3-D Surface Multiple Attenuation

Abstract

Surface multiples were removed from an irregularly sampled Gulf of Mexico 3-D seismic data set using a 2-D wave equation-based algorithm. Accomplishing that task required facing several challenges: the project's potentially high processing cost, the 2-D algorithm's inability to accommodate 3-D effects directly, and the algorithm's need to have an input wavefield with regular spatial sampling. The first two challenges were addressed by predicting surface multiples in a 2-D fashion for half of the 2-D lines in the 3-D data set. The predicted multiples were then subtracted from the original lines using a process called adaptive noise cancellation to correct the predictions for ignored 3-D effects such as those due to cable feathering and cross dip. In addition, the adaptive process was able to subtract multiples predicted for one subsurface line not only from that line, but also from an adjacent line, thereby almost halving the cost of the project. The spatial sampling problem, which arose mostly because the survey was designed without surface multiple prediction in mind, was solved by regularizing the 2-D lines with an assortment of interpolation processes before multiple prediction. The interpolation increased the processing cost by about a factor of four.

The impact of field‐survey characteristics on surface‐relatedmultiple attenuation

“…Thus, there can be no missing shots or receivers in the field data set, and the shot and receiver spatial sampling intervals must be equal. Because these requirements are not often met in recorded data, significant preprocessing must be applied to regularize the data in preparation for multiple prediction (Dragoset 2000). These problems are well recognized within the seismic data processing community and various processing schemes are in place to deal with them.…”

Practical, 3D surface‐related multiple prediction (SMP)

“…It should be noted that although the prediction process is 2D, the process may be applied to 3D data, e.g. on a subsurface line basis (Dragoset, 2000). This is not the same as a 3D prediction.…”