Full-Waveform Inversion of Time-Lapse Crosshole GPR Data Using Markov Chain Monte Carlo Method

Wang, Shengchao; Han, Liguo; Gong, Xun; Zhang, Shaoyue; Huang, Xingguo; Zhang, Pan

doi:10.3390/rs13224530

Cited by 3 publications

(1 citation statement)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A sequential Gibbs sampler is proposed to serve as a black box algorithm to sample a priori information defined by any geostatistical algorithm [12,13]. In this way, sampling the a posteriori probability density becomes very flexible [14][15][16][17].…”

Section: Introductionmentioning

confidence: 99%

MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study

et al. 2022

Self Cite

View full text Add to dashboard Cite

Ground-penetrating radar (GPR) crosshole tomography is widely applied to subsurface media images. However, the inadequacies of ray methods may limit the resolution of crosshole radar images, since the ray method is a type of high-frequency approximation. To solve this problem, the full waveform method is introduced for GPR inversion. However, full waveform inversion is computationally expensive. In this paper, we introduce a trained neural network that can be evaluated very quickly to replace a computationally intensive forward model. Additionally, the forward error of the trained neural network can be statistically analyzed. We demonstrate a methodology for a full waveform inversion of crosshole ground-penetrating radar data using the Markov chain Monte Carlo (MCMC) method. An accurate forward model based on Maxwell’s equations is replaced by a quickly trained neural network. This method achieves a high computation efficiency, which is four orders of magnitude faster than the accurate forward model. The inversion result of the synthetic waveform data shows a good performance of the trained neural network, which greatly improves the calculation efficiency.

show abstract

Section: Introductionmentioning

confidence: 99%

MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

Semisupervised Deep Neural Network-Based Cross-Frequency Ground-Penetrating Radar Data Inversion

Liu,

Wang,

Zhang

et al. 2023

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

2.5D transient electromagnetic inversion based on the unstructured quadrilateral finite-element method and a geologic statistics-driven Bayesian framework

Dai,

Wu,

Yue

2024

GEOPHYSICS

View full text Add to dashboard Cite

The uncertainty of subsurface resistivity can be determined using Bayesian inversion during electromagnetic exploration. However, the extensive sampling necessary for Bayesian inversion may result in high computational costs and prevent its application in complex, large-scale 2D and 3D electromagnetic inversions. For the first attempt, we implemented a 2.5D (i.e., 3D source with 2D geological model) Bayesian inversion to analyze ground transient electromagnetics method (TEM) and made some adaptations and improvements. To be specific, we employed geostatistical-based random modeling as a prior constraint strategy and incorporated additional prior information through model mapping to improve its performance in adapting to complex terrains. To perform pre-inversion of prior parameters, we utilized a spatially-constrained trans-dimensional Bayesian inversion approach, extracting priors for the formal inversion from the posterior of the pre-inversion. This could not only reduce computation time but also lessen the formal Bayesian inversion's reliance on prior selection. More importantly, to obtain more realistic uncertainty assessments, we designed a likelihood suited to the noise characteristics of TEM. The extensive statistical analysis of posterior results provides richer insights than previous methods, showing Bayesian inversion's superiority in both uncertainty reliability and resistivity recovery effectiveness over deterministic inversion and continuous 1D Bayesian inversion. Comparative studies with multiple simulated models and applications to complex terrain goaf areas validate the effectiveness and superiority of the proposed algorithm.

show abstract

Full-Waveform Inversion of Time-Lapse Crosshole GPR Data Using Markov Chain Monte Carlo Method

Cited by 3 publications

References 36 publications

MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study

MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study

Semisupervised Deep Neural Network-Based Cross-Frequency Ground-Penetrating Radar Data Inversion

2.5D transient electromagnetic inversion based on the unstructured quadrilateral finite-element method and a geologic statistics-driven Bayesian framework

Contact Info

Product

Resources

About