A modified event synchronization method based on event clustering

A core task in extreme event research is to evaluate the temporal similarity of events across different network nodes. The Event Synchronization (ES) method has been widely used to construct functional networks from extreme event series. However, the ES method has limitations in identifying event synchronization when extreme events exhibit obvious clustering patterns over time. To address this issue, we propose the concept of event clustering group and the Modified Event Synchronization (MES) method. By modifying the dynamic delay, it ensures that synchronization detection remains effective even during periods of high event density. Taking the mobile communication traffic data from a certain city as an example, we conduct a comparative analysis of the synchronization detection results between MES and ES under conditions of high event clustering. We also examine the sensitivity of the methods to parameter variations. The results demonstrate that under high clustering scenarios, MES effectively mitigates the degradation of synchronization identification and yields more stable outcomes in terms of both the number of identified synchronized events and the quantification of synchronization strength. In addition, the MES algorithm can be executed in parallel, which is compatible with multi-core and multi-node High-Performance Computing (HPC) platforms. This feature accelerates the processing of large-scale time-series data and supports the real-time analysis of extreme events.