Machine learning classifiers for attributing tephra to source volcanoes: an evaluation of methods for Alaska tephras

Matthew S.M. Bolton, Britta J.L. Jensen, Kristi Wallace, Nore Praet, David Fortin, Darrell Kaufman, Marc De Batist

Research output: Contribution to journalArticle

Abstract

Glass composition-based correlations of volcanic ash (tephra) traditionally rely on extensive manual plotting. Many previous statistical methods for testing correlations are limited by using geochemical means, masking diagnostic variability. We suggest that machine learning classifiers can expedite correlation, quickly narrowing the list of likely candidates using well-trained models. Eruptives from Alaska's Aleutian Arc-Alaska Peninsula and Wrangell volcanic field were used as a test environment for 11 supervised classification algorithms, trained on nearly 2000 electron probe microanalysis measurements of glass major oxides, representing 10 volcanic sources. Artificial neural networks and random forests were consistently among the top-performing learners (accuracy and kappa > 0.96). Their combination as an average ensemble effectively improves their performance. Using this combined model on tephras from Eklutna Lake, south-central Alaska, showed that predictions match traditional methods and can speed correlation. Although classifiers are useful tools, they should aid expert analysis, not replace it. The Eklutna Lake tephras are mostly from Redoubt Volcano. Besides tephras from known Holocene-active sources, Holocene tephra geochemically consistent with Pleistocene Emmons Lake Volcanic Center (Dawson tephra), but from a yet unknown source, is evident. These tephras are mostly anchored by a highly resolved varved chronology and represent new important regional stratigraphic markers.

Original languageEnglish (US)
JournalJournal of Quaternary Science
DOIs
StateAccepted/In press - Jan 1 2019

Fingerprint

tephra
volcano
lake
glass
Holocene
volcanic ash
image classification
electron probe analysis
artificial neural network
chronology
oxide
Pleistocene
method
evaluation
machine learning
Classifier
Volcano
Tephra
Evaluation
Machine Learning

Keywords

  • Alaska
  • classification
  • glass geochemistry
  • machine learning
  • tephra

ASJC Scopus subject areas

  • Arts and Humanities (miscellaneous)
  • Earth and Planetary Sciences (miscellaneous)
  • Palaeontology

Cite this

Machine learning classifiers for attributing tephra to source volcanoes : an evaluation of methods for Alaska tephras. / Bolton, Matthew S.M.; Jensen, Britta J.L.; Wallace, Kristi; Praet, Nore; Fortin, David; Kaufman, Darrell; De Batist, Marc.

In: Journal of Quaternary Science, 01.01.2019.

Research output: Contribution to journalArticle

Bolton, Matthew S.M. ; Jensen, Britta J.L. ; Wallace, Kristi ; Praet, Nore ; Fortin, David ; Kaufman, Darrell ; De Batist, Marc. / Machine learning classifiers for attributing tephra to source volcanoes : an evaluation of methods for Alaska tephras. In: Journal of Quaternary Science. 2019.
@article{0685edd9b58447fcba9b41c27bff5116,
title = "Machine learning classifiers for attributing tephra to source volcanoes: an evaluation of methods for Alaska tephras",
abstract = "Glass composition-based correlations of volcanic ash (tephra) traditionally rely on extensive manual plotting. Many previous statistical methods for testing correlations are limited by using geochemical means, masking diagnostic variability. We suggest that machine learning classifiers can expedite correlation, quickly narrowing the list of likely candidates using well-trained models. Eruptives from Alaska's Aleutian Arc-Alaska Peninsula and Wrangell volcanic field were used as a test environment for 11 supervised classification algorithms, trained on nearly 2000 electron probe microanalysis measurements of glass major oxides, representing 10 volcanic sources. Artificial neural networks and random forests were consistently among the top-performing learners (accuracy and kappa > 0.96). Their combination as an average ensemble effectively improves their performance. Using this combined model on tephras from Eklutna Lake, south-central Alaska, showed that predictions match traditional methods and can speed correlation. Although classifiers are useful tools, they should aid expert analysis, not replace it. The Eklutna Lake tephras are mostly from Redoubt Volcano. Besides tephras from known Holocene-active sources, Holocene tephra geochemically consistent with Pleistocene Emmons Lake Volcanic Center (Dawson tephra), but from a yet unknown source, is evident. These tephras are mostly anchored by a highly resolved varved chronology and represent new important regional stratigraphic markers.",
keywords = "Alaska, classification, glass geochemistry, machine learning, tephra",
author = "Bolton, {Matthew S.M.} and Jensen, {Britta J.L.} and Kristi Wallace and Nore Praet and David Fortin and Darrell Kaufman and {De Batist}, Marc",
year = "2019",
month = "1",
day = "1",
doi = "10.1002/jqs.3170",
language = "English (US)",
journal = "Journal of Quaternary Science",
issn = "0267-8179",
publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - Machine learning classifiers for attributing tephra to source volcanoes

