GSA Annual Meeting in Seattle, Washington, USA - 2017

Paper No. 87-21
Presentation Time: 9:00 AM-5:30 PM


MCCARTHY, Andrew L., Geosciences Department, Colorado State University - Warner College of Natural Resources, 1482 Campus Delivery, Fort Collins, CO 80523-1482,

Stochastic simulation of sedimentary facies requires a modeler to navigate multiple scales of heterogeneity and variable stratigraphic surface and bedding plane orientations. As outcrop studies produce more statistically-based bed-scale descriptions (e.g., thickness and length-scale distributions, transition probabilities, stacking patterns and orientations, etc.), incorporating this information into facies models to generate geologically-realistic outcomes remains challenging. This study 1) presents bed-scale statistics from Late Cretaceous Horseshoe Canyon Formation point bar deposits that outcrop in southeastern Alberta, and 2) incorporates the statistics into a geologically-realistic stochastic facies simulation. Grain size, sedimentary structures, and bedding characteristics were documented in 40 stratigraphic sections, and provide the basis for facies classification. Internal point bar architecture was mapped with dGPS surveys; these surfaces provide an essential spatial framework for facies architecture statistics. Vertical and horizontal facies proportions and transition probabilities were calculated from measured sections, and characterize the probability of encountering a facies at a specific position in a point bar deposit. These data were transformed into facies probability cubes to guide stochastic interpolation of facies between known measured section data within bounding surfaces along bedding planes. Plurigaussian simulation was compared to a hierarchical truncated gaussian workflow, demonstrating both statistical fidelity of this technique as well as qualitative reproduction of outcrop geology. These algorithms employ variograms (spatial length-scale characterizations) to interpolate between outcrop data, but can also consider the probability cubes and facies transition statistics. Conditioning simulations to outcrop statistics in this fashion yields more geologically-sound realizations than those generated without such conditioning. This method strongly imparts outcrop analogue character on final simulation results, and facies architecture and bounding surface modeling choices and assumptions are the strongest controls in this regard. Comparative study of workflows addresses the range of uncertainty for these controls.