pith. sign in

arxiv: 2605.16783 · v1 · pith:44BBWI6Pnew · submitted 2026-05-16 · ⚛️ physics.geo-ph

Pretrain-to-alignment learning paradigm to improve geophysical AI applicability under scarce field labels and synthetic-to-field gaps: A case study of relative geologic time estimation in global shelf-edge clinothems

Hui Gao , Xinming Wu , Jiarun Yang , Zhixiang Gao , Yimin Dou This is my paper

Pith reviewed 2026-05-19 19:30 UTC · model grok-4.3

classification ⚛️ physics.geo-ph
keywords geophysical AIrelative geologic timepretrain-to-alignmentdomain adaptationsynthetic-to-field gapshelf-edge clinothemsseismic interpretationprogressive learning
0
0 comments X

The pith

A pretrain-to-alignment paradigm enables accurate geophysical AI despite scarce labels and synthetic-to-field gaps.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a pretrain-to-alignment learning paradigm that combines self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning into a single progressive workflow. This sequence is meant to build representations that are relevant to real field conditions, create task-specific mappings, enforce physical consistency, and allow adaptation to specific targets. The approach is demonstrated through a case study of relative geologic time estimation across global shelf-edge clinothems, with tests on 3,000 field datasets from multiple sedimentary basins showing accurate results and better capture of fine-scale stratigraphic and structural features. A sympathetic reader would care because the method directly tackles common barriers to using AI in geophysics, such as the high cost of obtaining labels and the mismatch between synthetic training data and actual surveys.

Core claim

The pretrain-to-alignment learning paradigm systematically integrates self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning into a unified progressive learning workflow. Geophysical AI models are developed through sequential stages that progressively build field-relevant representations, task-specific mapping capability, field consistency, and target-specific adaptability. Validation using cross-survey relative geologic time estimation in global shelf-edge clinothems on 3,000 field datasets demonstrates accurate, robust, and well-generalized performance across diverse field surveys while significantly improving fine-scale stratigraphic

What carries the argument

The pretrain-to-alignment learning paradigm, which unifies self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning into sequential stages that build field-relevant representations and mappings for geophysical tasks.

If this is right

  • The paradigm produces accurate, robust, and generalized relative geologic time estimates across diverse global field surveys.
  • It yields significant gains in fine-scale stratigraphic and structural detail compared with prior approaches.
  • The workflow provides a practical reference for applying AI to other geophysical problems such as interpretation, regression, and inversion under similar data constraints.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The staged approach could serve as a template for domain adaptation in other scientific machine-learning settings that rely heavily on synthetic data.
  • Reducing the need for large labeled field datasets might lower costs for subsurface analysis in exploration geophysics.
  • Isolating the contribution of each stage through ablation tests would clarify which steps drive the observed gains in generalization.

Load-bearing premise

That sequential integration of self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning will progressively build field-relevant representations and task-specific mapping without introducing inconsistencies or overfitting to the synthetic domain.

What would settle it

If a controlled test on a new field survey shows that the full paradigm produces no measurable gain in accuracy or fine-scale detail over a model trained only on synthetic data, the claim that the progressive stages deliver field-relevant improvements would be falsified.

Figures

Figures reproduced from arXiv: 2605.16783 by Hui Gao, Jiarun Yang, Xinming Wu, Yimin Dou, Zhixiang Gao.

Figure 8
Figure 8. Figure 8: The predicted RGT results exhibit strong stratigraphic and structural con [PITH_FULL_IMAGE:figures/full_fig_p021_8.png] view at source ↗
Figure 1
Figure 1. Figure 1: Conceptual overview of the proposed pretrain-to-alignment paradigm. The [PITH_FULL_IMAGE:figures/full_fig_p042_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Geological prior-driven pretrain-to-alignment AI framework for RGT esti [PITH_FULL_IMAGE:figures/full_fig_p043_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The loss and learning rate curves across different training stages. (a) In the [PITH_FULL_IMAGE:figures/full_fig_p044_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of MAE reconstruction results using the same masked input [PITH_FULL_IMAGE:figures/full_fig_p045_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Workflow for calculating the local stratigraphic isochronous loss ( [PITH_FULL_IMAGE:figures/full_fig_p046_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Domain-adaptation fine-tuning using unsupervised geological prior-driven [PITH_FULL_IMAGE:figures/full_fig_p047_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Progressive improvement of predicted RGT results and extracted isosur [PITH_FULL_IMAGE:figures/full_fig_p048_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Final RGT baseline results for global field clinothems datasets from various [PITH_FULL_IMAGE:figures/full_fig_p049_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: RGT predictions obtained excluding the self-supervised pretrain stage from [PITH_FULL_IMAGE:figures/full_fig_p050_9.png] view at source ↗
read the original abstract