T2 - an evaluation of methods for Alaska tephras

AU - Bolton, Matthew S.M.

AU - Jensen, Britta J.L.

AU - Wallace, Kristi

AU - Praet, Nore

AU - Fortin, David

AU - Kaufman, Darrell

AU - De Batist, Marc

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Glass composition-based correlations of volcanic ash (tephra) traditionally rely on extensive manual plotting. Many previous statistical methods for testing correlations are limited by using geochemical means, masking diagnostic variability. We suggest that machine learning classifiers can expedite correlation, quickly narrowing the list of likely candidates using well-trained models. Eruptives from Alaska's Aleutian Arc-Alaska Peninsula and Wrangell volcanic field were used as a test environment for 11 supervised classification algorithms, trained on nearly 2000 electron probe microanalysis measurements of glass major oxides, representing 10 volcanic sources. Artificial neural networks and random forests were consistently among the top-performing learners (accuracy and kappa > 0.96). Their combination as an average ensemble effectively improves their performance. Using this combined model on tephras from Eklutna Lake, south-central Alaska, showed that predictions match traditional methods and can speed correlation. Although classifiers are useful tools, they should aid expert analysis, not replace it. The Eklutna Lake tephras are mostly from Redoubt Volcano. Besides tephras from known Holocene-active sources, Holocene tephra geochemically consistent with Pleistocene Emmons Lake Volcanic Center (Dawson tephra), but from a yet unknown source, is evident. These tephras are mostly anchored by a highly resolved varved chronology and represent new important regional stratigraphic markers.

AB - Glass composition-based correlations of volcanic ash (tephra) traditionally rely on extensive manual plotting. Many previous statistical methods for testing correlations are limited by using geochemical means, masking diagnostic variability. We suggest that machine learning classifiers can expedite correlation, quickly narrowing the list of likely candidates using well-trained models. Eruptives from Alaska's Aleutian Arc-Alaska Peninsula and Wrangell volcanic field were used as a test environment for 11 supervised classification algorithms, trained on nearly 2000 electron probe microanalysis measurements of glass major oxides, representing 10 volcanic sources. Artificial neural networks and random forests were consistently among the top-performing learners (accuracy and kappa > 0.96). Their combination as an average ensemble effectively improves their performance. Using this combined model on tephras from Eklutna Lake, south-central Alaska, showed that predictions match traditional methods and can speed correlation. Although classifiers are useful tools, they should aid expert analysis, not replace it. The Eklutna Lake tephras are mostly from Redoubt Volcano. Besides tephras from known Holocene-active sources, Holocene tephra geochemically consistent with Pleistocene Emmons Lake Volcanic Center (Dawson tephra), but from a yet unknown source, is evident. These tephras are mostly anchored by a highly resolved varved chronology and represent new important regional stratigraphic markers.

KW - Alaska

KW - classification

KW - glass geochemistry

KW - machine learning

KW - tephra

UR - http://www.scopus.com/inward/record.url?scp=85076392592&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85076392592&partnerID=8YFLogxK

U2 - 10.1002/jqs.3170

DO - 10.1002/jqs.3170

M3 - Article

AN - SCOPUS:85076392592

JO - Journal of Quaternary Science

JF - Journal of Quaternary Science

SN - 0267-8179

ER -