Visualization of geological uncertainties along cartesian and stratigraphic grids
|Location||International Geological Congress,oslo 2008|
|Author||Viard, Thomas۱; Caumon, Guillaume۱; Levy, Bruno۲; Royer, Jean-Jacques۳|
|Holding Date||15 September 2008|
The characterization of a geological domain is a key step that conditions many studies in the Earth Sciences. However, very few data are usually available for subsurface description - typically, the ratio between the volume explored and the total volume is ~10-8. A correct assessment of the associated uncertainties is therefore critical. Uncertainty modeling generally involves multiple stochastic simulations. These simulations generate randomly a set of possible subsurface models from the data distribution and one or several variability models. Although many works address the quantitative use of realizations to make predictions (e.g. through flow simulation and/or using response surfaces), research on the visual perception of spatial uncertainty is still an emerging research field.
We propose several real-time 3D visualization tools dedicated to the display of spatial uncertainties. These tools are applied along sections of Cartesian and stratigraphic grids. Local uncertainties are quantified using classical statistical methods, i.e. standard deviation, residuals, probability density functions, percentiles and intercentile ranges. Our visualization methods use uncertainty isoline display, illumination, transparency, and geometry addition through two-dimensional histogram bars. All of these features can be interactively modified. These visualization tools incorporate both the data of interest and their associated uncertainties, and aims at minimizing interference between the two pieces of information.
This makes it possible to discriminate very quickly between reliable and unreliable model regions. Such tools are useful to apprehend subsurface uncertainties, thus be aware of the models limits before acquiring new data. The proposed techniques, implemented in the Gocad geomodeling package, are illustrated on a complex geological environment. Both the fitness-for-purpose and the general validity of our visualization tools are discussed using quality criteria from the perception community.