Analyzing extreme temperatures in the United States using complex network theory

Abstract

Extreme temperature events (ETEs), such as recent severe heat waves, are one of the consequences of climate change and global warming. To better understand their complex behavior and forecast climate and/or hydrologic dynamics, understanding and investigating their spatial and temporal patterns are essential. We, therefore, analyzed ETEs over the Contiguous United States (CONUS) for both summer and winter seasons using complex network theory. For this purpose, the daily maximum and minimum temperature data were collected from the Climate Prediction Center (CPC) database. To determine the level of similarity between two geographic nodes, we employed the event synchronization method and constructed the network of ETEs. The constructed networks were then corrected for boundary effects. Network measures i.e., degree centrality (DC), betweenness centrality (BC), clustering coefficient (CC), mean geographic distance (MGD) and long-ranged directedness (LD) were determined to analyze complex patterns within each network. Based on the network measures, we found that the ETEs are more synchronized in winters than those in summers. In addition, the BC and LD revealed that California plays an important role in the large-scale propagation of ETEs during summers, while Texas and the eastern region of New Mexico during winters. Furthermore, the evolution of ETEs from 1979 to 2022 uncovered increasing and decreasing trends for the summer and winter seasons, respectively. We also detected teleconnections over the CONUS. By applying the Louvain method for community detection, we detected different communities in the network of extreme temperature events for the summer and winter seasons.