Mutual interactions between secular trends and modes of variability in a changing climate

Theory indicates that the structure of intrinsic modes of interannual atmospheric variability is largely controlled by the mean state on which they are superimposed and so should change if the mean state is modified. On the other hand, all else being equal, any externally produced changes in the structure of the mean state are likely to project strongly onto the leading intrinsic interannual modes. In this study we examine a climate change experiment that is especially well-suited for testing these ideas regarding mutual interactions. By doing so we also produce a physical interpretation of the experiment's results.

The experiment we analyze is a 62 member ensemble of business-as-usual integrations that has been produced with NCAR's coupled climate model known as CCSM1.4. The large ensemble allows us to investigate in detail the geographical structure of the secular trend in the experiment as well as how the intrinsic atmospheric modes are affected by that trend. A striking result of the analysis is that indeed the secular trend projects strongly onto a natural mode of the system, namely the circumglobal waveguide mode described by Branstator (2002). Analysis of single ensemble members shows that erroneous conclusions about this structure would have been drawn if the large ensemble had not been available. A second outcome of the analysis is that the structure of this intrinsic mode changes during the experiment, and stochastic experiments with a linearization of the atmospheric component of the GCM indicate that this structural change can be traced to dynamical effects of the changing mean circulation. Interestingly, because the waveguide mode is instrumental in communicating tropical ENSO to the rest of the globe, a consequence of modifications to the modal structure is a change in the global response to El Nino during the experiment.

Further analysis of the experiment is performed to learn more about the key processes (including SST-induced heating and synoptic eddy fluxes) involved in formation of the circumglobal mode and to determine how those processes are affected by the changing climate. Moreover climate change experiments performed with other coupled GCMs are examined in a similar fashion to determine whether the waveguide mode and interactions between mean circulation and modes of variability are useful concepts for understanding the changing climate in those models as well.