Source and Receiver Deghosting

In conventional marine seismic surveys, the air-guns and streamer(s) are placed at certain predefined depths below the sea surface. The decision of source/receiver depths for any given survey is based on several conditions, such as weather conditions, desired frequencies, equipment type, etc. In addition to those conditions, the location of the notch in the frequency spectrum created by the surface ghost reflections is something that must be considered. Ghosting is a direct result of placing the source and receivers near a sharp discontinuity, which is the water/air contact in marine acquisition.

The wavefield initiated by the air gun explosion initiates pressure waves in all directions. The upward traveling source wavefield reflects from the sea surface and follows the down-going wavefield with a certain delay. When these waves are recorded at the receivers, all the reflections appear as double image. This is known as the source ghost. Additionally, the wavefield that arrives at the receiver locations continues its upward travel and reflects from the sea surface and is recorded again. This is known as the receiver ghost. Deghosting derives and applies a filter to remove both source and receiver ghosts in seismic data. Ghost free gathers are passed on for further processing.

We have developed a Deghosting method in the tau-p domain, in which the time-delay of ghosts is a constant for seismic events with the same dip angle. An optimal notch filter is found by grid searching of time delay of ghosts and water reflection coefficient for each p trace. The slant stacking process naturally handles the space and time variant feature of ghosts. In the meantime, the influence of random noise is reduced due to stacking power of seismic events. A time-space domain greedy Radon transform is utilized to achieve best performance of the algorithm by providing a high-resolution tau-p model, therefore it is robust at handling spatial aliasing.

Shot before Deghosting

Shot after Deghosting (Side lobe reduced & wavelet sharpened)