Sediment velocity estimation using iterative 3-D migrations of short offset seismic reflection data in deep water

In deep ocean settings where water depth greatly exceeds the source-to-receiver length, the geometry is insufficient for accurate determinations of velocity from reflection-moveout. However, velocities are crucial for estimates of physical properties and image processing. Focusing analyses with conventional post-stack two-dimensional migration improves images, but does not produce geologically meaningful velocities except in the special case of a two-dimensional earth. For the more general case of the three-dimensional earth there is no a priori method to determine the degree of geometrical complexity. We present a technique using a short-offset three-dimensional (3-D) data set over the 5 km deep trench west of the Lesser Antilles. These data illustrate highly sensitive post-stack 3-D focusing analyses (± 20 m s^–1 interval velocities), and the relationship of these seismically derived velocities to rock velocities. In our Barbados example we were able to establish the presence of a widespread 80-160 m thick low-velocity zone at and above the main low-angle fault. This observation suggests the water-rich décollement leaks water into the overlying sections. Also evident is a low-velocity section associated with turbidite sands. These results are confirmed with sparse logging data and well samples. Deep-water short offset 3-D experiments provide a potentially effective approach for velocity estimation, replacing the operational complexity of long-offsets with simpler short-offset techniques. In areas of structural complications and abundant diffracted energy, it is a surprisingly accurate method, utilizing the high fidelity 3-D wavefield and the information carried in zero-offset diffraction ellipsoids. The velocity used to properly collapse a diffraction ellipsoid is explicitly the velocity of propagation in the media since the travel path is known exactly. Thus, the derived velocities should closely represent rock velocities, unlike the 2-D case where the propagation geometry is not known.