Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework

arXiv:2602.07759 · physics.geo-ph · Submitted 2026-02-08 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data".

Mira: Here's the gist of the paper. Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework is a study that proposes a propagation-based framework where decomposed free surface ghost wavefields are treated as physically meaningful components that can be constructively reused as additional wavefield components This approach provides a processing-based complement to conventional ghost decomposition technologies and suggests a unified approach in which acquisition geometry and ghost processing are jointly optimised to maximise the usable wavefield information Theoretical Foundation of Ghost Interpretation Free surface ghosts arise from the reflection of seismic wavefields at the sea surface, which acts as an approximately pressure-release boundary in marine environments From a kinematic perspective, free surface ghosts can be represented as delayed replicas of the primary wavefields In the frequency domain, the superposition of primary and ghost wavefields modulates the observed wavefield spectrum, producing periodic spectral notches whose locations depend explicitly on the towing depths of the source and receivers Decomposition Methodology The study derives a recursive formulation from the finite inverse expansion of the ghost operator, providing the theoretical basis for recursive predictive decomposition The ghost operator is formulated in terms of a propagation operator acting on finite-length seismic records, allowing its inverse to be expressed as a finite expansion rather than an infinite recursive filter This leads to the observation that the propagation operator is nilpotent when defined over the finite recording interval, meaning it terminates after a finite number of terms The resulting inverse operator is therefore a finite-duration operator and is represented by the finite expansion Constructive Reuse Strategy The decomposed ghost wavefields may be transformed into equivalent ghost-free wavefields through polarity reversal and propagation correction The final direct-signature estimate is obtained by wavefield superposition of the direct estimate and the corrected ghost estimate The stacking operation consistently exhibits a lower RMS error and a higher correlation coefficient than the direct estimate alone over the entire range of input S/N ratios examined Multidimensional Extension and Application The finite inverse expansion can be naturally extended to multidimensional wavefields by generalizing the temporal delay representation to a multidimensional wavefield-extrapolation operator The decomposition is performed in two successive stages: sourceside decomposition in the common-shot domain followed by receiver-side decomposition in the common-receiver domain Survey sinking restores a common acquisition geometry, and wavefield superposition combines the decomposed components to produce a reconstructed S-R gather Practical Outcomes and Implications Field-data examples demonstrate that the decomposed ghost wavefields can be constructively reused to enhance coherent reflection amplitudes and improve reflector continuity The proposed framework suggests that deeper towing may become advantageous because of its improved signal-to-noise ratio, enhanced operational stability, and reduced sensitivity to environmental conditions The study concludes that free surface ghosts can be interpreted not solely as unwanted noise, but as reusable physical wavefield components for coherent signal enhancement Future Directions Potential applications include variable-depth streamer surveys, over/under streamer configurations, dual-sensor streamer technologies, and future broadband acquisition systems Future developments may involve the joint optimisation of acquisition geometry, source design, and ghost processing rather than treating them as independent stages of the seismic workflow The mirror-source interpretation suggests that marine seismic acquisition and processing need not be regarded as independent stages of the exploration workflow The proposed framework provides a unified physical interpretation of ghost decomposition, restoration, and constructive reuse The study demonstrates that free surface ghosts can be interpreted not solely as unwanted noise, but as reusable physical wavefield components for coherent signal enhancement The proposed framework may also contribute to the recovery of low-frequency energy attenuated by the free surface ghost response The proposed framework is applicable to both future broadband acquisitions and legacy marine seismic datasets References Amundsen, L., Zhou, H., Reitan, A., Weglein, A.

Kai: B.

Mira: (2013).

Kai: On seismic deghosting by spatial deconvolution Berkhout, A.

Title and authors: Kai: So this paper is titled "Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework". It sounds very technical, but at its core, it’s about how we can actually use the noise caused by the sea surface reflection.

Mira: Yeah, it’s tackling the idea that those ghosts aren't just junk; they might actually be physical components of the wavefield itself that we can use to make things better. It suggests a way to go beyond just cleaning up data.

Lev: From an error correction standpoint, I wonder if this framework is practical when you try to implement it on real hardware with limited resources. We’re talking about turning something that’s supposed to be noise into something we can use for actual imaging, and that's a big engineering lift.

