Abstract

An original Discrete Perfect Sets (DPS) algorithm was developed and applied to clustering of earthquake epicenters with M≥3.0

in California. The obtained clusters correspond well with the locations of the epicenters of strong earthquakes with M≥6.5.

This fact allows considering them as earthquake-prone areas for the magnitude M≥6.5. We compared the obtained clusters

with the areas recognized in 1976 using Earthquake-Prone Areas (EPA) method. The comparison shows that the epicenter clusters

recognized by DPS algorithm are mostly located within the EPA areas or continue them in a specific direction. At the same time,

the clusters cover significantly smaller zones, about 13% of the total area of the EPA zones. An important feature

of the performed DPS clustering is that it uses only earthquake epicenters data instead of a wide range of geophysical,

geomorphological and geological objects and parameters used by EPA technique. The efficiency of DPS clustering

for recognition of earthquake-prone areas is also illustrated by applying to the seismic region of Caucasus.