ASD Viz

SCEC has been awarded an Accelerated Scientific Discovery (ASD) allocation from the NCAR Computatioinal and Information Systems Laboratory (CISL) for research on NCAR Wyoming Supercomputer Center (NCSW). We held a telecon on 4 May with CISL vis to discuss visualization targets for October time frame using CISL resources.

Visualization Examples from SCEC/Princeton

We provided some example of existing research visualizations including the following:

• SCEC Visualization Projects
• Point source simulation of Japan earthquake: ShakeMovie Site
• SDSC M8 vis examples
• Powerpoint presentation from En-Jui Lee and Po Chen 3D describing their 3D tomography showing comparisons of models before and after improvements: F3DT
• Powerpoint presentation from Patrick Small showing difference plots for regional 3D models Difference Plots

Visualization Examples from NCAR

NCAR Visualizations (VETS) web site showing examples of NCAR visualizations

• NCAR Viz

Example SCEC 4D Data Set

• IEEE Viz Contest Data Set

Questions from NCAR ICLS (Responses from SCEC/Princeton)

• Can you provide suggestions on what would be of interest to visualize, and reasonable to do given the time frame, for NWSC grand opening?
  • We discussed several examples of existing visualizations including ones posted on this page

• Do you have representative visualizations already that you could share? What tools were used?
  • We will identify a small number of visualization targets that we think can be generated on NCAR system in before opening in October for possible use during that process.

• Can we have early access to sample/representative data sets?
  • The IEEE vis data set posted above is an example of time-varying volume data with vector values at each grid point produced on a regular grid by our simulations. We will provide examples of other data sets as needed once we have selected specific development targets.

• Description of model outputs: variables, file format, grid type and resolution, size.
  • Typically we generate large, binary files for 4D data sets. 3D datasets (like 3D velocity models) can be generated in netCDF.

• What, if any, post-processing will be required for viz?
  • This depends heavily on the type of visualizations we select for near term target. Leading-edge work for us is vector visualization of time-varying volumetric data such as the GlyphSea work from SDSC. However, this may be too advanced

• Who on your team should we work with directly?
  • The Science leads on the project include Tom Jordan, Jeroen Tromp, Kim Olsen, Ebru Bozdag, Po Chen. The IT group includes Yifeng Cui, Phil Maechling, Scott Callaghan, Patrick Small. Phil Maechling will help coordinate so the right groups will participate.

Participants

• Tim Scheitlin - NCAR
• Dan Nagle - NCAR
• Thomas H. Jordan - SCEC/USC
• Kim Olsen - SDSU
• Yifeng Cui - SDSC
• Ebru Bozdag - Princeton
• Philip Maechling - SCEC/USC

Meeting Outcomes

• SCEC/Princeton will provide additional example of visualizations including (1) examples of how we compare starting model and improved model, and (2) how we calculate differences and display difference plots.
• SCEC/Princeton will have an internal discussion about reasonable visualization targets for possible in the October time frame from data generated on Janus, Jellystone, or Yellowstone.
• NCAR will look at examples provided and identify any examples that fit their particular expertise and experience.
• We will try to select a small, but useful, visualization target that our groups can collaborate. Hoped for scope would be no more than 1 week work by either group. We'll use this target as a way to evaluate the issues we face bridging solid earth and atmosphere science and IT groups.
• SCEC/Princeton and NCAR will have a follow-up telecon before end of May as the next step in this collaborative viz development.

Frechet Kernel Images

Fig 1: 3D map view of Frechet Kernel in Los Angeles (Credit: )

Fig 2: Side view of Frechet Kernel in Los Angeles showing receiver (surface left) and earthquake hypocenter (middle right), Vs2500 isosurface (mesh), and banana doughnut frechet Kernel (yellow) (Credit: )