Kai: Right, the paper proposes a propagation-based method starting from a finite inverse expansion of the ghost operator to derive a recursive formulation for this reuse. It sets up this entire process as being data-adaptive through predictive decomposition.

Mira: What’s interesting is they extend this to multidimensional wavefields by using wavefield extrapolation and reciprocity, which means you can decompose the signal based on both the source and receiver domains simultaneously. That’s a big step from just looking at a one dimension.

Lev: If it works in that multidimensional sense, it implies we could be separating real reflections from mirrored ones more cleanly before even starting the final reconstruction phase. That would significantly reduce the computational overhead on actual hardware running these kinds of algorithms.

The paper's summary: Kai: They summarize this by saying that decomposing those ghost wavefields allows you to transform them into equivalent ghost-free wavefields through polarity reversal and propagation correction. Then, you superimpose the direct estimate with this corrected ghost estimate to get a final direct signature.

Mira: The key result they point to is that when you stack the direct estimate and this corrected ghost estimate together, the stacking operation consistently shows a lower root mean square error and a higher correlation coefficient than just using the direct estimate by itself across all input signal-to-noise ratios they tested.

Lev: That improvement in RMS error is what matters for real data processing; it means cleaner results even when the signal quality is poor, which is crucial for deep ocean surveys where noise gets bad fast.

Kai: The paper also shows synthetic examples demonstrating increased signal strength through this ghost reuse, and they can recover spectral amplitude notches that were degraded by the free surface ghost response, including frequencies right near DC.

Mira: So they aren't just cleaning up the signal; they’re actually recovering information that was previously lost in those specific frequency bands due to the sea surface effect. It suggests a deeper level of data recovery than traditional deghosting.

Lev: If you can recover spectral notches near DC, that points to potentially recovering low-frequency energy that gets totally attenuated by the free surface response, which is something we’ve always struggled with in marine applications.

The paper's improvements: Kai: The authors suggest several key improvements. First, they move away from simple recursive filters by deriving a finite inverse expansion where the ghost operator is formulated through a propagation operator acting on finite-length seismic records. This means the inverse is a finite expansion, not an infinite filter.

Mira: That finiteness is crucial because it allows for recursive predictive decomposition using an auxiliary wavefield, which they define as u of z equals one I plus r P of z cubed q of z, and the update rule for q of z is u of z minus q of z. It makes the process more tractable.

Lev: But then they introduce an adaptive propagation operator estimation. They say you can estimate this using a finite-length prediction error filter from the current residual wavefield to account for noise, wavelet distortion, and local variations in the effective ghost delay. That’s where I get cautious; estimating a good predictor reliably on noisy real data is always tough for hardware implementation.

Kai: They then show how you can extend this finite inverse expansion to multidimensional wavefields by generalizing the temporal delay representation into a multidimensional wavefield extrapolation operator, which allows decomposition based on source and receiver mirroring.

Mira: And they show that survey sinking restores a common acquisition geometry, and then combining the decomposed components through wavefield superposition produces a reconstructed S-R gather. This whole workflow is framed as a unified preprocessing module.

Lev: It’s ambitious to suggest that survey sinking can be used *within* this framework for constructive combination, rather than just as an external step to fix geometry issues. That moves the optimization of acquisition geometry and ghost processing into a single joint optimization problem.

Conclusion: Kai: To wrap up, the main point is that free surface ghosts can be reinterpreted not just as unwanted noise, but as deterministic propagation phenomena we can actually use for coherent signal enhancement. It links decomposition to mirror-image interpretation and survey sinking.

Mira: So what this means for the field is that they are proposing a unified physical interpretation where ghost decomposition, mirror-image interpretation, survey sinking, and wavefield superposition all work together in one preprocessing workflow.

Lev: For the real world of quantum hardware or high-fidelity seismic processing, it means we have a structured way to see how these effects propagate through the system instead of just fighting them with post-processing filters. It’s a new way to approach signal integrity that might actually be applicable even to legacy data sets.

Kai: Exactly. This framework offers a path toward enhancing coherent reflection amplitudes and reflector continuity in field-data examples, suggesting deeper towing might become better because of the improved signal-to-noise ratio they see.

