Title of Presentation: Multiresolution Data Access Within The VISIT Visualization Environment

Primary (Corresponding) Author: Andrew Foulks

Organization of Primary Author: University of New Hampshire

Co-Authors: R. Daniel Bergeron, David Benedetto, Ted M. Sparr


Abstract: Interactive visualization of very large data sets remains a challenging problem to the visualization community.  One promising solution involves representing the data at multiple resolution levels in both space and time.  Low resolution data is used to give the scientist a large scale overview of the data. Finer resolution data is used to show smaller regions of interest chosen by the scientist exploring the data. 

Our goal in this paper is to demonstrate our software data tools and our extensions to the VISIT visualization environment that give the scientist access to multiresolution and adaptive resolution data.   Our software includes a multiresolution data generation tool (STARgen), and a C++ library (STARdata) giving software developers access to the multiresolution data.  STARgen uses wavelet transformation algorithms to generate the coarser resolutions from ggcm data.

Integration with VISIT includes two plugin modules: (1) a database plugin whose responsibility is to access STARdata and make the multiple resolutions available to VISIT 'Plots', and (2) an Operator Attribute plugin module that provides the user with a widget to control and change the resolution of the data loaded from a STARdata source.

As part of our collaboration with physicists studying the earth's interaction with the solar wind, we demonstrate our extensions using ggcm data produced by MHD simulations sampled on a rectilinear grid.  The example simulation data consists of 480 timesteps, each step 345x120x120 spatial resolution (finest level).  The total size of the data is 126 GB, far too large to fit into memory of the average workstation.

Using our plugin, the scientist can view the data at a coarse resolution, can then select a region of interest, followed by a zoom into that region, viewing the data at a finer resolution.