Deghosting, Debubble & Designature — g-Platform
Deghosting, debubble & designature

Deghosting, debubble & designature software

Every marine record carries a ghost from the sea surface, a bubble oscillation from the airgun, and a source signature that drifts from shot to shot. g-Platform models each effect from the acquisition geometry — tow depths, water velocity, near- and far-field signatures — and corrects it before the data reaches deconvolution, so the wavelet feeding imaging is broadband and consistent trace to trace.

Download free trial How the ghost is modeled Watch Demos
Source & receiver
Ghost modeled from tow depth
Bubble & signature
Deterministic designature, shot by shot
CPU or GPU
Distributed across the cluster
Stack section before deghosting in g-Platform, with the ghost notch flattening the spectrum and blurring the reflectors
The same stack section after deghosting in g-Platform, with a broadband spectrum and sharper, better-resolved reflectors
Amplitude spectrum QC panels in g-Platform, comparing the frequency content across the deghosting sequence
In view
Before deghosting
Three effects, one wavefield, and they have to come off in the right order. The ghost is a sea-surface reflection a few metres behind the direct source or receiver arrival, notching the spectrum at frequencies set by the tow depth. The bubble is the airgun's own aftershock, a decaying oscillation from the collapsing air bubble that rings behind every primary pulse. The signature is what is left once both are accounted for — the effective wavelet radiated by the gun array, which varies shot to shot with pressure, depth and array geometry. g-Platform's modules model the ghost in the F-K or Tau-Pi domain and subtract it, attenuate the bubble from the known airgun response, and designature to a common zero-phase wavelet.
Deghosting

Model the ghost from tow depth, then take it away.

The sea surface is a near-perfect reflector, so every source and receiver carries a polarity-reversed image of itself a few milliseconds behind the primary arrival. That delay, set by the tow depth and the emergence angle, is what the ghost model is built from.

Source and receiver ghosts, both
The gun array and the streamer are each towed below the surface, so both sides of the trip carry their own ghost. Source and receiver ghosts are modeled and removed separately, since their tow depths and geometry differ.
Built in the F-K or Tau-Pi domain
The ghost model is constructed in F-K or Tau-Pi from the airgun and streamer tow depths and the water velocity, so the notch position and depth are tied to acquisition geometry rather than guessed from the spectrum alone.
L1-norm solve, adaptive subtraction
The model is fit with an L1-norm iteration, tolerant of the outliers a raw least-squares fit would over-correct for, and removed by adaptive subtraction so amplitude and phase match the local data before anything is taken out.
Variable-depth streamers included
Where the streamer is towed on a slant or at variable depth rather than flat, the depth profile per receiver feeds the same model instead of a single constant depth.
Low-frequency-rich output
Removing the notch restores energy the ghost had been suppressing, so the deghosted output is skewed low-frequency relative to the input. A Q filter or spectral balancing pass typically follows.
Debubble & designature

One source, one signature, shot after shot.

An airgun does not radiate a clean spike: the primary pulse is followed by a decaying bubble oscillation, and the array's directivity shapes the wavelet further. Debubble strips the oscillation; designature converts what remains to a single reference wavelet the whole survey shares.

BeforeShot gather before designature in g-Platform, with the airgun's bubble and array directivity still in the wavelet
AfterThe same shot gather after designature in g-Platform, converted to a single zero-phase wavelet
Bubble removal
Debubble

Attenuates the airgun's secondary and tertiary bubble oscillations from the known response of the array — gun volumes, pressure and firing depth — leaving the primary pulse without the ringing tail that would otherwise smear into later reflections.

Deterministic filter
Designature

Designs a frequency-domain operator, shot by shot, that maps the measured or modeled source signature onto a common target wavelet — typically zero phase — so array directivity and shot-to-shot variation in the gun array do not carry through into the stack.

Near-field
Near-field signature estimation

Builds the effective far-field signature from a near-field hydrophone array mounted close to the guns, accounting for bubble interaction between individual elements of the array.

Far-field
Far-field signature designature

Uses a recorded far-field signature directly where one was captured during acquisition, skipping the near-field reconstruction step and designaturing straight from the measured wavelet.

Zero phase
Zero-phase conversion

Converts the designatured wavelet to zero phase so events align with their true reflection time, which also makes later deconvolution and horizon picking more consistent across the survey.

Array directivity
Source array correction

Accounts for how the gun array's geometry shapes the radiated wavelet with offset and azimuth, so the designature operator stays valid away from the array's vertical axis.

Workflow

Five steps from a raw marine record to a broadband, zero-phase gather.

