Regional Hydrostratigraphic Unit Modelling Using Directional MCP Simulation
N. Benoit1,2, D. Marcotte2, A. Boucher3, D. D’Or4 and A. Bajc5
1Geological Survey of Canada, Québec, Canada
2École Polytechnique de Montréal, Canada
3Advanced Resources and Risk Technology, LLC, Denver, USA
4Ephesia Consult, Belgium
5Ontario Geological Survey, Sudbury, Canada
In groundwater flow modelling, geological uncertainty characterization is a key input for risk assessment. The most influential geological uncertainty is attributed to the relative proportions, properties and spatial arrangement of the hydrostratigraphic units (HSU). The directional ordering of HSU is a critical component of sedimentary environments. This feature controls, to a large extent, the response of the system under a given external stimulus. Geostatistical algorithms, such as Bayesian Maximum Entropy method (BME) and Markov-type Categorical Prediction (MCP), present interesting features to impose directional constraints in categorical data simulation. In this study, we illustrate the ability of the MCP method to allow for trends and directional ordering in simulation. The MCP approach was tested first on a simple synthetic model and then applied to a regional 3D model in the Innisfil Creek sub-watershed. The required asymmetric transition probabilities between categories for MCP were extracted by Fast Fourier Transform computation from the transitional deterministic model. This model includes 15 hydrostratigraphic units displaying lateral variability and defined vertical ordering. The MCP realizations all reproduced hydrostratigraphic units in logical arrangements and proportions. The set of realizations appear globally unbiased with variations around the deterministic model. The resulting ensemble of geological models allows for the assessment of uncertainty of groundwater flow and transport as well as aquifer vulnerability and the delineation of wellhead protection areas.
Keywords : units ordering, categorical simulation, model uncertainty
