This is an archival copy of the Visualization Group's web page 1998 to 2017. For current information, please vist our group's new web page.

Table of Contents



  • Leadership and Project Management. Bethel, who is the VACET coordinating PI and overall lead investigator, has overseen "start-up" activities in this first year of VACET operations. VACET's mission is to deliver production quality, petascale capable visual data analysis software to the DOE science community for use on DOE's Open Computing facilities. The most important first-year accomplishment is establishing scientific stakeholder projects: VACET presently has about a dozen such projects that form the basis for all research, development and deployment work. This organizational model gracefully accommodates opportunity for VACET team members to assume leadership duties over individual projects as well as long-term science stakeholder management. This first year of operations has been a phenomenal success:
    • Publications: approximately 20 journal articles in the visualzation and graphics field's premier forums.
    • VACET's parallel-capable software infrastructure, which includes features desired by science stakeholders, deployed at NERSC/LBNL and CCS/ORNL for use by a broader science community.
    • Active outreach, including a strong presence at the SciDAC 2007 meeting in Boston, contributions to the SciDAC Review magazine and ASCR Discovery website.
    • Project website ( for public consumption and wiki ( for internal use.
    • Activities coordinated with other SciDAC Centers, Institutes and Science Applications have produced meaningful contributions and measureable positive impact for the SciDAC program as a whole: top quality publications, production-quality visual analysis software infrastructure deployed at DOE's Open Computing facilities, leadership of meaningful and impactful visual data analysis activities in DOE.

  • Production Quality AMR Visualization. One of VACET's science stakeholders is the Applied Partial Differential Equations Center (APDEC). Historially, APDEC has internally developed and maintained its own AMR-capable visual data analysis tool. During the past year, VACET has worked closely with APDEC to identify and prioritize critical and desireable features that would allow APDEC to exit the visualization software development and maintenance business and adopt use of a production quality, parallel capable, externally supported and maintained visual data analysis tool. VACET has performed the research and development of many of these critical features released the software at DOE's Open Computing facilities. APDEC has adopted this new software infrastructure for many of its visual data analysis activities, with a full project-wide transition expected in the next year as additional critical features come online.

    • G.H. Weber, V. Beckner, H. Childs, T. Ligocki, M. Miller, B. van Straalen, E.W. Bethel. "Visualization Tools for Adaptive Mesh Refinement Data." In: W. Benger, R. Heinzel, W. Kapferer, W. Schoor, M. Tyagi, S. Venkataraman, G.H. Weber, eds., Proceedings of the 4th High End Visualization Workshop (Tyrol Austria, June 18-22, 2007). ISBN 978-3-86541-216-4, Lehmanns Media, pp. 12-25, 2007. LBNL-62954.

    VisIt is a production quality, parallel capable AMR visualization tool that runs on all modern computational platforms, including IBM's BG/L, Cray XT3/4, and Linux clusters. This example shows direct volume rendering of AMR data, with an x/y plot comparing two subsets taken from the 3D data. This full-featured software is in use by multiple DOE science projects that use AMR-based solvers.

  • Supernova Spectrum Synthesis Visual Data Analysis Another of VACET's stakeholders is the Computational Astrophysics Consortium SciDAC Science Application. VACET is the primary source of visual data analysis infrastructure for this SciDAC project. Their new primary new code for modeling supernovae explosions uses an AMR-based solver; VACET's work for APDEC is being leveraged to meet AMR visualization needs in this science area, and extended where needed (see below, 3D Line-Integral Convolution Vector Field Visual Data Exploration). Related, LBNL VACET staff are providing the research and development needed to perform visual data analysis of large collections of supernova spectra, which will be used to compare simulation output with observed phenomena. Startup activities for this latter activity include forming the VACET team for this stakeholder project, needs assessment and prioritization, research and development of initial prototype, interactions and iterations with the stakeholder team refine the design and implementation.

  • 3D Line-Integral Convolution Vector Field Visual Data Exploration Several VACET customers have requested "better" vector field visualization technology. In response, we have performed the applied research to evaluate if a technique called "line-integral convolution" (LIC) is suitable for use and implementation in a general-purpose, parallel capable visual data analysis software package. The initial results (image below) show a 3D LIC representation mapped onto an isosurface manifold, which is the specific stakeholder request (from astrophysics and combustion stakeholders). This project is still work-in-progress. When completed, it will be implemented in VisIt and made available to the broader scientific community. This is a good example of the strong interplay between VACET's efforts in research and development, software engineering, and production deployment.

    Here, a 3D line-integral convolution field is first computed from a velocity field from a CFD simulation, then mapped onto an isosurface manifold. This novel implementation uses hardware acceleration for rendering, so is very fast and is applicable to all geometry-based visualization techniques (slices, manifolds, glyphs, and some forms of direct volume rendering).

    Note: this image is under embargo for "public showing" because it is unpublished work.

  • NERSC Analytics

    Statistical modeling of high-dimensional, time-varying, non-stationary time series enables the discovery of temporal, spatial, and multivariate statistical dependencies in large-scale climate simulations. Using latent variable models of tropical storm trajectories, we can infer the statistical relationships between space-time-varying atmospheric variables along a storm trajectory and the likelihood of the storm evolving into an intense hurricane. Hierarchical, stochastic models can predict the influence of global-scale, long-term climate patterns on such short-term, local events.
    Statistical machine learning techniques enable automatic spectra classification. This capability helps eliminate manual classification steps and increases scientific insight.