Processing-based Deghosting

In the case of a single-sensor streamer (flat or slanted) and a conventional source, PGS has developed processing-based deghosting methods that require information about the sea-surface and actual receiver and source depths.

In 2007, PGS launched GeoStreamer® which provides a field deliverable without any receiver ghost thanks to its dual-sensor streamer technology. More recently, PGS addressed the source ghost problem by introducing GeoSource® which is a multi-level source system. These technologies offer a non-compromise acquisition solution to the ghost issue.

PGS' best-in-class processing flows can be augmented with processing-based receiver and source deghosting. For GeoStreamer dual-sensor data, only the source deghosting is applied.

Single-Sensor Receiver Deghosting

For single-sensor recordings, no complementary measurements are available to recover signal in the notches, where up- and down-going waves interfere destructively. Using variable depth recordings (where hydrophones are positioned at offset-dependent depths) may reduce the interference effects, as the notches in the recorded data will be more diverse and less deep in character.

PGS uses a deghosting solution based on the wave-equation, where a forward operator that describes how all seismic events that are reflected by the sea-surface will create a ghost, is inverted for. For constant depth recordings, a spectral division operator is used. For variable depth recordings, a mixed domain integral equation is solved for.

Regardless of the geometry of the streamer shape, it is assumed that the sea-surface is perfectly flat, and that the sea-surface acts as a perfect mirror.

Although the application of these operators is mathematically correct, there are two key shortcomings that can be identified:

  • The deghosting operator aims to amplify the signal when the destructive interference between the up- and down-going waves occurs. This occurs at so-called notches. When the up-and down-going waves cancel each other, there is no signal to be amplified. If no signal is recorded, it cannot be reconstructed.
  • When the up-and down-going waves cancel out completely, the noise that is recorded (in notches) will be amplified nonetheless. This means that after the application of the equation, the noise will be enhanced in the data and can easily be mistaken for weak, underlying signal.

To ensure the enhancement is not too severe, a modified operator is used to stabilize the results. The solution uses two mechanisms for operator stabilization. Firstly, the computations are carried out in the Laplace frequency domain, where frequencies have both real and imaginary parts. Secondly, additional stabilization is used around the notches to avoid the noise being enhanced too severely, by increasing the complex frequency component locally, thereby dampening the operator further.

Finally, the assumptions made about the sea-surface indicate that for high fidelity processing (e.g. 4D monitoring), the deghosted results will be compromised in an unknown way. As such, single-sensor recording solutions may be suitable for large-scale exploration purposes, but not for applications requiring uncompromised results.

The processing-based receiver deghosting is known in PGS software environment as RBO (Receiver-base Bandwidth Optimization).

Source Deghosting

The use of acquisition-based deghosting solutions like GeoSource is likely to increase over time, however, the vast majority of data sets will still be acquired using conventional sources.

Sources are typically fired at a shallower depth than the streamer. Their actual depth is also more stable as the source array generally follows the sea-surface. This makes the source deghosting process more effective.

The source ghost typically occurs at much higher frequencies than the receiver ghost, due to the shallow tow depth of the sub-arrays. To deliver broadband seismic data up to very high frequencies, the source ghost must be removed.

Similar to receiver deghosting, the wave-equation processing deghosting technique can be applied to suppress the ghost and enhance the signal bandwidth for conventional sources. It also takes into consideration variations in source depth and emergence angle.

As sources and receivers are at different depths, this effectively means that we assume a local 1D earth. Of course, this is an assumption which is not valid by default; however the errors made are believed to be small enough to be acceptable.

The processing-based source deghosting is known in PGS software environment as SBO (Source-base Bandwidth Optimization).

The figures below show the application of receiver and source deghosting. The receiver depth was 15 m, leading to notches at around 0, 50, 100, 150 and 200 Hz. The source depth was 7 m, leading to notches at around 0 and 107 Hz.

The aim of deghosting is to achieve a resulting amplitude spectrum as linear as possible, comparable with the earth's response. Note the successful retrieval of a linear trend in the spectrum obtained after source and receiver deghosting.

 Figure 1 - raw shot gatherRaw shot gather in t-x and f-k domain acquired with a streamer depth of 15 m and a source depth of 7 m

 Figure 2 - RBO Shot GatherRBO shot gather in t-x and f-k domain. Receiver ghost has been suppressed.

 Figure 3 - RBO + SBO Shot GatherRBO + SBO shot gather in t-x and f-k domain. Receiver and source ghosts have been suppressed.

 De-ghostingAmplitude spectra extracted from the data. Note the successful retrieval of a linear trend in the spectrum obtained after source and receiver deghosting (orange line).