| The notion of climate predictability is introduced in terms of the relative importance of initial and boundary conditions to the evolution of climate-like dynamical systems. The dominance of the latter is identified by a newly introduced time scale called the climate predictability time. Because climate data sets defined with averages shorter than the predictability time are especially susceptible to internal variability, special attention was given to investigating the sensitivity that the climate predictability time exhibits to various physical processes. To accomplish this, the statistical dynamical model (SDM) framework is combined with turbulence closure theory to create simplified climate models whose predictability properties can easily be deduced. A quasigeostrophic model of intermediate complexity (MIC) is formulated using statistical closure theory and closed using the direct interaction approximation (DIA). A preliminary set of experiments indicate that that the predictability time is extended considerably by the presence of a more slowly evolving medium, such as an ocean. These experiments also indicate that climate statistics in ocean-atmosphere coupled systems can be just as sensitive to initial conditions in the atmosphere as they are to initial conditions in the ocean. These results suggest that the ocean may play an important role in amplifying errors in the atmosphere and that special care may be necessary when initializing the atmospheric component of general circulation models (GCMs) to avoid contaminating climate projections with chaotic error growth. More generally, this work details a robust application of statistical closure theory to the climate problem. It demonstrates how the SDM framework can be flexibly applied to studies of climate and, most importantly, provides a highly flexible implementation of an SDM, which can continue to serve as a tool to help fill the gap in our current understanding of climate dynamics. |
PDF Full Text Request