|
Activity Number:
|
294
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #309001 |
|
Title:
|
A Bayesian Model for Inversion of Geophysical Seismic and Electromagnetic Data
|
|
Author(s):
|
Jinsong Chen*+ and Susan Hubbard and Michael Hoversten and Alan Hubbard
|
|
Companies:
|
Lawrence Berkeley National Laboratory and Lawrence Berkeley National Laboratory and Chevron Energy Technology Company and University of California, Berkeley
|
|
Address:
|
1 Cyclotron Road, Berkeley, CA, 94720,
|
|
Keywords:
|
Bayesian model ; MCMC ; Joint inversion ; Geophysics ; Parameter estimation ; Slice sampling
|
|
Abstract:
|
Reliable estimates of gas saturation need to combine the commonly used seismic imaging techniques with the recently developed controlled-source electromagnetic (CSEM) methods. However, conventional methods for inversion of seismic and CSEM data are limited because such methods depends on the choice of the initial model parameters and cannot provide adequate information about uncertainty in the estimation. In this study, we develop a Bayesian model to jointly invert seismic and CSEM data and use a mixed sampling method that includes single and multivariate Metropolis-Hastings and slice sampling methods to explore the joint probability distribution function. Synthetic and field case studies show that the developed Bayesian model is effective for integration of seismic and CSEM data resulting in better estimates of gas estimation.
|
