14B.4
Using Metrics to Explore the Relationship Between Simulated Mean Climate and Variability
Peter J. Gleckler, LLNL, Livermore, CA; and K. R. Sperber and K. E. Taylor
Objective measures of climate model performance are proposed and used to assess simulations performed in the most recent phase of the Coupled Model Intercomparison Project (CMIP3; used in the IPCC AR4). We quantify how some models clearly simulate the present day climate better than others. However, with the diversity of fields considered the results are varied, as all models exhibit strengths and weaknesses. The issue of ranking overall model performance is further complicated by the demonstration of how common metrics can be sensitive to observational uncertainty, the domain of interest (e.g., the tropics versus extratropics), simulation of internal variability, and the spatial scales examined. For the extratropics, our multivariate examination of large scales suggests a relationship between how well a given model (relative to others) simulates the mean climate versus interannual variability. Such a relationship is less clear in the tropics. By further isolating fields that are relevant to underlying processes, we discuss the relationship between the fidelity of the simulated mean state and variability specific to several modes of variability (ENSO, NAO, and MJO). Additionally, this process-oriented perspective will be used for evaluating the fidelity with which these modes are simulated.
Session 14B, Climate Modeling and Diagnostics Part II
Thursday, 24 January 2008, 11:00 AM-12:15 PM, 217-218
Previous paper Next paper