Voxel Density: Enhancing attributes using a local estimate of confidence

Abstract

Attributes calculated from 3D seismic data volumes are commonly noisy and chaotic in their representation of geologic trends. The complex morphology and expression of geologic features can result in inconsistent performance of a given attribute for highlighting features of interest. Structural and diagenetic overprinting can also complicate attribute results. The handling of noise and regularization of uneven attribute performance is an important research goal. Voxel Density is a novel way to score the local significance of data trends within a 3D seismic volume. Significant regions can then then be enhanced, while insignificant regions can be suppressed or filtered out. Voxel Density is a “running window” algorithm, where a 3D operator is convolved with the entire input data volume. For each operator position, the number of data points that fall within a given range of values are counted; yielding a density score. Areas of high density score are considered to have high confidence. Conversely, areas of low density score are assumed to be noise and are filtered out or deemphasized. Noise can be filtered by removing data points from regions of low density score. Alternatively, the density score can be used to control the application of smoothing algorithms. By smoothing high confidence regions less aggressively, significant edges can be preserved during smoothing. Volume contrast can also be enhanced in an attribute volume; boosting the signal‐to‐noise (S/N) ratio.