[S-SS06] New trends in data acquisition, analysis and interpretation of seismicity

[SSS06-P04] Determination of focal mechanism of earthquakes in and around the Korean Peninsula using P-wave first-motion polarity analysis based on deep learning

*Mikyung Choi¹, Kyungmin Min¹, Ah-Hyun Byun¹, Enyoung Jo¹, Sun-Cheon Park¹ (1.Korea Meteorological Administration)

Keywords:Focal mechanism, Deep learning, P-wave first-motion polarity

The fault-plane orientation and slip directions of earthquakes can provide important information about fault structure. For earthquake with magnitude 3.5 or greater in and around the Korean peninsula, we automatically analyze moment tensor using a long-period waveform inversion. For small earthquakes (M < 3.5), earthquakes occur frequently and are important for characterizing the regional structure, but automatic moment tensor analysis cannot be applied. These earthquakes are typically determined by focal mechanism derived from manually identified P-wave first-motion polarities. Recently, research on artificial intelligence has been used in various fields, and neural network model has been developed to identify P-wave first-motion polarity using local earthquake data from South Korea. In this study, we tested automatic determination of the focal-mechanism of earthquakes with magnitude 2.0 or greater that occurred in and around Korean peninsula using a deep learning model for P-wave first motion polarity identification.