CRISTINA MASOLLER - What have we learned about the Earth climate dynamics by using complex networks and nonlinear data analysis tools?

Climate networks are located at the cross roads of three highly active interdisciplinary fields: climate dynamics, complex networks and time-series analysis. In this talk I will summarise the results obtained in our group within the framework of the LINC project: we have focused in constructing climate networks by using nonlinear data analysis tools, and I will discuss their interpretation in terms of known climate phenomena. Specifically, climate networks have been constructed by using ordinal mutual information, conditional mutual information and a Granger causality estimator. I will discuss the methods used to infer lag-times and to estimate links’ significance. Our analysis has uncovered correlations among different geographical regions, some of them very well limited in space and in time scale. I will try to integrate “the results” (the inferred networks) and the “causes” (known atmospheric and oceanic phenomena): I will show that the uncovered networks are fully consistent with the present understanding of climate phenomena. Finally, I will discuss relevant open issues for future work.