arxiv: 2605.01677 · v1 · submitted 2026-05-03 · ⚛️ physics.geo-ph

Recognition: unknown

A Review of Modeling and Waveform Inversion for Marine Seismic Data

Guoxin Chen

Authors on Pith no claims yet

Pith reviewed 2026-05-09 16:49 UTC · model grok-4.3

classification ⚛️ physics.geo-ph

keywords marine seismicwaveform inversionfull-waveform inversionelastic inversionartificial intelligencenumerical modelingdata preprocessingreservoir characterization

0 comments

The pith

Intelligent interpolation and AI-guided inversion solve key bottlenecks in marine seismic data

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

This review establishes that recent advances in marine seismic modeling and waveform inversion combine physics-based methods with artificial intelligence to address practical challenges. It argues that intelligent interpolation, multi-source joint inversion, low-frequency recovery with cycle-skipping suppression, physics-guided deep learning, and wide-band velocity modeling provide effective solutions across ocean-bottom, streamer, and passive-source surveys. If these hold, the result is a complete workflow from theory through engineering applications that supports resource exploration and environmental monitoring. Readers should care because marine seismic data underpins offshore safety, resource discovery, and climate-related projects such as carbon storage.

Core claim

The paper claims that integrating full-waveform inversion, elastic inversion, numerical modeling, and artificial intelligence is shifting marine seismic work from purely physics-driven methods to hybrid physics-constrained and data-driven modes. Reviewing work across data preprocessing, forward modeling, inversion, reservoir characterization, and imaging, it concludes that intelligent interpolation, multi-source joint inversion, low-frequency recovery and cycle-skipping suppression, physics-guided deep learning inversion, and wide-band velocity modeling form the main solutions to industrial bottlenecks in OBN/OBC, streamer, and passive-source cases. These create an end-to-end system that can

What carries the argument

Hybrid physics-guided deep learning inversion that merges traditional waveform modeling with artificial intelligence to recover low frequencies and suppress cycle skipping

If this is right

Enables reliable velocity models for deep-water exploration surveys
Supports joint inversion from multiple sources to improve data coverage and quality
Allows physics-guided deep learning to reduce cycle-skipping problems in inversion
Delivers wide-band velocity models useful for reservoir characterization and hazard detection
Creates an integrated workflow from modeling through imaging that supports carbon sequestration monitoring

Where Pith is reading between the lines

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

Similar hybrid techniques might transfer to land seismic or other geophysical inversion problems where low-frequency data is scarce.
Real-time application of these methods could improve dynamic monitoring of offshore structures.
The emphasis on physics constraints may guide development of more stable AI models in related wave-propagation fields.

Load-bearing premise

The selected papers accurately capture the main solutions in the field without selection bias or overgeneralization from their source context.

What would settle it

Applying the reviewed techniques to a known marine dataset with persistent cycle-skipping artifacts and observing no improvement in inversion results would show they are not effective key solutions.

read the original abstract

Marine seismic exploration is a core technology supporting marine resource exploration, seabed detection, carbon sequestration monitoring, and offshore engineering safety. The integration of full-waveform inversion (FWI), elastic inversion, numerical modeling, and artificial intelligence is driving a paradigm shift from physics-driven to physics-constrained and data-driven hybrid mode. Based on the JMSE special issue Modeling and Waveform Inversion of Marine Seismic Data, this paper systematically reviews 11 papers across six areas: data preprocessing, forward modeling, FWI, elastic inversion, reservoir characterization, and migration imaging. Results show that intelligent interpolation, multi-source joint inversion, low-frequency recovery and cycle-skipping suppression, physics-guided deep learning inversion, and wide-band velocity modeling are key solutions to industrial bottlenecks in OBN/OBC, streamer, and passive-source scenarios. These achievements form a complete system from theory to engineering application, supporting deep-water exploration, seabed hazard detection, and carbon sequestration monitoring. This paper also introduces the new JMSE special issue Marine Geophysical Exploration in the Era of Artificial Intelligence, summarizes recent AI-based advances, and prospects future trends of AI and marine seismic integration.

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.

Desk Editor's Note private letter to a colleague

This is a narrow-scope review summarizing 11 papers from one JMSE special issue with no new methods or analysis.

read the letter

This paper is a review that pulls together 11 articles from a JMSE special issue on marine seismic data modeling and inversion. It doesn't present any new methods, data, or derivations of its own. It organizes the material into six categories—data preprocessing, forward modeling, full-waveform inversion, elastic inversion, reservoir characterization, and migration imaging—and flags some recurring techniques such as intelligent interpolation, multi-source joint inversion, low-frequency recovery, physics-guided deep learning, and wide-band velocity modeling as practical responses to bottlenecks in OBN/OBC, streamer, and passive-source data. The structure is clear and the abstract does a reasonable job pointing readers toward the original papers and a follow-on special issue on AI integration. The main limitation is that the review offers no justification for selecting those 11 papers, no comparison with work published outside the special issue, and no critical assessment of where the highlighted methods fall short or how they compare quantitatively to alternatives. The claim that they form a complete system supporting deep-water exploration and carbon monitoring therefore rests on an unexamined assumption of representativeness. This kind of paper could be useful for someone entering the subfield who wants a quick map of that particular JMSE collection, but it adds little synthesis or perspective for specialists already following the literature. I would send it to peer review for a journal that publishes reviews, with the expectation that referees will ask for broader context and selection criteria.