Artificial intelligence (AI) has been increasingly applied to various geophysical scenarios, yet its practical deployment remains limited by scarce field labels, pronounced synthetic-to-field domain gaps, and insufficient physical consistency under complex and variable field conditions. To address these challenges, we propose a pretrain-to-alignment learning paradigm that systematically integrates self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning into a unified progressive learning workflow. In this paradigm, geophysical AI models are developed through sequential stages that progressively build field-relevant representations, task-specific mapping capability, field consistency, and target-specific adaptability. We validate this paradigm using cross-survey relative geologic time (RGT) estimation in global shelf-edge clinothems as a representative case study. Results from 3,000 field datasets spanning multiple sedimentary basins demonstrate that the proposed paradigm achieves accurate, robust, and well-generalized performance across diverse field surveys, while significantly improving fine-scale stratigraphic and structural details. More broadly, this study provides a practical methodological reference for a broader range of geophysical AI tasks, such as interpretation, regression, and inversion problems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces a pretrain-to-alignment learning paradigm that sequentially integrates self-supervised pretraining, synthetic supervision, prior-driven refinement, and domain-adaptation fine-tuning to enhance geophysical AI models facing scarce field labels and synthetic-to-field gaps. Validated through a case study on relative geologic time (RGT) estimation in global shelf-edge clinothems, the approach is tested on 3,000 field datasets from multiple sedimentary basins, with claims of accurate, robust, and well-generalized performance that improves fine-scale stratigraphic and structural details.

Significance. If substantiated, the paradigm could significantly advance the practical deployment of AI in geophysics by providing a structured way to leverage synthetic data while aligning to field conditions. The large-scale validation across diverse surveys strengthens the potential generalizability, offering a reference for related tasks such as interpretation and inversion.

major comments (3)
  1. [§4.2] §4.2 (Ablation experiments): No results are shown for a pipeline variant that omits the prior-driven refinement stage; without this, it is impossible to confirm that the sequential stages progressively build field-relevant representations rather than introducing uncorrected biases from earlier synthetic supervision.
  2. [Table 3] Table 3 (Cross-basin generalization metrics): The reported improvements on the 3,000 field datasets lack per-stage performance breakdowns or inconsistency checks when synthetic priors conflict with field observations; this directly undermines the central claim that the full workflow maintains physical consistency.
  3. [§5.1] §5.1 (Domain-adaptation fine-tuning): The description does not detail the mechanism for detecting and correcting overfitting to the synthetic domain; if any stage introduces uncorrectable bias, the generalization results on diverse basins cannot be attributed to the progressive alignment.
minor comments (2)
  1. [Abstract] The abstract introduces the 'pretrain-to-alignment learning paradigm' without a single-sentence operational definition; adding this would clarify the contribution for readers.
  2. [Figure 4] Figure 4 captions omit quantitative metrics (e.g., mean absolute error on fine-scale features) for the visual improvements shown in the RGT maps.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough and constructive review. The comments highlight important aspects of our ablation studies, generalization metrics, and domain-adaptation details that we will address to strengthen the manuscript. Below we respond point by point.

read point-by-point responses

  1. Referee: [§4.2] §4.2 (Ablation experiments): No results are shown for a pipeline variant that omits the prior-driven refinement stage; without this, it is impossible to confirm that the sequential stages progressively build field-relevant representations rather than introducing uncorrected biases from earlier synthetic supervision.

    Authors: We agree that demonstrating the incremental benefit of the prior-driven refinement stage is important for validating the progressive alignment claim. In the revised manuscript we will add a new ablation variant in §4.2 that removes this stage entirely, reporting quantitative metrics (e.g., RGT accuracy and stratigraphic continuity) on the same field datasets to show its specific contribution to bias correction and field-relevant representations. revision: yes

  2. Referee: [Table 3] Table 3 (Cross-basin generalization metrics): The reported improvements on the 3,000 field datasets lack per-stage performance breakdowns or inconsistency checks when synthetic priors conflict with field observations; this directly undermines the central claim that the full workflow maintains physical consistency.

    Authors: We acknowledge the value of per-stage breakdowns and explicit inconsistency checks. We will expand Table 3 to include performance metrics after each stage across the 3,000 datasets and add a dedicated analysis subsection that quantifies physical consistency (e.g., via stratigraphic continuity scores and conflict detection between synthetic priors and field observations) to directly support the physical-consistency claim. revision: yes

  3. Referee: [§5.1] §5.1 (Domain-adaptation fine-tuning): The description does not detail the mechanism for detecting and correcting overfitting to the synthetic domain; if any stage introduces uncorrectable bias, the generalization results on diverse basins cannot be attributed to the progressive alignment.

