrate theimprovement in terms of imaging and data fit achieved when accounting for attenuation.


Sign in / Sign up

Export Citation Format

Share Document