Case Study: A Large 3D Wide-Azimuth Ocean-Bottom Survey in the Deepwater Gulf of Mexico

Abstract

Abstract We present the P-wave processing and imaging of a large 3D ocean bottom node survey, consisting of 1628 node locations, in deepwater Gulf of Mexico over the Atlantis field. The data quality is good, and the preliminary imaging results using both the primary reflections and the receiver ghost appear promising. Introduction In the fall of 2005 Fairfield Industries was the primary contractor of a large, wide azimuth survey (Ross and Beaudoin, 2006) for BP over the Atlantis field in the deepwater GOM. The water depth in the area varied from approximately 1400 metres to 2300 metres. The survey used 902 four-component ocean bottom nodes (Mitchell and Grisham, 2006), which BP commissioned Fairfield to build. The purpose of the survey was wide azimuth, sub-salt imaging using P-waves (Beaudoin and Michell, 2006). The shot coverage was also on a hexagonal grid with a 50m spacing over an area of about 730km2. The survey was acquired in two patches, with approximately 1700 node placements in total. The time between shots was 11.2 seconds, with a 12 second trace length, although the nodes actually recorded continuously. The nodes were placed on the seafloor using an ROV. A dual Hydroacoustic Aided Inertial Navigation (HAIN) system (Marc et al, 2006) provided a positioning accuracy of approximately 0.5% of water depth (about 10m in this case). The node spacing was ~400m in both the x and y directions, using a hexagonal grid (Bardan, 1997) and covered an area of about 250 km2. Once placed on the seafloor the nodes were completely autonomous. The battery life and memory capacity were sufficient for about 28 days of continuous recording at a 2ms sample rate. The geophones were fixed (non-gimballed) and orthogonal, and the nodes were placed as much as possible on horizontal ground, such that one component was approximately vertical. Discussion The preliminary processing of the data to date has all been performed on common receiver gathers. Each gather is well sampled, and contains approximately 160,000 traces per component. The processing flow was very basic: node positioning, noise removal, quality control, geophone orientation, PZ summation and difference, and common receiver wave equation depth migration. The overall positioning accuracy of the nodes was very good (Figure 1). Using the direct arrival to determine the node locations confirmed the positions derived from the HAIN system to within 10 metres for 95% of the nodes, and to within 20 metres for over 99%. About a third of the nodes had a bulk time shift due to a clock synchronization error, but this was easily detected and corrected. Figure 1: The difference in the positions of the nodes calculated by a least squares inversion of the direct arrival times and the positions calculated using the HAIN system on the ROV. (available in full paper) Random noise was attenuated by spectral thresh-holding and balancing in the time-frequency space (see program tfdnoise at http://www.freeusp.org). Care was taken to avoid damaging both the direct arrival and the receiver side water bottom multiples.