Mira: It’s about treating the ghosts as reusable physical components rather than something to be discarded, which is a fundamental shift in how we approach data quality problems in marine seismic imaging.

Lev: I think what stands out is that it suggests that marine seismic acquisition and processing don't have to be treated as completely independent stages anymore; it’s about joint optimization from the start.

Arc Geo Support Co., Ltd. · Kyoto University

physics.geo-ph

Submitted: 2026-02-08

Updated: 2026-11-11

Comments: 43 pages, 12 figures, 2 tables. Revised figures and corrected minor typographical errors. The theoretical formulation and principal conclusions remain unchanged

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 85/100

The gist: Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework is a study that proposes a propagation-based framework where decomposed free surface ghost…

Key concepts

Free Surface Ghosts
These arise from seismic waves reflecting off the sea surface, which acts like a pressure-release boundary. They appear in the observed data as periodic spectral notches and can attenuate low-frequency energy, limiting usable bandwidth.
Finite Inverse Expansion
This is a mathematical method derived from the ghost operator that expresses its inverse as a finite sum rather than an infinite filter. This allows for recursive decomposition of ghost components by terminating after a specific number of terms related to the recording interval.
Constructive Reuse Strategy
Decomposed ghost wavefields are transformed using polarity reversal and propagation correction. Superimposing the corrected ghost estimate with the direct estimate results in a final signature that shows improved signal strength and better reflector continuity compared to using only the direct estimate.

Terminology

Summary

Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework is a study that proposes a propagation-based framework where decomposed free surface ghost wavefields are treated as physically meaningful components that can be constructively reused as additional wavefield components This approach provides a processing-based complement to conventional ghost decomposition technologies and suggests a unified approach in which acquisition geometry and ghost processing are jointly optimised to maximise the usable wavefield information

Theoretical Foundation of Ghost Interpretation

Free surface ghosts arise from the reflection of seismic wavefields at the sea surface, which acts as an approximately pressure-release boundary in marine environments From a kinematic perspective, free surface ghosts can be represented as delayed replicas of the primary wavefields In the frequency domain, the superposition of primary and ghost wavefields modulates the observed wavefield spectrum, producing periodic spectral notches whose locations depend explicitly on the towing depths of the source and receivers

Decomposition Methodology

The study derives a recursive formulation from the finite inverse expansion of the ghost operator, providing the theoretical basis for recursive predictive decomposition The ghost operator is formulated in terms of a propagation operator acting on finite-length seismic records, allowing its inverse to be expressed as a finite expansion rather than an infinite recursive filter This leads to the observation that the propagation operator is nilpotent when defined over the finite recording interval, meaning it terminates after a finite number of terms The resulting inverse operator is therefore a finite-duration operator and is represented by the finite expansion

Constructive Reuse Strategy

The decomposed ghost wavefields may be transformed into equivalent ghost-free wavefields through polarity reversal and propagation correction The final direct-signature estimate is obtained by wavefield superposition of the direct estimate and the corrected ghost estimate The stacking operation consistently exhibits a lower RMS error and a higher correlation coefficient than the direct estimate alone over the entire range of input S/N ratios examined

Multidimensional Extension and Application

The finite inverse expansion can be naturally extended to multidimensional wavefields by generalizing the temporal delay representation to a multidimensional wavefield-extrapolation operator The decomposition is performed in two successive stages: sourceside decomposition in the common-shot domain followed by receiver-side decomposition in the common-receiver domain Survey sinking restores a common acquisition geometry, and wavefield superposition combines the decomposed components to produce a reconstructed S-R gather

Practical Outcomes and Implications

Field-data examples demonstrate that the decomposed ghost wavefields can be constructively reused to enhance coherent reflection amplitudes and improve reflector continuity The proposed framework suggests that deeper towing may become advantageous because of its improved signal-to-noise ratio, enhanced operational stability, and reduced sensitivity to environmental conditions The study concludes that free surface ghosts can be interpreted not solely as unwanted noise, but as reusable physical wavefield components for coherent signal enhancement

Future Directions

Potential applications include variable-depth streamer surveys, over/under streamer configurations, dual-sensor streamer technologies, and future broadband acquisition systems Future developments may involve the joint optimisation of acquisition geometry, source design, and ghost processing rather than treating them as independent stages of the seismic workflow The mirror-source interpretation suggests that marine seismic acquisition and processing need not be regarded as independent stages of the exploration workflow

