Yet different geologists, and geologic disciplines, may yield very different maps of the same area, and legitimate professional disagreements regarding stratigraphy, lumping/splitting, structure, are routine. Problems are magnified for 3D geologic maps (3DGMs), where geology forms the foundation of quantitative process models whose output can be sensitive to small changes in the 3DGM, yet uncertainty regarding geologic features in the map increases over 2D due to lower data density, increased reliance on models with uncertain parameters, and potentially contradictory input from multiple disciplines.

Challenges occur when the societal question requires data resolution finer than the resolution of the 3DGM, when a structure is inadequately represented, or when geologic alternatives cannot be ruled out. For example, hydrologic flow modeling may require precise knowledge of the boundaries and the internal variability of the porosity and permeability of a geologic unit at depth, data which likely will never be directly available. Fault surfaces are unlikely to be the single, smooth, dipping surfaces typically modeled. More likely they are replete with splays, horses, and smaller, physically disconnected secondary faults. Such complexity cannot be mapped with certainty. Yet answers to questions involving subsurface fluid flow can be quite sensitive to these complexities. Such challenges may be approached by including stochastic elements in the 3DGM to accommodate the impact of geologic alternatives, scale, problem sensitivity, and data model on subsequent process modeling. One strategy is to choose two geologically distinct end-members (e.g. normal vs. reverse fault), construct several 3DGMs that range between the two (e.g. changing fault dip) and examine the sensitivity to the changes. Uncertain rock properties can also be included in 3DGMs using geostatistical representations. Such maps provide a spectrum of answers to a specific question, so that stakeholders may define how to act optimally, and also decide how to “hedge their bets.” An understanding of the challenges of mapping in 3D will help guide us to creating more robust and predictive 3DGMs.