Petroleum resource potential modeling seeks to characterize undiscovered petroleum resources. This information from the modeling can contribute to a reduction in corporate risk while characterizing the commercial potential of the undiscovered resources. Such models consider different types of variable dependencies arising from geologic risk evaluation, volumetric calculation, and resource aggregation to higher geographic levels. Commonly, the available data are not sufficient to specify such variable correlations or interdependencies, particularly in frontier regions. It is also a challenge to formulate variable correlations in resource calculations because geologic variables have to be fit to a multivariate lognormal distribution or other specific multivariate distributions with an appropriate correlation structure. However, variable correlations are common among the geologic variables, and ignoring the interdependencies may lead to a serious bias in the resource potential estimation and the uncertainty range. Recent methodological developments in statistics indicate that the use of copulas permits more flexibility for the consideration and incorporation of variable interdependency, thus analogs can be introduced to problems where estimating correlation structures are impossible and wider choices of statistical distributions become available. This article proposes the use of copulas for handling variable dependency in petroleum resource assessment. The methods and procedures are illustrated using examples from a hypothetical data set and the crude oil resource appraisal of Tertiary clastic plays in Beaufort-Mackenzie Basin in Arctic Canada. Comparisons of crude oil resource estimates obtained using different correlation scenarios for these plays suggest that when positive correlations are used, the mean value of the oil resource is increased about 1.6 times that estimated, assuming a complete independence among the input variables.

You do not currently have access to this article.