next up previous [pdf]

Next: Method Up: Chen: Deblending using SVMF Previous: Chen: Deblending using SVMF

Introduction

The technique of simultaneous-source (sometimes called multisource) acquisition involves firing more than one shot at nearly the same time, regardless of source interference. In conventional acquisition, either the temporal shooting intervals (or ignition interval (Wapenaar et al., 2012a)) or the spatial sampling intervals are large enough that the crosstalk between successive shots can be left out. The novel simultaneous-source technique can reduce the acquisition period and at the same time can improve data quality by decreasing the spatial-sampling interval (Berkhout, 2008). The benefits from simultaneous-source acquisition are compromised by the challenges in removing blending interference. Because of its economic benefits and technical challenges, this technique has attracted the attention of researchers in both industry and academia (Moore et al., 2008; Huo et al., 2012; Mahdad et al., 2011).

Two main ways exist to address the challenges of simultaneous-source acquisition. The first way is to use a first-separate and second-process strategy (Chen et al., 2014; Spitz et al., 2008; Abma et al., 2012; Mahdad et al., 2011), also known as "deblending" (Doulgeris et al., 2012). The other is to use direct imaging and waveform inversion by applying some constraint to eliminate the artifacts caused by interference (Dai et al., 2012; Jiang and Abma, 2010; Tang and Biondi, 2009; Dai and Schuster, 2011). Although the direct imaging approach has achieved some encouraging results, the currently preferred way is still to focus on the separation of blended data into individual sources. Currently existing deblending methods fall into two categories. The first is based on filtering (Hampson et al., 2008; Huo et al., 2012), which treats the deblending problem simply as a noise attenuation problem. This treatment is logical because the blending noise, although coherent in the common-shot domain, has been demonstrated to be incoherent in other domains, such as common-receiver, common-offset, and common-midpoint domains (Berkhout, 2008; Beasley, 2008). Thus, all the conventional denoising algorithms can be used in the same way. The second category includes methods based on inversion, converting the separation problem into an inversion problem. Because of the ill-posed property of the blending equation, some specific constraints must be added. The inversion problem can be solved either by directly inverting the matrix of the forward modeling operator (Wapenaar et al., 2012b,a), or by using an iterative framework that iteratively estimates the useful signal and subtracts the blending noise (Doulgeris and Bube, 2012; Chen et al., 2014; Mahdad et al., 2012,2011).

Median filter (MF) is well-known for its ability to remove spiky noise in a seismic profile after NMO or with relatively flatter events. MF has been successfully utilized in the deblending process (Huo et al., 2012; Mahdad et al., 2012). In this paper, I demonstrate the use of a space-varying median filter (SVMF) instead of a conventional MF to remove blending noise. I squeeze or stretch the length of conventional MF according to the signal reliability (SR) as a signal point. The SR is defined as the local similarity between the data initially filtered using MF and the original noisy data. Different SRs correspond to different window length for filtering. I use both synthetic and field blended common-receiver gathers to demonstrate the better energy-preserving property of SVMF. I also use both synthetic and field prestack time-shot-offset (TSO) domain blended data cube to demonstrate the efficient and effective performance of the proposed deblending approach using SVMF.


next up previous [pdf]

Next: Method Up: Chen: Deblending using SVMF Previous: Chen: Deblending using SVMF

2015-11-23