Underlying Low-Order Dynamics of Nonlinear Interaction among Northern Hemisphere Teleconnection Patterns and Its Association with the AO/NAM

Abstract

Abstract Studies on the nonlinear natures of the spatiotemporal structure of the Arctic Oscillation/Northern Hemisphere annular mode (AO/NAM) in the context of nonlinear interaction among the North Atlantic Oscillation (NAO), the Pacific–North American pattern (PNA), and the stratospheric polar vortex (SPV) are performed. The non-Gaussianity of the multivariate probability density function (PDF) in the phase space spanned by their indices is examined first. Five local maxima potentially related to circulation regimes are identified from the so-called angular PDF. One opposite pair of these regimes is found to correspond to the positive and negative phases of the AO/NAM. Since the authors are not sure that, due to uncertainty as suggested by statistical tests, some of the above regimes are non-Gaussian, the nonlinearity of phase-space tendency is employed as an assistant measure to identify them as nonlinear modes. It seems phase-space tendency traditionally estimated from time difference failed to be effective because of its dependence on Δt. To overcome this drawback a low-order stochastic dynamical model is established empirically from the indices. The investigation on the basic deterministic dynamics of this model suggests that the existence of regimes, such as those associated with the AO/NAM, can primarily be explained by its nonlinear deterministic part. However, two problems still remain unsolved: 1) one of the local maxima was almost not identified and 2) life cycles of the basic deterministic dynamics are too long to be related to the low-frequency variability. By introducing a multilevel approach of modeling, further insight into the residual noise of the above stochastic model can address these two issues quite well.