Visualization Showcase

Download a comprehensive PDF of the Visualization Showcase.

The Visualization Showcase provides a digital gallery of the powerful, evocative imagery associated with the TeraGrid's most exciting and compelling results. Featured projects have used TeraGrid resources to generate data, to produce the visualization, or both, and are the result of work accomplished within the past year.

Greg Foss, Jacobo Bielak, Ricardo Taborda and Haydar Karaoglu. Chino Hills- A Highly Computationally Efficient Earthquake Simulation
Abstract: The Quake Group of Carnegie Mellon performed the Chino Hills simulation on Kraken at the National Institute for Computational Sciences. The run was completed in 17 hours 45 minutes using 4096 processors with Hercules, the finite element parallel software they developed for earthquake simulations.

Amit Chourasia, Emmett Mcquinn, Bernard Minster and Jurgen Schulze. GlyphSea
Abstract: Vector visualization is an important area to investigate scientific data in many domains. Prior visualization strategies have primarily focused on maps and volumetric rendering of scalar fields, which do not permit a full characterization of underlying phenomena. We have developed an interactive glyph visualization application that allows scientists to explore vector fields. We implement novel technique of procedural dipole and cross mark texturing to encode and display vector data. We employ a novel lattice method to show neighborhood, which also enables to distinguish glyphs. We further enhance visualization by using screen space ambient occlusion, jitter, outline halos and displacement. By combining contextual information with glyphs, such as geographic maps and representations of crustal structure with three-dimensional isosurfaces, scientists are able examine vector fields in a rich and interactive multivariate environment. We demonstrate the results for time-dependent vector data from a three different simulations including the "Big One" earthquake simulation affecting the Los Angeles region.
These new visualizations clearly depict the obvious features as well those missed in previous investigations.

Amit Chourasia and Renyue Cen. Visualization of Tiger Simulation
Abstract: Cosmological Reionization simulation was performed based on the Tiger hydrodynamic code of the Princeton Cosmology Group. The visualizations shows the density distribution of cosmic gas as time evolves in a 100 megaparsec box. The filamentary structure, the cosmic "web", is a result of gravitational collapse of small fluctuations implanted during the very early inflationary epoch.
The simulation was carried jointly at NCSA and NASA Columbia system and the visualization was carried out on Cobalt at NCSA.

Peter Kristof, Jin Ryong Kim, Bedrich Benes, Lan Zhao and Carol X. Song. Large-scale volumetric visualization of Doppler reflectivity data
Abstract: The super resolution NEXRAD (Next-Generation Radar) Level II data, aka Weather Surveillance Doppler Radar 1988 (WSR-88D) data, provides critical information on reflectivity, wind velocity and spectrum width for the entire United States. The goal of this work is to develop a framework that enables multiple users to interactively access, analyze and visualize the Doppler reflectivity data in 3D to study near real-time weather events. To provide interactive high-quality volumetric weather visualization, we combined two approaches dealing with large-scale storage of global weather data and out-of-core volume rendering using CUDA ray casting. The results of our work show that the reflectivity data from multiple radars can be preprocessed into a data format that is efficient for large-scale volumetric visualization in near-real time and requires minimal run-time processing.

Micheal Norman, Robert Harkness, Pascal Paschos, Rick Wagner, Mark Hereld, Joseph Insley, Eric Olson and Michael Papka. Searching for Baryon Acoustic Oscillations in Intergalactic Absorption, Project StarGate Application Driver
Abstract: This simulation follows the growth of density perturbations in both gas and dark matter components in a volume 1 billion light years on a side beginning shortly after the Big Bang and evolved to half the present age of the universe. It calculates the gravitational clumping of intergalactic gas and dark matter modeled using a computational grid of 64 billion cells and 64 billion dark matter particles. The simulation uses a computational grid of 40963 cells and took over 4,000,000 CPU hours to complete.
The visualization shows the density value, whose range spans over 6 orders of magnitude. The animation of the evolution was made from 148 terabytes of data generated on Kraken at the National Institute for Computational Sciences and transferred to Argonne over ESnet. The rendering was performed on Eureka at the Argonne Leadership Computing Facility.

Gregory P. Johnson. Ozone Concentrations in Eastern Texas Due to Emissions in Houston
Abstract: The visualization shows simulations of ozone concentrations over eastern Texas. These simulations utilize various sources of data including satellite observations and land-based measurements. The first case shows simulated ozone concentrations over the state considering all contributions. The second case shows only ozone concentrations due to emissions in Houston. The visualization shows that emissions in Houston cause ozone concentrations to be elevated in other parts of the state, particularly in central and north Texas. It is thus apparent that emissions in one city can affect ozone levels in an entire region.

