Subsalt Marchenko imaging: A Gulf of Mexico example

Marchenko redatuming allows one to use surface seismic reflection data to generate the seismic response at any point in the subsurface due to sources at the surface. Without requiring much information about the earth’s properties, the seismic response generated by Marchenko redatuming contains accurate estimates of not only the primaries, but also internal multiples. A target-oriented imaging method, referred to as Marchenko imaging, was implemented for imaging complex structures of the earth using the seismic response obtained through Marchenko redatuming. Taking account of the contribution of both primaries and internal multiples, Marchenko imaging provides more accurate information about the earth’s structures than standard imaging methods (e.g. Reverse Time Migration). In order to apply Marchenko redatuming and imaging to field seismic data, we investigate two important aspects of the Marchenko framework: 1) the missing near offsets in marine shot records, and 2) the calibration of the reflection data. With numerical experiments using a Gulf of Mexico salt model (inspired by a Gulf of Mexico field dataset), we present a criterion to determine whether the missing near offsets of the towed-streamer dataset need to be reconstructed. We also demonstrate the importance of proper calibration for the amplitudes of the reflection data in removing the artifacts caused by the mismatch between the amplitudes of the two inputs, the reflection response and the first arrival, for Marchenko redatuming. Finally, we suggest a work flow for processing the marine towed-streamer field dataset acquired at the Gulf of Mexico, and present a complete theoretical and practical framework to produce a target-oriented subsalt image using the Marchenko methods. The images obtained from Marchenko imaging are consistent and comparable, for the most part, with conventional migration methods. However, Marchenko imaging achieves improvements in the continuity of the geological structures and in suppressing the artifacts, which we think are caused by internal multiples.