Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

**Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data.** / Asif, Muhammad Rizwan; Foged, Nikolaj; Maurya, Pradip Kumar et al.

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

Asif, MR, Foged, N, Maurya, PK, Grombacher, DJ, Christiansen, AV, Auken, E & Larsen, JJ 2022, 'Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data', *Geophysics*, vol. 87, no. 4, pp. E177-E187. https://doi.org/10.1190/geo2021-0335.1

Asif, M. R., Foged, N., Maurya, P. K., Grombacher, D. J., Christiansen, A. V., Auken, E., & Larsen, J. J. (2022). Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data. *Geophysics*, *87*(4), E177-E187. https://doi.org/10.1190/geo2021-0335.1

Asif MR, Foged N, Maurya PK, Grombacher DJ, Christiansen AV, Auken E, Larsen JJ. 2022. Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data. Geophysics. 87(4):E177-E187. https://doi.org/10.1190/geo2021-0335.1

Asif, Muhammad Rizwan et al. "Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data". *Geophysics*. 2022, 87(4). E177-E187. https://doi.org/10.1190/geo2021-0335.1

Asif MR, Foged N, Maurya PK, Grombacher DJ, Christiansen AV, Auken E et al. Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data. Geophysics. 2022 Jul;87(4):E177-E187. doi: 10.1190/geo2021-0335.1

Asif, Muhammad Rizwan ; Foged, Nikolaj ; Maurya, Pradip Kumar et al. / **Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data**. In: Geophysics. 2022 ; Vol. 87, No. 4. pp. E177-E187.

@article{177062f3a90a4e6bae1e24b8f16be7d7,

title = "Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data",

abstract = "Airborne time-domain electromagnetic surveys produce extremely large data sets with thousands of line kilometers of data and millions of possible models to explain the data. Inversion of such data sets to obtain the resistivity structures of the subsurface is computationally intensive and involves calculation of a significant number of forward and derivative responses for solving the least-squares inverse problem. The flight altitude of the airborne system needs to be included in the modeling, which adds further complexity. We propose to integrate neural networks in a damped iterative least-squares inversion framework to expedite the inversion process. We train two separate neural networks to predict the forward responses and partial derivatives independently for a broad range of resistivity structures and flight altitudes. Data inversion is not only used for producing the final subsurface models but also used during data processing, or to produce intermediate results during a survey. With these purposes in mind, we provide three inversion schemes with a tunable balance between computational time and modeling accuracy: (1) numerical forward responses used initially in combination with neural network derivatives, and the derivatives switched to a numerical solution in final iterations, (2) numerical forward responses in combination with neural network derivatives used throughout the inversion, and (3) only neural network forward responses and derivatives used in inversion. Experiments on field data find that we improve inversion speed without any loss in modeling accuracy with our first approach, whereas the second scheme gives a significant speedup at the cost of minor and often acceptable deviations in the inversion results from the conventional nonlinear inversion. The last approach is the fastest and captures the overall resistivity structures quite well. Therefore, depending on the modeling accuracy, inversion speedup factors of up to 50 are realized by using the proposed schemes. ",

keywords = "Airborne survey, electromagnetics, inversion, least squares, neural networks",

author = "Asif, {Muhammad Rizwan} and Nikolaj Foged and Maurya, {Pradip Kumar} and Grombacher, {Denys James} and Christiansen, {Anders Vest} and Esben Auken and Larsen, {Jakob Juul}",

note = "Publisher Copyright: {\textcopyright} 2022 Society of Exploration Geophysicists.",

year = "2022",

month = jul,

doi = "10.1190/geo2021-0335.1",

language = "English",

volume = "87",

pages = "E177--E187",

journal = "Geophysics",

issn = "0016-8033",

publisher = "Society of Exploration Geophysicists",

number = "4",

}

TY - JOUR

T1 - Integrating neural networks in least-squares inversion of airborne time-domain electromagnetic data

AU - Asif, Muhammad Rizwan

AU - Foged, Nikolaj

AU - Maurya, Pradip Kumar

AU - Grombacher, Denys James

AU - Christiansen, Anders Vest

AU - Auken, Esben

AU - Larsen, Jakob Juul

N1 - Publisher Copyright: © 2022 Society of Exploration Geophysicists.

PY - 2022/7

Y1 - 2022/7

N2 - Airborne time-domain electromagnetic surveys produce extremely large data sets with thousands of line kilometers of data and millions of possible models to explain the data. Inversion of such data sets to obtain the resistivity structures of the subsurface is computationally intensive and involves calculation of a significant number of forward and derivative responses for solving the least-squares inverse problem. The flight altitude of the airborne system needs to be included in the modeling, which adds further complexity. We propose to integrate neural networks in a damped iterative least-squares inversion framework to expedite the inversion process. We train two separate neural networks to predict the forward responses and partial derivatives independently for a broad range of resistivity structures and flight altitudes. Data inversion is not only used for producing the final subsurface models but also used during data processing, or to produce intermediate results during a survey. With these purposes in mind, we provide three inversion schemes with a tunable balance between computational time and modeling accuracy: (1) numerical forward responses used initially in combination with neural network derivatives, and the derivatives switched to a numerical solution in final iterations, (2) numerical forward responses in combination with neural network derivatives used throughout the inversion, and (3) only neural network forward responses and derivatives used in inversion. Experiments on field data find that we improve inversion speed without any loss in modeling accuracy with our first approach, whereas the second scheme gives a significant speedup at the cost of minor and often acceptable deviations in the inversion results from the conventional nonlinear inversion. The last approach is the fastest and captures the overall resistivity structures quite well. Therefore, depending on the modeling accuracy, inversion speedup factors of up to 50 are realized by using the proposed schemes.

AB - Airborne time-domain electromagnetic surveys produce extremely large data sets with thousands of line kilometers of data and millions of possible models to explain the data. Inversion of such data sets to obtain the resistivity structures of the subsurface is computationally intensive and involves calculation of a significant number of forward and derivative responses for solving the least-squares inverse problem. The flight altitude of the airborne system needs to be included in the modeling, which adds further complexity. We propose to integrate neural networks in a damped iterative least-squares inversion framework to expedite the inversion process. We train two separate neural networks to predict the forward responses and partial derivatives independently for a broad range of resistivity structures and flight altitudes. Data inversion is not only used for producing the final subsurface models but also used during data processing, or to produce intermediate results during a survey. With these purposes in mind, we provide three inversion schemes with a tunable balance between computational time and modeling accuracy: (1) numerical forward responses used initially in combination with neural network derivatives, and the derivatives switched to a numerical solution in final iterations, (2) numerical forward responses in combination with neural network derivatives used throughout the inversion, and (3) only neural network forward responses and derivatives used in inversion. Experiments on field data find that we improve inversion speed without any loss in modeling accuracy with our first approach, whereas the second scheme gives a significant speedup at the cost of minor and often acceptable deviations in the inversion results from the conventional nonlinear inversion. The last approach is the fastest and captures the overall resistivity structures quite well. Therefore, depending on the modeling accuracy, inversion speedup factors of up to 50 are realized by using the proposed schemes.

KW - Airborne survey

KW - electromagnetics

KW - inversion

KW - least squares

KW - neural networks

UR - http://www.scopus.com/inward/record.url?scp=85130096825&partnerID=8YFLogxK

U2 - 10.1190/geo2021-0335.1

DO - 10.1190/geo2021-0335.1

M3 - Journal article

AN - SCOPUS:85130096825

VL - 87

SP - E177-E187

JO - Geophysics

JF - Geophysics

SN - 0016-8033

IS - 4

ER -