Conclusions

The proposed framework provides a unified physical interpretation of ghost decomposition, restoration, and constructive reuse The study demonstrates that free surface ghosts can be interpreted not solely as unwanted noise, but as reusable physical wavefield components for coherent signal enhancement The proposed framework may also contribute to the recovery of low-frequency energy attenuated by the free surface ghost response The proposed framework is applicable to both future broadband acquisitions and legacy marine seismic datasets

References

Amundsen, L., Zhou, H., Reitan, A., Weglein, A. B. (2013). On seismic deghosting by spatial deconvolution Berkhout, A. p. 5

Where the paper says it

Constructive Reuse of Free-Surface Ghost where 𝜁 denotes the z-transform variable used in the one-dimensional formulation; the symbol 𝜁 is adopted to distinguish it from the vertical spatial coordinate 𝑧 introduced later. The ghost component is interpreted as a delayed replica of the direct signature generated by reflection at the free surface.

J. (1980). Seismic Migration: Imaging of Acoustic Energy by Wave Field Extrapolation Carlson, D. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

, Long, R., Loyd, J., Zaske, J., Day, C. L., Hickey, C. L. (2007). Broadband seismic acquisition and processing Claerbout, J. p. 2
Where the paper says it

The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey. The results demonstrate the practical feasibility of the proposed framework and its potential as a processing-based extension to conventional marine seismic acquisition for broadband imaging and inversion.

F. (1985). Imaging the Earth's Interior Dai, W., Schuster, G. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

T. (2013). Plane-wave least-squares reverse-time migration Dellinger, J. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

, Ross, A., Meaux, D., Brenders, A., Harper, M., Naranjo, J., Cole, R., Pool, R., Mitchell, S. (2016). Wolfspar an ‘FWI-friendly’ ultralow-frequency marine seismic source Ferber, R. p. 5
Where the paper says it

In conventional marine seismic processing, the recorded seismic trace is commonly interpreted as the convolution of the source wavelet with a ghost operator representing free surface reflections at the source and receiver. In the frequency domain, this operator produces depth-dependent interference patterns that appear as periodic spectral notches.

, Beasley, C. J. (2014). Simulating ultra-deep-tow marine seismic data for receiver deghosting Gazdag, J. p. 2
Where the paper says it

Starting from the finite inverse expansion of the ghost operator, the proposed framework is developed into a multidimensional wavefield propagation framework incorporating source- and receiver-side ghost decomposition, survey sinking, wavefield superposition, and the constructive reuse of decomposed ghost wavefields. The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey.

(1978). Wave equation migration with the phase-shift method Goupillaud, P. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

L. (1961). An approach to inverse filtering of near-surface layer effects from seismic records Groenestijn, G. p. 1
Where the paper says it

Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al. , 1992; van Groenestijn and Verschuur, 2009).

J. A. van, Verschuur, D. p. 1
Where the paper says it

Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al. , 1992; van Groenestijn and Verschuur, 2009).

J., 2009). Estimating primaries by sparse inversion and application to near-offset data reconstruction Itenburg, R. p. 0
Where the paper says it

Synthetic examples demonstrate increased signal strength through ghost reuse, spectral amplitude recovery of ghost-related notches including frequencies near DC, and clear decomposition of ghost-related wavefield components. Field-data examples acquired using a conventional shallow-towed airgun survey further demonstrate improvements in reflector continuity and coherent reflection energy, together with partial restoration of spectral amplitudes degraded by the free surface ghost, supporting the practical applicability of the proposed framework.

, Christensen, J. N., Söllner, R. J., 2006). Dual-sensor streamer technology Jamali Hondori, E. p. 1
Where the paper says it

Consequently, considerable effort has been devoted to extending the usable bandwidth through both acquisition and processing technologies. Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al.

