Date of Award
2023-05-01
Degree Name
Master of Science
Department
Geophysics
Advisor(s)
Aaron A. Velasco
Abstract
The Tehuantepec seismic gap, located off the southern shore of Oaxaca and Chiapas, Mexico, was previously thought to be an aseismic zone due to no significant event in 100 years. The September 8, 2017 (M8.2) Tehuantepec earthquake disproved this idea and added many questions surrounding the Mexican subduction zone. Specifically, the earthquake did not occur at the subduction megathrust. It ruptured the subducting plate below the megathrust and appeared to stop at the megathrust. Following this event, as well as the September 19, 2017 (M7.1) Morelos-Puebla earthquake, researchers from the University of Texas at El Paso (UTEP), Universidad Autónoma Cuidad Jaurez (UACJ), and Servicio Sismólogico Nacional (SSN) recorded aftershocks from both permanent and temporary seismic networks, providing us with an immense amount of raw waveform data. The SSN alone has cataloged over 30,000 aftershock events six months following the Tehuantepec earthquake, presenting a problem with finding an effective and efficient way of reviewing the data recorded from these networks. To better understand the nature of the rupture zone of the Tehuantepec earthquake, accurate earthquake depths and locations are needed. Previous studies applying traditional automated algorithms for earthquake phase detection and association resulted in messy seismic phase picks and shallow aftershock locations, illustrating the complexity of determining earthquake hypocenters for this aftershock sequence. In this thesis, we present a practical workflow for earthquake detection and monitoring using two machine learning algorithms, PhaseNet, a picking algorithm for picking arrival times of P and S waves, and the Gaussian Mixture Model Association (GaMMA), an association method based on a Bayesian Gaussian Mixture Model used for phase association. Evaluating both models against cataloged phase picks and events from SSN, we determined a promising approach to accurately and efficiently pick and locate earthquakes in the Tehuantepec region.
Language
en
Provenance
Recieved from ProQuest
Copyright Date
2023-05
File Size
50 p.
File Format
application/pdf
Rights Holder
Marc Adrian Garcia
Recommended Citation
Garcia, Marc Adrian, "Enhancing Earthquake Detection Through Machine Learning- An Application to The 2017 Mw 8.2 Tehuantepec Earthquake in Mexico" (2023). Open Access Theses & Dissertations. 3910.
https://scholarworks.utep.edu/open_etd/3910