--> An Improved Seismic Coherence Technology for Characterizing Faults

AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

An Improved Seismic Coherence Technology for Characterizing Faults

Abstract

Faults play a vital role in the development of oil and gas reservoirs, traps and hydrocarbon migration and accumulation. Using seismic attributes to characterize faults can greatly improve the efficiency of seismic interpretation and, more importantly, can enable the effective characterization of faults with small throws that sometimes cannot be easily recognised on seismic sections. Currently, with a focus on the discontinuity, the coherence has been developed for the characterization of faults. If the change in the shape of waveform at a fault is not obvious, which may occur when the fault throw and heave are small, and then it is not easy to detect the fault by coherence. In this study, we proposed an improved coherence-based method that considers both the discontinuity and the shapes of seismic reflectors. Firstly, we calculated the gradient ▽u at each analysis point, where u is amplitude. The gradient gives the orientation of a reflector. A more robust measure of the orientation is the use of mean value of ▽u, such as using a volume containing 3×3×3 samples. On the top and lower sides of the reflection, the gradients point downwards and upwards, respectively, and will cause cancellation when averaging them. So we can use the mean value of the structural tensor s, which is obtained by multiplying vector ▽u by its transpose (▽u)T, to define the orientation of the reflector. Secondly, the amplitude u serves as the weight factor that is multiplied by the structural tensor s, and let α denotes the product. We measured the difference D between each two values of α of the neighbouring traces in a running window and used the sum of the absolute value of the differences to define the coherence C. It is obvious that the large value of C indicates faults. For simplicity, we can use the spectral decomposition of the structural tensor s to find its eigenvectors, and use the eigenvector corresponding to the largest eigenvalue instead of s itself in the calculation of coherence. We selected 3-D seismic data from Tazhong area of the Tarim Basin to test the advantages of our method in characterising faults. The fault throws and heaves are both small in this area and many of the faults do not produce obvious waveform anomalies, and the previous coherence methods fail to detect some of them. But when applying our method, they are clearly characterized. So the proposed method is more capable of detecting subtle faults having quite small throws and heaves.