Nonlinear components in global climate teleconnections

Among the sources of temporal variability in the climate system, an important role belongs to internal variability modes – phenomena with oscillatory behavior ranging from predominantly sub-annual (e.g. North Atlantic Oscillation) or inter-annual (e.g. Southern Oscillation) to decadal or multidecadal variations (e.g. Pacific Decadal Oscillation, Atlantic Multidecadal Oscillation). These oscillations manifest themselves not only within their particular geographical areas of origin, but their effects are typically also transmitted through long-range teleconnections, affecting weather and climate patterns worldwide. Analysis of these relationships is often done assuming their linearity – but rarely is such assumption explicitly verified.

In this presentation, presence and magnitude of nonlinear components in long-range teleconnections associated with selected climate variability modes are studied through various time series analysis methods. Several nonlinearity-quantifying statistics, ranging from simple measures of asymmetry in the regression coefficients to outcomes of more formal surrogate data-based tests, are employed to investigate the teleconnection-related responses of local temperatures across the globe. It is shown that substantial variations exist in degree of manifested nonlinearity, subject to both the target location and type of the variability mode(s) considered. Potential of individual nonlinearity-sensitive techniques for more realistic capture of the teleconnection-related response patterns is also discussed, with an ultimate goal of construction of a more accurate model of variability transfer in the climate system.