Referee Report

2 major / 2 minor

Summary. This review summarizes 11 papers from the JMSE special issue on Modeling and Waveform Inversion of Marine Seismic Data. It organizes the work into six areas (data preprocessing, forward modeling, FWI, elastic inversion, reservoir characterization, and migration imaging) and concludes that intelligent interpolation, multi-source joint inversion, low-frequency recovery/cycle-skipping suppression, physics-guided deep learning inversion, and wide-band velocity modeling constitute key solutions that form a complete system addressing industrial bottlenecks in OBN/OBC, streamer, and passive-source scenarios, while also announcing a follow-on JMSE special issue on AI in marine geophysics.

Significance. If the synthesis accurately captures the 11 papers without selection bias, the review could serve as a concise entry point to recent hybrid physics-AI approaches in marine seismics, potentially informing work on deep-water exploration and monitoring applications. Its value is constrained by the narrow sourcing and absence of quantitative cross-validation against the wider literature.

major comments (2)

[Abstract] Abstract and implied Results section: the assertion that the five listed techniques 'are key solutions to industrial bottlenecks' and 'form a complete system' rests on the unstated premise that the 11 JMSE special-issue papers are representative of the field; no selection criteria, inclusion/exclusion rationale, or comparison to contemporaneous non-JMSE literature is supplied, rendering the generalization unsupported.
[Abstract] Abstract: the claim that these techniques resolve bottlenecks 'across OBN/OBC, streamer, and passive-source scenarios' lacks any tabulated performance metrics, benchmark comparisons, or falsifiable tests showing superiority over alternative methods; the review therefore provides no evidence that the highlighted approaches outperform existing solutions.

minor comments (2)

[Abstract] The abstract introduces the new JMSE special issue but supplies no scope statement, submission timeline, or explicit linkage to the current review's findings.
The six-area organizational structure is announced but the manuscript provides no explicit mapping of the 11 papers to these areas or a summary table that would allow readers to trace individual contributions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our review manuscript. We address each major comment point by point below, indicating the specific revisions we will make to improve clarity and scope.

read point-by-point responses

Referee: [Abstract] Abstract and implied Results section: the assertion that the five listed techniques 'are key solutions to industrial bottlenecks' and 'form a complete system' rests on the unstated premise that the 11 JMSE special-issue papers are representative of the field; no selection criteria, inclusion/exclusion rationale, or comparison to contemporaneous non-JMSE literature is supplied, rendering the generalization unsupported.

Authors: The manuscript is explicitly presented as a review of the 11 papers from the JMSE special issue 'Modeling and Waveform Inversion of Marine Seismic Data'. The abstract opens with 'Based on the JMSE special issue...', indicating that the synthesis and conclusions are drawn directly from these contributions. We will revise the abstract and the final section to state clearly that the five techniques are identified as key solutions within the context of this special-issue collection. We will also add a sentence specifying the inclusion criteria (all papers accepted in the special issue) and note the limited scope, avoiding any implication of broader field representativeness. revision: yes
Referee: [Abstract] Abstract: the claim that these techniques resolve bottlenecks 'across OBN/OBC, streamer, and passive-source scenarios' lacks any tabulated performance metrics, benchmark comparisons, or falsifiable tests showing superiority over alternative methods; the review therefore provides no evidence that the highlighted approaches outperform existing solutions.

Authors: As a review summarizing existing papers, we do not introduce new benchmark tests, quantitative metrics, or cross-method comparisons. The claims reflect the scenarios and outcomes reported in the 11 papers, which collectively cover OBN/OBC, streamer, and passive-source cases. We will add a summary table to the manuscript that lists each reviewed paper, the data scenario it addresses, and the specific industrial bottleneck it targets according to the original authors. We will also moderate the abstract language to describe the techniques as contributing solutions demonstrated in the reviewed works, rather than asserting they 'form a complete system' or definitively outperform alternatives. This makes the evidential basis more transparent while remaining within the review's scope. revision: partial

Circularity Check

0 steps flagged

No circularity: literature review summarizes external papers without internal derivations or self-referential predictions

full rationale

The paper is explicitly a review of 11 papers from the JMSE special issue on Modeling and Waveform Inversion of Marine Seismic Data. It describes techniques such as intelligent interpolation and physics-guided deep learning inversion as key solutions based on those external contributions, without any new mathematical derivations, first-principles predictions, fitted parameters, or equations that could reduce to the paper's own inputs by construction. No self-citation chains, uniqueness theorems, or ansatzes are invoked to justify the central claims; the structure is descriptive summarization of cited literature rather than a derivation chain.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

