A stable and practical implementation of least-squares reverse time migration

Authors:
  • YZYu Zhang
    Yu Zhang
  • LDLian Duan
    Lian Duan
  • YXYi Xie
    Yi Xie

Abstract
Adapting reverse-time migration (RTM) and demigration (RTDM) as the imaging and modeling operators to maximize the cross-correlation between the simulated and observed seismic data, we introduce a new practical least-squares RTM (LSRTM) scheme and derive its Newton's method in seeking numerical solutions with fast convergence. Through both synthetic and real experiments, the proposed LSRTM is shown to provide high quality images with balanced amplitudes, improved focusing and enhanced resolution. It is also capable of removing free surface ghosts caused by towed streamer acquisition and reducing crosstalk noise associated with data shot with simultaneous sources.