MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study
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 meth...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/6/1320 |
| _version_ | 1797442766687436800 |
|---|---|
| author | Shengchao Wang Liguo Han Xiangbo Gong Shaoyue Zhang Xingguo Huang Pan Zhang |
| author_facet | Shengchao Wang Liguo Han Xiangbo Gong Shaoyue Zhang Xingguo Huang Pan Zhang |
| author_sort | Shengchao Wang |
| collection | DOAJ |
| description | 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. |
| first_indexed | 2024-03-09T12:46:45Z |
| format | Article |
| id | doaj.art-4df354c4122f42df9ffe76cb2dea958f |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T12:46:45Z |
| publishDate | 2022-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-4df354c4122f42df9ffe76cb2dea958f2023-11-30T22:11:03ZengMDPI AGRemote Sensing2072-42922022-03-01146132010.3390/rs14061320MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation StudyShengchao Wang0Liguo Han1Xiangbo Gong2Shaoyue Zhang3Xingguo Huang4Pan Zhang5College of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, ChinaCollege of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, ChinaCollege of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, ChinaCollege of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, ChinaCollege of Instrumentation & Electrical Engineering, Jilin University, Changchun 130026, ChinaCollege of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, ChinaGround-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.https://www.mdpi.com/2072-4292/14/6/1320ground penetrating radar (GPR) crossholeMCMCtrained neural networkfull waveform inversion (FWI) |
| spellingShingle | Shengchao Wang Liguo Han Xiangbo Gong Shaoyue Zhang Xingguo Huang Pan Zhang MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study Remote Sensing ground penetrating radar (GPR) crosshole MCMC trained neural network full waveform inversion (FWI) |
| title | MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study |
| title_full | MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study |
| title_fullStr | MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study |
| title_full_unstemmed | MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study |
| title_short | MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study |
| title_sort | mcmc method of inverse problems using a neural network application in gpr crosshole full waveform inversion a numerical simulation study |
| topic | ground penetrating radar (GPR) crosshole MCMC trained neural network full waveform inversion (FWI) |
| url | https://www.mdpi.com/2072-4292/14/6/1320 |
| work_keys_str_mv | AT shengchaowang mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy AT liguohan mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy AT xiangbogong mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy AT shaoyuezhang mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy AT xingguohuang mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy AT panzhang mcmcmethodofinverseproblemsusinganeuralnetworkapplicationingprcrossholefullwaveforminversionanumericalsimulationstudy |