HR: 0800h
AN: NG31B-0876    [Abstracts]
TI: Statistical Inversion for Quantifying Uncertainties in Climate Prediction
AU: * Jackson, C S
EM: charles@ig.utexas.edu
AF: Institute for Geophysics Jackson School of Geosciences The University of Texas at Austin, 4412 Spicewood Springs Rd., Bldg 600, Austin, TX 78759 United States
AU: Sen, M K
EM: mrinal@ig.utexas.edu
AF: Institute for Geophysics Jackson School of Geosciences The University of Texas at Austin, 4412 Spicewood Springs Rd., Bldg 600, Austin, TX 78759 United States
AU: Stoffa, P L
EM: pauls@ig.utexas.edu
AF: Institute for Geophysics Jackson School of Geosciences The University of Texas at Austin, 4412 Spicewood Springs Rd., Bldg 600, Austin, TX 78759 United States
AB: The effort to estimate uncertainties in climate prediction, as similar to many problems in geophysical inversion, is defined by the dual challenges of 1) the non-linear dependencies between sources of modeling uncertainty and 2) the computational expense of any single model evaluation. Our analyses have focused on Greedy importance sampling based on multiple Very Fast Simulated Annealing (VFSA) as an ideal method for efficiently meeting the dual objectives of identifying problem solutions and sampling a multidimensional probably distribution for describing solution uncertainties. One question we have addressed is the extent to which problem characteristics affect the efficiency of multiple VFSA. We have found that certain algorithmic choices can be set based on broader characteristics of a problem such as its dimensionality and not the degree to which nonlinearities are important. We shall also present our approach to estimating observational uncertainties from a climate model, and how this knowledge may be incorporated into the analysis and expressed within estimates of the uncertainty in the parameter inversion.
UR: http://www.ig.utexas.edu/people/staff/charles/uncertainties\_in\_model\_predictio.htm
DE: 3337 Numerical modeling and data assimilation
DE: 3260 Inverse theory
DE: 3309 Climatology (1620)
SC: Nonlinear Geophysics [NG]
MN: 2004 AGU Fall Meeting