Let *v*_{w} be the velocity of water. Assume that the seismic
velocities of sediments are greater than .The algorithm of eliminating the water-bottom multiples consists of three steps:

- 1.
- Applying the forward transformation with velocity to the
shot gather:
(7) *I*(*t*,_{0}*x*), the water bottom multiples are migrated to the left side of the shot while other events stay on the right side of the shot. - 2.
- Setting
*I*(*t*,_{0}*x*) to zero for all*x*smaller than*x*_{s}:(8) - 3.
- Applying the backward transformation with velocity to
:
(9) *v*_{w}.

Until now, I have always assumed that *D*(*t*,*x*_{r}) is a common shot gather
and the subsurfaces are horizontal reflectors. Actually, the same algorithm can
be applied to CMP gathers because the travel-time equations for
the reflection from a horizontal reflector are identical in a CMP gather and a CSG.
For a CSG, the assumption of horizontal reflectors is necessary
because the receiver array on
the right side of the shot can receive the reflected energy from a dipping bed
that is on the left side of the shot. Therefore, after forward transformation,
the image of the dipping bed could possibly appear on the left side of the shot.
The situation is different when the algorithm is applied to CMP gathers.
On a CMP gather, the moveout of a dipping reflector is the same as the
moveout of a flat reflector of high stacking velocity. After the
forward transformation, the image of a dipping reflector will stay in the
positive offsets. Thus the algorithm can adapt the reflection from a dipping
layer when it is applied to a CMP gather.

1/13/1998