Separation of signal and coherent noise by migration filtering

Abstract

We describe a new separation algorithm, denoted as migration filtering, that separates data arrivals according to their path of propagation. The algorithm accounts for the actual moveout and amplitudes of the data arrivals and is not restricted to uniformly weighted hyperbolic or linear moveout. This is accomplished by using forward, modeling operators for the computation of the signal and the coherent noise arrivals. The model estimates for both components are computed by a least-squares linerarized inversion scheme. The method is applied to synthetic and field data to demonstrate the strengths and limitations of this new separation method.