Analysis of the covariance matrix in FWI through density of covariance maps
Abstract
Full waveform inversion (FWI) is a tool for the inversion of seismic data. There are several sources of uncertainty in the results provided by FWI. The quantification of such uncertainties has been studied through the resolution matrix (Res), which rests on a quadratic approximation that interprets the Hessian matrix as the posterior covariance matrix. Despite efforts in the use of Res, there is no published analysis of the uncertainties contained in the full correlation matrix, (R). Our approach leads to build the full R matrix, which, at the end of the day, is the final quantity that includes all the information associated with uncertainties.
We focused on uncertainties related to variation in the starting models of the FWI, and thus propose a method to study the full R matrix, which is-called the Density of Correlation Map, D. By using the D map, we found that the highest uncertainty zones in the FWI inverted model are near the sources, the model boundaries, and the interfaces. We argue that D can be a complement for the study and estimation of uncertainties in FWI.
References
Biswas R, Sen M (2017) 2d full-waveform inversion and uncertainty estimation using the reversible jump Hamiltonian Montecarlo. SEG Technical Program Expanded Abstracts, DOI https://doi.org/10.1190/segam2017-17680416.1
Lailly P (1983) The seismic inverse problem as a sequence of before stack migrations. In: Bednar JB, Redner R, Robinson E, Weglein A (eds) Conference on inverse scattering: theory and application, SIAM, pp 206-220
Tarantola A (1984) Inversion of seismic reflection data in the acoustic approximation. Geophysics, 49(8):1259-1266
Brossier R, Operto S, Virieux J (2009) Seismic imaging of complex structures by 2d elastic frequency-domain full-waveform inversion. Geophysics, DOI https://doi.org/10.1190/1.3215771
Brenders A, Albertin U, Mika J (2012) Comparison of 3d time- and frequency-domain waveform inversion: Benefits and insights of a broadband, discrete-frequency strategy. SEG Technical Program Expanded Abstracts, DOI https: //doi.org/10.1190/segam2012-1299.1
Vigh D, Starr B, Kapoor J, Li H (2010) 3d full waveform inversion on a gulf of Mexico waz data set. SEG Technical Program Expanded Abstracts, DOI https://doi.org/10.1190/1.3513935
Fitchner A, Trampert J (2011) Resolution analysis in full waveform inversion. Geophys J Int, DOI https://doi.org/10.1111/j.1365-246X.2011.05218.x
Nocedal J (1980) Updating quasi-Newton matrices with limited storage. Mathematics of Computation, 35(151):773-782
Liu D, Nocedal J (1989) On the limited memory BFGS method for large scale optimization: Mathematical Programming. Springer, DOI https://doi.org/10.1007/BF01589116
Brossier R, Operto S, Virieux J (2010) Which data residual norm for robust elastic frequency-domain full waveform inversion? Geophysics, DOI https://doi.org/10.1190/1.3379323
Virieux J, Operto S (2009) An overview of full-waveform inversion in exploration geophysics. Geophysics, DOI https://doi.org/10.1190/1.3238367
Boonyasiriwat C, Valasek P, Routh P, Cao W, Schuster G, Macy B (2009) An efficient multiscale method for time- domain waveform tomography. Geophysics, DOI https://doi.org/10.1190/1.3151869
Kennett B, Nolet G (1978) Resolution analysis for discrete systems. Geophys J R astr Soc, 53:413-425
Williamson P (1991) A guide to the limits of resolution imposed by scattering in ray tomography. Geophysics, DOI https://doi.org/10.1190/1.1443032
Meles G, Greenhalgh S, Maurer H, Green A, van der Kruk J (2012) Gpr full-waveform sensitivity and resolution analysis using an fdtd adjoint method. IEEE Transactions on Geoscience and Remote Sensing, DOI 10.1109/IWAGPR.2011.5963863
Tao Y, Sen M (2013) Frequency-domain full wave inversion with scattering-intregral approach and its sensivity analysis. Journal of Geophysics and Engineering, DOI 10.1088/1742-2132/10/6/065008
Abreo, S. A., Ramírez- Silva, A. B., & Reyes- Torres, O. M. (2018). A GPU implementation of the second order adjoint state theory to quantify the uncertainty on FWI results. CT&F - Ciencia, Tecnología Y Futuro, 8(2). https://doi.org/10.29047/01225383.86
Tarantola A (2005) Inversion Problem Theory and methods for model parameter estimation. SIAM
dos Santos A, Pestana R (2015) Time-domain multiscale fullwaveform inversion using the rapid expansion method and efficient step-length estimation. Geophysics, DOI https: //doi.org/10.1190/geo2014-0338.1
Plessix RE (2006) A review of the adjoint-state method for computing the gradient of a functional with geophysical applications. Geophysical Journal International, DOI https://doi.org/10.1111/j.1365-246X.2006.02978.x
Nocedal J, Wright SJ (2006) Numerical optimization, 2edn. Springer
Ma Y, Hale D (2012) Quasi-newton full-waveform inversion with a projected hessian matrix. Geophysics, DOI https: //doi.org/10.1190/geo2011-0519.1
Leeuwen TV, Mulder WA (2010) A correlation-based misfit criterion for wave-equation traveltime tomography. Geophysical Journal International, DOI https://doi.org/10.1111/j.1365-246X.2010.04681.x
Madagascar Development Team (2012) Madagascar Software, Version~1.4. http://www.ahay.org/