1
Read the acquisition geometry
Source and streamer tow depths, water velocity, gun array geometry and any near- or far-field signature recordings feed straight from the project's geometry and headers.
2
Model source and receiver ghosts
Build the ghost model in F-K or Tau-Pi from tow depth and water velocity, separately for the source side and the receiver side.
3
Subtract and rebalance
Remove the ghost model by adaptive subtraction, then restore the spectrum with a Q filter or spectral balancing pass on the low-frequency-rich result.
4
Debubble and designature
Attenuate the bubble oscillation, then design and apply the designature operator, shot by shot, from the near- or far-field signature.
5
QC the spectrum, not just the section
Compare amplitude spectra before and after: a flat, broadband spectrum with no residual notch is the check, alongside the difference gather.
Correction & scale

Modeling the ghost is only half the job. Removing it cleanly is the other half.

A ghost or bubble model rarely cancels by plain arithmetic. Adaptive subtraction adjusts the model to the local character of the data before it is taken out, the same way it does for demultiple.

Ghost & bubble correction
How the model is built and applied
Ghost model
Built in F-K or Tau-Pi from source and streamer tow depths and water velocity, solved with an L1-norm iteration, source and receiver side handled independently.
Streamer depth
Flat or variable-depth / slant streamers, with a per-receiver depth profile where the tow is not constant.
Debubble input
Gun volumes, pressure and firing depth per element of the array, or a recorded near-field signature where available.
Designature input
Recorded far-field signature, a near-field reconstruction, or a modeled signature when neither was captured; applied shot by shot.
Target wavelet
Zero phase by default, so downstream deconvolution and picking work from an aligned, consistent wavelet.
Post-correction balance
Q compensation or spectral balancing to counter the low-frequency skew that deghosting leaves behind.
Running it on a survey
Compute, limits and the QC that comes out
CPU or GPU
Ghost, debubble and designature all run on either, selected per module, moving between a workstation and an accelerated node without a workflow change.
Distributed
Calculation spread across the processing server with a per-machine bulk size in megabytes, thread limits and an affinity tag to find the job in the server QC interface.
Calculation area
Restrict a run by sequence number or by inline and crossline range, so a parameter test costs a few lines rather than the whole survey.
Outputs
The ghost or bubble model, the corrected gathers, and the designatured trace, plus the untouched input saved alongside for comparison.
QC displays
Amplitude spectra before and after, difference gathers, and the applied designature operator against the measured or modeled signature.
Bad-sample policy
Corrupted samples are fixed, reported and stopped on, or passed through — your choice per module, so a production job does not fail on one gather.
Modules in this group
The deghost, debubble and designature modules named on this page — not the full list
Deghost Debubble Designature Near-field signature estimation Far-field signature designature Zero-phase conversion Adaptive subtraction
More g-Platform capabilities

Deghost is one link in the processing chain.

The same project, the same batch processing and the same cluster carry a survey from field tape through to depth. These are the other topics in the g-Platform workflow.

g-Platform overview Signal processing & denoise Multiple attenuation Deconvolution Velocity analysis Stack imaging Time imaging Depth imaging FWI & RTM VSP processing Refraction statics Residual statics First-break picking Geometry assignment & QC Data & geometry formats
FAQ

Questions, answered.

What is the difference between deghosting, debubble and designature?
They correct three different parts of the same marine source-and-receiver wavefield. Deghosting removes the sea-surface reflection that follows every source and receiver a few metres behind the direct arrival, notching the spectrum. Debubble removes the secondary and tertiary oscillations left by the airgun's collapsing air bubble, a ringing that trails the primary pulse. Designature takes what is left — ghost, bubble and array directivity combined — and converts it to a single known wavelet, typically zero phase, so every trace in the survey starts from the same signature before imaging.
How does g-Platform model and remove the ghost?
The source and receiver ghost model is built in the F-K or Tau-Pi domain from the airgun and streamer tow depths and the water velocity, solved with an L1-norm iteration, and adaptively subtracted from the data. The result is low-frequency-rich, so a Q filter or spectral balancing pass usually follows to bring the spectrum back into shape.
What does a designature operator need before it can run?
A signature to design against — a near-field hydrophone measurement from the gun array, a recorded far-field signature, or a modeled one where neither was captured. The operator is a deterministic frequency-domain filter that maps the measured signature onto a target wavelet, applied shot by shot so that variation in the source array from one shot to the next does not leak into the stack.
Why does deghosted data need a Q filter afterward?
Removing the ghost notch boosts the low end of the spectrum that the notch had been suppressing, so the output is low-frequency-rich relative to the input. A Q compensation or spectral-balancing pass afterward restores a flat, broadband spectrum before the data goes on to deconvolution and imaging.
Get started

Try deghost, debubble and designature on your own gathers.

Model the ghost, strip the bubble, and designature to a common wavelet — on marine or even on land data. Talk to Geomage about a demo, or take g-Platform for a trial run.

Download free trial Request pricing