Spatiotemporal analysis of extreme precipitation events in the United States at mesoscale: Complex network theory

Analyzing the complex behavior of extreme precipitation events (EPEs) is essential to better understand the effect of climate change on surface and subsurface water resources and forecast extreme hydrologic events. In this study, we applied complex network theory to analyze precipitation data in the contiguous United States (CONUS) and investigate synchronization patterns of EPEs across the nation. We constructed the complex network of EPEs using the CPC database, including the precipitation time series available at 3,276 grid points all over the CONUS. To compute the level of similarity between any two grid points, we analyzed their EPE series using two methods: (i) event synchronization (ES) and (ii) time warp edit distance (TWED). The constructed precipitation networks (PNs) for the summer and winter seasons were analyzed, and network measures i.e., degree centrality (DC), partial degree (PD), and mean geographic distance (MGD) were determined. By analyzing the precipitation data from 1979 to 2021, we found that the number of EPEs in the CONUS has been increasing over the past four decades with similar slopes in the summer and winter seasons. We also observed both local and regional connections as well as teleconnections demonstrating EPE synchronizations across scales. Results of the network measures revealed important geographic locations in terms of their connection patterns to other locations for both seasons. Using the Louvain method, we detected seven communities in the network for the summer season and five for the winter season by means of the ES. Similar results were obtained using the TWED.