I propose to reduce the costs of both standard RTM and least-squares RTM by calculating the blended
migration operator
,
and then save it on disk. When the blended migration kernel is ready,
the migration image can be obtained by a dot product of
and the recorded shot gathers in the
domain; here,
denotes the inverse Fourier transform.
Because the migration kernels for the whole model space is too large to be saved, I phase-encode the
migration kernels
as well as phase-encode the recoded shot gathers before doing the dot product.
This reduces the memory cost by at least an order-of-magnitude.
Once the blended migration kernel is saved, it does not need to be recomputed at each LSM iteration so this can result in almost two orders-of-magnitude reduction in cost for iterative least-squares migration (Nemeth et al., 1999; Aoki and Schuster, 2009). A further decrease in cost can be achieved by phase-encoding and multisource migration.
In this chapter, I briefly introduce the least-squares phase-encoded GDM theory first, followed by numerical tests on the 2D SEG/EAGE salt model that demonstrate the effectiveness of this method. Conclusions are drawn at the end.