ABSTRACT: Understanding the geologic meaning of seismic data - Facies identification and classification using trace shape
Darmon, Loz, and Randy Phillips , Flagship Geosciences, L.L.C, Houston, TX
The relationship between seismic data and geology is a key issue in the characterization of hydrocarbon reservoirs. 3-Dimensional seismic data has the advantage of providing large amounts of data that will normally areally and vertically encompass the reservoir. Unfortunately, the relationship between the character of the seismic trace and the reservoir geology / content is often hard to define. One unique approach to this problem - the idea of using trace shape in conjunction with an automatic pattern-recognition method - is illustrated in this paper. The assumption is made that changes in trace shape are related to changes in the geology and/or content of the reservoir, and that this approach may yield better results than the more traditional magnitude-based approaches. The paper outlines the basic approach of this technique, which is to use an unsupervised Neural Network to identify and classify patterns in the data. The advantage of this approach is that, although complex in operation, it is fairly easy to understand, as it duplicates the differentiation process as used by the human brain - i.e., we tend to separate items based on visual recognition of shape. A good interpreter can quickly recognize a geological feature on a 2D seismic line - it is not so easy to repeat this process on a 3D volume. Several examples of this approach will be given, each example tailored to a different geological scenario. The power of this method when also combined with traditional approaches will also be shown via the use of examples.
AAPG Search and Discovery Article #90913©2000 AAPG International Conference and Exhibition, Bali, Indonesia