• Number 310  |
• April 26, 2010

Quieting your data

Noisy or irrelevant data can distract scientists and consume expensive storage space and computing time, especially when studying ultra-large, dynamic, incomplete climate data sets to predict how regulations will influence the global climate. Scientist at DOE’s Pacific Northwest National Laboratory, along with a host of collaborators, created a new approach that can discern between noise and nonlinear events. This new data-reduction approach, called Stochastic Proper Orthogonal Decomposition (SPOD), quantifies the uncertainty and extracts the useful information from the large-scale noisy data sets. This approach helps scientists to calibrate climate models using SPOD filtered large-scale noisy climate data, which results in more accurate prediction of climate change. The collaborators in this research project are Courant Institute, New York University, Brown University, and Louisiana State University. This work is part of a larger research effort in mathematical analysis of petascale data sets funded by DOE’s Advanced Scientific Computing Research.

[Kristin Manke, 509.372.6011
Kristin.manke@pnl.gov]