Colombian climate teleconnections from complex networks and information transference

Abstract

ABSTRACT: This study leverages complex network analysis and information transfer quantifiers to characterize the climate variability teleconnection spatial patterns across Colombia, based on causal relationships. Causality between six oceanic indices (Niño 3.4, Niño 3, SOI, TNA, NTA, and CAR) and three local climatic variables (total precipitation [TP], two-meter temperature [T2M], and vertical convergence of moisture flux [VIMFLUX]) is estimated through the Liang-Kleeman Maximum Likelihood Estimator (LKMLE), which works as a quantifier of information transfer between time series. Six climate complex networks are built based on the strength of these teleconnections, summarizing the overall causal influence of each index on the chosen variables. The spatial distribution of teleconnections between El Niño-Southern Oscillation (ENSO) indices and TP captures the strong connectivity over eastern and southeastern Colombia, as in previous studies. Nevertheless, our findings suggest that ENSO does not have a statistically significant causal relationship with T2M. On the other hand, Atlantic Ocean-related indices are found to have significant causal teleconnections with northern, western, and northwestern Colombia, especially over the Orinoco and Amazon’s basins for TP, in agreement with previous studies. Our model results in simple, undirected, and unweighted graphs. The presence of several isolated vertices for all indices suggests sparse networks. This indicates that most of the territory exhibits limited teleconnection-based connectivity. Despite this sparseness, the properties of the graphs hint at a highly complex underlying network structure that may not readily fit into the currently established categories of canonical complex networks.