Delineation of seismic sources using an optimized fuzzy clustering

Seismic source zones have an important role in hazard assessment in probabilistic seismic hazard analysis. These zones often determined according to judgments by experts are, in most cases, non-uniform across a specific region and typically controversial. Thus, most of the uncertainty in probabilistic seismic hazard analysis can be related to the delineation of seismic sources. Another problem of probabilistic seismic hazard analysis is the way earthquakes are associated with the faults. Even though it is well-known that earthquakes happen on faults, but most of them are still unknown, this constrains the realization and assessment of seismic risks by experts. This paper attempts to determine seismic sources and associate events to faults using a fuzzy particle swarm optimization clustering algorithm. The algorithm works based on the minimization of two objective functions: distance from events to fault, and distance from events to their center of density (i.e. cluster center). The algorithm is applied on seismic data acquired from northwest of Iran, and its performance is evaluated based on the events assigned to the faults by previous researchers. Comparing associated earthquakes to faults by the algorithm in northwest of Iran with known and documented earthquakes, reveals that, 79.2% of the events are correctly induced by faults. Final result shows that, this methodology will help seismological engineers take a step forward in hazard analysis by determining seismic sources and assigning earthquakes to different active faults.