Thickness Delineation Using Frequency Dependent Principal Component Spectral Attribute
University of Houston
Spectral decomposition is among the most popular methods for resolving subtle geologic feature such as thin layer. However, it often generate multiple iso-frequency seismic volumes of identical size to the original data. Handling this large amount of data is a great challenge to interpreters. One solution of this problem could be to consider it a task to reduce dimension of multiple attribute data.We propose applying a principal component based algorithm to generate frequency dependent attributes from spectral components while preserving most information. Following extraction of the first few principal components, we rotate them to meet the varimax criteria so that they are most interpretable. As a result, each of the rotated principal component is easily relates to a band of seismic frequency. Directed by interpretation, we generate frequency dependent principal component attributes, each of which indicates thin layer tuning effect by seismic wave that has a particular range of wavelength.
The frequency dependent principal component spectral attribute method is tested on data from synthetic wedge model and field survey in FortWorth Basin, North Texas of USA. Both cases show that proposed approach superior narrow band seismic and traditional principal component spectral attribute in terms of resolution to subtle structure. In the field data example, frequency dependent attribute reveals that layer thickness varies in concentric pattern around a deep karst depression, indicating layer thinning around collapse area due to strata stretching and/or faulting.
AAPG Search and Discovery Article #90182©2013 AAPG/SEG Student Expo, Houston, Texas, September 16-17, 2013