    Authors: We agree that the current description of the domain-adaptation stage is insufficiently detailed on overfitting mitigation. In the revised §5.1 we will explicitly describe the detection and correction mechanisms, including domain-discrepancy monitoring (e.g., via adversarial or MMD losses), field-specific regularization terms, and early-stopping criteria based on field validation performance, thereby clarifying how progressive alignment is preserved. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical claims rest on external field data validation

full rationale

The paper describes a four-stage progressive learning workflow (self-supervised pretraining, synthetic supervision, prior-driven refinement, domain-adaptation fine-tuning) and asserts performance via results on 3,000 field datasets across basins. No equations, fitted parameters renamed as predictions, self-definitional constructs, or load-bearing self-citations appear in the provided text that would reduce the central claims to inputs by construction. The validation is presented as external empirical evidence rather than an internal redefinition or renaming of known patterns, rendering the derivation self-contained against benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The central claim rests on the untested premise that the four-stage workflow reliably transfers knowledge from synthetic to field domains without explicit quantification of remaining domain gap or physical consistency violations.

axioms (2)
  • domain assumption Self-supervised pretraining on unlabeled geophysical data produces representations that are useful for downstream RGT estimation after adaptation.
    Invoked when describing the first stage of the paradigm in the abstract.
  • domain assumption Synthetic supervision plus prior-driven refinement can enforce physical consistency under complex field conditions.
    Stated as part of the unified progressive learning workflow.
invented entities (1)
  • pretrain-to-alignment learning paradigm no independent evidence
    purpose: Unified workflow to address scarce labels and synthetic-to-field gaps in geophysical AI.
    Newly named staged process introduced in the abstract; no independent falsifiable prediction outside the paper is provided.

pith-pipeline@v0.9.0 · 5754 in / 1381 out tokens · 31379 ms · 2026-05-19T19:30:36.028819+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

147 extracted references · 147 canonical work pages · 2 internal anchors

  1. [1]

    Deep learning for seismic inverse problems: Toward the acceleration of geophysical analysis workflows

    Amir Adler, Mauricio Araya-Polo, and Tomaso Poggio. Deep learning for seismic inverse problems: Toward the acceleration of geophysical analysis workflows. IEEE signal processing magazine, 38 0 (2): 0 89--119, 2021

    work page 2021

  2. [2]

    Basin analysis: Principles and application to petroleum play assessment

    Philip A Allen and John R Allen. Basin analysis: Principles and application to petroleum play assessment. John Wiley & Sons, 2013

    work page 2013

  3. [3]

    Current state and future directions for deep learning based automatic seismic fault interpretation: A systematic review

    Yu An, Haiwen Du, Siteng Ma, Yingjie Niu, Dairui Liu, Jing Wang, Yuhan Du, Conrad Childs, John Walsh, and Ruihai Dong. Current state and future directions for deep learning based automatic seismic fault interpretation: A systematic review. Earth-Science Reviews, 243: 0 104509, 2023

    work page 2023

  4. [4]

    The quantifiable clinothem--types, shapes and geometric relationships in the plio-pleistocene giant foresets formation, taranaki basin, new zealand

    Ingrid Anell and Ivar Midtkandal. The quantifiable clinothem--types, shapes and geometric relationships in the plio-pleistocene giant foresets formation, taranaki basin, new zealand. Basin Research, 29: 0 277--297, 2017

    work page 2017

  5. [5]

    Machine learning in microseismic monitoring

    Denis Anikiev, Claire Birnie, Umair bin Waheed, Tariq Alkhalifah, Chen Gu, Dirk J Verschuur, and Leo Eisner. Machine learning in microseismic monitoring. Earth-Science Reviews, 239: 0 104371, 2023

    work page 2023

  6. [6]

    Deep-learning tomography

    Mauricio Araya-Polo, Joseph Jennings, Amir Adler, and Taylor Dahlke. Deep-learning tomography. The Leading Edge, 37 0 (1): 0 58--66, 2018

    work page 2018

  7. [7]

    Scott Bullimore, Sverre Henriksen, Frode M Liest l, and William Helland-Hansen. Clinoform stacking patterns, shelf-edge trajectories and facies associations in tertiary coastal deltas, offshore norway: Implications for the prediction of lithology in prograding systems. Norwegian Journal of Geology/Norsk Geologisk Forening, 85, 2005

    work page 2005

  8. [8]

    Embedded physical constraints in machine learning to enhance vegetation phenology prediction

    Mengying Cao and Qihao Weng. Embedded physical constraints in machine learning to enhance vegetation phenology prediction. GIScience & Remote Sensing, 61 0 (1): 0 2426598, 2024

    work page 2024

  9. [9]

