Site selection for Leg 178 was based on multichannel and single-channel seismic reflection profiles, collected in recent years either as dedicated site surveys or as part of large-scale regional reconnaissance surveys (data compilation took place within the ANTOSTRAT Antarctic Peninsula Regional Working Group: see Cunningham et al., 1995). The data were collected with a variety of configurations and have variable vertical and horizontal resolution. The principal multichannel, single-channel, and deep-tow boomer data sets and their acquisition parameters are summarized in Tables T9 and T10.
Multichannel (MCS) seismic data sets were processed to common-midpoint stack using standard procedures, which include spherical divergence compensation, predictive deconvolution, common midpoint gather, and normal move-out correction before stacking. Migration was applied to selected lines (e.g., IT92-109: see "Seismic Stratigraphy," in the "Site 1096" chapter). Attenuation of the seafloor multiple reflection on profiles from the continental shelf was achieved by application of f-k filter before stack (British Antarctic Survey profiles) or application of a median filter on common shot gathers after move-out with water velocity (Osservatorio Geofisico Sperimentale profiles). Further details of the seismic data recorded at each site are given in the "Seismic Stratigraphy" sections of site chapters.
During the interpretation procedure, seismic units are defined between reflectors that either are unconformities or mark distinct changes in seismic character or both, following procedures described, for instance, in Vail et al. (1977) and Posamentier et al. (1988). Boundaries on seismic reflection profiles are correlated at each site with logging data and physical properties measurements through the production of synthetic seismograms (see "Synthetic Seismograms"). A detailed seismic stratigraphic background for each of the Leg 178 sites can be found in the individual site chapters.
Synthetic seismograms were computed for most of the sites using density and compressional wave velocity from all available sources. These included MST whole-core and individual sample measurements, and downhole logging data. Density (D) and P-wave velocity (v) together define the acoustic impedance (Z):
The acoustic impedance contrast (reflectivity, R1/2) determines how much of the wave energy is reflected and how much is transmitted at a boundary between layers of impedance Z1 and Z2:
Here, we assume that the travel path of the wave is perfectly vertical, we do not consider multiple reflections between layers, and we ignore attenuation and dispersion.
A wavelet representative of the seismic data is derived either from a far-field signature in water or from the seafloor reflection next to the drill site (see "Seismic Stratigraphy" sections in site chapters for details). The synthetic seismogram is calculated by convolving the source wavelet with a reflectivity model (Kearey and Brooks, 1984; Bender, 1985).
An example (Fig. F15) shows a simplified earth model with four reflectors. To convolve a source signal (with a wavelength of several meters) with a reflectivity profile in length units (for the MST data, 2-cm spacing), the reflectivity profile is transformed into a time-dependent profile with the same time resolution as the source wavelet (Mayer, 1994).
Before creating a synthetic seismogram, the raw data are edited and corrected for artifacts that could have a dominant effect on the results. Impedance values that are larger or smaller than a defined percentage above or below a multiple polynomial regression curve are removed and replaced by interpolated values. The correction is done in two steps. During the first step, lower order polynomials and wider thresholds are used. In a second step, more restrictive correction terms can be applied. Special care is taken for values at the beginning and end of the series. An example of an application of the method to a 20-m-long data section is shown in Figure F16. After calculation of the synthetic trace, appropriate band-pass filtering is applied to make the synthetic seismogram comparable to the site-survey profiles (see "Seismic Stratigraphy" sections in site chapters).