Multiple Attenuation Overview

Abstract This paper is an overview of existing methods for multiple attenuation in seismic data processing. We describe the methods following a classification carried out according to the characteristics of the multiples upon which the attenuation methods are based. We present a brief description of the types of multiple problem appropriate for each method. Introduction Multiple reflections are seismic events that undergo more than one subsurface reflection in the path from source to receiver. They are one of the most studied subjects in seismic data processing because of their impact on the quality of the subsurface image. Although multiples carry information about the subsurface, they are treated as noise in most current processing schemes, and if not well attenuated, will interfere with the identification of structures and corrupt the amplitude information at primary reflections. Therefore, the goal of any seismic data processor is to deliver a multiple free data set tothe interpreter. Since multiples are present in any seismic survey, the amount of multiple 'noise' in the data at the end of the processing flow will depend on the intrinsic characteristics of the multiples and the multiple attenuation processes applied to the data. Refs. 1 and 2 aim to give a systematic view of multiple attenuation techniques. The first presents the methods that were well established at the time and proposes a strategy for attenuating multiples. The second, while giving a classification of type of multiples according to selected criteria, concentrates most of its attention on the comparison of wave-equation based methods developed in the last decade. In this paper, we present an overview of the vast subject of multiple attenuation, describe existing methods, and relate them to the characteristics of the multiples. It is not our intention to make this either a tutorial or a collection of strategies for multiple attenuation. Rather, our goal is to present a picture of multiple attenuation techniques that are currently available to the seismic processor. Ref. 2 divides the methods into two broad categories: filtering methods and wavefield prediction and subtraction methods. The filtering methods rely on 'a feature or property that differentiates primary from multiples'. Deconvolution, stacking, f-k filtering, and Radon filtering are placed in this category. The wavefield prediction and subtraction methods use a wave-equation based concept to generate an estimate of the multiples which are then subtracted from the input data. The inverse-scattering and feedback-model methods belong to this category. In this paper, we will follow this classification and categorize some recently developed methods. Multiple and the Attenuation Methods Multiples may have one or more of the main characteristics listed below that are exploited by the multiple attenuation methods. The first characteristic is based on the difference in moveout between the travel time curves of multiples and primaries. This difference is caused by the different trajectories undergone by the events in the subsurface. Methods exploiting this characteristic are classified as based on separability.