## 3D Demultiple

For pre-stack modeling of multiples, PGS proposes 3D solutions which are fully data-driven and utilize powerful parallel computing resources.

Demultiple is a critical step in the pre-conditioning of data for advanced imaging and velocity model building. In many processing and imaging methods, these multiple reflections have to be modeled and removed from the recorded data in order to obtain a clearer image of the sub-surface.

- At the imaging stage, multiples interfere with primary reflections and mask geological features of the target.
- At the velocity estimation stage, multiples introduce errors in the velocity estimation, often lowering velocity estimates. This can lead to reduced imaging capability.

The geological complexity of the sub-surface requires pre-stack 3D solutions to multiple modeling. Moreover, modern acquisition geometries yield richer azimuthal information eliminating the need for assumptions in conventional algorithms such as 2D approximations or a 1D earth model.

Multiple reflections can be classified in two main categories:

- Surface related multiples, whose amplitudes are generally much stronger than interfering primaries. At least one of the reflected ray paths is down-going from the sea surface
- Internally scattered multiples, where all the scattering reflections occur in the subsurface

Surface Related Multiples

Surface Related Multiples

PGS offers comprehensive 3D pre-stack demultiple solutions for any range of water depths and acquisition geometries. The workflows outlined in the table below are based on two key methodologies which prove to be complementary:

- 3D wavefield extrapolation based Surface Related Multiple Elimination (SRME) is performed by wavefield extrapolation in common 3D shot gathers for towed-streamer acquisition and in common 3D receiver gathers for ocean bottom acquisition (OBS, nodes, PRM).
- 3D convolutional SRME uses cross-convolution of the recorded data either with itself or an estimate of the primary wavefield.

The table below summarizes the standard approaches and combinations for attenuating surface related multiples in various water depths:

Water Depths |
> 700 m | > 700 m | ~150 m and < 700 m |
< 150 m |
< 150 m |

Technology |
Convolutional | Wavefield Extrapolation + Convolutional |
Wavefield Extrapolation + Iterative Convolutional |
Wavefield Extrapolation + Muted Convolutional |
Wavefield Extrapolation + Seabed Model Convolutional + Muted Convolutional |

Reflectivity Model |
N/A | The reflectivity model grid 12.5 m x 12.5 m can be: - Migrated near angle stack - Synthetic reflectivity model of the seabed - SWIM true reflectivity cube |

'Convolutional' refers to 3D convolutional SRME and 'Wavefield Extrapolation' refers to 3D Wavefield Extrapolation Based Surface Related Multiple Elimination (SRME)

**Internally Scattered Multiples**

The remaining multiples are addressed once the surface related multiples have been successfully attenuated in the recorded data.

We can distinguish two types of internally scattered multiples.

- Internal reflection multiples, where scattering events are reflections in the subsurface
- Diffracted multiples, where at least one scattering events is a diffraction in the subsurface

Note that when these events generate surface related multiples, they will generally be attenuated by the 3D SRME techniques.

Internal multiple reflections are modeled pre- or post-stack using a single or multi-dimensional internal multiple elimination technique. The method can be used several times to target various depth ranges of possible internal multiple generators.

Finally, the remaining diffracted multiples which are not attenuated by the SRME or Internal Multiple Elimination (IME) techniques are addressed by using specific detection and filtering techniques. The method relies on their clear differentiation in bandwidth and amplitude in comparison to primary reflections arriving at the same time.