    A pretraining and fine-tuning framework for 3d fault detection based on global-local feature learning

    Shengguang Chu, Kewen Li, and Guangyue Zhou. A pretraining and fine-tuning framework for 3d fault detection based on global-local feature learning. Expert Systems with Applications, 292: 0 128641, 2025

    work page 2025

  10. [10]

    Seismic fault detection in real data using transfer learning from a convolutional neural network pre-trained with synthetic seismic data

    Augusto Cunha, Axelle Pochet, H \'e lio Lopes, and Marcelo Gattass. Seismic fault detection in real data using transfer learning from a convolutional neural network pre-trained with synthetic seismic data. Computers & Geosciences, 135: 0 104344, 2020

    work page 2020

  11. [11]

    Petrophysical properties prediction from prestack seismic data using convolutional neural networks

    Vishal Das and Tapan Mukerji. Petrophysical properties prediction from prestack seismic data using convolutional neural networks. Geophysics, 85 0 (5): 0 N41--N55, 2020

    work page 2020

  12. [12]

    Convolutional neural network for seismic impedance inversion

    Vishal Das, Ahinoam Pollack, Uri Wollner, and Tapan Mukerji. Convolutional neural network for seismic impedance inversion. Geophysics, 84 0 (6): 0 R869--R880, 2019

    work page 2019

  13. [14]

    Learning-based seismic velocity inversion with synthetic and field data

    Stuart Farris, Robert Clapp, and Mauricio Araya-Polo. Learning-based seismic velocity inversion with synthetic and field data. Sensors, 23 0 (19): 0 8277, 2023

    work page 2023

  14. [15]

    Fast structural interpretation with structure-oriented filtering

    Gijs C Fehmers and Christian FW H \"o cker. Fast structural interpretation with structure-oriented filtering. Geophysics, 68 0 (4): 0 1286--1293, 2003

    work page 2003

  15. [16]

    A foundation model enpowered by a multi-modal prompt engine for universal seismic geobody interpretation across surveys

    Hang Gao, Xinming Wu, Luming Liang, Hanlin Sheng, Xu Si, Hui Gao, and Yaxing Li. A foundation model enpowered by a multi-modal prompt engine for universal seismic geobody interpretation across surveys. Information Fusion, page 103437, 2025 a

    work page 2025

  16. [17]

    Clinoformnet-1.0: stratigraphic forward modeling and deep learning for seismic clinoform delineation

    Hui Gao, Xinming Wu, Jinyu Zhang, Xiaoming Sun, and Zhengfa Bi. Clinoformnet-1.0: stratigraphic forward modeling and deep learning for seismic clinoform delineation. Geoscientific Model Development, 16 0 (9): 0 2495--2513, 2023

    work page 2023

  17. [18]

    A geologically informed and data-driven ai workflow for fully seismic stratigraphic interpretation of sedimentary basin

    Hui Gao, Xinming Wu, and Xuesong Ding. A geologically informed and data-driven ai workflow for fully seismic stratigraphic interpretation of sedimentary basin. IEEE Transactions on Geoscience and Remote Sensing, 63: 0 1--13, 2025 b . doi:10.1109/TGRS.2025.3583211

    work page doi:10.1109/tgrs.2025.3583211 2025

  18. [19]

    Massive-scale unlabeled field and labeled synthetic seismic datasets of global shelf-edge clinothems

    Hui Gao, Xinming Wu, Jintao Li, Xiaoming Sun, and Jiarun Yang. Massive-scale unlabeled field and labeled synthetic seismic datasets of global shelf-edge clinothems, 2026 a . URL https://arxiv.org/abs/2604.17235

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  19. [21]

    Deep learning for relative geologic time and seismic horizons

    Zhicheng Geng, Xinming Wu, Yunzhi Shi, and Sergey Fomel. Deep learning for relative geologic time and seismic horizons. Geophysics, 85 0 (4): 0 WA87--WA100, 2020

    work page 2020

  20. [22]

    Structure-oriented smoothing and semblance

    Dave Hale. Structure-oriented smoothing and semblance. CWP report, 635 0 (635), 2009

    work page 2009

  21. [23]

    Geomorphological and sequence stratigraphic variability in wave-dominated, shoreface-shelf parasequences

    Gary J Hampson and Joep EA Storms. Geomorphological and sequence stratigraphic variability in wave-dominated, shoreface-shelf parasequences. Sedimentology, 50 0 (4): 0 667--701, 2003

    work page 2003

  22. [24]

