3D geologic subsurface modeling within the Mackenzie Plain, Northwest Territories, Canada

Abstract: Three-dimensional (3D) models are widely used within the geosciences to provide scientists with conceptual and quantitative models of the earth’s subsurface. As a result, 3D geologic modelling is a growing field, such that scientific research often cannot keep up with technical advancements. Literature often shows conflicting results with respect to which interpolation algorithms produce the best surfaces, and it is not always clear which methods are most appropriate for a particular geological setting. This study looks at three commonly used interpolation techniques – Inverse Distance Weighting (IDW), kriging and triangulation – and assesses their effectiveness at capturing geologic structures in the subsurface. The study uses a modified Horizons method to create a solid 3D stratigraphic model of the subsurface of an area within the Mackenzie Plain, NWT, in Canada. The Horizons method involves interpolating individual stratigraphic surfaces, or horizons, representing their depositional sequence. Surface intersections are corrected where necessary, and a solid model is built by extruding each surface down to the top of the surface below. Triangulation produced the most geologically appropriate surfaces, whereas IDW produced surfaces with a stronger bullseye effect; although kriging produced some surfaces well, it did not result in acceptable surfaces where discontinuities were present. Structural features such as folds in the subsurface were captured only where the data density was sufficient. Large folds spanning the majority of the study area were visible in the modelled surfaces; however smaller folds and monoclines were not visible. It was possible to model a thrust fault in the subsurface by creating two separate stratigraphic models on either side of the fault, cutting them at the fault plane and merging them together after each side was converted to a solid model. The model produced in this study showed promise as a basis for future modelling and further 3D model refinements. The modelling process was capable of highlighting areas where surface mapping did not match up with subsurface measurements, and conversely, highlighted areas where subsurface measurements did not match with observed surface features. However, comparing the modelled results with seismic survey images in the region showed that the specific locations of the subsurface structures (e.g. fold troughs) were not captured in the correct location. As well, the presence of discontinuous surfaces made it necessary for manual edits to be performed, in order to accurately represent the known subsurface geology – complicating the model and making replication more difficult. The results presented in this thesis can be used to guide methods in other similar 3D modelling exercises. As well, it will help us apply the most appropriate methods for any given geologic setting, resulting in more accurate and efficient 3D models.