The Climate Network: Analyzing the Complex Connections of Earth’s Climate (2012)

Undergraduate: Melissa Wrzesien

Faculty Advisor: Peter Mucha
Department: Environmental Science

---

Areas all over the globe are linked through long-range climate connections, which have far-reaching impacts beyond a single ocean or continent. El Nino occurs in the tropical Pacific yet affects the moisture in the Southeast United States. One way to study these interactions is through network analysis, an interdisciplinary science that studies relationships between points in a system. Using air temperature data, we reproduced a 2009 study by Donges et al. that sought to understand how the globe is connected. We used the Pearson correlation and the with mutual information to determine whether most relationships occurred in a linear or nonlinear fashion. We calculated network metrics on various spatial scales, such as area weighted connectivity on the local scale and betweenness on the global scale, in order to obtain a visual representation of our global information. Once our results matched the 2009 study, we worked with a weighted network instead of a binary one. Both types of networks led to similar results. While network analysis on climate data is a new field, we were able to draw some conclusions. Some areas appear to be hotspots of activity in terms of the global network, such as the tropical Pacific Ocean. This region had high values in most metrics. Additionally, it appears as if the linear and nonlinear approaches yield similar results, which indicates that the easier-to-calculate linear relationships may be sufficient to explaining some global patterns.