T140 Complexity in Modeling: How Much Is Too Much?

Monday, 10 October 2011
1:30 PM - 5:30 PM
Room 101DE

MANUAL AND SEMI AUTOMATED 3D VOXEL MODELING OF COMPLEX GEOLOGY USING GEOSCENE3D

Paper number: T140-4
Program: Technical Sessions
Time: 2:20 PM – 2:35 PM

Authors

Abstract

Hydrogeologic models often consists of a fixed number of continuous stacked layers also known as a layer cake model. The tops and bottoms of these layers are defined by regular grid surfaces interpolated from points defined by the modeling geologist. This type of model is relatively simple to build and easy to comprehend. However, there are some serious limitations. Layers are not allowed to overlap vertically and in most case the number of layers are limited. Because of the limitations, modeling complex geology using this approach often result in loss of detail.

A model build from three dimensional grids consisting of equally sized 3D grid cells also known as voxels do not have the limitations of the layer cake model. Each voxel is assigned a lithology or other type of property. Geologic structures are modeled using volumes consisting of neighboring voxels with identical properties. Volumes representing geologic structures can range in size from a single voxel to an unlimited number of voxels. Each voxel can be assigned a virtually unlimited number of different lithologies or other type of property.

Multiple software packages offer automated voxel modeling using a statistical approach. Automated modeling is fast but the intuition and experience of the modeling geologist may be lost to some degree. In GeoScene3D voxel models can be build manually or semi automated allowing the geologist to use his/her intuition and experience directly.

Voxel modeling in GeoScene3D requires spatial selection of voxels and then assigning a property to the selected voxels. GeoScene3D offers multiple tools for manual and semi automated selection of voxels. It is possible to select voxels in one 3D grid and then use the selection to assign geologic properties in another or multiple other 3D grids. As an example the selection may be done in a 3D grid representing a geophysical or geochemical property. The selection is specified by a spatial geometry and/or a value interval. Isolated volumes of neighboring voxels within a specified value range can be selected by pointing out a seed voxel.

The presentation will present some of the ideas behind this approach.