In the first preprocessing step, the raw data were corrected for constant velocity spherical divergence by multiplying each trace with a time-variant amplitude scale function of the form A(t)=t. Following the divergence correction, we performed a normal moveout correction with an initial estimate of the best stacking velocity function. A windowed section of the first CMP gather can be seen in Figure .
A significant change in shape and amplitude of the source wavelet is visible at the transition zone between the 50 m and 100 m group spacings, where the wavelet exhibits very low frequency content and weak amplitude. In the direction of the near offsets (to the right in Figure ), the wavelet sharpens again and regains its large amplitude. These waveform shape and amplitude changes with offset can seriously degrade any subsequent AVO analysis. In order to regularize the wavelet as much as possible, a single trace source wavelet deconvolution before NMO correction was performed. Although the application of single trace, as opposed to surface-consistent, deconvolution can result in undesired changes in amplitude and waveform, the Blake Outer Ridge seismic data were of high enough quality that single trace deconvolution with careful quality control gave very consistent results. The deconvolved data were then bandpass filtered to the original data bandwidth to remove spurious deconvolutional high frequency noise, and again NMO corrected. After having accounted for the change of the waveform shape at the transition zone, the amplitude change also required compensation. In order to do this, two main assumptions were made. First, it was assumed that an offset-dependent rather than an angle-dependent amplitude correction was sufficient, since the difference between the maximal angle of incidence at the seafloor (33) and the BSR reflection (30) is negligibly small. Second, we assumed a functional form for the AVO response of the seafloor reflection, based on equal positive P and S impedance contrasts, adjusted here to 0.4 respectively, via linearized Zoeppritz modeling. Based on these assumptions, the amplitude calibration was performed by scaling each trace to the seabottom amplitude to match the predicted seafloor AVO, as a function of offset. Figure shows a windowed CMP section after the deconvolution and amplitude corrections.
Comparing this result with the previous figure, the improvement of the waveforms is obvious. There are only small variations in the shape of the wavelet left and the amplitudes are physically consistent across the entire offset range.
A linear interpolation at the near offsets was used to even the group spacings to 50 m at each offset and thus eliminated the traveltime kink introduced by the nonlinear streamer. After applying an inverse NMO correction to the interpolated data, a high resolution NMO stacking velocity analysis was performed. The velocity scans showed a nearly linear rms velocity function at each CMP, which varies as a function of depth below the seafloor (Figure ). A 2-D rms velocity model was built based on the velocity scan information along the line. It assumes a water velocity of 1.48 km/s and increases linearly with depth with a constant gradient of 0.06 km/s/s (Figure ).
After having derived a good stacking velocity model for the data, they were processed again by using the new velocity model. Since it is essential for the subsequent impedance contrast estimation that the reflector moveout is very flat after NMO correction, an additional static shift was applied before the amplitude scaling to correct for some small non-hyperbolic offset-dependent moveout in the CMP gathers. The final deconvolved, NMO, static, amplitude and group spacing corrected data are shown in Figure .
This figure clearly indicates the nearly constant AVO amplitudes of the seafloor reflection and the increasing negative amplitudes with increasing offset of the hydrate reflection. This result can be compared with the theoretical AVO curves for hydrate over sediment and hydrate over gas which were shown in Figure . The comparison of the observed behavior with the predicted one already suggests that there might be free gas present beneath the bottom of the hydrate stability field.