A new paper is added to the collection of reproducible documents: Random noise attenuation by a selective hybrid approach using f-x empirical mode decomposition
Empirical mode decomposition (EMD) becomes attractive recently for random noise attenuation because of its convenient implementation and ability in dealing with non-stationary seismic data. In this paper, we summarize the existing use of EMD in seismic data denoising and introduce a general hybrid scheme which combines $f-x$ EMD with a dipping-events retrieving operator. The novel hybrid scheme can achieve a better denoising performance compared with the conventional $f-x$ EMD and selected dipping event retriever. We demonstrate the strong horizontal-preservation capability of $f-x$ EMD that makes the EMD based hybrid approach attractive. When $f-x$ EMD is applied to a seismic profile, all the horizontal events will be preserved, while leaving few dipping events and random noise in the noise section, which can be dealt with easily by applying a dipping-events retrieving operator to a specific region for preserving the useful dipping signal. This type of incomplete hybrid approach is termed as selective hybrid approach. Two synthetic and one post-stack field data examples demonstrate a better performance of the proposed approach.
A new paper is added to the collection of reproducible documents: Deblending using a space-varying median filter
Deblending is a currently popular method for dealing with simultaneous-source seismic data. Removing blending noise while preserving as much useful signal as possible is the key to the deblending process. In this paper, I propose to use space-varying median filter (SVMF) to remove blending noise. I demonstrate that this filtering method preserves more useful seismic reflection than does the conventional version of median filter (MF). In SVMF, I use signal reliability (SR) as a reference to pick up the blending spikes and increase the window length in order to attenuate the spikes. When useful signals are identified, the window length is decreased in order to preserve more energy. The SR is defined as the local similarity between the data initially filtered using MF and the original noisy data. In this way, SVMF can be regionally adaptive, instead of rigidly using a constant window length through the whole profile for MF. Synthetic and field-data examples demonstrate excellent performance for my proposed method.