    Self-supervised pre-training and few-shot finetuning for gas-bearing prediction

    Long Han, Xinming Wu, Renjie Chen, Yunhua Shi, Zhanxuan Hu, and Huijing Fang. Self-supervised pre-training and few-shot finetuning for gas-bearing prediction. Journal of Geophysical Research: Machine Learning and Computation, 2 0 (2): 0 e2025JH000631, 2025

    work page 2025

  23. [25]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll \'a r, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000--16009, 2022

    work page 2022

  24. [26]

    Nicholas E Holgate, Gary J Hampson, Christopher A-L Jackson, and Steen A Petersen. Constraining uncertainty in interpretation of seismically imaged clinoforms in deltaic reservoirs, troll field, norwegian north sea: Insights from forward seismic models of outcrop analogs. AAPG Bulletin, 98 0 (12): 0 2629--2663, 2014

    work page 2014

  25. [28]

    Mapping full seismic waveforms to vertical velocity profiles by deep learning

    Vladimir Kazei, Oleg Ovcharenko, Pavel Plotnitskii, Daniel Peter, Xiangliang Zhang, and Tariq Alkhalifah. Mapping full seismic waveforms to vertical velocity profiles by deep learning. Geophysics, 86 0 (5): 0 R711--R721, 2021

    work page 2021

  26. [29]

    Deep learning for simultaneous seismic image super-resolution and denoising

    Jintao Li, Xinming Wu, and Zhanxuan Hu. Deep learning for simultaneous seismic image super-resolution and denoising. IEEE Transactions on Geoscience and Remote Sensing, 60: 0 1--11, 2021

    work page 2021

  27. [30]

    Cigvis: An open-source python tool for the real-time interactive visualization of multidimensional geophysical data

    Jintao Li, Yunzhi Shi, and Xinming Wu. Cigvis: An open-source python tool for the real-time interactive visualization of multidimensional geophysical data. Geophysics, 90 0 (1): 0 F1--F10, 2025

    work page 2025

  28. [31]

    Machine learning for subsurface geological feature identification from seismic data: Methods, datasets, challenges, and opportunities

    Lei Lin, Zhi Zhong, Chenglong Li, Andrew Gorman, Hao Wei, Yanbin Kuang, Shiqi Wen, Zhongxian Cai, and Fang Hao. Machine learning for subsurface geological feature identification from seismic data: Methods, datasets, challenges, and opportunities. Earth-science reviews, 257: 0 104887, 2024

    work page 2024

  29. [32]

    Deep transfer learning for seismic characterization of strike-slip faults in karstified carbonates from the northern tarim basin

    Jiawei Liu, Guanghui Wu, Lixin Chen, Xiaoguo Wan, Bingshan Ma, Ransong Zhang, Chen Qiu, and Xupeng Wang. Deep transfer learning for seismic characterization of strike-slip faults in karstified carbonates from the northern tarim basin. Scientific Reports, 15 0 (1): 0 9242, 2025

    work page 2025

  30. [34]

    Seismic features and automatic discrimination of deep and shallow induced-microearthquakes using neural network and logistic regression

    S Mostafa Mousavi, Stephen P Horton, Charles A Langston, and Borhan Samei. Seismic features and automatic discrimination of deep and shallow induced-microearthquakes using neural network and logistic regression. Geophysical Journal International, 207 0 (1): 0 29--46, 2016

    work page 2016

  31. [35]

    Earthquake transformer—an attentive deep-learning model for simultaneous earthquake detection and phase picking

    S Mostafa Mousavi, William L Ellsworth, Weiqiang Zhu, Lindsay Y Chuang, and Gregory C Beroza. Earthquake transformer—an attentive deep-learning model for simultaneous earthquake detection and phase picking. Nature communications, 11 0 (1): 0 3952, 2020

    work page 2020

  32. [36]

    Clinoforms and clinoform systems: Review and dynamic classification scheme for shorelines, subaqueous deltas, shelf edges and continental margins

    Stefano Patruno and William Helland-Hansen. Clinoforms and clinoform systems: Review and dynamic classification scheme for shorelines, subaqueous deltas, shelf edges and continental margins. Earth-Science Reviews, 185: 0 202--233, 2018

    work page 2018

  33. [37]

    Automatic channel detection using deep learning

    Nam Pham, Sergey Fomel, and Dallas Dunlap. Automatic channel detection using deep learning. Interpretation, 7 0 (3): 0 SE43--SE50, 2019

    work page 2019

  34. [38]

    Clinoform development by advection-diffusion of suspended sediment: Modeling and comparison to natural systems

    Carlos Pirmez, Lincoln F Pratson, and Michael S Steckler. Clinoform development by advection-diffusion of suspended sediment: Modeling and comparison to natural systems. Journal of Geophysical Research: Solid Earth, 103 0 (B10): 0 24141--24157, 1998

    work page 1998

  35. [39]

