High-dimensional co-occurrence matrix: A new tool for 3D seismic target visualization and interpretation

Abstract

The interpretation of seismic attributes has been developed for more than 40 years, but it still cannot be fully matched with the study of geologic sedimentology. There are two main bottlenecks. One is that almost no attributes can completely characterize amplitudes from a 3D perspective. Another is the influence of wavelets (frequency and phase). Accordingly, a new volumetric attribute extraction technique is presented for solving these two problems as much as possible. A high-dimensional co-occurrence matrix (HCM), as a statistical representation of the spatial structural relationship between data points, can be one of the solutions to extracting 3D attributes. By considering the global spatial distribution, the HCM fixes the directional distortion of the existing co-occurrence matrix methods and effectively suppresses vertical wavelet illusions. By realizing a rapid transformation of seismic data to 3D geologic target visualizations, this method does not rely heavily on structural information such as dip and azimuth and becomes a convenient bridge from seismic to geologic interpretation. We have applied the methodology to channel delineation in a 3D seismic survey acquired in Ordos, China. The results indicate that the attribute volume extracted from HCM expresses the geologic target convincingly and eliminates the illusions caused by the wavelets. The deposition history of the channels could be observed continuously through geologic time.