A theoretical development of machine learning for geophysical inversions: An application to volcano imaging

Rashtbehesht, Majid

Full Metadata

Overview

Year:: 2022
Contributor:: Rashtbehesht, Majid (creator); Huber, Christian (Advisor); Fischer, Karen (Reader); Tsai, Victor (Reader); Liang, Yan (Reader); Naif, Samer (Reader); Brown University. Department of Earth, Environmental, and Planetary Sciences (sponsor)
Genre:: theses
Subject:: Machine Learning; Applied mathematics; Geophysics; Seismology; Deep Learning; Bayesian Inversion; Volcanology
Extent:: 11, 178 p.

Files

Description

Abstract:: Volcanic hazards monitoring relies heavily on constraints offered by geophysical inversions to decipher the structure of subvolcanic magmatic systems. Over the past decades a multidisciplinary effort involving petrologists and geophysicists has allowed to refine models of magma storage and crustal transport consistent with both types of data. Beyond the progress that this effort has allowed, it has also revealed a lack of quantitative assessment of the limitations of geophysical imaging in resolving the finer scale features that are now recognized to play a major role on the dynamics of magmatic systems. In this thesis, I first provide an analysis of the limitations of seismic imaging to resolving thin melt-rich layers in the shallow crust with a synthetic study. This work highlights the trade-off between resolving layer thickness and melt fraction and indicates that more accurate inversion require more data in the inversion process (waveform modeling). The rest of the thesis then focuses on two directions for improvement (1) using complementary datasets (joint inversions) to improve the imaging capability of seismic waveform inversion and (2) develop machine learning inspired techniques to improve the waveform inversion process. Through various synthetic case studies, I analyze the performance of these new methods and compare them to more traditional approaches.
Notes:: Thesis (Ph. D.)--Brown University, 2022

Content

Citation

Rashtbehesht, Majid, "A theoretical development of machine learning for geophysical inversions: An application to volcano imaging" (2022). Earth, Environmental and Planetary Sciences Theses and Dissertations. Brown Digital Repository. Brown University Library. https://repository.library.brown.edu/studio/item/bdr:y7yyzkss/

Relations

Collection:

Earth, Environmental and Planetary Sciences Theses and Dissertations

Theses and Dissertations for the Earth, Environmental and Planetary Sciences department.
...