    Seismic geomorphology--an overview

    Henry W Posamentier, Richard J Davies, Joseph A Cartwright, and Lesli Wood. Seismic geomorphology--an overview. 2007

    work page 2007

  36. [40]

    A large-scale benchmark on geological fault delineation models: Domain shift, training dynamics, generalizability, evaluation, and inferential behavior

    Jorge Quesada, Chen Zhou, Prithwijit Chowdhury, Mohammad Alotaibi, Ahmad Mustafa, Yusufjon Kumakov, Mohit Prabhushankar, and Ghassan AlRegib. A large-scale benchmark on geological fault delineation models: Domain shift, training dynamics, generalizability, evaluation, and inferential behavior. IEEE Access, 13: 0 215110--215131, 2025

    work page 2025

  37. [41]

    Seismic foundation model: A next generation deep-learning model in geophysics

    Hanlin Sheng, Xinming Wu, Xu Si, Jintao Li, Sibo Zhang, and Xudong Duan. Seismic foundation model: A next generation deep-learning model in geophysics. Geophysics, 90 0 (2): 0 IM59--IM79, 02 2025. ISSN 0016-8033. doi:10.1190/geo2024-0262.1

    work page doi:10.1190/geo2024-0262.1 2025

  38. [42]

    Saltseg: Automatic 3d salt segmentation using a deep convolutional neural network

    Yunzhi Shi, Xinming Wu, and Sergey Fomel. Saltseg: Automatic 3d salt segmentation using a deep convolutional neural network. Interpretation, 7 0 (3): 0 SE113--SE122, 2019

    work page 2019

  39. [43]

    An all-in-one seismic phase picking, location, and association network for multi-task multi-station earthquake monitoring

    Xu Si, Xinming Wu, Zefeng Li, Shenghou Wang, and Jun Zhu. An all-in-one seismic phase picking, location, and association network for multi-task multi-station earthquake monitoring. Communications Earth & Environment, 5 0 (1): 0 22, 2024

    work page 2024

  40. [44]

    Unwrapping instantaneous phase to generate a relative geologic time volume

    Tracy J Stark. Unwrapping instantaneous phase to generate a relative geologic time volume. In SEG International Exposition and Annual Meeting, pages SEG--2003. SEG, 2003

    work page 2003

  41. [45]

    Relative geologic time (age) volumes—relating every seismic sample to a geologically reasonable horizon

    Tracy J Stark. Relative geologic time (age) volumes—relating every seismic sample to a geologically reasonable horizon. The Leading Edge, 23 0 (9): 0 928--932, 2004

    work page 2004

  42. [46]

    Sequence stratigraphic models for exploration and production: Evolving methodology, emerging models and application histories, volume 22

    RJ Steel, T Olsen, JM Armentrout, and NC Rosen. Sequence stratigraphic models for exploration and production: Evolving methodology, emerging models and application histories, volume 22. 2002

    work page 2002

  43. [47]

    Stacked shelf-edge delta reservoirs of the columbus basin, trinidad, west indies

    Johan C Sydow, Joe Finneran, and Andrew P Bowman. Stacked shelf-edge delta reservoirs of the columbus basin, trinidad, west indies. 2013

    work page 2013

  44. [48]

    Estimators for orientation and anisotropy in digitized images

    Lucas J Van Vliet and Piet W Verbeek. Estimators for orientation and anisotropy in digitized images. In ASCI, volume 95, pages 16--18, 1995

    work page 1995

  45. [49]

    Multiscale structural similarity for image quality assessment

    Zhou Wang, Eero P Simoncelli, and Alan C Bovik. Multiscale structural similarity for image quality assessment. In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, volume 2, pages 1398--1402. Ieee, 2003

    work page 2003

  46. [50]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing, 13 0 (4): 0 600--612, 2004

    work page 2004

  47. [51]

    Faultseg3d: Using synthetic data sets to train an end-to-end convolutional neural network for 3d seismic fault segmentation

    Xinming Wu, Luming Liang, Yunzhi Shi, and Sergey Fomel. Faultseg3d: Using synthetic data sets to train an end-to-end convolutional neural network for 3d seismic fault segmentation. Geophysics, 84 0 (3): 0 IM35--IM45, 2019

    work page 2019

  48. [52]

    Building realistic structure models to train convolutional neural networks for seismic structural interpretation

    Xinming Wu, Zhicheng Geng, Yunzhi Shi, Nam Pham, Sergey Fomel, and Guillaume Caumon. Building realistic structure models to train convolutional neural networks for seismic structural interpretation. Geophysics, 85 0 (4): 0 WA27--WA39, 2020

    work page 2020

  49. [53]

