Visualization of gridded scalar data with uncertainty in geosciences

Characterization of the earth's subsurface involves the construction of 3D models from sparse data and so leads to simulation results that involve some degree of uncertainty. This uncertainty is often neglected in the subsequent visualization, due to the fact that no established methods or available software exist. We describe a visualization method to render scalar fields with a probability density function at each data point. We render these data as isosurfaces and make use of a colour scheme, which intuitively gives the viewer an idea of which parts of the surface are more reliable than others. We further show how to extract an envelope that indicates within which volume the isosurface will lie with a certain confidence, and augment the isosurfaces with additional geometry in order to show this information. The resulting visualization is easy and intuitive to understand and is suitable for rendering multiple distinguishable isosurfaces at a time. It can moreover be easily used together with other visualized objects, such as the geological context. Finally we show how we have integrated this into a visualization pipeline that is based on the Visualization Toolkit (VTK) and the open source scenegraph OpenSG, allowing us to render the results on a desktop and in different kinds of virtual environments. (C) 2010 Elsevier Ltd. All rights reserved.