In recent years, data-driven multiple prediction methods and wavefield extrapolation methods have proven to be powerful methods to attenuate multiples from data acquired in complex 3-D geologic environments. These methods make use of a two-stage approach, where first the multiples (surface-related and / or internal) multiples are predicted before they are subtracted from the original input data in an adaptively. The quality of these predicted multiples often raises high expectations for the adaptive subtraction techniques, but for various reasons these expectations are not always met in practice. Standard adaptive subtraction methods use the well-known minimum energy criterion, stating that the total energy after optimal multiple attenuation should be minimal. When primaries and multiples interfere , the minimum energy criterion is no longer appropriate. Also, when multiples of different orders interfere, adaptive energy minimization will lead to a compromise between different amplitudes corrections for the different orders of multiples. This paper investigates the performance of two multiple subtraction schemes for a real data set that exhibits both interference problems. Results from an adaptive subtraction in the real curvelet domain, separating primaries and multiples, are compared to those obtained using a more conventional adaptive subtraction method in the spatial domain.
展开▼
