sub header2

[View PDF]

 

Kenneth Moreland, SNL

Objective

  • The Climate Science Computational End Station (CCES) aims to predict future climates by using scenarios of anthropogenic emissions and other changes resulting from U.S. energy policy decisions.

Technology

  • NSF/DOE Community Atmosphere Model (CAM5): A 3D global model used to predict atmospheric behavior and its effect on other climate systems.
  • Demonstrated a two-year simulation at 1/8 degree global resolution with full prognostic aerosols and monthly varying sea surface temperatures derived from observations.
    • Performed on 64K cores of Intrepid (25 million total core-hours), running at 0.25 simulated years/day
  • ParaView generated Animation sifts through 916GB of results data.
Total Precipitable Water Visualization

Impact

  • Visualization clearly shows hurricanes in Atlantic and Pacific.
    • Top right: a Katrina-like hurricane makes landfall while another prepares to travel along the eastern U.S. coast.
    • Direct right: Two typhoons interacting in the Pacific.
  • Presented by Warren Washington, INCITE PI, at the 92nd American Meteorological Society Meeting.
    • “We are on the threshold of simulating the global high-resolution atmosphere circulation on decadal and century time scales and this animation demonstrates this new capability.”

Hurricanes in Atlantic and Pacific

Notes:

For this particular visualization the challenge is simply addressing the increasing scale of climate simulations and demonstrating the readiness of our capability visualization. The visualization helps identify and observe weather patterns that I expect are difficult to characterize.  But I believe the main intention of the animation is for communication, which is important in the climate community.

http://www.alcf.anl.gov/articles/new-groundbreaking-global-climate-visualization-debuts-92nd-american-meteorological-society
http://www.kitware.com/news/home/browse/Paraview?2012_02_08&ParaView+Used+in+Atmospheric+Climate+Modeling

Acknowledgements: Visualization Joseph Insley (Argonne) and Andy Bauer (Kitware); Model Setup and Integration Mark Taylor (Sandia)

Technical details

  • 27 files, each ~34G containing 720 time steps (equivalent to one hour).  Thus each file contained 30 days.
  • Total over 2 years.  Animation is abbreviated to 2 months.
  • All 27 months of data were rendered, but it is ~12 minutes of animation.  Released animation abbreviated.
  • Rendering used pvbatch serial, farmed out on 81 nodes of Eureka to render everything in 8 wall clock hours (648 CPU hours).
  • All movies available: http://www.mcs.anl.gov/research/fl/flinternal/CCESDev/

file: famip5_cam2_2004-07-30_01_960x540.mp4
Aug 5 - 10: A "katrina" like hurricane makes landfall, and another
hurricane travels along the eastern coast.

file: famip5_cam2_2005-08-24_01_960x540.mp4
Sept 2- Sept 10 (two typhoons interacting- that Warren showed)