This project uses an unstable periodic orbit analysis, applied to simplified models of atmospheric dynamics, to investigate the nature of time-dependent atmospheric dynamics. The choice of technique is based on the idea that many recurrent circulation patterns exhibit coherent, organized behavior for some period of time. The investigation will determine whether such an analysis can provide insights into what patterns are likely to occur and why, shed light on the nature of transitions between different atmospheric states, and help identify what type of model states should be most amenable to long-range prediction. The approach will then be used to examine how such properties of the atmosphere react to changes in external forcing.
The two models that will be investigated are a barotropic model and a two-level quasigeostrophic model. The first part of the project will involve experimenting with ways of successfully finding unstable periodic orbits in such models, building on studies of unstable periodic orbits that have been made in other fields. Once found, the unstable periodic orbits will be compared with the aperiodic life-cycles of prominent circulation patterns such as the Pacific-North American pattern and the North Atlantic Oscillation. The project will continue with an attempt to use an expansion in unstable periodic orbits to approximate the response operator that describes the sensitivity of the atmospheric model to small changes in external forcing. The results will be compared with the approximate response operator derived from the fluctuation-dissipation theorem. Lastly, the predictability of states close to points on the least unstable periodic orbits will be examined with the intent of determining whether such states have anomalous predictability.
If successful, the work may help understand variations in predictability in long-range atmospheric forcing, and provide a better understanding of uncertainty in some types of long-range forecasts.
