Data Analysis Services Group - May 2014

News and Accomplishments

VAPOR Project

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

TG GIG PY6 Award:

XD Vis Award:

KISTI Award:

Scott worked on the CAM dataset importer:

My work on implementing CAM’s vertical coordinate into the VDC finished during May.  I was able to re-use the ROMS algorithms with minimal changes, which means that CAM data can be easily added to the VDCWizard during the 2.4 release. One issue I ran into during my work was with inadequate texture memory being allocated to display my test dataset.  After confirming that my elevation layers were monotonically increasing and that their calculations were sound, the memory issue was pinned down.  I believe we will be increasing the default value for the next release.

An additional issue that I encountered was during the VDC (not VDF) creation; when the conversion algorithms map the VDF time data to timestamps in the netCDF files.  I initially contacted Steve Emerson of Unidata about this because we suspected the problem resided within udunits.  However I came to find that CAM requires a higher degree of precision for its time variables than the models we are currently converting, because they are denoted as ‘days since 1979-01-01 00:00:00’.  The fact that the units are in days (not seconds) necessitates the use of a higher decimal precision in our VDF files.  The changes that I made for this fix in the NetCDFCFCollection were tested against our PERL suite; however the fix causes our test scripts to break during the VDF creation phase, as our current converters now have higher precision in our XML timestamp logs.

2.3 Development:

• Alan made changes to the isoline renderer so that users can color-map the isolines based on a color mapping of the isoline domain (as in a transfer function).
• Scott gathered requirements for the text renderer. 
• Miles did some minor restructuring, moving the time routines used through vapor into the common library.
• Miles continued work on the GLFlow class:

  • Added profiling to test_glflow

  • Added responses for most test_glflow parameters

  • test_glflow tubes all rendering properly, with stride, etc.

  • successful cone rendering

  • single, manually constructed arrow rendering

3.0 Development:

• Alan updated several 3.0 features. He has completed session load/save support. Alan also implemented a mechanism for supporting Undo/Redo of operations that cannot be performed simply by modifying the XML representation.  The mechanism for supporting undo/redo of instance creation and removal is now working.
• We are making the 3.0 Doxygen documentation easier to navigate by separating the developer API docs from other docs.
• John continued working on the WASP and VDCNetCDF API implementations:
  • The forward and inverse transforms were threaded. John notes that he explored OpenMP as an option but frustratingly discovered that it is still not supported on the Mac
  • Added support for non-block-aligned data
  • Tested reading and writing WASP files.
  • Completed first phase of integration of WASP into VDCNetCDF (WASP files can now be read/written from VDCNetCDF API)

Administrative:

John wrote annual reviews and had goal setting meetings with staff

John was on an interview panel for an open Admin 1 position in CISL

Education and Outreach:

Software Research Projects

Feature Tracking:

Climate data compression:

Production Visualization Services & Consulting

• Pete Johnsen of Cray, working with Mel Shapiro, produced 36 hours of a new 500m WRF run for ERICA IOP4 (a severe Atlantic storm in January 1989) on Blue Waters.  Alan converted and visualized several variables from this data.
• Alan generated several high resolution images of ERICA based on recent simulations by Pete Johnsen on Blue Waters.  Mel Shapiro will be using these in an upcoming NSF lecture about the value of ultra-high resolution in understanding intense storm systems.
• John started discussion with Peter Sullivan about visualizing Peter's turbulence data.

ASD Support

• xxx

Publications, Papers & Presentations

• xxx

Proposals

John and Shawn worked with UCSD on another proposal aimed extending VAPOR's PDA data model to support biomedical data. The proposal is a response to an NIH FOA, due in June. 

SCIparCS Summer Internship

• In preparation for Sunni, Alan and Scott researched how the model renderer operates and decided that Sunni would be primarily writing .vms files to correspond with observational data.  Scott discovered a field campaign called FRAPPE that happened to be closely knit with scientists that Alan has worked with before (Gabi Phister and Mary Barth).  It was true serendipity how this worked out.  Alan reached out to them and now we are working to visualize their field campaign.  Scott also spent some time creating .obj files of NCAR’s C-130 and NOAA’s BAO in Blender, which are both massive instruments being used in this field campaign.   Joey Mandoza was a great resource for questions while I was building my models.
• Sunni arrived on May 19 and is making excellent progress.  She has displayed aircraft trajectories in Google Earth and in VAPOR (using the Model renderer).
• We met with Pfister and Barth on June 2 to see how Sunni will collaborate with the FRAPPE project (which will be collecting pollution data based on C130 flight paths in Colorado) Sunni is writing Python code to map the flight path data to visualize in VAPOR. We also expect that Sunni will use the Google Earth capabilities that Mohammad developed last summer
• Alan made changes to the Model renderer so that we can attach colors to each object in a vms file.  This is useful in color-mapping the aircraft flight-path from FRAPPE.

Sunni worked on the following:

• Developed python code that reads aircraft data from ICARTT files and generates a .vms file for the VAPOR model renderer
• Additional capabilities of the code:
  • Assigns a color to the model object based on the value of the user-specified measurement variable
  • Parses VAPOR .vtf and .vdf files for projection, extent, and transfer function information
• Results in VAPOR model renderer:
  • A glider plane and a trail of spheres (representing 10 minutes of historical data), colored according to a user-specified color scale
• Generated .ncl files with Mohammed’s code from last year to verify python output files
• Worked with Alan and Scott to generate a video demo of data rendering in VAPOR.
• Video demo will be presented at meeting with Gabi and Mary on Tuesday, June 17, 2014 at FL0