    Sensing prior constraints in deep neural networks for solving exploration geophysical problems

    Xinming Wu, Jianwei Ma, Xu Si, Zhengfa Bi, Jiarun Yang, Hui Gao, Dongzi Xie, Zhixiang Guo, and Jie Zhang. Sensing prior constraints in deep neural networks for solving exploration geophysical problems. Proceedings of the National Academy of Sciences, 120 0 (23): 0 e2219573120, 2023

    work page 2023

  50. [54]

    Parametric convolutional neural network-domain full-waveform inversion

    Yulang Wu and George A McMechan. Parametric convolutional neural network-domain full-waveform inversion. Geophysics, 84 0 (6): 0 R881--R896, 2019

    work page 2019

  51. [55]

    Seismic fault detection with convolutional neural network

    Wei Xiong, Xu Ji, Yue Ma, Yuxiang Wang, Nasher M AlBinHassan, Mustafa N Ali, and Yi Luo. Seismic fault detection with convolutional neural network. Geophysics, 83 0 (5): 0 O97--O103, 2018

    work page 2018

  52. [56]

    Machine learning-based seismic subsurface characterization: The state of the art and future perspectives

    Minghui Xu, Luanxiao Zhao, Mingliang Liu, and Jianhua Geng. Machine learning-based seismic subsurface characterization: The state of the art and future perspectives. Journal of Geophysical Research: Machine Learning and Computation, 2 0 (4): 0 e2025JH000846, 2025 a

    work page 2025

  53. [57]

    Geological knowledge-guided dual-branch deep learning model for identification of geochemical anomalies related to mineralization

    Ying Xu, Renguang Zuo, and Yang Bai. Geological knowledge-guided dual-branch deep learning model for identification of geochemical anomalies related to mineralization. Journal of Geophysical Research: Machine Learning and Computation, 2 0 (1): 0 e2024JH000468, 2025 b

    work page 2025

  54. [58]

    3d saltseg-cl: Unsupervised embedding characterization based multi-task dense prediction method for 3d salt bodies

    Zhifeng Xu, Kewen Li, Ruonan Yin, Yating Fan, and Jian Ma. 3d saltseg-cl: Unsupervised embedding characterization based multi-task dense prediction method for 3d salt bodies. Expert Systems with Applications, 267: 0 126249, 2025 c

    work page 2025

  55. [59]

    Deep-learning inversion: A next-generation seismic velocity model building method

    Fangshu Yang and Jianwei Ma. Deep-learning inversion: A next-generation seismic velocity model building method. Geophysics, 84 0 (4): 0 R583--R599, 2019

    work page 2019

  56. [60]

    Seismic resolution enhancement using physics-assisted seismic deconvolution network and domain adaptation

    Yang Yang, Zhuo Wang, Naihao Liu, Yuxin Zhang, Rongchang Liu, and Jinghuai Gao. Seismic resolution enhancement using physics-assisted seismic deconvolution network and domain adaptation. Geophysics, 90 0 (3): 0 R113--R125, 2025

    work page 2025

  57. [61]

    Fault2seisgan: A method for the expansion of fault datasets based on generative adversarial networks

    Shuo Zhao, Renwei Ding, Tianjiao Han, YiLin Liu, Jinwei Zhang, and Lihong Zhao. Fault2seisgan: A method for the expansion of fault datasets based on generative adversarial networks. Frontiers in Earth Science, 11: 0 1091803, 2023 a

    work page 2023

  58. [62]

    Artificial intelligence for geoscience: Progress, challenges, and perspectives

    Tianjie Zhao, Sheng Wang, Chaojun Ouyang, Min Chen, Chenying Liu, Jin Zhang, Long Yu, Fei Wang, Yong Xie, Jun Li, et al. Artificial intelligence for geoscience: Progress, challenges, and perspectives. The Innovation, 5 0 (5), 2024

    work page 2024

  59. [63]

    Few-shot learning for seismic facies segmentation via prototype learning

    Yunhe Zhao, Bianfang Chai, Liangxun Shuo, Zenghao Li, Heng Wu, and Tianyi Wang. Few-shot learning for seismic facies segmentation via prototype learning. Geophysics, 88 0 (3): 0 IM41--IM49, 2023 b

    work page 2023

  60. [64]

    Fault-pls-pcl: Cross-domain seismic fault detection via pseudo-label selection and prototype contrastive learning

    Guangyue Zhou, Kewen Li, Ruonan Yin, and Shengguang Chu. Fault-pls-pcl: Cross-domain seismic fault detection via pseudo-label selection and prototype contrastive learning. Pattern Recognition, page 112960, 2025

    work page 2025

  61. [65]

