Leveraging spatial metadata in machine learning for improved objective quantification of geological drill core
Leveraging spatial metadata in machine learning for improved objective quantification of geological drill core
Here we present a method for using the spatial x–y coordinate of an image cropped from the cylindrical surface of digital 3D drill core images and demonstrate how this spatial metadata can be used to improve unsupervised machine learning performance. This approach is applicable to any data set with known spatial context, however, here it is used to classify 400 m of drillcore imagery into 12 distinct classes reflecting the dominant rock types and alteration features in the core. We modified two unsupervised learning models to incorporate spatial metadata and an average improvement of 25% was achieved over equivalent models that did not utilize metadata. Our semi-supervised workflow involves unsupervised network training followed by semi-supervised clustering where a support vector machine uses a subset of M expert labeled images to assign a pseudolabel to the entire data set. Fine-tuning of the best performing model showed an f
1 (macro average) of 90%, and its classifications were used to estimate bulk fresh and altered rock abundance downhole. Validation against the same information gathered manually by experts when the core was recovered during the Oman Drilling Project revealed that our automatically generated data sets have a significant positive correlation (Pearson's r of 0.65–0.72) to the expert generated equivalent, demonstrating that valuable geological information can be generated automatically for 400 m of core with only ∼24 hr of domain expert effort.
Oman drilling program, geoscience, hydrothermal alteration, machine learning, mining, neural networks
Grant, Lewis J.C.
482d8a7a-c0e7-489b-8757-e7e44f40bd82
Massot‐Campos, Miquel
6d2b0c16-899c-4f69-8c8d-9434188a30b8
Coggon, Rosalind M.
78a3f775-17b5-415e-97f2-26a0deffc444
Thornton, Blair
8293beb5-c083-47e3-b5f0-d9c3cee14be9
Rotondo, Francesca C.
7e05d184-fd35-44db-9b29-b093aeabe708
Harris, Michelle
bdbb7272-e462-4941-9db5-441a71a70ece
Evans, Aled D.
41a3083e-fb13-4f18-a35b-c0763afa7716
Teagle, Damon A.H.
396539c5-acbe-4dfa-bb9b-94af878fe286
14 March 2024
Grant, Lewis J.C.
482d8a7a-c0e7-489b-8757-e7e44f40bd82
Massot‐Campos, Miquel
6d2b0c16-899c-4f69-8c8d-9434188a30b8
Coggon, Rosalind M.
78a3f775-17b5-415e-97f2-26a0deffc444
Thornton, Blair
8293beb5-c083-47e3-b5f0-d9c3cee14be9
Rotondo, Francesca C.
7e05d184-fd35-44db-9b29-b093aeabe708
Harris, Michelle
bdbb7272-e462-4941-9db5-441a71a70ece
Evans, Aled D.
41a3083e-fb13-4f18-a35b-c0763afa7716
Teagle, Damon A.H.
396539c5-acbe-4dfa-bb9b-94af878fe286
Grant, Lewis J.C., Massot‐Campos, Miquel, Coggon, Rosalind M., Thornton, Blair, Rotondo, Francesca C., Harris, Michelle, Evans, Aled D. and Teagle, Damon A.H.
(2024)
Leveraging spatial metadata in machine learning for improved objective quantification of geological drill core.
Earth and Space Science, 11 (3), [e2023EA003220].
(doi:10.1029/2023EA003220).
Abstract
Here we present a method for using the spatial x–y coordinate of an image cropped from the cylindrical surface of digital 3D drill core images and demonstrate how this spatial metadata can be used to improve unsupervised machine learning performance. This approach is applicable to any data set with known spatial context, however, here it is used to classify 400 m of drillcore imagery into 12 distinct classes reflecting the dominant rock types and alteration features in the core. We modified two unsupervised learning models to incorporate spatial metadata and an average improvement of 25% was achieved over equivalent models that did not utilize metadata. Our semi-supervised workflow involves unsupervised network training followed by semi-supervised clustering where a support vector machine uses a subset of M expert labeled images to assign a pseudolabel to the entire data set. Fine-tuning of the best performing model showed an f
1 (macro average) of 90%, and its classifications were used to estimate bulk fresh and altered rock abundance downhole. Validation against the same information gathered manually by experts when the core was recovered during the Oman Drilling Project revealed that our automatically generated data sets have a significant positive correlation (Pearson's r of 0.65–0.72) to the expert generated equivalent, demonstrating that valuable geological information can be generated automatically for 400 m of core with only ∼24 hr of domain expert effort.
Text
Earth and Space Science - 2024 - Grant - Leveraging Spatial Metadata in Machine Learning for Improved Objective
- Version of Record
More information
Accepted/In Press date: 5 February 2024
Published date: 14 March 2024
Keywords:
Oman drilling program, geoscience, hydrothermal alteration, machine learning, mining, neural networks
Identifiers
Local EPrints ID: 490326
URI: http://eprints.soton.ac.uk/id/eprint/490326
ISSN: 2333-5084
PURE UUID: a9ee3e16-5c5f-4210-8c84-a00a31609a94
Catalogue record
Date deposited: 23 May 2024 16:50
Last modified: 12 Jun 2024 02:03
Export record
Altmetrics
Contributors
Author:
Miquel Massot‐Campos
Author:
Rosalind M. Coggon
Author:
Michelle Harris
Author:
Aled D. Evans
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics