MACHINE LEARNING FOR FAST ANALYSIS AND VALID SOURCE POINT ESTIMATION IN EARLY WARNING OF EARTH QUAKES

Abstract: EARTHQUAKE hypocenter localization is essential in the field of seismology and plays a critical role in a variety of seismological applications such as tomography, source characterization, and hazard assessment. This underscores the importance of developing robust earth quake monitoring systems for accurately determining the event origin times and hypocenter locations. In addition, the rapid and reliable characterization of ongoing earthquakes is a crucial, yet challenging, task for developing seismic hazard mitigation tools like earthquake early warning (EEW) systems. While classical methods have been widely adopted to design EEW systems, challenges remain to pinpoint hypo center locations in real-time largely due to limited information in the early stage of earthquakes. Among various key aspects of EEW, timeliness is a crucial consideration and additional efforts are required to further improve thehypo center location estimates with minimum data from the first few seconds after the P-wave arrival and the first few seismograph stations that are triggered by the ground shaking.

Keywords: Earth quake early(EEW), Random Forest(RF), seismic hazard mitigation.