FEMTIC-DABIC is a 3-D magnetotelluric (MT) inversion code developed based on FEMTIC (v4.2). It is designed to determine and use the statistically optimal regularization parameter (α) in each iteration, thereby controlling the overall smoothness of the final subsurface resistivity model.
FEMTIC-DABIC leverages Akaike’s Bayesian Information Criterion (ABIC) to quantify the magnitude of the marginal likelihood function: a key metric in Bayesian theory, where higher values indicate a more optimal hyperparameter (here, α) that balances data fitting and prior model information. To render ABIC computationally tractable for 3-D MT inversion, we propose a data-space variant of ABIC, termed DABIC. The 3-D MT data-space inversion workflow built on this DABIC variant is formally referred to as the D-DABIC inversion method.
For details on the methodology and workflow of the D-DABIC inversion method, please refer to:
Song, H., Yu, P., Usui, Y., Uyeshima, M., Diba, D., & Zhang, L. Three-dimensional Magnetotelluric Inversion based on a Data Space variant of Akaike’s Bayesian Information Criterion. Geophysics (2025), https://doi.org/10.1190/geo-2025-0233.
For more information about the FEMTIC code, please refer to the FEMTIC repository: https://github.com/yoshiya-usui/femtic.
v1.4 Jan. 12, 2026: I've revised the D-DABIC workflow to ensure it is capable of incorporating the distortion correction functionality.
v1.3 Sep. 13, 2025: Added Minimum Norm (MN) Stabilizer with Depth of Investigation (DOI) Support. Introduced a new regularization option (|m - m_r|) to constrain inversion toward a reference model (m_r); the primary purpose of this option (for now) is to enable DOI analysis for model appraisal.
v1.2 Sep. 11, 2025: Reference Model (m_r) Configuration Option. Added support for defining a user-provided reference model (m_r), enabling physics-based constraints in the inversion.
v1.1 Dec. 30, 2024: Laplacian Filter (LF) for Marginal Likelihood Maximization. Enabled the LF as an alternative regularization during the D-DABIC optimization.
v1.0 Nov. 28, 2024: Core FEMTIC-DABIC Framework. Implemented a 3-D data-space inversion method using a data-space variant of the Akaike’s Bayesian Information Criterion (D-DABIC).
If you encounter any issues, have questions, or want to provide feedback, please feel free to contact:
Han SONG: 1831736@tongji.edu.cn; han.song@tu-berlin.de
Yoshiya USUI: yusui@eri.u-tokyo.ac.jp
You may also report issues via the project's repository (e.g., GitHub Issues).