Data Reduction Techniques for Scientific Visualization and Data Analysis

Date and time: 
Wednesday, January 11, 2017 - 10:00
220 Deschutes
Samuel Li
University of Oregon
  • Hank Childs (Chair)
  • Allen Malony
  • Boyana Norris


The classic paradigm for scientific visualization and data analysis is post-hoc, where simulation codes write results on the file system and visualization routines read them back to operate. This paradigm sees file I/O as an increasing bottleneck, in terms of both transfer rates and storage capacity. Worse, the I/O bottleneck also jeopardizes the turnaround times for visualization and data analysis routines, which is a precondition for successful explorations.

Data reduction is a means to tackle the above-mentioned difficulties. This paper surveys common techniques used for visualization and analysis of scientific data sets.

We identify three important characteristics of these techniques: 1) lossy or lossless, 2) reduce on write or reduce on read, and 3) resulting in original memory footprint or reduced memory footprint.

We also survey existing use cases with data reduction integrated into their workflow with respect to the three characteristics.

Finally, this survey serves as a starting point for a visualization scientist as well as a simulation scientist to explore data reduction options to tackle his I/O constraints.