Network similarity and statistical analysis of earthquake seismic data

TL;DR

This paper analyzes the structural similarities of earthquake networks from different regions using spectral methods and investigates regional earthquake dynamics through statistical measures, identifying key hub regions and predicting earthquake occurrences.

Contribution

It introduces a spectral similarity approach for earthquake networks and provides a statistical framework for understanding regional earthquake dynamics and hub identification.

Findings

Hierarchical clustering reveals structural similarities among earthquake networks.

Directed networks are strongly connected, enabling statistical analysis.

Conditional probabilities help predict earthquake occurrences in different regions.

Abstract

We study the structural similarity of earthquake networks constructed from seismic catalogs of different geographical regions. A hierarchical clustering of underlying undirected earthquake networks is shown using Jensen-Shannon divergence in graph spectra. The directed nature of links indicates that each earthquake network is strongly connected, which motivates us to study the directed version statistically. Our statistical analysis of each earthquake region identifies the hub regions. We calculate the conditional probability of the forthcoming occurrences of earthquakes in each region. The conditional probability of each event has been compared with their stationary distribution.