Multi-azimuth Streamer Acquisition - Initial data analysis

Abstract

Abstract A thin but complex layer of partially eroded anhydrite and other facies lie at a depth of around 3km across large areas of the Nile Delta in the Mediterranean. Wavefield distortion, attenuation and the generation of complex multiple diffraction noise cause the quality of the underlying seismic image to be highly variable. Multi-azimuth (MAZ) seismic can help resolve these issues and improve the deep pre-Messinian image. Here we discuss the processing flow and initial data analysis of a MAZ streamer dataset. The main elements of the flow used are a standard streamer demultiple followed by Kirchhoff PSTM. The initial data analysis shows that MAZ greatly improves general image quality, signal-to-noise ratio and lateral resolution, and suppresses diffracted and other multiples effectively, despite some of the obvious limitations of the processing flow. Issues and challenges around this approach are discussed. Introduction Multi-azimuth or wide-azimuth seismic is not a new technology, and has been with us for many years in the form of land and ocean bottom surveys. The literature is rich with examples of how high-fold multi-azimuth data can produce stunning improvements over their single azimuth 3D equivalents (Rogno et al., 1999, Keggin et al., 2002, Gaus and Hegna, 2003, Arntsen and Thompson, 2003, Riou et al., 2005, Manley et al., 2005, Michell et al., 2006). We know from theory and case histories that multi-azimuth data will lead to improved signal to noise, improved multiple attenuation and improved illumination. However, because of approximations in current processing technology, the processing of multi-azimuth data will leave errors in the final imaged results, both kinematic and dynamic. Simple stacking of the data, though surprisingly robust in most situations, makes assumptions about data consistency between surveys and will likely not result in the most optimal image. This paper shows how multi-azimuth (MAZ) towed streamer data is processed in the Nile Delta, looks at some of the issues highlighted above and discusses our initial attempts to improve the image of the combined datasets. The multi-azimuth (MAZ) survey A 630 sq km five azimuth towed streamer survey was acquired in late 2004. The legacy 2000 dataset was used as the 6th azimuth. The objective of the survey was to provide the 'best possible' seismic data quality over a pre-Messinian discovery for appraisal and development purposes. Processing flow and its limitations In terms of large velocity contrasts, part of the Nile Delta can be characterized by a dipping, complex sea floor and the thin, but complex Messinian layer that lies at a depth of around 3km. This layer can be characterized by anhydrite deposits, channel systems and a rugose base, where the total layer thickness is commonly less than 200m. Gas hydrates near the waterbottom and the Messinian layer are major multiple generators. In these relatively deep waters surface related multiple elimination (SRME, Verschuur et al., 1992) has been the preferred demultiple tool, typically followed by an additional Radon demultiple. With the rugosity of the seafloor, the near seafloor gas hydrate scatterers, and the Messinian, we know that our multiples have a strong 3-D character. Therefore the used SRME+Radon flow only provides a partial solution. In a 3-D earth with large lateral velocity changes, different acquisition directions will sample different parts of the subsurface, hence could potentially experience dramatically different velocities. The way to image these data properly is using prestack depth migration, where the true raypaths of the seismic energy are followed. Gaus and Hegna, 2003 and Riou et al., 2005 describe the PSDM method and its application to multi-azimuth data. The complexity of the Messinian layer, however, is such that a robust velocity model cannot be derived, neither by current tomographic solutions, nor by a traditional Gulf of Mexico interpretation approach. This is because the complexity of the Messinian layer is not 'simply' structural, but the internal velocity distribution in the relatively thin layer is also highly variable.