Effect of random noise in ground roll suppression via adaptive singular value decomposition
Seyed Ahmad Mortazavi, Abdolrahim Javaherian · 2012
Adaptive Singular Value Decomposition (ASVD) is a coherency-based filter which decomposes data into its eigenimages and can detect horizontal events in the first eigenimages. By using the adaptive method, the ground roll is converted to a horizontal event in each window. By zeroing the first eigenvalues which present the ground roll, it can be suppressed. By increasing the number of eigenvalues to be zeroed, the ground roll is better attenuated but more signals are damaged as well. In this paper, an ASVD filter, in MATLAB code, was applied to synthetic data containing the ground roll, refractors and reflectors over an earth model with one weathering layer and a real shot record from the South West of Iran. Results show that the ASVD can attenuate the ground roll with minimum harm to the signals. The ASVD method has two important advantages: (1) it can calculate the velocity for the best rotation to make the ground roll as horizontal as possible in each window and (2) the results of applying the filter to the synthetic data with various signal-to-noise ratios (SNR) show that this filter is not sensitive to SNR.