OpenGL Visualization of Hurricane Isabel Gregory P. Johnson, Christopher A. Burns Texas Advanced Computing Center The University of Texas at Austin
Goals Visualize the Hurricane Isabel dataset with the following three criteria: Multiple Characteristics Interactivity Exploration
Our Solution Standalone C++/OpenGL application Standard language, APIs ==> Portability Pthreads for multithreading FLTK Toolkit for GUI widgets Make use of modern GPU hardware Our solution runs on a standard desktop PC in real time
Multiple Characteristics Clouds Billboarded quads Opacity, luminance are functions of data Precipitation Color-coded spheres for snow, rain, and grauppel Easily distinguished from clouds Wind Horizontal slicing plane Colored either by wind direction, or wind magnitude Enable or disable any combination of the three visualizations
Interactivity Challenges Limited I/O Bandwidth Very large files Load data in a separate thread Prefetch the next timestep Sorting of cloud particle quads Correct alpha blending requires sorting, sublinear performance Organize voxel data in memory as 3D tiles Sort quads in each tile, then sort the tiles Good memory performance
Exploration Camera can move anywhere at any time Transfer functions user-adjustable Can pause simulation at any time, or advance to any timestep Slicing plane altitude is user-adjustable
Wind vector colors are mapped to the compass directions above Gallery Warmer wind vector colors indicate wind magnitude. Notice the sharp falloff of wind magnitude in the eye, heavy precipitation to the north. Wind vector colors are mapped to the compass directions above