東華大學圖書館 |

The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection // Hee Choi.
作者:	Choi, Hee,
面頁冊數:	1 electronic resource (129 pages)
附註:	Source: Dissertations Abstracts International, Volume: 87-04, Section: B.
Contained By:	Dissertations Abstracts International87-04B.
標題:	Lithosphere. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32289558
ISBN:	9798297667013

The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection /
Choi, Hee,

The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection /Hee Choi. - 1 electronic resource (129 pages)

Source: Dissertations Abstracts International, Volume: 87-04, Section: B.

Although the timing of Earth's first subduction events remains controversial, the rock record suggests that felsic crust formed before subduction began. Geologic evidence from some of Earth's earliest cratons has been interpreted as reflecting the formation of initialcontinental blocks by non‐subduction processes, with subduction initiating later at their margins. Several numerical modeling studies using a commonly cited mechanism for generating subduction called a pseudo-plastic rheology have suggested that thick Archean cratons could concentrate stress along their margins, potentially leading to subduction initiation. This thesis explores how subduction zones formed and evolved at continental margins during early Earth, using a combination of numerical modeling and machine learning.Chapter 2 explores the influence of early continents on the initiation of subduction using mantle convection models with grain-damage rheology, which is a different mechanism to contrast with pseudo-plasticity. Unlike pseudo-plastic models, grain-damage rheology accounts for weak zone memory, and attempts to better represent the microphysical processes leading to lithospheric weakening. My model results indicate that subduction initiation does not necessarily occur at continental margins, but instead often initiates elsewhere and migrates towards the continent. Scaling analysis shows that early continents impart only a minor increase in lithospheric stress and are not capable of independently triggering subduction. These findings suggest that the choice of rheological model significantly influences whether continents play an important role in subduction initiation. It also suggests a reinterpretation of geological evidence, with observed features more likely representing interactions with pre-existing subduction zones rather than direct initiation of new subduction.Building on this idea, Chapter 3 explores the establishment of sustained subduction zones at continental margins. This study identifies two distinct subduction styles: "transient", in which the subduction zone initiates in one location and migrates over time, and "persistent", where the subduction zone remains relatively stationary. I demonstrate that transient subduction zones impacted early continents, ultimately leading to the formation of stable, persistent subduction zones at their margins. Characteristic time scales for this transition are calculated, with approximately 0.5 billion years required from the continent formation to the development of persistent subduction. This timeframe aligns with multiple lines of geological evidence, supporting the interpretation that these observations may reflect the migration of subduction zones toward continents rather than direct initiation at their margins.Chapter 4 presents the development of a Fully Convolutional Network (FCN) to track subduction zones in numerical mantle convection models. This tool was motivated by the need to accurately track subduction zone locations over time for quantitative analysis. Previously, tracking relied on surface velocity divergence and required arbitrary thresholds to exclude minor convergence zones that are not true subduction zones. By detecting subduction zones directly from model images, my FCN approach eliminates the need for these thresholds, improving accuracy. This chapter outlines the deep learning framework and compares the FCN's results with prior methods. The tool's accuracy opens up possibilities for applying machine learning to other geophysical problems and provides a new approach for analyzing subduction zones in both early and modern settings.Together, these chapters reveal that subduction initiation on early Earth was likely influenced more by the rheological models used than by the presence of continents alone, challenging previous assumptions about continent-driven subduction initiation. The study also demonstrates that impact between subduction zone and continent can lead to the formation of persistent subduction zones over time, with characteristic time scales aligning with geological evidence. Additionally, by developing a machine learning tool to track subduction zones in numerical models, this work provides a more precise approach for analyzing tectonic processes, which could be applied to a variety of geophysical studies.

English

ISBN: 9798297667013Subjects--Topical Terms:

