← Back to Sessions

Phasehunter: Seismic Wave Onset Time Determination Through Probabilistic Deep Learning Regression

Description: 

Seismic phase picking is the process of identifying the onset time of seismic waves in a seismogram. The onset time is one of the most important measurements in seismology because it is used in a wide range of applications such as earthquake location, seismic tomography, source discrimination, and earthquake early warning. The detection and picking of seismic phases is a difficult and labor-intensive task when performed manually, and it becomes particularly challenging during times of high seismic activity. We have developed a deep learning method that measures the onset time of seismic waves through probabilistic deep regression. Previous deep learning methods that have been developed for this purpose treat onset time determination as a classification problem operating on discretely sampled time series. Instead, we treat it as a regression problem, which enables - at least in principle - sub-sample accuracy. Our method also estimates onset time uncertainty, which can be used to increase the reliability of phase association and location accuracy.

Session: Opportunities and Challenges for Machine Learning Applications in Seismology

Type: Oral

Date: 4/19/2023

Presentation Time: 10:45 AM (local time)

Presenting Author: Artemii Novoselov

Student Presenter: Yes

Invited Presentation: No

Authors