Analysis of a six-component atmospheric spectral model: chaos, predictability and vacillation

The properties of a low-order atmospheric spectral model are investigated by using a bifurcation analysis of its steady states and periodic solutions. The time-asymptotic behaviour of the model is either stationary, periodic, quasi-periodic or chaotic, depending on the parameter values and initial conditions. Different scenarios are found leading to the generation of strange attractors. Some include the occurrence of homoclinic orbits, such that for nearby parameter values chaotic orbits exist moving in small tubes around the homoclinic orbits. The chaotic motion describes an irregular flow, predictable on a time scale given by the reciprocal of its positive Lyapunov exponent. However, the model cannot describe transitions between different preferent flow regimes. This is due to the severe truncation of the spectral expansions.