Model‐based removal of water‐layer multiple reflections

Abstract

We have developed a procedure to attenuate water‐layer multiple reflections. We estimate the sea‐bottom reflectivity function and use it plus calculated amplitude functions to model all order water‐layer multiple reflections, taking into account both amplitude and waveform shape. We model the primary and multiple reflections from the sea bottom in the frequency‐slowness domain. The amplitude function in the data modeling includes the source directivity function, source ghost response, receiver array directivity function, receiver ghost response, and offset‐dependent geometrical spreading. For small offsets we can assume that the seabed reflectivity depends only on frequency, and it is estimated using a least‐squares algorithm. An unknown scaling constant in the data is estimated using the amplitude of the primary and first multiple reflection from the sea bed. The composite sea‐bottom reflectivity is estimated as a function of frequency for each common midpoint (CMP) position. We apply the algorithm to high‐resolution seismic data from the North Sea. The modelled data match the recorded data well, and the estimated primary reflectivity is more geologically meaningful than the stacked trace. By comparison with Radon transform multiple removal applied to the same data, the model‐based method was more computationally efficient and left less residual multiple energy.