Page 532 - Acquisition and Processing of Marine Seismic Data

11.7 PRESTACK MIGRATION 523

FIG. 11.33 Schematic illustration of velocity analysis in the case of conflicting dips. (A) If the velocity of A (1500 m/s) is
picked (left), the event A is flattened and B is overcorrected after NMO correction (right). (B) If the velocity of B (2000 m/s) is
picked (left), the event B is flattened and A is undercorrected after NMO correction (right). In both cases, only one of the reflec-
tions can be represented in the stack section with its proper amplitude value.

after NMO correction, since 1500 m/s is too slow efficient than a conventional prestack migration
to flatten the event B (Fig. 11.33A). On the con- algorithm.
trary, if the velocity at point B (2000 m/s) is In the presence of conflicting dips, stack sec-
picked in favor of event B, this time the event tions are no longer identical to zero-offset sec-
B is flattened and A is undercorrected after tions. Because all of the poststack migration
NMO correction, since 2000 m/s is too fast to algorithms assume a zero-offset section as an
flatten the event A (Fig. 11.33B). In any case, input, poststack migrations do not provide cor-
either event A or event B is represented by lower rect subsurface images, and simply prestack
amplitude than what it actually has in the final migration is required. Prestack migration is
stack section, since only one of them can be flat- done by summing the amplitudes along the non-
tened after NMO correction. zero offset diffraction times. As in the zero-offset
In case of conflicting dips and strong lateral case, seismic velocities determine the shape of
velocity variations, prestack migration should the hyperbola (the migration operator) along
be used to obtain an accurate subsurface image. which the summation is performed. Prestack
In addition, in order to solve the conflicting dips migration is applied to the common offset sec-
issue, dip moveout (DMO), explained in tions, and all common offsets are processed in
Section 11.10, can also be implemented, since this manner and finally summed up to obtain
dip moveout is much more computationally the migrated image.