An Improved Coherence Algorithm with Robust Local Slope Estimation

Yuchen Wang,Wenkai Lu,Peng Zhang

DOI: https://doi.org/10.1016/j.jappgeo.2015.01.015

IF: 1.845

2015-01-01

Journal of Applied Geophysics

Abstract:During recent years, dip scanning technique has been applied to traditional coherence algorithms to bring higher stability and robustness. In our research, we find that the slope information used for coherence estimation should trace layers correctly but contain few fluctuations and details about discontinuities. The second demand is useful for coherence algorithms to detect small discontinuities of structures. However, dip scanning technique cannot trace layers correctly if the slope variation of structures is large. Furthermore, estimated slopes may contain fluctuations and details which make small discontinuities invisible in coherence cube. In addition, dip scanning leads to huge computation cost when applied to seismic data with large slope variation. We introduce complex seismic trace analysis combined with L1 norm minimization to slope estimation to improve the coherence estimation. Slopes estimated with complex seismic trace analysis contain fewer fluctuations and details about discontinuities and can be used to trace layers more accurately compared with dip scanning. We introduce the L1 norm minimization to overcome the drawbacks of low stability and low robustness of Hilbert transform. By estimating slopes with supertrace data, we enhance the algorithm's ability to detect small scale discontinuities and furthermore, its robustness to noise. In addition, the computational efficiency is also improved by reducing the dip scan range. Applications on synthetic and real data sets show that our coherence can highlight faults and channels effectively and reveal more details than the original cross-correlation-based coherence algorithm with dip scanning with lower runtime overhead does.