Spatial, Temporal and Spatio-Temporal Clustering in Multi-Fault Seismic Zones: A review

This review paper focuses on a specialized problem that of seismic spatio temporal clustering, i.e. an analysis of the grouping of earthquake events based on space proximity and time distribution. It is a useful technique to identify the patterns that increase the understanding of the tectonic processes, seismic hazard assessment and earthquake forecasting. Fault lines and tectonic boundaries are grouped together spatially with events, while sequential events such as aftershocks are grouped based on time. Spatio temporal clustering of seismicity helps in regions of concentrated seismic activity or high seismic risk. To detect and analyze these clusters, Epidemic Type Aftershock Sequence (ETAS) model and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) are applied through different statistical models and machine learning techniques. Further improving seismic hazard assessments by considering fault interaction is identifying multi fault seismic zones where multiple faults systems interact. Many of the studies involving multiple fault dynamics are in regions encountering such settings, including California's San Andreas Fault, subduction zones in Japan and Chile as well as the southern front of the Hellenic seismic arc. The problem from analyzing multi-fault zones is in the complexities of the data and the unpredictability of how faults interact, but recent break throughs in machine learning promise to address these challenges.