Previous   |   Next

doi:10.2204/iodp.proc.311.110.2006

Seismic data

Acquisition

Cruise PGC9902 used a 40 inch³ (0.65 L) sleeve gun with a wave shape kit to reduce the bubble ghost. This source was towed at 2 m depth, 20 m behind the ship, and fired at a nominal spacing of 20 m. The COAMS streamer has 102 hydrophones at variable, increasing spacings, with offsets ranging from 120 to 1140 m. This MCS streamer consists of five segments, each followed by a depth sensor. Before and after the hydrophones were 100 m long vibration isolating modules. The COAMS streamer was towed with a 60 m long cable. The SCS Teledyne array consists of 50 hydrophones along a 25 m long streamer that was towed 42 m behind the vessel.

The two-dimensional (2-D) MCS lines of Cruise PGC9902 were shot along four 50 km long margin-perpendicular transects (Fig. F1), two of them continuous (Lines ODP-2 and 3), one split into two parts (Lines ODP-1 and 7), and one split into three segments (Lines ODP-4, 5, and 6). These transects were intersected by five 10 km long margin-parallel lines (CAS-2, 3, 4, 5A, and 5B). COAMS profiles Inline 27 and Inline 38 were part of a pseudo 3-D component of this survey, in which 40 margin-parallel lines were acquired at a separation of 100 m.

Additional Teledyne SCS lines were shot across alternate or additional drill locations that were not covered with the COAMS data. The cold vent surveys of 1999 and 2000 also collected pseudo-3-D data, this time along lines that were only 25 m apart. The seismic source was again the 40 inch³ sleeve gun. Only the 2004 Teledyne survey used a 40 inch³ air gun.

Preprocessing

The seismic data that we present in this paper stem mainly from the COAMS streamer, and so the data processing focuses on these survey data. The first processing step was to establish the geometry of the seismic recording system. The data were then processed in a conventional manner to produce high-resolution structural images.

One key processing step was localization of the hydrophones of the COAMS streamer. Along the streamer, four depth sensors constantly monitor the depth of the streamer, and these revealed that the streamer did not lie flat in the water. A heavy tow-cable in the front and lead ballast on the last 300 m of the streamer created down-pull at either end, whereas the central section floated high because of the buoyancy of the oil-filled cable. In order to determine the position of each hydrophone, an array element localization (AEL) method developed by Dosso and Riedel (2001) was applied. This method uses the arrival times of the direct wave and the seafloor reflection and matches these with ray-tracing through a known or assumed (in our case measured) water velocity profile over the recorded bathymetry. A linear inversion results in the smoothest streamer shape that fits the measured arrivals. The known hydrophone spacing along the cable and the depth sensor data of the streamer provide a priori information that is also included in the inversion. Slight modification of the software was necessary. The search algorithm to identify the correct reflection segment of the seafloor was improved for this study to allow for highly varying water depth. The result of this inversion is a realistic streamer configuration in the form of offset from the shot and depth below sea surface of each hydrophone.

Initially, this AEL of the streamer was done for every shot along one line. As the shape of the streamer appeared not to vary greatly during the survey, one representative solution of the hydrophone locations was used for the entire survey. In the case of any deviation of the true hydrophone location from the calculated location, subsequent automatic residual static calculations during the data processing were assumed to correct for any time shifts caused by location errors.

Main processing

After the geometry was assigned for each trace, the data were processed in the manner outlined in Table T1. First, a bandpass filter with smooth ramping was applied, either with a frequency range of 80–250 Hz for a high-frequency analysis or with a low-frequency range of 30–140 Hz. Spectral analysis of the data showed that the sleeve gun produced signals up to 250 Hz. Lower-frequency data, however, produce clearer images of the bottom-simulating reflector (BSR) as shown by Spence et al. (2000) and Chapman et al. (2002). In order to map both fine structures as well as the BSR, two frequency bands were chosen, and separate images were produced for each frequency band.

Regardless of the frequency range, a predictive deconvolution (Table T1) was applied, which did well in suppressing the strong surface ghost. This was followed by the same bandpass filter as applied before. Thereafter, a spatial linear signal detection (in practice, a velocity filter in the tau-p domain), with subsequent rho filtering to restore the frequency spectrum, was applied.

Velocity analysis was done in a loop with residual static corrections. As mentioned before, our AEL preprocessing solution may not have always taken care of small undulations in the streamer geometry during the entire survey; therefore, automatic alignment of reflectors after normal moveout (NMO) correction produced common midpoint (CMP) gathers that were highly constructive for stacking. Then the NMO correction was removed, and another round of velocity analysis followed. If necessary, a second residual static correction was calculated, which required another velocity analysis.

After stacking the data, despiking with optional automatic gain control (AGC) produced final sections that were then time-migrated either with a Stolt or finite-difference (FD) migration. Stolt migration produced better results with respect to the sediment structure, whereas FD migration was better for amplitude preservation. For clearer display, another AGC was applied if amplitude preservation was not required.

Some of the seismic data shown below come from the Teledyne SCS recording. These data were processed with bandpass filter, deconvolution, and simple migration with an assumed realistic velocity field.

Three-dimensional binning

Parts of the PGC9902 survey area were covered with seismic lines at 100 m spacing, enabling a 3-D data analysis. In addition to the processing above, a 3-D grid was created using a bin size of 9.8 m along the in-line orientation and 100 m for the cross lines. Because of the large cross-line spacing, 3-D migration of the data produced undesirably strong spatial aliasing; therefore, only 2-D time migration was applied to the data prior to binning.

For the 3-D Teledyne survey data, similar binning considerations as for the 3-D COAMS data were necessary, this time with smaller bins (25 m by 10 m).

3.5 kHz data

The 3.5 kHz data required no processing; only the amplitude envelope was used for plotting.

Top of page   |   Previous   |   Next