Multiple Suppression on Marine Data Recorded in The Middle East
Jock Drummond1, Riaz Alá'i1, Stan Morris1
(1) Anadarko Petroleum Corporation, The Woodlands, TX
Multiple removal from marine seismic data recorded in very shallow water areas is known to be problematical. In areas where the sea-bottom is very shallow and when the acoustic impedance response is high, the data is generally contaminated with a lot of reverberations between the sea-surface and the sea-bottom.
For optimal interpretation of geological structures, the data needs to be pre-processed to suppress these “reverberations”. As generally known, with a shallow sea-bottom, industry standard methodologies for multiple removal are challenging. In this paper the well-known method of predictive deconvolution has been applied and illustrated to marine data recorded in the Middle East.
This dataset appeared to be a good example to validate the suppression of the reverberations and obtaining data quality improvement. The successful removal of strong aliased linear noise and surface-related multiples has resulted in a dataset, which is more suitable for further processing with primaries that are more readily identifiable.
The strong aliased linear noise has been suppressed by determining slope filters that have been applied in the F-K domain. For the prediction of the surface-related multiples, autocorrelations were calculated on shot gathers to validate identification of the reverberation periods. Based on the identified periods, the original data was bulk shifted by the reverberation lag time. The bulk shifted shot gathers have been subtracted from the original data (in the shot domain) using an optimized adaptive least-squares algorithm which uses diverse 1-D temporal filters for diverse order of multiples recorded in the data.