Nonlinear Laplacian spectral analysis for time series: Capturing intermittency and low-frequency variability
Dimitris Giannakis, CIMS

Many processes in science and engineering develop multiscale temporal and spatial patterns, with complex underlying dynamics and time-dependent external forcings. Because of the possible advances in our understanding and prediction of these phenomena, extracting the salient modes of variability empirically from incomplete observations is a problem of wide contemporary interest. In this talk, we present a technique for analyzing high-dimensional, complex time series that exploits the geometrical relationships between the observed data points to recover features characteristic of strongly nonlinear dynamics (such as intermittency), which are not accessible to classical singular spectrum analysis. The method employs Laplacian eigenmaps, evaluated after suitable time-lagged embedding, to produce a reduced representation of the observed samples, where standard tools of matrix algebra can be used to perform truncated singular value decomposition despite the nonlinear geometrical structure of the data set. We illustrate the utility of the method in capturing intermittent modes associated with the Kuroshio current in the North Pacific sector of a comprehensive climate model, and dimensionality reduction of a low-order model for the atmosphere featuring chaotic regime transitions.