Aarhus University Seal / Aarhus Universitets segl

Machine learning based fast forward modelling of ground-based time-domain electromagnetic data

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Standard

Machine learning based fast forward modelling of ground-based time-domain electromagnetic data. / Bording, Thue Sylvester; Asif, Muhammad Rizwan; Barfod, Adrian S.; Larsen, Jakob Juul; Zhang, Bo; Grombacher, Denys James; Christiansen, Anders Vest; Engebretsen, Kim Wann; Pedersen, Jesper Bjergsted; Maurya, Pradip Kumar; Auken, Esben.

In: Journal of Applied Geophysics, Vol. 187, 104290, 04.2021.

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Harvard

APA

CBE

MLA

Vancouver

Author

Bibtex

@article{bf107ee8b31248b9a1f1ccfe3b1b712a,
title = "Machine learning based fast forward modelling of ground-based time-domain electromagnetic data",
abstract = "Inversion of large-scale time-domain electromagnetic surveys are computationally expensive and time consuming. Deterministic or probabilistic inversion schemes usually require calculations of forward responses, and often thousands to millions of forward responses are computed. We propose a machine learning based forward modelling approach as a computationally feasible alternative to approximate numerical forward modelling where a neural network is employed to model the relationship between the resistivity models and corresponding forward responses. For training of the neural network, we generated forward responses using conventional numerical algorithm for 93,500 resistivity models derived from different surveys conducted in Denmark representing typical resistivities of sedimentary geological layers. The input resistivity models and the network target outputs, i.e. forward responses, are scaled using a novel normalization strategy to ensure each gate is equally prioritized. The performance of the network is evaluated on two test datasets consisting of 8942 resistivity models by comparing the forward responses generated by the neural network and the conventional algorithm. We also measure the performance for the time derivatives of forward responses, i.e. dB/dt, by incorporating a system response. The results show that the proposed strategy is at least 13 times faster than commonly used accurate modelling methods and achieves an accuracy of 98% within 3% relative error, which is comparable to data uncertainty. Additional experiments on surveys from two other continents show that the results generalize in similar geological settings. Thus, under certain geological constraints, the proposed methodology may be incorporated into the pre-existing inversion structures, allowing for significantly faster inversion of large datasets.",
keywords = "Forward responses, Inversion modelling, neural networks, Transient electromagnetic",
author = "Bording, {Thue Sylvester} and Asif, {Muhammad Rizwan} and Barfod, {Adrian S.} and Larsen, {Jakob Juul} and Bo Zhang and Grombacher, {Denys James} and Christiansen, {Anders Vest} and Engebretsen, {Kim Wann} and Pedersen, {Jesper Bjergsted} and Maurya, {Pradip Kumar} and Esben Auken",
note = "Funding Information: The data from Ladysmith, South Africa was collected during a project funded by Umgeni Water , JG Africa, and the Poul Due Jensens Foundation . Funding Information: This work was supported by Innovation Fund Denmark under the {\textquoteleft}MapField{\textquoteright} project . Publisher Copyright: {\textcopyright} 2021 Elsevier B.V. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.",
year = "2021",
month = apr,
doi = "10.1016/j.jappgeo.2021.104290",
language = "English",
volume = "187",
journal = "Journal of Applied Geophysics",
issn = "0926-9851",
publisher = "Elsevier BV",

}

RIS

TY - JOUR

T1 - Machine learning based fast forward modelling of ground-based time-domain electromagnetic data

AU - Bording, Thue Sylvester

AU - Asif, Muhammad Rizwan

AU - Barfod, Adrian S.

AU - Larsen, Jakob Juul

AU - Zhang, Bo

AU - Grombacher, Denys James

AU - Christiansen, Anders Vest

AU - Engebretsen, Kim Wann

AU - Pedersen, Jesper Bjergsted

AU - Maurya, Pradip Kumar

AU - Auken, Esben

N1 - Funding Information: The data from Ladysmith, South Africa was collected during a project funded by Umgeni Water , JG Africa, and the Poul Due Jensens Foundation . Funding Information: This work was supported by Innovation Fund Denmark under the ‘MapField’ project . Publisher Copyright: © 2021 Elsevier B.V. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.

PY - 2021/4

Y1 - 2021/4

N2 - Inversion of large-scale time-domain electromagnetic surveys are computationally expensive and time consuming. Deterministic or probabilistic inversion schemes usually require calculations of forward responses, and often thousands to millions of forward responses are computed. We propose a machine learning based forward modelling approach as a computationally feasible alternative to approximate numerical forward modelling where a neural network is employed to model the relationship between the resistivity models and corresponding forward responses. For training of the neural network, we generated forward responses using conventional numerical algorithm for 93,500 resistivity models derived from different surveys conducted in Denmark representing typical resistivities of sedimentary geological layers. The input resistivity models and the network target outputs, i.e. forward responses, are scaled using a novel normalization strategy to ensure each gate is equally prioritized. The performance of the network is evaluated on two test datasets consisting of 8942 resistivity models by comparing the forward responses generated by the neural network and the conventional algorithm. We also measure the performance for the time derivatives of forward responses, i.e. dB/dt, by incorporating a system response. The results show that the proposed strategy is at least 13 times faster than commonly used accurate modelling methods and achieves an accuracy of 98% within 3% relative error, which is comparable to data uncertainty. Additional experiments on surveys from two other continents show that the results generalize in similar geological settings. Thus, under certain geological constraints, the proposed methodology may be incorporated into the pre-existing inversion structures, allowing for significantly faster inversion of large datasets.

AB - Inversion of large-scale time-domain electromagnetic surveys are computationally expensive and time consuming. Deterministic or probabilistic inversion schemes usually require calculations of forward responses, and often thousands to millions of forward responses are computed. We propose a machine learning based forward modelling approach as a computationally feasible alternative to approximate numerical forward modelling where a neural network is employed to model the relationship between the resistivity models and corresponding forward responses. For training of the neural network, we generated forward responses using conventional numerical algorithm for 93,500 resistivity models derived from different surveys conducted in Denmark representing typical resistivities of sedimentary geological layers. The input resistivity models and the network target outputs, i.e. forward responses, are scaled using a novel normalization strategy to ensure each gate is equally prioritized. The performance of the network is evaluated on two test datasets consisting of 8942 resistivity models by comparing the forward responses generated by the neural network and the conventional algorithm. We also measure the performance for the time derivatives of forward responses, i.e. dB/dt, by incorporating a system response. The results show that the proposed strategy is at least 13 times faster than commonly used accurate modelling methods and achieves an accuracy of 98% within 3% relative error, which is comparable to data uncertainty. Additional experiments on surveys from two other continents show that the results generalize in similar geological settings. Thus, under certain geological constraints, the proposed methodology may be incorporated into the pre-existing inversion structures, allowing for significantly faster inversion of large datasets.

KW - Forward responses

KW - Inversion modelling, neural networks

KW - Transient electromagnetic

UR - http://www.scopus.com/inward/record.url?scp=85101802204&partnerID=8YFLogxK

U2 - 10.1016/j.jappgeo.2021.104290

DO - 10.1016/j.jappgeo.2021.104290

M3 - Journal article

AN - SCOPUS:85101802204

VL - 187

JO - Journal of Applied Geophysics

JF - Journal of Applied Geophysics

SN - 0926-9851

M1 - 104290

ER -