Full text loading...
-
Recursive-Iterative Zero-Phase Filtering Via Singular Spectrum Analysis
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 79th EAGE Conference and Exhibition 2017, Jun 2017, Volume 2017, p.1 - 5
Abstract
We present a recursive-iterative Singular Spectrum Analysis (RI-SSA) algorithm which explores the time-correlation between reflected events. The RI-SSA algorithm depends on the first eigenimage of the SVD of the data matrix only. It is formed by letting each column be the data vector shifted one place down. The first eigenimage is related to the part of the signal with most strong correlation along the time variable and may be transformed to a time signal, which mainly consists of the low-frequency part of the input signal. We show that this corresponds to filtering the data with a symmetric zero-phase filter, which is the autocorrelation of the first eigenvector associated to the time variable. The computational implementation may be done using the power-method in a recursive scheme, increasing the order of the data matrix, by increasing the number of shifted traces. This improves the separation of the input data in a low-frequency and high-frequency component. This separation may be further improved by adding iterations. The output of the RI-SSA algorithm is the low and high frequency part of the signal. We illustrate the effectiveness of this new approach to the prediction and subtraction of the ground-roll.