, Katou, M., Tara, K., Asakawa, E., Mikada, H. (2019). Mirror reverse time migration using vertical cable seismic data for methane hydrate exploration Lu, S. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

, Whitmore, D. p. 2
Where the paper says it

Although the constructive reuse of free surface ghost wavefields has not previously been investigated, wavefield components previously considered to be unwanted interference have been shown to be constructively reused to improve the accuracy of subsurface imaging. These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al.

p. 2
Where the paper says it

Although the constructive reuse of free surface ghost wavefields has not previously been investigated, wavefield components previously considered to be unwanted interference have been shown to be constructively reused to improve the accuracy of subsurface imaging. These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al.

N., Valenciano, A. p. 2
Where the paper says it

These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al. , 2010; Valenciano et al.

p. 2
Where the paper says it

These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al. , 2010; Valenciano et al.

A., Chemingui, N., 2015). Separated wavefield imaging using primary and multiple energy Moldoveanu, N. p. 2
Where the paper says it

Although the constructive reuse of free surface ghost wavefields has not previously been investigated, wavefield components previously considered to be unwanted interference have been shown to be constructively reused to improve the accuracy of subsurface imaging. These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al.

, Combee, L., Egan, M., Hampson, G., Sydora, L., Abriel, W. (2007). Over/under towed-streamer acquisition: A method to extend seismic bandwidth to both higher and lower frequencies Ozasa, H. p. 1
Where the paper says it

Consequently, considerable effort has been devoted to extending the usable bandwidth through both acquisition and processing technologies. Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al.

, Mikada, H., Takekawa, J., 2025). Development and experimental results of a marine seismic vibrator Posthumus, B. p. 2
Where the paper says it

The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey. The results demonstrate the practical feasibility of the proposed framework and its potential as a processing-based extension to conventional marine seismic acquisition for broadband imaging and inversion.

J. (1993). Deghosting using a twin streamer configuration Robinson, E. p. 1
Where the paper says it

Consequently, considerable effort has been devoted to extending the usable bandwidth through both acquisition and processing technologies. Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al.

A., Treitel, S. p. 1
Where the paper says it

, 2016; Ronen and Chelminski, 2017; Tellier et al. Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al.

, 1980). Geophysical Signal Analysis Ronen, S. p. 1
Where the paper says it

More recently, these developments have been further advanced by the introduction of ultra-low-frequency marine seismic sources (Dellinger et al. , 2016; Ronen and Chelminski, 2017; Tellier et al.

, Chelminski, S. p. 1
Where the paper says it

More recently, these developments have been further advanced by the introduction of ultra-low-frequency marine seismic sources (Dellinger et al. , 2016; Ronen and Chelminski, 2017; Tellier et al.

, 2017). Tuned Pulse Source—A new low frequency seismic source Soubaras, R. p. 1
Where the paper says it

, 2007), and variable-depth streamers (Soubaras and Dowle, 2010; Soubaras and Lafet, 2013). More recently, these developments have been further advanced by the introduction of ultra-low-frequency marine seismic sources (Dellinger et al.

, 1996). Ocean bottom hydrophone and geophone processing Soubaras, R. p. 1
Where the paper says it

Consequently, considerable effort has been devoted to extending the usable bandwidth through both acquisition and processing technologies. Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al.

, Dowle, R. p. 1
Where the paper says it

, 2006), over/under streamer acquisition (Moldoveanu et al. , 2007), and variable-depth streamers (Soubaras and Dowle, 2010; Soubaras and Lafet, 2013).

, 2010). Variable-depth streamer—a broadband marine solution Soubaras, R. p. 1
Where the paper says it

, 2006), over/under streamer acquisition (Moldoveanu et al. p. 1

Where the paper says it

Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al. , 2006), over/under streamer acquisition (Moldoveanu et al.

, 2007), and variable-depth streamers (Soubaras and Dowle, 2010; Soubaras and Lafet, 2013).

, Lafet, Y. p. 1
Where the paper says it

, 2006), over/under streamer acquisition (Moldoveanu et al. , 2007), and variable-depth streamers (Soubaras and Dowle, 2010; Soubaras and Lafet, 2013).

, 2013). Variable-depth streamer acquisition: Broadband data for imaging and inversion Tellier, N. p. 2
Where the paper says it

The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey. The results demonstrate the practical feasibility of the proposed framework and its potential as a processing-based extension to conventional marine seismic acquisition for broadband imaging and inversion.