    Seismic fault detection with iterative deep learning

    Ruoshui Zhou, Yufei Cai, Fucai Yu, and Guangmin Hu. Seismic fault detection with iterative deep learning. In SEG International Exposition and Annual Meeting, page D033S077R006. SEG, 2019

    work page 2019

  62. [66]

    Learning from unlabelled real seismic data: Fault detection based on transfer learning

    Ruoshui Zhou, Xingmiao Yao, Guangmin Hu, and Fucai Yu. Learning from unlabelled real seismic data: Fault detection based on transfer learning. Geophysical Prospecting, 69 0 (6): 0 1218--1234, 2021

    work page 2021

  63. [67]

    Phasenet: a deep-neural-network-based seismic arrival-time picking method

    Weiqiang Zhu and Gregory C Beroza. Phasenet: a deep-neural-network-based seismic arrival-time picking method. Geophysical Journal International, 216 0 (1): 0 261--273, 2019

    work page 2019

  64. [68]

    A novel data-knowledge dual-driven model coupling artificial intelligence with a mineral systems approach for mineral prospectivity mapping

    Renguang Zuo, Fanfan Yang, Qiuming Cheng, and Oliver P Kreuzer. A novel data-knowledge dual-driven model coupling artificial intelligence with a mineral systems approach for mineral prospectivity mapping. Geology, 2024

    work page 2024

  65. [69]

    Sequence stratigraphic models for exploration and production: Evolving methodology, emerging models and application histories , author=

    work page

  66. [70]

    Sedimentology , volume=

    Geomorphological and sequence stratigraphic variability in wave-dominated, shoreface-shelf parasequences , author=. Sedimentology , volume=. 2003 , publisher=

    work page 2003

  67. [71]

    Earth-Science Reviews , volume=

    Clinoforms and clinoform systems: Review and dynamic classification scheme for shorelines, subaqueous deltas, shelf edges and continental margins , author=. Earth-Science Reviews , volume=. 2018 , publisher=

    work page 2018

  68. [72]

    Seismic geomorphology--an overview , author=

    work page

  69. [73]

    Journal of Geophysical Research: Solid Earth , volume=

    Clinoform development by advection-diffusion of suspended sediment: Modeling and comparison to natural systems , author=. Journal of Geophysical Research: Solid Earth , volume=. 1998 , publisher=

    work page 1998

  70. [74]

    , author=

    Clinoform stacking patterns, shelf-edge trajectories and facies associations in Tertiary coastal deltas, offshore Norway: Implications for the prediction of lithology in prograding systems. , author=. Norwegian Journal of Geology/Norsk Geologisk Forening , volume=

    work page

  71. [75]

    Stacked shelf-edge delta reservoirs of the Columbus Basin, Trinidad, West Indies , author=

    work page

  72. [76]

    AAPG Bulletin , volume=

    Constraining uncertainty in interpretation of seismically imaged clinoforms in deltaic reservoirs, Troll field, Norwegian North Sea: Insights from forward seismic models of outcrop analogs , author=. AAPG Bulletin , volume=

    work page

  73. [77]

    2013 , publisher=

    Basin analysis: Principles and application to petroleum play assessment , author=. 2013 , publisher=

    work page 2013

  74. [78]

    SEG International Exposition and Annual Meeting , pages=

    Seismic stratigraphic interpretation from a geological model-A north sea case study , author=. SEG International Exposition and Annual Meeting , pages=. 2011 , organization=

    work page 2011

  75. [79]

    Geophysics , volume=

    Predictive painting of 3D seismic volumes , author=. Geophysics , volume=. 2010 , publisher=

    work page 2010

  76. [80]

    Journal of Applied Geophysics , volume=

    Fast seismic horizon reconstruction based on local dip transformation , author=. Journal of Applied Geophysics , volume=. 2013 , publisher=

    work page 2013

  77. [81]

    Geophysics , volume=

    Least-squares horizons with local slopes and multigrid correlations , author=. Geophysics , volume=. 2018 , publisher=

    work page 2018

  78. [82]

    SEG International Exposition and Annual Meeting , pages=

    Unwrapping instantaneous phase to generate a relative geologic time volume , author=. SEG International Exposition and Annual Meeting , pages=. 2003 , organization=

    work page 2003

  79. [83]

    The Leading Edge , volume=

    Relative geologic time (age) volumes—Relating every seismic sample to a geologically reasonable horizon , author=. The Leading Edge , volume=. 2004 , publisher=

    work page 2004

  80. [84]

    Geophysics , volume=

    Deep learning for relative geologic time and seismic horizons , author=. Geophysics , volume=. 2020 , publisher=

    work page 2020

Showing first 80 references.