HR: 14:55h
AN: IN33D-06    [Abstracts]
TI: Computing Information-Theoretic Quantities in Large Climate Data Sets
AU: * Knuth, K H
EM: knuth@email.arc.nasa.gov
AF: Intelligent Systems Division, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Castle, J P
EM: pcastle@email.arc.nasa.gov
AF: Education Associates, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Curry, C T
EM: ctc@email.arc.nasa.gov
AF: Education Associates, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Curry, C T
EM: ctc@email.arc.nasa.gov
AF: Department of Applied Mathematics and Statistics, University of California Santa Cruz, Santa Cruz, CA 95064 United States
AU: Gotera, A
EM: agotera@mail.arc.nasa.gov
AF: Education Associates, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Gotera, A
EM: agotera@mail.arc.nasa.gov
AF: Department of Mathematics, California State University East Bay, Hayward, CA 94542 United States
AU: Huyser, K A
EM: khuyser@email.arc.nasa.gov
AF: Education Associates, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Wheeler, K R
EM: kevin.r.wheeler@nasa.gov
AF: Intelligent Systems Division, NASA Ames Research Center, Moffett Field, CA 94035 United States
AU: Rossow, W B
EM: wrossow@giss.nasa.gov
AF: NASA Goddard Institute for Space Studies, 545 West 112th St, New York, NY 10025 United States
AB: Information-theoretic quantities, such as mutual information, allow one to quantify the amount of information shared by two variables. In large data sets, the mutual information can be used to identify sets of co-informative variables and thus are able to identify variables that can act as predictors of a phenomenon of interest. While mutual information alone does not distinguish a causal interaction between two variables, another information-theoretic quantity called the transfer entropy can indicate such possible causal interactions. Together, these quantities can be used to identify causal interactions among sets of variables in large distributed data sets. We are currently developing a suite of computational tools that will allow researchers to calculate, from data, these useful information-theoretic quantities. Our software tools estimate these quantities along with their associated error bars, the latter of which are critical for describing the degree of uncertainty in the estimates. In this presentation we demonstrate how mutual information and transfer entropy can be applied so as to allow researchers not only to identify relations among climate variables, but also to characterize and quantify their possible causal interactions.
UR: http://www.huginn.com/knuth/index.html
DE: 0429 Climate dynamics (1620)
DE: 0520 Data analysis: algorithms and implementation
DE: 3252 Spatial analysis (0500)
DE: 3270 Time series analysis (1872, 4277, 4475)
DE: 3310 Clouds and cloud feedbacks
SC: Earth and Space Science Informatics [IN]
MN: Fall Meeting 2005