As a review paper, it does not introduce new free parameters, axioms, or invented entities. It relies on the standard assumptions and methods of marine seismic geophysics as described in the reviewed papers.

pith-pipeline@v0.9.0 · 5486 in / 1133 out tokens · 38723 ms · 2026-05-09T16:49:55.582210+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

19 extracted references · 2 canonical work pages

[1]

Earth and Space Science 2025, 12(8): e2025EA004225

A denoising method for salt and pepper noise in remote sensing based on Swin-Transformer convolution U-Net and filtering-FSCU-Net. Earth and Space Science 2025, 12(8): e2025EA004225. Chen, G., et al.,

2025
[2]

arXiv preprint arXiv:2603.17701

Deep Learning-Based 3D Seismic Velocity Inversion Under Dual- Domain Sparse Representation. arXiv preprint arXiv:2603.17701. Chen, G

work page arXiv
[3]

arXiv preprint arXiv:2407.19419

Accurate background velocity model building method based on iterative deep learning in sparse transform domain. arXiv preprint arXiv:2407.19419. Chen, G. et al.,

work page arXiv
[4]

Geophysical Journal International 2024, 237(3):1674–1698

Joint Model and Data-Driven Simultaneous Inversion of Velocity and Density. Geophysical Journal International 2024, 237(3):1674–1698. Chen G., Wu R.S., Chen, S.C

2024
[5]

85th EAGE Conference and Exhibition, Oslo, Norway, 2024.6.10- 6.13

Simultaneous high-fidelity inversion of velocity and density based on deep learning assistance. 85th EAGE Conference and Exhibition, Oslo, Norway, 2024.6.10- 6.13. Chen, J. et al

2024
[6]

IEEE Transactions on Geoscience and Remote Sensing 2024, 62:1-14

Efficient Seismic Data Denoising via Deep Learning with Improved MCA-SCUNet. IEEE Transactions on Geoscience and Remote Sensing 2024, 62:1-14. Chen, G,

2024
[7]

Journal of Earth Science 2025, 36(2):830-834

High-Precision Sub-Seafloor Velocity Building Based on Joint Tomography and Deep Learning on OBS Data in the South China Sea. Journal of Earth Science 2025, 36(2):830-834. Du, R. et al.,

2025
[8]

Numerical Modeling on Ocean-Bottom Seismograph P-Wave Receiver Function to Analyze Influences of Seawater and Sedimentary Layers. J. Mar. Sci. Eng. 2024, 12,

2024
[9]

Han, G.; Li, Y .; Huang, J. 2025a. Hierarchical Joint Elastic Full Waveform Inversion Based on Wavefield Separation for Marine Seismic Data. J. Mar. Sci. Eng. 2025, 13,

2025
[10]

Han, B.; Liang, Q.; Mao, W.; Chen, G. 2025b. Imaging Ocean-Bottom Seismic Data with Acoustic Kirchhoff Pre-Stack Depth Migration: A Numerical Investigation of Migration Responses and Crosstalk Artifacts. J. Mar. Sci. Eng. 2025, 13,

2025
[11]

Integrating Multimodal Deep Learning with Multipoint Statistics for 3D Crustal Modeling: A Case Study of the South China Sea. J. Mar. Sci. Eng. 2024, 12,

2024
[12]

Wave Equation Reflection Inversion for Wide- Spectrum Velocity Model Reconstruction in Marine Seismic. J. Mar. Sci. Eng. 2025, 13,

2025
[13]

Seismic denoising via multi-round SCU-Net-SSIM

Qi, Y ., et al., 2025a. Seismic denoising via multi-round SCU-Net-SSIM. IEEE Geoscience and Remote Sensing Letters 2025, 23, 1-5, 2026, Art no. 7500805. Qi, Y . L., et al. 2025b. Multi-round deep learning based on SCU-Net for seismic data denoising. 86th EAGE Annual Conference & Exhibition. Qu, S.; Hu, Y .; Huang, X.; Fang, J.; Jiang, Z

2025
[14]

Elastic Wave Phase Inversion in the Local-Scale Frequency–Wavenumber Domain with Marine Towed Simultaneous Sources. J. Mar. Sci. Eng. 2025, 13,

2025
[15]

Improving the Seismic Impedance Inversion by Fully Convolutional Neural Network. J. Mar. Sci. Eng. 2025, 13,

2025
[16]

Reservoir Characterization Based on Bayesian Amplitude Versus Offset Inversion of Marine Seismic Data. J. Mar. Sci. Eng. 2025, 13,

2025
[17]

Full Waveform Inversion of Irregularly Sampled Passive Seismic Data Based on Robust Multi-Dimensional Deconvolution. J. Mar. Sci. Eng. 2025, 13,

2025
[18]

Intelligent Interpolation of OBN Multi-Component Seismic Data Using a Frequency-Domain Residual-Attention U-Net. J. Mar. Sci. Eng. 2026, 14,

2026
[19]

Physics-Guided Self-Supervised Learning Full Waveform Inversion with Pretraining on Simultaneous Source. J. Mar. Sci. Eng. 2025, 13,

2025