Multi-waveform classification for seismic facies analysis. (April 2017)

• Record Type: Journal Article
• Authors: Song, Chengyun; Liu, Zhining; Wang, Yaojun; Li, Xingming; Hu, Guangmin
• Is Part Of: Computers & geosciences. Volume 101(2017)
• Page Start: 1
• Page End: 9
• Publication Date: 2017-04
• Subjects: Multi-waveform classification -- Seismic facies analysis -- SOM -- Multilinear subspace learning; Environmental policy -- Periodicals; 550.5
• DOI: 10.1016/j.cageo.2016.12.014
• Languages: English
• ISSNs: 0098-3004

Abstract

Abstract: Seismic facies analysis provides an effective way to delineate the heterogeneity and compartments within a reservoir. Traditional method is using the single waveform to classify the seismic facies, which does not consider the stratigraphy continuity, and the final facies map may affect by noise. Therefore, by defining waveforms in a 3D window as multi-waveform, we developed a new seismic facies analysis algorithm represented as multi-waveform classification (MWFC) that combines the multilinear subspace learning with self-organizing map (SOM) clustering techniques. In addition, we utilize multi-window dip search algorithm to extract multi-waveform, which reduce the uncertainty of facies maps in the boundaries. Testing the proposed method on synthetic data with different S/N, we confirm that our MWFC approach is more robust to noise than the conventional waveform classification (WFC) method. The real seismic data application on F3 block in Netherlands proves our approach is an effective tool for seismic facies analysis. Highlights: Classifying the multi-waveform in a 3D window can suppress the effect of data noise. Multi-window dip search algorithm is used to extract multi-waveform. Multilinear subspace learning is used to reduce dimension of multi-waveform.