A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration

Hejun Zhu, Siwei Li, Sergey Fomel, Georg Stadler, Omar Ghattas

Research output: Contribution to journalArticle

Abstract

Full-waveform inversion (FWI) enables us to obtain highresolution subsurface images; however, estimating model uncertainties associated with this technique is still a challenging problem. We have used a Bayesian inference framework to estimate model uncertainties associated with FWI. The uncertainties were assessed based on an a posteriori covariance operator, evaluated at the maximum a posteriori model. For the prior distribution, we have used a spatially nonstationary covariance operator based on a plane-wave construction with local dips measured from migrated images. Preconditioned frequencydomain FWI was used to estimate the maximum a posteriori model. Efficient manipulation of the posterior covariance was based on a low-rank approximation of the data misfit Hessian preconditioned by the prior covariance operator. The strong decay of the singular values indicated that data were mostly informative about a low-dimensional subspace of model parameters. To reduce computational cost of the randomized singular value decomposition, we have used a Hessian approximation based on point-spread functions. The 2D numerical examples with synthetic data confirmed that the method can effectively estimate uncertainties for FWI. Visual comparisons of random samples drawn from the prior and posterior distributions have allowed us to evaluate model uncertainties. Furthermore, we found out how statistical quantities, such as means and pointwise standard deviation fields, can be efficiently extracted from the prior and posterior distributions. These fields helped us to objectively assess subsurface images provided by FWI.

Original languageEnglish (US)
Pages (from-to)R307-R323
JournalGeophysics
Volume81
Issue number5
DOIs
StatePublished - Nov 18 2015

Fingerprint

waveforms
inversions
estimates
operators
Optical transfer function
Singular value decomposition
point spread functions
approximation
inference
Uncertainty
inversion
manipulators
dip
standard deviation
plane waves
estimating
decomposition
costs
decay
cost

ASJC Scopus subject areas

  • Geophysics
  • Geochemistry and Petrology

Cite this

A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration. / Zhu, Hejun; Li, Siwei; Fomel, Sergey; Stadler, Georg; Ghattas, Omar.

In: Geophysics, Vol. 81, No. 5, 18.11.2015, p. R307-R323.

Research output: Contribution to journalArticle

Zhu, Hejun ; Li, Siwei ; Fomel, Sergey ; Stadler, Georg ; Ghattas, Omar. / A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration. In: Geophysics. 2015 ; Vol. 81, No. 5. pp. R307-R323.
@article{c724680b9a1c462598de7918b76a36e1,
title = "A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration",
abstract = "Full-waveform inversion (FWI) enables us to obtain highresolution subsurface images; however, estimating model uncertainties associated with this technique is still a challenging problem. We have used a Bayesian inference framework to estimate model uncertainties associated with FWI. The uncertainties were assessed based on an a posteriori covariance operator, evaluated at the maximum a posteriori model. For the prior distribution, we have used a spatially nonstationary covariance operator based on a plane-wave construction with local dips measured from migrated images. Preconditioned frequencydomain FWI was used to estimate the maximum a posteriori model. Efficient manipulation of the posterior covariance was based on a low-rank approximation of the data misfit Hessian preconditioned by the prior covariance operator. The strong decay of the singular values indicated that data were mostly informative about a low-dimensional subspace of model parameters. To reduce computational cost of the randomized singular value decomposition, we have used a Hessian approximation based on point-spread functions. The 2D numerical examples with synthetic data confirmed that the method can effectively estimate uncertainties for FWI. Visual comparisons of random samples drawn from the prior and posterior distributions have allowed us to evaluate model uncertainties. Furthermore, we found out how statistical quantities, such as means and pointwise standard deviation fields, can be efficiently extracted from the prior and posterior distributions. These fields helped us to objectively assess subsurface images provided by FWI.",
author = "Hejun Zhu and Siwei Li and Sergey Fomel and Georg Stadler and Omar Ghattas",
year = "2015",
month = "11",
day = "18",
doi = "10.1190/GEO2015-0641.1",
language = "English (US)",
volume = "81",
pages = "R307--R323",
journal = "Geophysics",
issn = "0016-8033",
publisher = "Society of Exploration Geophysicists",
number = "5",

}

TY - JOUR

T1 - A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration

AU - Zhu, Hejun

AU - Li, Siwei

AU - Fomel, Sergey

AU - Stadler, Georg

AU - Ghattas, Omar

PY - 2015/11/18

Y1 - 2015/11/18

N2 - Full-waveform inversion (FWI) enables us to obtain highresolution subsurface images; however, estimating model uncertainties associated with this technique is still a challenging problem. We have used a Bayesian inference framework to estimate model uncertainties associated with FWI. The uncertainties were assessed based on an a posteriori covariance operator, evaluated at the maximum a posteriori model. For the prior distribution, we have used a spatially nonstationary covariance operator based on a plane-wave construction with local dips measured from migrated images. Preconditioned frequencydomain FWI was used to estimate the maximum a posteriori model. Efficient manipulation of the posterior covariance was based on a low-rank approximation of the data misfit Hessian preconditioned by the prior covariance operator. The strong decay of the singular values indicated that data were mostly informative about a low-dimensional subspace of model parameters. To reduce computational cost of the randomized singular value decomposition, we have used a Hessian approximation based on point-spread functions. The 2D numerical examples with synthetic data confirmed that the method can effectively estimate uncertainties for FWI. Visual comparisons of random samples drawn from the prior and posterior distributions have allowed us to evaluate model uncertainties. Furthermore, we found out how statistical quantities, such as means and pointwise standard deviation fields, can be efficiently extracted from the prior and posterior distributions. These fields helped us to objectively assess subsurface images provided by FWI.

AB - Full-waveform inversion (FWI) enables us to obtain highresolution subsurface images; however, estimating model uncertainties associated with this technique is still a challenging problem. We have used a Bayesian inference framework to estimate model uncertainties associated with FWI. The uncertainties were assessed based on an a posteriori covariance operator, evaluated at the maximum a posteriori model. For the prior distribution, we have used a spatially nonstationary covariance operator based on a plane-wave construction with local dips measured from migrated images. Preconditioned frequencydomain FWI was used to estimate the maximum a posteriori model. Efficient manipulation of the posterior covariance was based on a low-rank approximation of the data misfit Hessian preconditioned by the prior covariance operator. The strong decay of the singular values indicated that data were mostly informative about a low-dimensional subspace of model parameters. To reduce computational cost of the randomized singular value decomposition, we have used a Hessian approximation based on point-spread functions. The 2D numerical examples with synthetic data confirmed that the method can effectively estimate uncertainties for FWI. Visual comparisons of random samples drawn from the prior and posterior distributions have allowed us to evaluate model uncertainties. Furthermore, we found out how statistical quantities, such as means and pointwise standard deviation fields, can be efficiently extracted from the prior and posterior distributions. These fields helped us to objectively assess subsurface images provided by FWI.

UR - http://www.scopus.com/inward/record.url?scp=84984623467&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84984623467&partnerID=8YFLogxK

U2 - 10.1190/GEO2015-0641.1

DO - 10.1190/GEO2015-0641.1

M3 - Article

VL - 81

SP - R307-R323

JO - Geophysics

JF - Geophysics

SN - 0016-8033

IS - 5

ER -