Improving Geological Models through Statistical Integration of Borehole Data and Geologists’ Cross-Sections

摘要

When working in complex subsurface conditions, uncertainty exists due to natural variability in the rock. This is due to the nature of the rock’s deposition or emplacement, or due to processes (e.g. tectonic or metamorphic) post-emplacement. Currently, geological models created for a rock engineering project are used as primary drivers of decision-making, and rely on limited borehole data and knowledge of the regional geology. Geological variability leads to uncertainty in these models, which lowers confidence in their use. While the simplest way to increase confidence is to add boreholes, at a certain point this becomes cost prohibitive, as knowledge gained from adding a new borehole becomes small relative to the scale of the project. In this work, the authors seek to optimize the combined effect of borehole data and a geologist’s confidence in their cross-section by utilizing sequential indicator cosimulation, treating boreholes as primary data and the geologist’s cross-section as secondary data. Although the combined cosimulation of the primary and secondary data failed to produce a result that was better than geologist’s cross-sections, this project provides valuable insight into the quantification of spatial uncertainty, which could be used in future works for value of information analyses of potential additional data.