, Large, J., Ronen, S. p. 1
Where the paper says it

More recently, these developments have been further advanced by the introduction of ultra-low-frequency marine seismic sources (Dellinger et al. , 2016; Ronen and Chelminski, 2017; Tellier et al.

, Aznar, J., 2021). A new wave in marine seismic source technology Valenciano, A. p. 2
Where the paper says it

Starting from the finite inverse expansion of the ghost operator, the proposed framework is developed into a multidimensional wavefield propagation framework incorporating source- and receiver-side ghost decomposition, survey sinking, wavefield superposition, and the constructive reuse of decomposed ghost wavefields. The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey.

A., Crawley, S., Klochikhina, E., Chemingui, N., Lu, S., Whitmore, D. (2014). Separated Wavefield Imaging (SWIM) of Complex Structures Verschuur, D. p. 2
Where the paper says it

Although the constructive reuse of free surface ghost wavefields has not previously been investigated, wavefield components previously considered to be unwanted interference have been shown to be constructively reused to improve the accuracy of subsurface imaging. These approaches, known as Separated Wavefield Imaging (SWIM), have shown that selected wavefield components can be constructively utilised (Whitmore et al.

J., Berkhout, A. J., Wapenaar, C. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

P. A., 1992). Adaptive surface-related multiple elimination Wapenaar, K. p. 1
Where the paper says it

, 2016; Ronen and Chelminski, 2017; Tellier et al. Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al.

, Thorbecke, J., van der Neut, J. p. 1
Where the paper says it

Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al. , 1992; van Groenestijn and Verschuur, 2009).

, Broggini, F., Slob, E., Snieder, R., 2014). Marchenko imaging Whitmore, N. p. 1
Where the paper says it

Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data. Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al.

D., Valenciano, A. A., Söllner, W., Lu, S., 2010). Imaging of primaries and multiples using a dual-sensor towed streamer Yilmaz, Ö. p. 2
Where the paper says it

The proposed methodology is validated using both synthetic and field-data examples acquired with a conventional shallow-towed marine seismic survey. The results demonstrate the practical feasibility of the proposed framework and its potential as a processing-based extension to conventional marine seismic acquisition for broadband imaging and inversion.

, 2001). Seismic Data Analysis: Processing, Inversion, and Interpretation of Seismic Data

--- Page 1 ---

Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework Constructive Reuse of Free-Surface Ghost Wavefields in Marine Seismic Data: A Propagation-Based Framework Hitoshi Mikada12 1 Arc Geo Support Co., Ltd., Osaka, Japan 2 Professor Emeritus, Kyoto University, Kyoto, Japan Abstract Recent advances in marine seismic acquisition and processing have substantially improved the decomposition of free surface ghost wavefields Building upon these developments, this study proposes a propagation-based framework in which decomposed ghost wavefields are treated as physically meaningful components that can be constructively reused as additional wavefield components Starting from a finite inverse expansion of the ghost operator, a recursive formulation is derived and implemented through adaptive predictive decomposition The formulation is subsequently extended to multidimensional wavefields through wavefield extrapolation and reciprocity, enabling decomposition into components associated with real and mirrored sources and receivers The decomposed wavefields are then restored to a common acquisition geometry by survey sinking and combined through wavefield superposition, forming a preprocessing module that can be incorporated into conventional marine seismic processing workflows Synthetic examples demonstrate increased signal strength through ghost reuse, spectral amplitude recovery of ghost-related notches including frequencies near DC, and clear decomposition of ghost-related wavefield components Field-data examples acquired using a conventional shallow-towed airgun survey further demonstrate improvements in reflector continuity and coherent reflection energy, together with partial restoration of spectral amplitudes degraded by the free surface ghost, supporting the practical applicability of the proposed framework Keywords free surface ghost, ghost reuse, wavefield decomposition, predictive decomposition, survey sinking

--- Page 4 ---

Constructive Reuse of Free-Surface Ghost 1 Introduction Recent advances in wave-equation-based seismic imaging and inversion have substantially increased the demand for broadband seismic data Modern waveequation-based methods, including full-waveform inversion (FWI; Virieux and Operto, 2009), least-squares reverse-time migration (LSRTM; Dai and Schuster, 2013), and, more recently, Marchenko-based wavefield retrieval (Wapenaar et al., 2014) require accurate low- and high-frequency wavefield information for reliable subsurface imaging and inversion Free surface ghosts have long been recognised as one of the principal factors limiting the usable bandwidth of marine seismic data because they introduce characteristic spectral notches and attenuate low-frequency energy (Yilmaz, 2001; Carlson et al. p. 1

