Extreme precipitation events (EPEs) pose significant threats to human society and the economy, causing damage to infrastructure and disrupting economic activity. The increasing frequency and intensity of EPEs due to global warming are a major concern. While deep learning has improved our understanding and prediction of EPEs, exploring their propagation patterns and underlying mechanisms remains a challenge. Climate networks offer a powerful tool for studying EPEs and their teleconnections. Previous research using event synchronization (ES) and climate networks has shown promise in forecasting EPEs in specific regions, such as the central Andes. However, a comprehensive, comparative investigation of EPE propagation across diverse global regions is lacking. This study addresses this gap by establishing and evaluating a global directed network of EPEs to investigate preferred propagation pathways and their underlying physical mechanisms. The study also analyzes the potential predictability of EPEs along these pathways, providing valuable insights for improving EPE forecasts.

Climate networks have emerged as a powerful tool for studying extreme precipitation events (EPEs) and their associated teleconnections. Boers et al. demonstrated the effectiveness of combining event synchronization (ES) with climate networks to forecast EPEs, achieving over 60% accuracy in the central Andes. This approach has also been applied to reveal global extreme precipitation teleconnection patterns. Other studies have used ES and similarity measures to investigate EPE patterns in specific regions, such as the Ganga River basin and the United States. These studies suggest the existence of preferred spatiotemporal patterns in EPE occurrence, but they primarily focus on precipitation coherence under specific local weather systems or are limited to specific regions. A comparative investigation of EPEs across diverse global regions has been lacking. This study aims to fill this gap by conducting a comprehensive analysis of EPE propagation patterns across the globe.

This study utilizes daily precipitation data from the NOAA Climate Prediction Center (CPC) Global Unified Gauge-Based Analysis (1980-2020) with a 0.5° x 0.5° spatial resolution. Extreme precipitation events (EPEs) are defined as days exceeding the 95th percentile of daily precipitation. The event synchronization (ES) measure, a nonlinear synchronization measure accounting for dynamical time lags, is used to quantify the co-occurrence of EPEs at different locations. A functional climate network is constructed, where nodes represent grid cells, and edge weights represent the strength of ES. The directionality of edges indicates the temporal order of EPEs between locations. Significance tests are applied to prune weak edges, and robustness tests are conducted by varying the maximum delay parameter (τmax) to ensure the results are robust. Network divergence (ΔS) is calculated to identify source and sink regions, which represent the start and end points of EPE propagation pathways. Outward strength is also used to identify temporal order and the dynamics of EPEs. The weighted mean azimuth is calculated to determine the direction of propagation. To investigate underlying atmospheric dynamics, composite anomalies of 850 mb geopotential height and wind fields from NCEP-NCAR Reanalysis 1 are analyzed. The potential predictability of EPEs is assessed using frequency statistics, calculating the probability of an EPE at a subsequent grid cell within 3 days of an EPE at the initial grid cell. The Antarctic region is excluded due to sparse station data. MATLAB was used for coding and plotting.

The study identified 16 major EPE propagation pathways globally. These pathways are often spatially confined by geographical features such as mountains. The analysis revealed that the propagation of EPEs is influenced by various factors including: * **Regional Weather Systems:** Short-distance edges in the network are attributed to regional weather systems, such as frontal systems or convective systems. * **Topography:** Orographic effects significantly influence EPE propagation, particularly in mountainous regions like the Andes and Appalachians. * **Rossby Waves:** The eastward movement of Rossby waves plays a crucial role in the dynamics of EPE propagation, often interacting with low-level jets and enhancing moisture transport. Specific examples include: * **North America:** EPEs propagate along the Appalachian Mountains, driven by eastward-moving Rossby wave activity and the Great Plains Low-Level Jet (GPLLJ). * **Australia:** EPEs propagate along the coast and Great Dividing Range, influenced by atmospheric rivers and the Australian Low-Level Jet. The study also found that the potential predictability of EPEs along these pathways is variable, with probabilities decaying with distance from the source. North-Central Europe and North America show higher predictability than other regions. For the 16 pathways, the average probability of an EPE occurring within 3 days of an initial EPE is 0.45. On average, over 32% of EPEs follow the identified propagation patterns, reaching as high as 65.81% in specific regions. Seasonal variations in EPE propagation also exist, with chain events mostly occurring during summer/autumn in the Northern Hemisphere and winter/spring in the Southern Hemisphere. The study noted that ENSO has a weak influence on spatial pathways but can affect predictability.

The findings of this study significantly advance our understanding of EPE propagation patterns on a global scale. The identification of 16 preferred pathways, coupled with the analysis of underlying physical mechanisms (topography, Rossby waves, regional weather systems), provides valuable insights into the spatiotemporal dynamics of EPEs. The demonstration of substantial EPE predictability in certain areas along these pathways has important implications for flood and landslide early warning systems. The results emphasize the importance of considering both regional weather systems and large-scale atmospheric teleconnections when forecasting EPEs. The study's use of climate networks provides a novel framework for analyzing extreme climate events, potentially extending to other extreme events such as heatwaves and cold waves.

This study presents a comprehensive analysis of extreme precipitation event (EPE) propagation pathways using climate network approaches. The identification of 16 major propagation patterns, influenced by topography and Rossby waves, offers crucial insights for EPE forecasting. The demonstrated predictability along certain pathways suggests improved early warning capabilities. Future research could focus on integrating these findings into ensemble forecasting and machine learning models, investigating the influence of other atmospheric teleconnections, and extending this framework to other types of extreme climate events.

The study's predictability estimates are based on the occurrence of EPEs, not their intensity. The potential predictability, therefore, does not equate to realized predictability and should be considered prior knowledge to integrate into future ensemble forecasts. The accuracy of EPE predictability may vary depending on the selection and position of grid cells. Additionally, while the study explored the influence of ENSO, more detailed investigation into the relationship between atmospheric teleconnections and EPE propagation pathways is needed. The study also excluded the Antarctic region due to data sparsity.