EVOLVING NEEDS FOR GEOLOGIC MAPPING USED IN ENGINEERING RISK AND RELIABILITY PROJECTS

A geologic map is an artistic representation of one interpretation of geologic features and relationships inferred from limited observations of the distribution of rock types and structures, often without supporting subsurface data or laboratory test results. Traditional geologic maps tend to be used by engineers in one of three ways: 1) unit boundaries are accepted as deterministic truth, 2) the overall range of geologic conditions is estimated and applied to a site as a single random variable, or 3) the mapping is ignored and quantitative subsurface geotechnical data are used as a surrogate for stratigraphy. Engineers characterizing sites for projects that require rigorous risk and reliability determinations must apply a theoretical framework that addresses uncertainty explicitly. Reliability-based design is formulated in probability theory, but characterizing geologic material variability remains a challenge in geotechnical practice.

Sources of uncertainty in geologic mapping have been addressed to some degree in the context of error in digital elevation models and global positioning satellite receivers. However, it appears that uncertainty associated with formation contact confidence and display of "certain, concealed, and inferred" contacts and faults is not yet quantified, not to mention gradational contacts and facies changes. Even if such uncertainty were quantified, it is unclear how it would be used. The Outcrop Confidence Level described in the Rock Material Field Classification System (National Engineering Handbook, Natural Resources Conservation Service) addresses the relative measure of predictability or homogeneity of the structural domain and lithology of rock units over the site of investigation and appears to be a useful starting point in quantifying uncertainty in geologic maps suitable for use in reliability-based design. Geotechnical engineers are struggling to characterize subsurface conditions using sophisticated approaches (i.e., first-order reliability methods or Monte Carlo simulations) to satisfy code requirements. Geologists must transform geologic maps to support reliability-based design or risk being marginalized by computer scientists and statisticians who will translate the maps using random field theory without knowledge of geologic principles.