For 3D seismic imaging in structurally complex areas, the use of migration by wavefield extrapolation has become widespread. By its very nature, this family of migration methods operates on data sets that satisfy a wave equation in the context of a single, physically realizable field experiment, such as a common-shot record. However, common-shot migration of data recorded over dipping structures requires a migration aperture much larger than the recording aperture, resulting in extra computations. A different type of wave-equation record, the response to a linear or planar source, can be synthesized from all the common-shot records. Synthesizing these records from common-shot records involves slant-stack processing, or applying delays to the various shots; we call these records delayed-shot records. Delayed-shot records don't suffer from the aperture problems of common-shot records since their recording aperture is the length of the seismic survey. Consequently, delayed-shot records hold potential for efficient, accurate imaging by wavefield extrapolation. We present a formulation of delayed-shot migration in 2D and 3D (linear sources) and its application to 3D marine streamer data. This formulation includes a discussion of sampling theory issues associated with the formation of delayed-shot records. For typical marine data, 2D and 3D delayed-shot migration can be significantly more efficient than common-shot migration. Synthetic and real data examples show that delayed-shot migration produces images comparable to those from common-shot migration.