2055717
Lithosphere.

The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection /
LDR:05800nmm a22004093i 4500 001 2427649
005 20260911085216.5
006 m o d
007 cr|nu|||||||||
008 260929s2025 miu||||||m |||||||eng d
020 $a 9798297667013
035 $a (MiAaPQD)AAI32289558
035 $a (MiAaPQD)PennState22133hxc5400
035 $a AAI32289558
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Choi, Hee, $e author. $3 3814635
245 1 0 $a The Origin and Evolution of Subduction in Early Earth: Insights From Geodynamic Models and Machine Learning-Based Detection / $c Hee Choi.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (129 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 87-04, Section: B.
500 $a Advisors: Foley, Bradford Committee members: Fisher, Donald; Marone, Chris; Smye, Andrew; Elsworth, Derek; Reimink, Jesse.
502 $b Ph.D. $c The Pennsylvania State University $d 2025.
520 $a Although the timing of Earth's first subduction events remains controversial, the rock record suggests that felsic crust formed before subduction began. Geologic evidence from some of Earth's earliest cratons has been interpreted as reflecting the formation of initialcontinental blocks by non‐subduction processes, with subduction initiating later at their margins. Several numerical modeling studies using a commonly cited mechanism for generating subduction called a pseudo-plastic rheology have suggested that thick Archean cratons could concentrate stress along their margins, potentially leading to subduction initiation. This thesis explores how subduction zones formed and evolved at continental margins during early Earth, using a combination of numerical modeling and machine learning.Chapter 2 explores the influence of early continents on the initiation of subduction using mantle convection models with grain-damage rheology, which is a different mechanism to contrast with pseudo-plasticity. Unlike pseudo-plastic models, grain-damage rheology accounts for weak zone memory, and attempts to better represent the microphysical processes leading to lithospheric weakening. My model results indicate that subduction initiation does not necessarily occur at continental margins, but instead often initiates elsewhere and migrates towards the continent. Scaling analysis shows that early continents impart only a minor increase in lithospheric stress and are not capable of independently triggering subduction. These findings suggest that the choice of rheological model significantly influences whether continents play an important role in subduction initiation. It also suggests a reinterpretation of geological evidence, with observed features more likely representing interactions with pre-existing subduction zones rather than direct initiation of new subduction.Building on this idea, Chapter 3 explores the establishment of sustained subduction zones at continental margins. This study identifies two distinct subduction styles: "transient", in which the subduction zone initiates in one location and migrates over time, and "persistent", where the subduction zone remains relatively stationary. I demonstrate that transient subduction zones impacted early continents, ultimately leading to the formation of stable, persistent subduction zones at their margins. Characteristic time scales for this transition are calculated, with approximately 0.5 billion years required from the continent formation to the development of persistent subduction. This timeframe aligns with multiple lines of geological evidence, supporting the interpretation that these observations may reflect the migration of subduction zones toward continents rather than direct initiation at their margins.Chapter 4 presents the development of a Fully Convolutional Network (FCN) to track subduction zones in numerical mantle convection models. This tool was motivated by the need to accurately track subduction zone locations over time for quantitative analysis. Previously, tracking relied on surface velocity divergence and required arbitrary thresholds to exclude minor convergence zones that are not true subduction zones. By detecting subduction zones directly from model images, my FCN approach eliminates the need for these thresholds, improving accuracy. This chapter outlines the deep learning framework and compares the FCN's results with prior methods. The tool's accuracy opens up possibilities for applying machine learning to other geophysical problems and provides a new approach for analyzing subduction zones in both early and modern settings.Together, these chapters reveal that subduction initiation on early Earth was likely influenced more by the rheological models used than by the presence of continents alone, challenging previous assumptions about continent-driven subduction initiation. The study also demonstrates that impact between subduction zone and continent can lead to the formation of persistent subduction zones over time, with characteristic time scales aligning with geological evidence. Additionally, by developing a machine learning tool to track subduction zones in numerical models, this work provides a more precise approach for analyzing tectonic processes, which could be applied to a variety of geophysical studies.
546 $a English
590 $a School code: 0176
650 4 $a Lithosphere. $3 2055717
650 4 $a Continental dynamics. $3 3175374
650 4 $a Geology. $3 516570
650 4 $a Physics. $3 516296
650 4 $a Plate tectonics. $3 542702
650 4 $a Viscosity. $3 1050706
650 4 $a Rheology. $3 570780
650 4 $a Geochemistry. $3 539092
650 4 $a Heat. $3 573595
650 4 $a Metamorphism. $3 3566240
650 4 $a Deformation. $2 lcstt $3 3267001
690 $a 0592
690 $a 0996
690 $a 0406
690 $a 0372
690 $a 0605
710 2 $a The Pennsylvania State University. $e degree granting institution. $3 3765934
720 1 $a Foley, Bradford $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 87-04B.
790 $a 0176
791 $a Ph.D.
792 $a 2025
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32289558