| Title | Retrieval of subsurface resistivity from magnetotelluric data using a deep-learning-based inversion technique |
| |
| Author | Liu, X ; Craven, J
A; Tschirhart, V |
| Source | Minerals vol. 13, issue 4, 2023 p. 1-16, https://doi.org/10.3390/ min13040461 |
| Image |  |
| Year | 2023 |
| Alt Series | Natural Resources Canada, Contribution Series 20210009 |
| Publisher | MDPI |
| Document | serial |
| Lang. | English |
| Media | on-line; digital |
| Subjects | mathematical and computational geology; Science and Technology; magnetotelluric surveys; magnetotelluric interpretations; Learning |
| Illustrations | diagrams; models; flow charts; tables; plots; geological sketch maps |
| Program | Targeted Geoscience Initiative (TGI-6) Digital Geoscience and Method Development Project |
| Released | 2023 03 29 |
| Abstract | Inversion is a fundamental step in magnetotelluric (MT) data routine analysis to retrieve a subsurface geoelectrical model that can be used to inform geological interpretations. To reduce the effect of
non-uniqueness and local minimum trapping problems and improve calculation speeds, a data-driven mathematical method with a deep neural network was developed to estimate the subsurface resistivity. In this study, a deep learning (DL) inversion
technique using a revised multihead convolutional neural network (CNN) architecture was investigated for MT data analysis. We created synthetic datasets consisting of 100,000 random samples of resistivity layers to train the network's parameters. The
trained model was validated with independent noised datasets, and the predicted results displayed reasonable accuracy and reliability, which demonstrates the potential application of DL inversion for real-world MT data. The trained model was used to
analyze MT data collected in the southwestern Athabasca Basin, Canada. The calculated results from the DL method displayed a detailed subsurface resistivity distribution compared to traditional iterative inversion. Since this approach can predict a
resistivity model without multiple forward modeling operations after the CNN model is created, this framework is suitable to speed up the computation of multidimensional MT inversion for subsurface resistivity. |
| Summary | (Plain Language Summary, not published)
The magnetotelluric (MT) method uses natural electromagnetic signals to image resistivity distribution of subsurface. It has been using in tectonic
boundary, earthquake study, exploration for natural resources, such as mineral and geothermal for decades. Inversion of MT data is a significant step for retrieving the subsurface resistivity information and following geological interpretation. In
order to overcome the drawback of traditional inversion methods, a potential alternative inversion technique has been emerged to estimate subsurface properties in many geophysical methods. This data-driven method using deep neural network starts be
popular along with the computer hardware improvement in recent years. In this study, we proposed a deep learning (DL) inversion method using multi-headed convolutional neural network (CNN) architecture for one dimensional MT data inversion. We
created synthetic datasets for training the CNN model weight parameters, and validated the trained DL model. The accepted CNN model was operated on real MT data from the Athabasca Basin. Because this approach can predict resistivity model without
multiple forward modeling operations after CNN model is created, this framework opens up for three dimensional MT inversion to predict the more complex resistivity structure instantly and will save lots of computation time. This advantage may make DL
inversion to be a significant tool for processing EM survey data and be benefit to mineral exploration. We created synthetic datasets for training the CNN model weight parameters, and validated the trained DL model. The accepted CNN model was
operated on real MT data from the Athabasca Basin. Because this approach can predict resistivity model without multiple forward modeling operations after CNN model is created, this framework opens up for three dimensional MT inversion to predict the
more complex resistivity structure instantly and will save lots of computation time. This advantage may make DL inversion to be a significant tool for processing EM survey data and be benefit to mineral exploration. We created synthetic datasets for
training the CNN model weight parameters, and validated the trained DL model. The accepted CNN model was operated on real MT data from the Athabasca Basin. Because this approach can predict resistivity model without multiple forward modeling
operations after CNN model is created, this framework opens up for three dimensional MT inversion to predict the more complex resistivity structure instantly and will save lots of computation time. This advantage may make DL inversion to be a
significant tool for processing EM survey data and be benefit to mineral exploration. this framework opens up for three dimensional MT inversion to predict the more complex resistivity structure instantly and will save lots of computation time. This
advantage may make DL inversion to be a significant tool for processing EM survey data and be benefit to mineral exploration. this framework opens up for three dimensional MT inversion to predict the more complex resistivity structure instantly and
will save lots of computation time. This advantage may make DL inversion to be a significant tool for processing EM survey data and be benefit to mineral exploration. |
| GEOSCAN ID | 328223 |
|
|