Spatial sampling effects of a multi-source multi-cable recording configuration

Abstract

Acquiring large volumes of marine 3-D seismic data using conventional single-source single-streamer recording vessels is no longer considered cost effective. By comparison a dual-source dual-cable recording configuration quadruples the number of subsurface kilometres recorded for each sail line. Taken to the extreme, a 'quadquad' system can record twelve common-midpoint (CMP) lines per sail line. In this manner large data volumes can be acquired economically over a shorter time interval. This efficient method of data capture does have the unfortunate characteristic of low-fold coverage and non-uniform offset distribution for each subsurface line. For a dual-source dual-receiver configuration the offset distribution, within the CMP gather of any single subsurface line, repeats every four CMPs. For a 'quad-quad' system the CMP offset distribution repeats every eighth CMP Data acquired in the above manner deviates from the stack-array criterion proposed by Anstey, and as a consequence a residual of the coherent noise present in the raw data will pass through to the stack. It is demonstrated that the 'herring-bone' pattern observed on some stacked data is directly related to the acquisition configuration. The North West Shelf of Australia is characterized by a hard limestone water bottom. This reflector generates a series of strong periodic multiple events which dominate the amplitude spectrum of the data. The paper will illustrate that the wavelength of the residual multiple is related to both the recording configuration and the water depth. Synthetic models are used to simulate data collected using a multi-source multi-receiver recording configuration. f-k domain displays of synthetic stacked data are used to illustrate why the residual multiple energy is concentrated in particular areas of f-k space. Having established the origin of the residual noise trains on the stacked data, we suggest methods to remove them from the data. Traditional multiple-attenuation techniques, post-stack f-k notch filters and shot-record interpolation are suggested as possible processing techniques which will attenuate the 'herring-bone' signature from the stacked section.