Spatiotemporal Organization of Extreme Rainfall Events During Early and Late Monsoon Onset Using Event Synchronization and Climate Networks

Abstract

The interannual variability of the Indian Summer Monsoon (ISM) onset which is categorized into early and late onset has a significant influence on the spatiotemporal organization of Extreme Rainfall Events (EREs). Traditional techniques such as Principal Component Analysis (PCA) and linear correlation have been limited by assumptions of linearity and a focus on continuous field variance, failing to capture the discrete, nonlinear, and delayed dependencies inherent in episodic rainfall. This thesis prepares a framework that integrates nonlinear Event Synchronization (ES), Network Divergence (ND), Propagation Analysis, and Community Detection using a hierarchical Stochastic Block Model (SBM) to understand the climate structure of the monsoon transition. By using ES with a defined τmax = 5 days, we capture connections that linear methods usually miss. The computation of normalized Network Divergence identifies critical source-sink dynamics, distinguishing regions that initiate convective signals(source) from those that act as receivers(sink). Furthermore, our 14-day propagation analysis tracks the physical evolution of rainfall signals, while the hierarchical SBM which is a Bayesian generative model guided by the Minimum Description Length (MDL) principle helps identify stable spatial synchronization zones without the resolution limits of standard heuristics like the Louvain algorithm. Using the high-resolution Multi-Source Weighted-Ensemble Precipitation (MSWEP) dataset (1979-2020), we have successfully achieved a comprehensive climate network for both early and late monsoon onset years. Our comparative analysis reveals that there exists a fundamentally different topological pattern in each regime. We posited that this difference in climate structure can be explained by the hypothesis that these networks are modulated by the Boreal Summer Intraseasonal Oscillation (BSISO). Specifically, the distinct patterns of synchronization and propagation likely correspond to the varying BSISO modes namely, canonical northeastward, eastward-blocked, or quasi-stationary which are regulated by the background thermodynamic state of the Indo-Pacific. Overall, this research provides a clear framework and helps advance our understanding of how intraseasonal drivers reshape the topology of extreme weather.