Spatial Data Analytics-Assisted Subsurface Modeling: A Duvernay Case Study

Abstract

Data analytics facilitate the examination of spatial data sets by using multiple techniques to find and understand patterns to guide decision making. However, standard data analysis tools assume that the data are independent and identically distributed, an assumption that spatial data sets usually do not fulfill. Furthermore, the usual methods neglect spatial continuity and the inherent data paucity that should be considered in the data analytics workflow. We present a new approach that combines data analytics, geostatistics, and optimization techniques to provide an end-to-end workflow to analyze two-dimensional (2D) data sets. The proposed workflow identifies outliers based on their spatial location or distribution, models geological trends using a Gaussian kernel, models the semivariogram, and performs sequential Gaussian simulation applying kriging or cokriging for cosimulation. Moreover, it provides metrics and diagnostic plots to evaluate the goodness of the results at each step. It is also semiautomatic because it leverages the user’s judgment for subsequent operations. For optimization, the workflow uses Bayesian optimization and evolutionary algorithms. We demonstrate the use of the workflow by analyzing 1,152 wells over the Duvernay Formation in Canada. The examples include the simulation of density-porosity as the secondary feature and the cosimulation of total organic content constrained by the former. The proposed workflow helps focus more on interpreting the results than the modeling parameters, reducing workforce time and subjective errors. Moreover, the spatial simulation includes multiple realizations to assess uncertainty and support decision making in data paucity scenarios. Overall, the proposed workflow is a valuable and complementary tool for evaluating uncertainty in mature geospatial data.