Characterization of climate indices in models and observations using Hurst Exponent and Reyni Entropy Techniques

Because models are intrinsically incomplete and evolving, multiple methods are needed to characterize how well models match observations and were their weaknesses lie. For the study of climate, global climate models (GCM) are the primary tool. Therefore, in order to improve our climate modeling confidence and our understanding of the models weakness we need to apply more and more measures of various types until one finds differences. Then we can decide if these differences have important impacts on ones results and what they mean in terms of the weaknesses and missing physics in the models. In this work, we investigate a suite of National Center for Atmospheric Research (NCAR) Community Climate System Model (CCSM3) simulations of varied complexity, from fixed sea surface temperature simulations to fully coupled T85 simulations. Climate indices (e.g. NAO), constructed from the GCM simulations and observed data, are analyzed using Hurst Exponent (R/S) and Reyni Entropy methods to explore long-term and short-term dynamics (i.e. temporal evolution of the time series). These methods identify clear differences between the models and observations as well as between the models. One preliminary finding suggests that fixing midlatitude SSTs to observed values increases the differences between the model and observation dynamics at long time scales.