3D Visualization of Sparse Geophysical Data Representing Uncertainty

Abstract

Geophysical data are sparse and by nature difficult to analyze. Usually domain experts use "mental models" to infer missing data according to the surrounding data and their own knowledge. The main goal of this work is to explore the best way to represent uncertainty in geophysical data. Given the sparse nature of the represented data, it is important to provide a 3D volumetric representation of the whole subsoil, based on a geostatistical process. We use kriging interpolation to generate a structured grid from the original sparse data. However, the analysis of such an interpolated representation must be careful, since the uncertainty varies significantly according to the distance to real measurements. We use different representations to emphasize data uncertainty during the analysis stage. The different visualization techniques implemented in our prototype, as well as methods used to simultaneously visualize resistivity and uncertainty information, are presented.