Where the paper says it

Among various exploration environments, marine seismic surveys present a particularly challenging case because the usable bandwidth has long been limited by free surface ghosts. Free-surface ghosts have long been recognised as one of the principal factors limiting the usable bandwidth of marine seismic data because they introduce characteristic spectral notches and attenuate low-frequency energy (Yilmaz, 2001; Carlson et al.

, 2007) Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al. p. 1
Where the paper says it

Consequently, considerable effort has been devoted to extending the usable bandwidth through both acquisition and processing technologies. Acquisition-based approaches have evolved through several complementary technologies for wavefield decomposition and broadband acquisition, including twin-streamer deghosting (Posthumus, 1993), PZ wavefield separation (Soubaras, 1996; Itenburg et al.

, 2006), over/under streamer acquisition (Moldoveanu et al., 2007), and variable-depth streamers (Soubaras and Dowle, 2010; Soubaras and Lafet, 2013) Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al. p. 1
Where the paper says it

, 2016; Ronen and Chelminski, 2017; Tellier et al. Processing-based approaches include deterministic inverse filtering and predictive deconvolution (Claerbout, 1985; Robinson and Treitel, 1980), as well as wavefieldbased methods such as surface-related multiple elimination (Verschuur et al.

, 1992; van Groenestijn and Verschuur, 2009)

--- Page 5 ---

Constructive Reuse of Free-Surface Ghost Although the constructive reuse of free surface ghost wavefields has not previously been investigated, wavefield components previously considered to be unwanted interference have been shown to be constructively reused to improve the accuracy of subsurface imaging Building upon these advances, the present study proposes a propagation-based framework in which free surface ghosts are reinterpreted as deterministic propagation phenomena rather than simply as unwanted wavefield components The formulation is subsequently extended to multidimensional wavefields through reciprocity, enabling source- and receiver-side ghost decomposition, survey sinking (Claerbout, 1985), wavefield superposition, and constructive reuse of ghost wavefields within a unified preprocessing workflow

--- Page 6 ---

Constructive Reuse of Free-Surface Ghost From a one-dimensional source-ghost model considered in this study. Figure 1 illustrates the one-dimensional source-ghost model considered in this study The observed signature (x(z)) is represented as the sum of the direct signature (w(z)) and the ghost signature (g(z)) x(z) = w(z) + g(z) (1) where z denotes the z-transform variable used in the one-dimensional formulation; the symbol z is adopted to distinguish it from the vertical spatial coordinate z introduced later The ghost component is interpreted as a delayed replica of the direct signature generated by reflection at the free surface Introducing the ghost operator (E(z)), the ghost signature can be expressed as g(z) = -E(z)w(z) (2)

--- Page 7 ---

Constructive Reuse of Free-Surface Ghost Accordingly, the observed signature is expressed as x(z) = 1I − E(z) 3w(z) (3) where z denotes the z-transform variable used in the one-dimensional formulation; the symbol z is adopted to distinguish it from the vertical spatial coordinate z introduced later This operator representation forms the basis for the finite inverse expansion developed in the following sections

Finite Inverse Expansion

In this study, the ghost operator is instead formulated in terms of a propagation operator acting on finite-length seismic records, allowing its inverse to be expressed as a finite expansion rather than an infinite recursive filter For the one-dimensional model, the ghost operator may be expressed as the product of a free surface reflection coefficient and a propagation operator P(z), E(z) = rP(z) (4) where r denotes the magnitude of the free surface reflection coefficient The propagation operator reduces to a simple delay expressed as P(z) = z n! where n denotes the propagation delay in samples, corresponding to the ghost delay τ = 2d/V Because the free surface acts as a pressure-release boundary, the ghost signature has opposite polarity and may therefore be written as g(z) = -rP(z)w(z) The observed signature is therefore represented by x(z) = w(z) + g(z) = 1I − rP(z) 3w(z) (3)

Recursive Adaptive Decomposition

The finite inverse expansion may alternatively be interpreted as a recursive process rather than as a finite polynomial Introducing an auxiliary wavefield u(z) = 1I + rP(z) 3q.(z), the recursive update is written as q. p. 9

Where the paper says it

(𝜁) = 11 + 𝑟𝑃(𝜁)3𝑞. (𝜁), the recursive update is written as 𝑞.