Micheal Norman, Robert Harkness, Pascal Paschos, Rick Wagner, Mark Hereld, Joseph Insley, Eric Olson and Michael Papka. Searching for Baryon Acoustic Oscillations in Intergalactic Absorption, The Expanding Universe
Abstract: This simulation follows the growth of density perturbations in both gas and dark matter components in a volume 1 billion light years on a side beginning shortly after the Big Bang and evolved to half the present age of the universe. It calculates the gravitational clumping of intergalactic gas and dark matter modeled using a computational grid of 64 billion cells and 64 billion dark matter particles. The simulation uses a computational grid of 4096^3 cells and took over 4,000,000 CPU hours to complete.
The visualization shows the density value, whose range spans over 6 orders of magnitude. The animation illustrates the expansion of the universe over time, and highlights how individual structures (such as galaxy clusters) collapse due to gravity, while simultaneously being pushed further and further from each other. The animation was made from 148 terabytes of data generated on Kraken at the National Institute for Computational Sciences and transferred to Argonne over ESnet. The rendering was performed on Eureka at the Argonne Leadership Computing Facility.

Maria Esteva, Suyog D Jain and Weijia Xu. Visualization of Large Electronic Records Collection
Abstract: The National Archives and Records Administration (NARA) is charged with ensuring continuous access to government records. Digital archival collections are diverse in nature, presenting multiple media types organized in diverse arrangements that serve the purposes of the many authors, software, and hardware involved in their creation. To preserve and provide access to electronic records collections, archivists need first to conduct a series of analysis with purposes of discovering their structure and content, and to make decisions about their long-term preservation needs. Our research examines information visualization for archival analysis and long-term preservation planning of terabyte size collections.
Using structural and technical metadata extracted from the Transcontinental Persistent Archives Prototype at NARA as well as file format categorizations and file format risk scorings, we created a visualization that uses relational database management systems (RDBMs), alignment algorithms, WordNet, and treemap to enable interactive visual analysis of large and heterogeneous electronic records collections. In this representation, each square is a directory that can be shown at different hierarchical levels in which level one is a Records group and the subsequent levels show how the directories and files within are organized in series and sub-series bearing diverse file formats and arrangements. As shown in the video that illustrates this presentation, user selected views present the properties of the data (file types, number of files, size of files, preservation risk), and the way in which the different directories are organized at different levels of aggregation and abstraction, which are represented by color ranges. The visualization allows transforming data belonging to126 Records Groups from different Federal Agencies comprised by 3,067,000 files and totaling 4.5 terabytes into information that allows learning about the collection and making long-term preservation decisions.

Brandt Westing. Web Based Visualization using TACC's Longhorn Visualization Portal and Envision
Abstract: The Longhorn Visualization Portal provides an easy to use and approachable interface for researchers to use high performance resources. In addition, it provides a semi-guided process to visualize scientific data. Researchers can work from their offices over the web from anywhere with an internet connection and interactively visualize data.
EnVision is the visualization software that allows researchers to develop interactive visualizations intuitively. It handles data importation and supports visualization techniques that users expect such as volume rendering, isosurfacing, cutting planes, contour slices, and plotting capabilities.

Greg Foss, Ming Xue, Jili Dong, Christopher Phillips and David O'Neal. Selections from Violent Storm Simulation Animations shown at the workshop Data Visualization: Taking the Presentation of Methods and Results to the Next Level, Imiloa Astronomy Center, Hilo, Hawai'i, 2009
Abstract: Shown will be a series of five or more animations generated from numerical data produced from two storm simulations run with ARPS (Advanced Regional Prediction System) by the Center for Analysis and Prediction of Storms: a supercell thunderstorm spawning a tornado, and a hurricane initialized with data from hurricane Ike. This is a selection from fifteen animations created for Kelvin Droegemeier's (Vice President for Research, U of Oklahoma) talk at the workshop Data Visualization: Taking the Presentation of Methods and Results to the Next Level. The animations were originally projected in 3D stereo, some at the front and others stretched overhead, in the planetarium at the Imiloa Astronomy Center in Hilo, Hawai'i, November 2009.

Contact Visualization Showcase Chair Eric Wernert (IU).