Evaluation of uncertainty propagation in seismic facies classification

Abstract

The aim of this work is to present a new methodology for evaluating the formation and propagation of uncertainty in seismic facies classification. We extend the idea of Grana and Della Rossa (2010) further and introduce the entropy to directly depict the uncertainty of facies classification, since the variable that represents facies is categorical. Firstly, we define log-facies based on petrophysical properties, and subsequently compute the corresponding entropy. Then, we derive facies probability conditioned on different properties such as elastic data at different scale and seismic data step by step. We evaluate the uncertainty in each step by the corresponding entropy which is derived from facies probability. The increment of entropy in each step can be considered to be the introduced uncertainty of the current properties. We also quantify the quality of facies classification by the reconstruction rate and evaluate the corresponding uncertainty by the mean value of entropy in each step. Using synthetic examples, the results not only clearly illustrate the influence degree of uncertainty in each step on seismic facies classification, but also analyze the component of final uncertainty. The methodology provides a vivid illustration for the formation and propagation of uncertainty in seismic facies classification as well as the great value for risk analysis and optimal decision-making.