UNCERTAINTY OVERSHADOWED – AN OVERVIEW OF COMMON MISCONCEPTIONS IN THREE-DIMENSIONAL MODELING FOR RESERVOIR CHARACTERIZATION AND SOURCE MASS DELINEATION

Three-dimensional modeling programs are commonly used for geologic data analysis to interpret and visualize complex field data for reservoir characterization and source mass delineation. The fundamental purpose of utilizing modeling programs is to overcome the burden of data limitations to derive a close approximation of the expected value at unsampled spaces with a specific area of concern. Three dimensional geologic models are also commonly used as a tool during data collection efforts to minimize unnecessary expenditures and to direct the allocation of available resources for each phase of the investigation.

Numerous interpolation algorithms and methodologies have been incorporated into several different modeling platforms, all intended to facilitate the process of predicting these unknowns; however the highly heterogeneous, anisotropic, and non-stationary nature of the object being sampled creates a source of uncertainty that cannot be easily predicted. Too often, the resultant model is awarded a level of confidence that is not justified.

This paper is intended to represent gross error in prediction that can result when a single algorithm is used for subsurface interpretation. This paper will discount the theory that a single superior interpolation algorithm exists for all datasets and will highlight the misconceptions attributed to honoring data points. Examples will be provided that will highlight the fundamentals of the non-uniqueness problem and offer a method to address the absence of expert opinion in existing numerical solutions that are currently on the market.

Overshadowing the significance of recognizing and presenting uncertainty in geologic interpretations can result in staggering consequences for the end user. Without adequate understanding of the uncertainty of each prediction, the probability of misdirecting project goals is unacceptable.