.png){kind=icon}
.png){kind=icon}
Return
Three-Dimensional Unfolding and Unfaulting for Structural Interpretation Using Self-Supervised Learning
.png){kind=icon}
PRE
AI
DOI：10.1029/2024JB031069.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Accurate identification of isochronal surfaces is essential for interpreting stratigraphy, analyzing deformations, and advancing geological modeling. However, complex geological deformations over geological timescales challenge stratigraphic layer interpretation. Transforming deformed geological structures into flattened space simplifies their interpretation and enables the analysis of entire volumes of structures. Traditional single-plane methods often fail to capture the fault complexities or avoid area distortion. We present a deep learning framework that restores structure from deformed to flattened states using a 3-D lightweight neural network, which computes shifts to realign layers and conserve geometry. The predicted shifts are constrained by partial differential equations to confirm structural orientations, while dynamic time warping further enhances continuity across faults. Applied to several highly deformed field examples, our approach outperforms conventional methods, precisely aligning geological layers in complex faulting and folding regions. This framework integrates geological insights into restoration, offering fresh perspectives on 3-D structural deformation mechanisms.
Keywords:
structural restoration
deep learning
geological modeling
Journal
J
IF:
0
Papers:
218
Citations:
0