Dip-dividing Multiple Matching and Separation Based on F-x Empirical Mode Decomposition

Adaptively subtraction for multiple plays an important role in multiple attenuation. Errors between original data and predicted multiples have a extreme impact on multiple elimination. Conventional multiple matching and substacting, which are made by minimizing the residual between the original data and the predicted multiples in a least-squares sense, lead to unsmooth adjacent sample points especially in high wavenumber events. In this paper, we propose a dip-dividing multiple matching and separation method based on window dip decomposition. Because of the dip selection property of f-x EMD, this method can preserve totally the horizontal or low-dip-angle primary overlapped by multiple and enhance the matching and substracting result for high-dip-angle primary. Synthetic data examples demonstrate the effectiveness of the proposed method.