GEOSCAN Search Results: Fastlink


TitleHighlight Box 15-2. Evaluating uncertainty in the Oak Ridges Moraine Geological Model
AuthorLogan, C E; Russell, H A J; Sharpe, D R
Alt SeriesNatural Resources Canada, Contribution Series 20170325
Mediapaper; on-line; digital
ProgramGroundwater Geoscience, Aquifer Assessment & support to mapping
AbstractThere are numerous aspects of error propagation in numeric modelling (e.g., Beven 2016). Evaluation of uncertainty in geological modelling has often focused on reducing and assessing the aleatory uncertainty in interpolation. Reliable numeric geological models, however, have defensible conceptual geological models. To account for various components of epistemic uncertainty, the Oak Ridges Moraine study emphasized ground-truth data collection (e.g., mapping, drilling and geophysical investigations) to define subsurface stratigraphic architecture to test the event-stratigraphic model (e.g., Sharpe et al. 2002). A numeric GIS model was then created using an ¿Expert System¿ approach based on this stratigraphic model. Primary geological and geophysical data were manually coded (training data) and then used within a rules-based system to constrain and adjust the stratigraphic classification of abundant low-quality archival data, such as water well records (Fig. 1) (e.g., Logan et al. 2006). To estimate model uncertainty, a hybrid approach was implemented that estimated the accuracy of any grid cell value of an interpolated stratigraphic surface, related to the proximity and relative quality of supporting data points. Confidence grids were produced for each stratigraphic surface to provide an estimation of potential errors.
Summary(Plain Language Summary, not published)
Short 500 word summary of uncertainty estimation in the Oak Ridges Moraine Geological Model.