-)(z) = u.(z) − q.(z) (8) This immediately yields the recursive representation of the finite inverse expansion given by Equation (7), which is precisely the recursive representation of the finite inverse expansion given by Equation (6)

Where Gains Come From

The adaptive propagation operator may be estimated from a finite-length prediction error filter estimated from the current residual wavefield The recursive implementation provides a practical, data-adaptive realisation of the finite inverse expansion derived in Section 2.3 The final direct-signature estimate is obtained by summing all partial contributions, wˆ (z) = ∑ q. p. 9

Where the paper says it

Thus, the ghost-free wavefield is obtained as the accumulated sum of the recursively generated partial wavefields. Although the synthetic example shown in Figure 2 uses a source signature for clarity, the recursive formulation is equally applicable to arbitrary seismic wavefields, including reflected events.

(z)./0.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Adaptive Prediction Operator Estimation: The AI system can estimate an adaptively estimated prediction operator u(z) from a finite-length prediction error filter estimated from the current residual wavefield to account for observational noise, wavelet distortion, and local variations in effective ghost delay.

  2. Recursive Predictive Decomposition: The system can implement a recursive predictive decomposition that repeatedly applies the prediction-and-subtraction process to the residual wavefield, extending beyond conventional single-step predictive deconvolution by incorporating higher-order propagation terms.

  3. Constructive Reuse of Ghost Wavefields: The AI system can perform constructive reuse by superposing the estimated primary wavefield with a polarity reversed and time shifted ghost estimate, resulting in an improved direct signature that exhibits an approximately 6 dB increase in coherent reflection amplitude relative to the true primary.

  4. Multidimensional Wavefield Decomposition and Restoration: The system can decompose recorded data into four physically interpretable wavefield classes (S-R, S-MR, MS-R, and MS-MR) based on source and receiver mirroring, followed by restoring these components to a common acquisition geometry via survey sinking for constructive combination.

  5. Unified Propagation Framework: The AI system can interpret free surface ghosts as deterministic propagation phenomena rather than simply as unwanted wavefield components, providing a unified physical interpretation that bridges ghost decomposition, mirror-image interpretation, survey sinking, and wavefield superposition.

  6. Legacy Data Enhancement: The processing framework can be applied to extensive legacy marine seismic datasets acquired before broadband acquisition technologies were widespread to recover additional bandwidth and enhance low-frequency content without requiring re-acquisition.

Abstract

Recent advances in marine seismic acquisition and processing have substantially improved the decomposition of free surface ghost wavefields. Building upon these developments, this study proposes a propagation-based framework in which decomposed ghost wavefields are treated as physically meaningful components that can be constructively reused as additional wavefield components. Starting from a finite inverse expansion of the ghost operator, a recursive formulation is derived and implemented through adaptive predictive decomposition. The formulation is subsequently extended to multidimensional wavefields through wavefield extrapolation and reciprocity, enabling decomposition into components associated with real and mirrored sources and receivers. The decomposed wavefields are then restored to a common acquisition geometry by survey sinking and combined through wavefield superposition, forming a preprocessing module that can be incorporated into conventional marine seismic processing workflows. Synthetic examples demonstrate increased signal strength through ghost reuse, spectral amplitude recovery of ghost-related notches including frequencies near DC, and clear decomposition of ghost-related wavefield components. Field-data examples acquired using a conventional shallow-towed airgun survey further demonstrate improvements in reflector continuity and coherent reflection energy, together with partial restoration of spectral amplitudes degraded by the free surface ghost, supporting the practical applicability of the proposed framework. The proposed framework therefore provides a processing-based complement to conventional ghost decomposition technologies and suggests a unified approach in which acquisition geometry and ghost processing are jointly optimised to maximise the usable wavefield information.

Related papers