Inferring long memory processes in the climate network via ordinal pattern analysis

Abstract

We analyze climatological data from a complex networks perspective, using techniques of nonlinear time series symbolic analysis. Specifically, we employ ordinal patterns and binary representations to analyze monthly averaged surface air temperature (SAT) anomalies. By computing the mutual information of the time series in regular grid points covering the Earth’s surface and then performing global thresholding, we construct climate networks that uncover short-term memory processes, as well as long ones (5–6 yr). Our results suggest that the time variability of the SAT anomalies is determined by patterns of oscillatory behavior that repeat from time to time with a periodicity related to intraseasonal variations and to El Niño on seasonal to interannual time scales. The present work is located at the triple intersection of three highly active interdisciplinary research fields in nonlinear science: symbolic methods for nonlinear time series analysis, network theory, and non linear processes in the earth climate. While a lot of effort is being done in order to improve our understanding of natural complex systems, with many different methods for mapping time series to network representations being investigated and employed in complex systems such as the human brain, our work is the first one aimed at characterizing the global climate network in terms of oscillatory patterns that tend to repeat from time to time, with various time scales. By mapping these processes into a global network, using ordinal patterns and binary representations, we find that the structure of the network changes drastically at different time scales.