Data Analysis Services Group - October 2013

News and Accomplishments

VAPOR Project

Project information is available at: http://www.vapor.ucar.edu

TG GIG PY6 Award:

XD Vis Award:

The final project report was written and submitted to the TACC.

KISTI Award:

John visited KISTI collaborators in Seoul and attended the Korean Supercomputing Conference (KSC). He gave a VAPOR tutorial to to a group from the Korean Meteorological Center, and a talk on VAPOR at KSC. The Koreans indicated their intent to continue funding the VAPOR project for the foreseeable future. Another visitor from KISTI will most likely arrive in the Spring.

  • Scott continued work on the vdcwizard data translation GUI:
  • Added ability to create a new .vdf file (not just overwrite an existing one).
  • Creating a warning prompt if the user is about to overwrite pre-existing VDC content.
  • Added option for user to exit VDCWizard upon task completion, or remain within the application for further data conversions.
  • Added WRF support (still in progress).
  • Removed superfluous user options (such as refinement level).
  • Implemented .xpm images for logos/icons, removed compiler warnings
  • Implemented progress bar during VDC file creation
  • Added single timestep/variable processing to Copy2VDF.cpp and dataholder.cpp
  • Removed start-time and number-of-timesteps options from vdfCreate page
  • Fixed memory leak happening during mom/pop/roms VDC creation.  (Still working on wrf memory leak)
  • Forced .vdf file extension on user specified file name (since Vapor only reads VDF files with this extension)

Scott also got acquainted with the perl-based VAPOR regression suite, and made some minor usability improvements to the command line translation utilities.

Finally, Scott started evaluating the ESMF re-regridding tools as a possible replacement for our own simplistic and fragile re-gridding code. In addition to finding the ESMF code highly challenging to build, thus far the performance of the ESMF code is an order of magnitude slower than our own. These results are preliminary and further investigation is warranted.

2.3 Development:

We have started work on the vapor 2.3 release in earnest, with a planned RC0 drop targeted for December 15, the end of the KISTI Year Three PoP.  Alan has started work on several planned features, implementing a discrete color mapping and a transfer function in the flow panel.  He has also been fixing bugs in key-framing.

  • John rebuilt VAPOR's third-party library dependencies with correct path names on the new glade system.
  • A number of minor bugs were fixed in the vdf and IDL libraries
  • Command line options for the various model data command line conversion utilities were made consistent with one another.
  • A simplified level-of-detail and refinement control interface was designed to help users get started with the VDC2 data format.
  • John continued testing the new vdcwizard GUI
  • Miles started adding support in vaporgui for the version checking software he prototyped earlier.

3.x Development:

Alan continued to develop prototype code for Vapor 3.0, based on the functionality we have been discussing in our 3.0 design meetings.  The initial prototype will only support the most basic functionality, such as rendering barbs. Alan has implemented simplified Params and Render libraries to support this functionality, removing all Qt dependence.  This prototype now includes a working barbs renderer.  This work has been put on hold while we work on the 2.3.0 release.

Yannick Polius, Scott's predecessor, has volunteered his time to help out with various rendering related tasks. John and Yannick have been reviewing literature and making plans to prototype a GPU implementation of a ray caster that would support curvilinear and stretched grids (stretched grids are currently supported but the rendering quality can be quite poor if the grids are highly anisotropic).

Miles produced prototype code that demonstrated a method for cleanly interrupting a long running task under Qt.

Administrative:

Education and Outreach:

Scott gave a Vapor demonstration to Andreas Muschinski and his team at the Northwest Research Associates.  They are currently showing interest in how to use Vapor for their research, most importantly a 8192^3 DNS turbulence model; currently the largest turbulence model ever produced.

Scott and John attended IEEE Vis Week in Atlanta.

Software Research Projects

Feature Tracking:

Climate data compression:

Production Visualization Services & Consulting

  • Peter Johnsen is in the process of creating an even higher-resolution simulation of hurricane Sandy on the Blue Waters Cray machine (100m resolution) that we expect to visualize.
  • Ryan Maue is creating an improved simulation of ERICA, a severe Atlantic storm from 1989, that we expect to visualize.
  • Kelly Gaither, chair of the SC2013 Video show, did not include our Sandy animation due to a defect in the show submission review software.  She instead will include the video in several high-profile events such as at the SC2013 Keynote and the Walk-in videos.

ASD Support

John worked with Peter Ireland to address reviewer's comments for their SC13 Visualization Theater submission. John and Peter also reworked their animation and submitted to the 2013 APS DFD Gallery of Fluid Motion.

Publications, Papers & Presentations

  • xxx

Systems Projects

Data Services

  • xxx

Accounting & Statistics

  • xxx

Security & Administration

  • xxx

System Monitoring

  • xxx

System Support

ML - Data Analysis & Visualization Clusters

  • xxx

GLADE Storage Cluster

  • xxx

Data Transfer Cluster

  • xxx

Experimental Clusters

  • xxx

Test Clusters

Storage Usage Statistics

dasg:NWSC+GLADE+Usage+Report

Other

  • xxx
  • No labels