Complex climate networks of nonlinearly correlated time series

A climate anomaly is defined as the difference between a climate variable and a baseline, which is often the climate normal. In climate change studies, climate anomalies are more important than the average climate. Positive and negative extremes of climate anomalies are equally important, because they represent quite opposite climate events. Therefore, extreme event based synchronization is a proper choice for measuring the similarity of event-like series. However, the traditional event synchronization method cannot incorporate positive and negative extremes simultaneously. In this study, a newly proposed event synchronization (ES) measure is adopted as similarity measure of climate anomaly time series, where both positive and negative extremes are identified as extreme events. Then, global complex climate networks based on this similarity measure of surface air temperature (SAT) anomaly time series have been constructed and analyzed. Exponential function and power function have been fitted to the empirical degree distribution of positive and negative climate networks, respectively. The prominent atmospheric teleconnection pattern (North Atlantic Oscillation, NAO) as well as the remote impacts of ENSO have been correctively detected by the global climate networks. The advantages of ES-based complex networks have also bee discussed. This study provides an illustrative example of constructing complex climate network model for nonlinearly correlated climate time series with both positive and negative extremes.