Future work: finding the velocity

The left panel in Figure 2 shows that even after successful elimination of coherent non-thermohaline events, large amounts of random noise are left in the data. To be able to obtain RMS velocities in a time-efficient manner, with a high-amplitudes autopicker, we obtain the semblance of each individual CMP gather. The prior knowledge of the limits of velocity variations in water allows for a small-range, high-resolution transformation. The result is seen in Figure 4. The presence of velocity variations with depth and midpoint is apparent. The absolute value of these variations does not surpass 30 m/s.

Velocity departures from the background can be seen in the left panel of Figure 5 as residual curvatures in angle-domain common image gathers (ADCIGs). This figure contains only the angles between $10^\circ$ and $30^\circ$. Because of the small depth of the thermohaline reflections and of the missing near-offset information, information from incidence angles smaller than $10^\circ$ was obtained only from the offset-continuation fill. We discarded it, since it was characterized completely by our estimate of constant velocity. The angles larger than $30^\circ$ were unusable because of a loss of bandwidth during the transformation to ADCIGs. The small curvatures place the velocity anomalies well within the limits of the Born approximation. This means that WEMVA would be a suitable tool for resolving them.

WEMVA is an iterative inversion scheme that attempts to optimize the focusing of the migrated image Biondi and Sava (1999). Specifically, the result of wavefield-continuation migration is transformed to ADCIGs, the gathers are flattened, the difference from the unflattened image is taken to obtain an image perturbation, which is finally inverted into a velocity update. We plan to perform this procedure in the future, using moveout shifts computed by dip field integration Guitton (2003a) to flatten the gathers for the image perturbation.

 

stxfx
Figure 4 Semblance scan volume denoised with the procedure described in Vlad (2003). In the right panel, the vertical line delimits the 1500 m/s point on the slowness axis.

 

Cuben4aT Figure 5 ADCIGs between $10^\circ$ and $30^\circ$ after migration with a constant velocity of 1520 m/s.