A key issue in wavefield separation is to find a domain where the signal an
d coherent noise are well separated from one another. A new wavefield separ
ation algorithm, called migration filtering, separates data arrivals accord
ing to their path of propagation and their actual moveout characteristics.
This is accomplished by using forward modeling operators to compute the sig
nal and the coherent noise arrivals. A linearized least-squares inversion s
cheme yields model estimates for both components: the predicted signal comp
onent is constructed by forward modeling the signal model estimate. Synthet
ic and field data examples demonstrate that migration filtering improves se
paration of P-wave reflections and surface waves, P-wave reflections and tu
be waves, P-wave diffractions, and S-wave diffractions. The main benefits o
f the migration filtering method compared to conventional filtering methods
are better wavefield separation capability, the capability of mixing any t
wo conventional transforms for wavefield separation under a general inversi
on framework, and the capability of mitigating the signal and coherent nois
e crosstalk by using regularization. The limitations of the method may incl
ude more than an order of magnitude increase in computation costs compared
to conventional transforms and the difficulty of selecting the proper model
ing operators for some wave modes.