Alex Pang, Suresh Lodha, and Craig Wittenbrink

Baskin Center for Computer Engineering and Information Sciences
University of California
Santa Cruz, CA 95064


Alex Pang, CIS Board, University of California, Santa Cruz, CA 95064

voice: (408) 459-2712, fax: (408) 459-4829, email:



Virtual Environments.


Visualization, Uncertainty, Data quality, Confidence, Error, Accuracy, ...


The primary objective of this research is to design new and improved methods of visualizing uncertainty in scientific data. A secondary objective of this research is to identify and calculate uncertainty from data and algorithms. Uncertainty encompasses several concepts of the data including error, accuracy, noise, data quality, data validity, and confidence level. The definition and derivation of uncertainty depends upon sources including instrument acquisition, interpolation techniques, and visualization. Almost all scientific data contain uncertainty, which are as important to the proper interpretation of the results as the original data. However, very few visualization techniques depict uncertainty.

The researchers hope to improve data comprehension by focusing on methods for viewing data together with uncertainty information. The challenge lies in incorporating uncertainty features into scientific data displays without overly increasing the visual complexity. There are two approaches to combining uncertainty into a visualization: mapping uncertainty information as an additional piece of data (as has been done in the past), and creating new visualization primitives and abstractions. For the first method, the researchers will develop untried mapping techniques including bump mapping, texture mapping, contours of uncertainty, and surface probes (an uncertainty indicator embedded throughout a surface). For the second method, the researchers will create visualization primitives that cannot be visually separated in interpretation, a data modifier. Many uncertainties cannot use a modifier approach, but, for several types of uncertainty, this novel approach is clearly superior. The modifier encodes the uncertainty in the primitives that are modified, for example spatial displacement indicates spatial uncertainty. Modifier approaches include fractal surfaces and new types of symbols that encode information. Data from three main categories will be used: measured, calculated, and modelled. The researchers will use experimentation, metrics, and application expert's evaluations to find the right combination of techniques.

Visualization is a valuable tool in understanding large amounts of data and the phenomena represented by these data. Complete specification of data should include uncertainty. The proposed work will address this often neglected aspect of data visualization for environmental and graphics synthesis applications. The resulting techniques should significantly improve all visualization and graphics applications where data uncertainty is a concern. Integrated data and uncertainty visualizations will make more accurate interpretations and improved decisions possible for many disciplines.

We see several potential avenues for research and applications of the methods developed for visualizing uncertainty. Some of these include motion analyses tools to better understand the differences between human and simulated movements, and visualization tools to help scientists integrate model data versus measured data in a continuous process as found in data assimilation research. We are exploring this search space by initially focussing on three problems: design and evaluation of uncertainty glyphs for presenting environmental data; development and classification of different methods for comparing fixed 3D surface attributes; and visual comparison of different interpolated scattered data fields.


Alex Pang and Adam Freeman, "Methods for Comparing 3D Surface Attributes", to appear in SPIE Proceedings on Visual Data Exploration and Analysis III, 1996.

Suresh Lodha, Alex Pang, Bob Sheehan and Craig Wittenbrink, "Visual Comparison of Surfaces", UCSC Tecnical Report UCSC-CRL-95-46, 1995.

Craig Wittenbrink, "IFS Fractal Interpolation for 2D and 3D Visualization", To appear in IEEE Visualization `95, Atlanta, GA 1995, 1995.

Craig Wittenbrink, Alex Pang, and Suresh Lodha. "Verity Visualization: Visual Mappings", UCSC Technical Report UCSC-CRL-95-48, 1995.

Craig M. Wittenbrink, Elijah Saxon, Jeff J. Furman, Alex Pang and Suresh Lodha, "Glyphs for Visualizing Uncertainty in Environmental Vector Fields", SPIE Proceedings on Visual Data Exploration and Analysis II, Vol. 2410, 1995, pp. 87-100.

Alex Pang, Jeff Furman and Wendell Nuss, "Data Quality Issues in Visualization", SPIE Proceedings on Visual Data Exploration and Analysis I, Vol. 2178, 1994, pp. 12-23.


The visualization discipline is concerned with the transformation of numeric or abstract information into visual form to facilitate their understanding. To quote from the NSF panel report on Visualization in Scientific Computing: "Visualization offers a method for seeing the unseen. It enriches the process of scientific discovery and fosters profound and unexpected insights." The key goal of visualization is to obtain insight from the deluge of data that are commonly found in different disciplines today. Inherent in visualization methods is the innovative mapping or encoding of data into visual form so that the decoding of that representation is efficient and unambiguous for the human.

Visualization research has been driven primarily by a few key areas: Computational Fluid Dynamics -- whether it be supersonic flow over high performance aircraft or global climate simulation, and Biomedical Visualization -- driven both by higher resolution models of physical processes and newer imaging technologies. More recently, researchers have started to focus their attention on the visualization of more abstract data. A term that is emerging that captures this essence is Information Visualization. Other recent research concerns involve the interfacing of visualization systems with data bases, and visualization of very large data sets. Likewise it is also being enhanced with new technologies such as high speed communication infrastructures and more powerful computers.

Behind the glitz of visualization images, their success is measured by the amount of insight gained by the users. One of the shortcomings of most visualization work today is that it is difficult to obtain or attach a confidence level to different portions of the images. We see this as another opportunity for research within visualization which is the subject of our current investigation on visualizing uncertainty. The visualization pipeline often starts before the data is handed to the visualization practicioner. In fact, it often starts at the data acquisition stage (whether data are physically measured or mathematically modelled and simulated). Anywhere from the data acquisition stage, and intermediate data transformation stages, up to and including the final visualization stage, there is opportunity for corrupting the data. A truthful visualization must convey the effects of these steps in the final visualization.


B. McCormick, T. DeFanti, and M. Brown, "Visualization in Scientific Computing", Computer Graphics, volume 21, number 6, November 1987.

IEEE Visualization Proceedings, including symposia proceedings from Biomedical Visualization, InfoVis, and Parallel Rendering.

SPIE Proceedings on Visual Data Exploration and Analysis.

Siggraph proceedings, including course notes, Volume Visualization workshops, etc.


As visualization practitioners, one must be familiar with the application domain to understand how the data is obtained, and how to best present it for their users. Visualization caters to application areas that have difficulty analyzing their data. As such, it has been used in a wide variety of areas -- including areas where visualization tools traditionally come from such as Computer Graphics. We therefore think that there are opportunities for collaboration between visualization and all areas within ISP. Benefits could go both ways -- to the application domain, as well as to enriching the visualization suite of methods. We list some possible projects from each area in the next section.


Virtual Environments

Speech and Natural Language Understanding

Other Communication Modalities

Adaptive Human Interfaces

Usability and User-Centered Design

Intelligent Interactive Systems for Persons with Disabilities