Network Analysis of Atmospheric River Moisture Transport: Connectivity, Trend, and Climatology

Atmospheric Rivers (ARs) are lengthy, narrow atmospheric corridors that transport substantial moisture over great distances, often resulting in heavy precipitation upon landfall. Characterization of the connectivity and spatial organization of AR events across regions is highly challenging due to their dynamic nature and the complex nonlinear interactions governing moisture transport pathways. This study applies complex network theory to analyse AR moisture transport patterns and connectivity over the West Coast of North America. The ERA5 reanalysis data and 7 CMIP6 climate model outputs over the period 1970–2014 are studied. We first employ the Mann-Kendall trend analysis to examine long-term changes in integrated water vapor transport intensity, thereby establishing the temporal evolution of AR characteristics that the network analysis will contextualize. We next evaluate model performance through Spearman correlation analysis and develop a hybrid network construction methodology that integrates six different threshold selection techniques to determine the optimal correlation threshold for network construction. We then apply several network measures, including degree centrality, clustering coefficient, and closeness centrality, to characterize the organization of an AR system. The Mann-Kendall analysis reveals significant intensification on the West Coast (+0.5-1.0 kg/m/s per year), strengthening in the Gulf of Alaska (p < 0.05). Climatological composites reveal the primary AR corridor at 40–50°N, with peak intensities of 450–500 kg/m/s in the central Pacific and making landfall along the Northern California/Oregon coast at intensities exceeding 400 kg/m/s. Model evaluation identifies EC-Earth3 and EC-Earth3-CC as the best-performing (Spearman r > 0.40), substantially outperforming other CMIP6 models (r = 0.02–0.24). Network validation establishes optimal parameters at r = 0.35 correlation threshold and 5% edge density, with network stability exceeding 0.95 and >90% inter-model agreement on top 100 nodes. Network centrality analysis reveals a hierarchical organization with uniform clustering coefficients (0.6–0.8) across the North Pacific, a north-south gradient in degree centrality (0.08–0.11 in the northern, 0.02–0.04 in the subtropical region), and identifies a critical moisture transport hub at 30–40°N, 120–140°W. The northern Pacific storm corridor (40–55°N) dominates all network measures, confirming primary AR pathways, with the EC-Earth models reliably reproducing the observed patterns. These findings demonstrate that network theory offers a quantitative framework for understanding AR connectivity and organization, with applications for climate change assessment and water resource management.