Efficiency of coherency attributes in seismic data interpretation


Seismic coherency is a measure of lateral changes in acoustic impedance that are caused by variations in structure, stratigraphy, lithology, porosity, and fluid content. Seismic coherency is a geometrical attribute that establishes temporal and lateral relationships with other attributes. Seismic coherency can be defined by coherency attributes. When coherency attributes are applied to seismic data, they can define continuity between two or more traces within a seismic window. The rate of seismic continuity is a sign of geological continuity. In interpretation of 3-D seismic data, a coherency cube can be extremely effective in delineating geological continuity or discontinuity, such as minor faults.
There are three solutions to calculate coherency attributes. They are cross-correlation, semblance and eigenstructure. These approaches are based on the continuity of traces in a time or depth interval in which similar traces show high coherency, while non-similar traces show low coherency. Cross-correlation algorithm was proposed by Bahorich and Farmer (1995) then it was completed by Marfurt et al (1998). In this approach, to calculate the coherency three traces are chosen (one as a base and two others in the direction of in-line and x-line). First, coherency is calculated in a finite time interval along in-line then along x-line. Finally, coherency is achieved by multiplying the root of maximum coherency value in each time interval along in-line and x-line. Semblance algorithm was introduced by Marfurt et al (1998). This method is employed using, as narrow as possible, a temporal window analysis typically determined by the highest usable frequency in the input seismic data. Near-vertical structural features, such as faults are better enhanced when using a longer temporal analysis window. By this algorithm, we are able to balance the conflicting requirements between maximizing lateral resolution and increasing S/N ratio. Eigenstructure algorithm was presented by Grestenkorn and Marfurt (1999). This algorithm is based on the estimation of coherency using covariance matrix.
To study the ability of coherency attributes in delineating minor faults we generated several 3-D synthetic seismic cubes with horizontal, dipping, and cross dipping layers with minor faults. We also studied the effect of the dominant frequency, signal to noise ratio and the size of the analysis cube in calculating coherency attributes using Matlab software. Using a Ricker wavelet with dominant frequency of 30 Hz and signal to noise ratio of 1 we found that the analysis cube with size for horizontal layers and for dipping layers are appropriate for coherency calculation. Semblance and eigenstructure algorithms are useful to detect minor faults in 3-D synthetic seismic cubes.
We applied all three approaches of coherency attributes on 3-D real data. Seismic data were belonged to the Khangiran gas field in NE Iran. The main reservoir is the Mozdooran formation (limestone) and its cap rock is red siltstone. In seismic data, the sample interval was 4 ms with a distance between traces of 25 m along in-line and x-line directions with 101 in-lines and 71 x-lines and 1000 ms time interval. Coherency attributes proved to be very effective in defining minor geological discontinuity even of 4 ms.