On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes

Rasmus Bødker Madsen; Andrea Zunino; Thomas Mejer Hansen

doi:10.1190/segam2017-17725822.1

On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes

Rasmus Bødker Madsen, Andrea Zunino, Thomas Mejer Hansen

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › peer-review

7 Citations (Scopus)

Abstract

A realistic noise model is essential for trustworthy inversion of geophysical data. Sometimes, as in case of seismic data, quantification of the noise model is non-trivial. To remedy this, a hierarchical Bayes approach can be adopted in which properties of the noise model, such as the amplitude of an assumed uncorrelated Gaussian noise model, can be inferred as part of the inversion. Here we demonstrate how such an approach can lead to substantial overfitting of noise when inverting a 1D reflection seismic NMO data set. We then argue that usually the noise model is correlated, and suggest to infer the amplitude of a correlated Gaussian noise model. This provides better results than assuming an uncorrelated model. In general though, the results suggest that care should be taken using the hierarchical Bayes approach to infer the noise model.

Original language	English
Title of host publication	SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas
Publisher	Society of Exploration Geophysicists
Pages	601-606
Number of pages	6
DOIs	https://doi.org/10.1190/segam2017-17725822.1
Publication status	Published - 17 Aug 2017
Externally published	Yes
Event	Society of Exploration Geophysicists International Exposition and 87th Annual Meeting - George R. Brown Convention Center, Houston, United States Duration: 24 Sept 2017 → 29 Sept 2017 Conference number: 87

Publication series

Name	SEG Technical Program Expanded Abstracts
Publisher	Society of Exploration Geophysicists
Volume	2017
ISSN (Print)	1052-3812
ISSN (Electronic)	1949-4645

Conference

Conference	Society of Exploration Geophysicists International Exposition and 87th Annual Meeting
Abbreviated title	SEG 2017
Country/Territory	United States
City	Houston
Period	24/09/17 → 29/09/17

Keywords

Noise
Inference
Bayes
Inversion
Correlated noise

Programme Area

Programme Area 3: Energy Resources

Access to Document

10.1190/segam2017-17725822.1

Cite this

Madsen, R. B., Zunino, A., & Hansen, T. M. (2017). On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes. In SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas (pp. 601-606). (SEG Technical Program Expanded Abstracts; Vol. 2017). Society of Exploration Geophysicists. https://doi.org/10.1190/segam2017-17725822.1

@inproceedings{7bbd319d809944e58e40a2da2c85a65e,

title = "On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes",

abstract = "A realistic noise model is essential for trustworthy inversion of geophysical data. Sometimes, as in case of seismic data, quantification of the noise model is non-trivial. To remedy this, a hierarchical Bayes approach can be adopted in which properties of the noise model, such as the amplitude of an assumed uncorrelated Gaussian noise model, can be inferred as part of the inversion. Here we demonstrate how such an approach can lead to substantial overfitting of noise when inverting a 1D reflection seismic NMO data set. We then argue that usually the noise model is correlated, and suggest to infer the amplitude of a correlated Gaussian noise model. This provides better results than assuming an uncorrelated model. In general though, the results suggest that care should be taken using the hierarchical Bayes approach to infer the noise model.",

keywords = "Noise, Inference, Bayes, Inversion, Correlated noise",

author = "Madsen, {Rasmus B{\o}dker} and Andrea Zunino and Hansen, {Thomas Mejer}",

note = "Publisher Copyright: {\textcopyright} 2017 SEG.; Society of Exploration Geophysicists International Exposition and 87th Annual Meeting, SEG 2017 ; Conference date: 24-09-2017 Through 29-09-2017",

year = "2017",

month = aug,

day = "17",

doi = "10.1190/segam2017-17725822.1",

language = "English",

series = "SEG Technical Program Expanded Abstracts",

publisher = "Society of Exploration Geophysicists",

pages = "601--606",

booktitle = "SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas",

address = "United States",

}

Madsen, RB, Zunino, A & Hansen, TM 2017, On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes. in SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas. SEG Technical Program Expanded Abstracts, vol. 2017, Society of Exploration Geophysicists, pp. 601-606, Society of Exploration Geophysicists International Exposition and 87th Annual Meeting, Houston, Texas, United States, 24/09/17. https://doi.org/10.1190/segam2017-17725822.1

On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes. / Madsen, Rasmus Bødker; Zunino, Andrea; Hansen, Thomas Mejer.

SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas. Society of Exploration Geophysicists, 2017. p. 601-606 (SEG Technical Program Expanded Abstracts; Vol. 2017).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › peer-review

TY - GEN

T1 - On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes

AU - Madsen, Rasmus Bødker

AU - Zunino, Andrea

AU - Hansen, Thomas Mejer

N1 - Conference code: 87

PY - 2017/8/17

Y1 - 2017/8/17

N2 - A realistic noise model is essential for trustworthy inversion of geophysical data. Sometimes, as in case of seismic data, quantification of the noise model is non-trivial. To remedy this, a hierarchical Bayes approach can be adopted in which properties of the noise model, such as the amplitude of an assumed uncorrelated Gaussian noise model, can be inferred as part of the inversion. Here we demonstrate how such an approach can lead to substantial overfitting of noise when inverting a 1D reflection seismic NMO data set. We then argue that usually the noise model is correlated, and suggest to infer the amplitude of a correlated Gaussian noise model. This provides better results than assuming an uncorrelated model. In general though, the results suggest that care should be taken using the hierarchical Bayes approach to infer the noise model.

AB - A realistic noise model is essential for trustworthy inversion of geophysical data. Sometimes, as in case of seismic data, quantification of the noise model is non-trivial. To remedy this, a hierarchical Bayes approach can be adopted in which properties of the noise model, such as the amplitude of an assumed uncorrelated Gaussian noise model, can be inferred as part of the inversion. Here we demonstrate how such an approach can lead to substantial overfitting of noise when inverting a 1D reflection seismic NMO data set. We then argue that usually the noise model is correlated, and suggest to infer the amplitude of a correlated Gaussian noise model. This provides better results than assuming an uncorrelated model. In general though, the results suggest that care should be taken using the hierarchical Bayes approach to infer the noise model.

KW - Noise

KW - Inference

KW - Bayes

KW - Inversion

KW - Correlated noise

UR - http://www.scopus.com/inward/record.url?scp=85074452157&partnerID=8YFLogxK

U2 - 10.1190/segam2017-17725822.1

DO - 10.1190/segam2017-17725822.1

M3 - Conference article in proceedings

T3 - SEG Technical Program Expanded Abstracts

SP - 601

EP - 606

BT - SEG International Exposition and Annual Meeting, 24-29 September, Houston, Texas

PB - Society of Exploration Geophysicists

T2 - Society of Exploration Geophysicists International Exposition and 87th Annual Meeting

Y2 - 24 September 2017 through 29 September 2017

ER -

On inferring the noise in probabilistic seismic AVO inversion using hierarchical Bayes

Abstract

Publication series

Conference

Keywords

Programme Area

Access to Document

Fingerprint

Cite this