Structure-preserving Seismic Impedance Inversion Using Spatio-temporal Coupled MCMC
Open Access
- 28 May 2026
- journal article
- Published by Oxford University Press (OUP) in Journal of Geophysics and Engineering
- Vol. 23 (5) , 1707-1717
Abstract
Seismic impedance inversion is a key technology for oil and gas reservoir prediction. However, linear inversion is inherently limited in accuracy, while nonlinear approaches remain computationally intensive; furthermore, neither effectively resolves geological boundaries without sufficient prior constraints. Therefore, we propose an inversion method termed Spatio-temporally Coupled Adaptive Delayed-Rejection Relativity-of-Gaussian Markov Chain Monte Carlo (MCMC). The method relies on a Hidden Markov Model (HMM) to capture the spatio-temporal evolution of impedance. By facilitating appropriate state transitions, this HMM provides essential prior-informed constraints for the posterior distribution within the MCMC framework. Furthermore, a two-stage delayed rejection algorithm incorporating adaptive covariance updating is utilized to enhance computational efficiency. Subsequently, the inversion results are processed using the Relativity-of-Gaussian method, which effectively preserves strata boundaries and structural features while significantly enhancing the sharpness of the inverted results. Numerical experiments and field applications validate the effectiveness of the proposed method, showing enhanced robustness and better-defined spatial features over conventional MCMC inversions.Keywords
Funding Information
- College Students’ Innovative Entrepreneurial Training Plan Program of China (202510613082)
- Sichuan Science and Technology Program (2026NSFSC0235)
- China National Petroleum Corporation (2024DQ03011)
This publication has 31 references indexed in Scilit:
- Deep Learning-based Self-supervised Multi-parameter InversionGeophysical Journal International, 2025
- Post-stack seismic inversion through probabilistic neural networks and deep forward neural networksEarth Science Informatics, 2024
- Seismic impedance inversion based on cycle-consistent generative adversarial networkPetroleum Science, 2022
- Improvement and application of preprocessing technique for multitrace seismic impedance inversionApplied Geophysics, 2021
- Robust deep learning seismic inversion witha prioriinitial model constraintGeophysical Journal International, 2021
- Structure-, stratigraphy- and fault-guided regularization in geophysical inversionGeophysical Journal International, 2017
- Stochastic and deterministic seismic inversion methods for thin-bed resolutionJournal of Geophysics and Engineering, 2012
- DRAM: Efficient adaptive MCMCStatistics and Computing, 2006
- Parsimonious Bayesian Markov chain Monte Carlo inversion in a nonlinear geophysical problemGeophysical Journal International, 2002
- Seismic Tomography1987