Comparison of multifrequency selection strategies for simultaneous-source full-waveform inversion

Abstract

The computational cost of full-waveform inversion (FWI) is a major obstacle to estimating the velocity model for large-scale problems. One way to reduce the overall cost of waveform inversion is by adopting a simultaneous-source strategy. In other words, multiple sources are simultaneously fired to simulate supershot gathers and thereby reduce the number of seismic modeling simulations that are performed during the inversion. However, the use of simultaneous sources introduces crosstalk artifacts that arise from the interference among the sources that constitute a supershot. We analyzed the influence of different simultaneous multifrequency selection strategies on crosstalk artifacts. Our analysis focused on a frequency-domain FWI algorithm that is implemented with simultaneous sources that are randomly encoded with random time shifts. In the multiscale conventional FWI strategies, a finite set of discrete frequencies was selected and the inversion was carried out sequentially from low- to high-frequency data components. First, the long wavelength components of model parameters were recovered from the low-frequency data, and then more details and features were extracted as the inversion proceeded to the higher frequency data. We examined six frequency selection strategies and tested the performance of the algorithm with encoded data sets. Numerical tests showed that high-fidelity results could be attained by inverting partially overlapped groups of temporal frequencies. Our FWI algorithm is based on a matrix-free Gauss-Newton method. To mitigate crosstalk artifacts during the numerical inversion, a new encoding was generated at every iteration. We also found that high-resolution images can be obtained by resampling new source positions and new encoding functions at every iteration of the FWI algorithm.

Estimation of elastic constants for HTI media using Gauss-Newton and full-Newton multiparameter full-waveform inversion

“…We could potentially also restrict the inverse Hessian construction to the diagonal elements of the off-diagonal blocks, i.e., the parameter-type approximation (Innanen, 2014a) for reducing the computational cost, for which an adjoint-state method would need to be developed. Phase-encoding methods have been widely studied for calculating the gradient (Vigh and Starr, 2008;Tang, 2009;Anagaw and Sacchi, 2014;Pan et al, 2014a) or Hessian approximations (Castellanos et al, 2015). The phase-encoding methods can also be used in multiparameter FWI to calculate the gradient and multiparameter Hessian for reducing the computational burden (Pan et al, 2015a).…”

Application of shifted-Laplace preconditioners for heterogenous Helmholtz equation- part 2: Full waveform inversion

“…To solve for the physical properties of the subsurface, we run Gauss-Newton method with and without preconditioners. All of the frequencies are inverted at the same time (For different frequency selection strategies see Anagaw and Sacchi (2014)). We used fixed mesh size for all of the frequency realizations.…”

“…In frequency domain algorithms only limited number of frequency realizations needs to be inverted. Moreover, there is a possibility of inverting the frequency realizations with different grouping strategies to have a better convergence rate (Anagaw and Sacchi, 2014). One can also use the grouping strategies in linearized inversion problems of seismic imaging (Kazemi and Sacchi, 2015).…”

Quantitative imaging using ultrasonic tomography based on full waveform inversion

“…More details of the Hessian used to improve the descent direction and the smoothing regularization constraint can be found in [189,195,196]. Noticeable artifacts are observed from both imaging in the experiment.…”