Multiresolution Visualization and Analysis of Turbulence using VAPOR

Multiresolution Visualization and Analysis of Turbulence using VAPOR Alan Norton NCAR/CISL Boulder, CO USA Turbulent Theory and Modeling: GTP Theme-of-Year Workshop February 28, 2008 This work is funded in part through a U. S. National Science Foundation, Information Technology Research program grant Alan Norton <alan@ucar. edu>

Outline • VAPOR project overview • VAPOR technical capabilities (new 1. 2 release) • Interaction techniques for understanding massive turbulence datasets – Six techniques that have been developed through scientific use of VAPOR – Visualization is a data exploration process • Lessons and future work Alan Norton <alan@ucar. edu>

VAPOR project overview • VAPOR is the Visualization and Analysis Platform for Oceanic, atmospheric and solar Research • Problem: Because of the recent growth in supercomputing performance, scientific datasets are becoming too large to interactively apply analysis and visualization resources. • Goal: Make it easier to analyze and visualize massive (Terabyte and greater) datasets – Provide interactive data access – Develop user interface customized for scientists • VAPOR is funded by NSF ITR: a collaboration with NCAR, UC Davis’ Institute for Data Analysis and Visualization, and Ohio State University’s Dept. of Computer and Information Sciences Alan Norton <alan@ucar. edu>

VAPOR Technical Approach Key components 1. Multiresolution data representation, enables interactive access: • Entire dataset available at lowered resolution • Regions of interest available at full resolution 2. Prioritize ease-of-use for scientific research 3. Integrate visualization and analysis, interactively steering analysis while reducing data handling 4. Exploit power of GPU Visual data browsing Combination of visualization with multiresolution data representation enables interactive discovery Refine Coarsen Quantitative analysis Data selection Alan Norton <alan@ucar. edu>

Principal Capabilities of VAPOR 1. 2 New features in version 1. 2 (Oct 2007) • Isosurfaces – Interactively generated using GPU • Spherical grid rendering (prototype) • Support for WRF (and terrain-following grids) Existing features: • Flow integration – Both steady and time-varying flow integration – Field line advection • Volume rendering – Interactive color/transparency editor • Interactive control of region size and data resolution • Bidirectional integration with IDL® for analysis • Data probing and contour planes – Interactive flow seed placement • Animation of time-varying data Alan Norton <alan@ucar. edu>

VAPOR data exploration examples Interactive visualization facilitates scientific discovery • Combining visualization with analysis of a vortex, in a solar hydrodynamic simulation (Mark Rast) • A ‘current roll’ in a multiterabyte MHD dataset (Pablo Mininni) • Advection of magnetic field lines in a velocity field (Pablo Mininni) • Advance of cold air mass in Georgia, April 2007 (Thara Prabhakaran) Alan Norton <alan@ucar. edu>

VAPOR’s Interaction Techniques for Understanding Massive Turbulence Datasets Interactive feedback is key to visual data understanding 1. Multiresolution data browsing – Enables interactive access to terabyte datasets 2. Visual color and opacity editing with histograms – Identify features of interest by color and opacity 3. Export/import data to/from analysis toolkit – Currently supporting IDL® 4. Use planar probe for visual flow seed placement – Local data values guide seed placement 5. Track structure evolution with field line advection – Time-evolution of structures shown by field line motion 6. Use the GPU for interactive rendering – Cartesian, Spherical, Terrain-following (WRF) grids Alan Norton <alan@ucar. edu>

Interaction Technique 1: Multiresolution data browsing Enabled by wavelet data representation • Interactively visualize full data at low resolution • Zoom in, increase resolution for detailed understanding Alan Norton <alan@ucar. edu>

Interaction Technique 2: Visual color/transparency editing Design developed with Mark Rast • Drag control points to define opacity and color mapping • Histogram used to guide placement • Continuous visual feedback in 3 D scene Alan Norton <alan@ucar. edu>

Interaction Technique 3: Export/import data to/from analysis toolkit Currently using IDL® • User specifies region to export to IDL session • IDL performs operations on specified region • Results imported as new variables in VAPOR Alan Norton <alan@ucar. edu>

Interaction Technique 4: Use planar probe for visual flow seed placement Useful to place flow seeds based on local data values • Planar probe provides cursor for precise placement in 3 D • Field lines are immediately reconstructed as seeds are specified Alan Norton <alan@ucar. edu>

Interaction Technique 5: Track structure evolution with field line advection Animates field lines in velocity field • Useful in tracking evolution of geometric structures (e. g. current sheets, flux tubes) • Based on algorithm proposed by Aake Nordlund Alan Norton <alan@ucar. edu>

Interaction Technique 6: Use the GPU for interactive data rendering • Modern GPU’s are cheap, fast, effective – GPU’s are SIMD clusters, efficiently traverse data arrays – Support for cartesian, spherical, terrain-following grids B. Brown, Solar MHD simulation T. Prabhakaran, April 2007 cold event in WRF Alan Norton <alan@ucar. edu>

VAPOR Lessons • Multiresolution methods are essential for understanding massive data sets. • Interactive analysis and visualization can indeed enable or accelerate scientific discovery • One-on-one interaction between scientists and software developers results in valuable interaction techniques • We are only beginning to exploit the power of GPU’s • Largest obstacles: – Wide diversity of data representations used in research – Data conversion effort Alan Norton <alan@ucar. edu>

VAPOR Plans • New features prioritized by the VAPOR steering committee and user input • Features under consideration include: – Mapping of variables to isosurface color/opacity – Support for 2 D data – Image-based flow visualization – Perform math operations on data – Keyframing and spin animation – Parallel data conversion on supercomputers – Wavelet data compression Send suggestions to vapor@ucar. edu Alan Norton <alan@ucar. edu>

VAPOR Availability • Version 1. 2. 2 software released in January 2008 • Runs on Linux, Irix, Windows, Mac • System requirements: – a modern (n. Vidia or ATI) graphics card (available for about $200) – ~1 GB of memory • Supported in NCAR visualization/analysis systems • Software dependencies: – IDL® http: //www. ittvis. com/ (only for interactive analysis) • Executables, documentation available (free!) at http: //www. vapor. ucar. edu/ • Source code, feature requests, etc. at http: //sourceforge. net/projects/vapor Alan Norton <alan@ucar. edu>

Acknowledgements • Steering Committee – – – • Developers Nic Brummell - CU Yuhong Fan - NCAR, HAO Aimé Fournier – NCAR, IMAGe Pablo Mininni, NCAR, IMAGe Aake Nordlund, University of Copenhagen Helene Politano - Observatoire de la Cote d'Azur Yannick Ponty - Observatoire de la Cote d'Azur Annick Pouquet - NCAR, ESSL Mark Rast - CU Duane Rosenberg - NCAR, IMAGe Matthias Rempel - NCAR, HAO Geoff Vasil, CU – – John Clyne – NCAR, CISL Alan Norton – NCAR, CISL Kenny Gruchalla – CU Victor Snyder - CSM • Research Collaborators – – Kwan-Liu Ma, U. C. Davis Hiroshi Akiba, U. C. Davis Han-Wei Shen, Ohio State Liya Li, Ohio State • Systems Support – Joey Mendoza, NCAR, CISL Alan Norton <alan@ucar. edu>