Machine learning of source spectra for large earthquakes

DOI number:10.1093/gji/ggac215

Affiliation of Author(s):中国科学技术大学地球和空间科学学院

Journal:Geophysical Journal International

Place of Publication:英国

Abstract:The shape of earthquake source spectra, traditionally fit by physics-based models, contains important parameters to constrain rupture dimension, duration and geometry. Here we apply machine learning (ML) to derive single- and double-variable data-driven models of source spectra from 3675 Mw > 5.5 global earthquakes, assuming that the Fourier transform of source time functions represent earthquake source spectra below 1 Hz. The single-variable ML model, in the same degree of freedom as the Brune model, improves the goodness of fit by 8.5 per cent. Specifically, the ML model fits the data wit

First Author:Shang Ma

Co-author:Zefeng Li, Wei Wang

Indexed by:Journal paper

Correspondence Author:Zefeng Li

Document Code:24

Discipline:Natural Science

First-Level Discipline:Geophysics

Document Type:SCI

Volume:231

Issue:1

Page Number:692-702

ISSN No.:0956-540X

Translation or Not:no

Date of Publication:2022-06-11

Links to published journals:https://doi.org/10.1093/gji/ggac215