The zero-velocity layer was introduced in Higginbotham et al. (1985) to increase the maximum dip imaging capability of finite-difference depth migration. Beasley and Lynn (1992) adapted the idea to improve the imaging, again using finite-difference depth migration, of seismic data acquired in areas of irregular topography. Beasley and Lynn's application improves upon the conventional method of processing, which is to time shift the data from the acquisition surface to a horizontal datum, and then migrate using the near-surface velocity above the surface and the best estimate of seismic velocity below the surface. The conventional procedure typically produces artifacts in the shallow part of the section that are characteristic of overmigration. To reduce these artifacts, velocities are often reduced for the migration step. The use of the zero-velocity layer overcomes the need to adjust the migration velocities. Here, a component of the migration velocity is set to zero in the layer between the datum and the surface. The function of the zero-velocity layer in migration is to remove the elevation-static correction applied in shifting the data to the flat datum. Only after the data have migrated through the zero-velocity layer to the irregular recording surface does the migration begin to act in its customary sense, moving energy from trace to trace.