John E. Eastwood

Interactive seismic facies classification using textural attributes and neural networks

In this study, we present an application of textural analysis to 3D seismic volumes. Specifically, we combine image textural analysis with a neural network classification to quantitatively map seismic facies in three-dimensional data.

Using legacy seismic data in an integrated time-lapse study: Lena Field, Gulf of Mexico

Seismic monitoring (time-lapse or 4-D seismic) has the potential to significantly increase recovery in existing and new fields. While future field developments should benefit from seismic acquisition designed for time-lapse monitoring, the portfolio of current seismic monitoring opportunities for most companies consists of existing fields for which one or more 3-D surveys have already been acquired. These legacy seismic data sets were not acquired for the purposes of seismic monitoring and are often very different in terms of acquisition and processing parameters. In addition, the seismic acquisition is rarely timed to optimally map reservoir changes or impact development decisions. Repeatability of the seismic data in the nonreservoir portion of the data volume and the robustness and credibility of the seismic difference anomaly within the reservoir are key issues for the application of legacy data in time-lapse analysis. Another issue is whether 4-D seismic differences can be interpreted in terms of reservoir production changes. If these issues can be effectively addressed, then legacy 4-D seismic may be used as a tool for reservoir surveillance or reservoir management. The objective of this paper is to understand the magnitude of the processing effort required to obtain reliable time-lapse differences and to interpret the seismic difference observed in the B80 reservoir of the Lena Field through the use of geologic modeling, flow simulation, and seismic modeling. The Lena Field (Mississippi Canyon Block 281) is south of the modern Mississippi delta in 1000 ft of water. The field is on the western flank of a salt dome within a fault-bounded intraslope basin. Hydro-carbon production is from six Pliocene-age sands. As shown in the seismic section (Figure 1), the B80 reservoir is about 10 500 ft below sea level at approximately 3 s TWT. The interval is interpreted as a low-stand fan systems tract representing deposition …

Introduction to special section: Interpretation for unconventional resources

Interpretation for unconventional resources introduces a broad range of issues and objectives that are markedly different from those encountered when interpreting for conventional plays. There are of course many commonalities. Structural interpretation is still a strong driver, particularly to

Interactive Seismic Facies Classification of Stack And AVO Data Using Textural Attributes And Neural Networks

We present an interactive method for volume-based seismic facies mapping using seismic textural attributes and probabilistic neural networks. Textural analysis can quantitatively describe many aspects of the classic seismic facies description preformed by the interpreter. Stratigraphically-steered seismic texture is a quantitative, multi-trace (imagebased) attribute that mimics the visual process of the seismic interpreter more effectively than traditional trace-based attribute analyses do. Probabilistic neural networks (PNNs) are parallel implementations of a standard Bayesian classifier that can efficiently perform pattern classification. A primary advantage of the PNN is that it does not require extensive training. In the case of seismic analysis, a reliable seismic facies classification can occur with as little as one example per facies class.