HR: 11:25h
AN: NG32A-05    [Abstracts]
TI: Relating Regime Structure to Probability Distribution and Preferred Structure of Small Errors in a Large Atmospheric GCM
AU: * Straus, D M
EM: straus@cola.iges.org
AF: George Mason University, COLA 4041 Powder Mill Rd, Suite 302, Calverton, MD 20705, United States
AB: The probability distribution (pdf) of errors is followed in identical twin studies using the COLA T63 AGCM, integrated with observed SST for 15 recent winters. 30 integrations per winter (for 15 winters) are available with initial errors that are extremely small. The evolution of the pdf is tested for multi-modality, and the results interpreted in terms of clusters / regimes found in: (a) the set of 15x30 integrations mentioned, and (b) a larger ensemble of 55x15 integrations made with the same GCM using the same SSTs. The mapping of pdf evolution and clusters is also carried out for each winter separately, using the clusters found in the 55-member ensemble for the same winter alone. This technique yields information on the change in regimes caused by different boundary forcing (Straus and Molteni, 2004; Straus, Corti and Molteni, 2006). Analysis of the growing errors in terms of baroclinic and barotropic components allows for interpretation of the corresponding instabilities.
DE: 1620 Climate dynamics (0429, 3309)
DE: 3235 Persistence, memory, correlations, clustering (3265, 7857)
DE: 3275 Uncertainty quantification (1873)
DE: 4410 Bifurcations and attractors
DE: 4468 Probability distributions, heavy and fat-tailed (3265)
SC: Nonlinear Geophysics [NG]
MN: 2007 Fall Meeting