Multigeophysical data integration using cluster analysis: assisting geological mapping in Trøndelag, Mid-Norway
Authors:
Open Access
- 2 December 2020
- journal article
- research article
- Published by Oxford University Press (OUP) in Geophysical Journal International
- Vol. 225 (2) , 1142-1157
Abstract
SUMMARY: Modern geophysical data acquisition technology makes it possible to measure multiple geophysical properties with high spatial density over large areas with great efficiency. Instead of presenting these co-located multigeophysical data sets in separate maps, we take advantage of cluster analysis and its pattern exploration power to generate a cluster map with objectively integrated information. Each cluster in the resulting cluster map is characterized by multigeophysical properties and can be associated with certain geological attributes or rock types based on existing geological maps, field data and rock sample analysis. Such a cluster map is usually high in resolution and proven to be more helpful than single-attribute maps in terms of assisting geological mapping and interpretation. In this paper, we present the workflow and technical details of applying cluster analysis to multigeophysical data of a study area in the Trøndelag region in Mid-Norway. We address the importance of carefully designed pre-processing procedures regarding the input data sets to ensure an unbiased data integration using cluster analysis. Random forest as a supervised machine learning method for classification/regression is strategically employed post-clustering for quality evaluation of the results. The multigeophysical data used for this study include airborne magnetic, frequency electromagnetic and radiometric measurements, together with ground gravity measurements. Due to the nature of these input data, the resulting cluster map carries multidepth information. When associated with available geological information, the cluster map can help interpret not only bedrock outcrops but also rocks underneath the sediment cover.Keywords
This publication has 18 references indexed in Scilit:
- Machine learning for data-driven discovery in solid Earth geoscienceScience, 2019
- Testing cluster analysis on combined petrophysical and geochemical data for rock mass classificationExploration Geophysics, 2017
- Automated pattern recognition to support geological mapping and exploration target generation – A case study from southern NamibiaJournal of African Earth Sciences, 2015
- Insights into magmatic processes and hydrothermal alteration of in situ superfast spreading ocean crust at ODP/IODP site 1256 from a cluster analysis of rock magnetic propertiesGeochemistry, Geophysics, Geosystems, 2014
- A few useful things to know about machine learningCommunications of the ACM, 2012
- Integrated data analysis for mineral exploration: A case study of clustering satellite imagery, airborne gamma-ray, and regional geochemical data suitesGeophysics, 2012
- Random ForestsMachine Learning, 2001
- The random subspace method for constructing decision forestsIEEE Transactions on Pattern Analysis and Machine Intelligence, 1998
- FCM: The fuzzy c-means clustering algorithmComputers & Geosciences, 1984
- A dendrite method for cluster analysisCommunications in Statistics, 1974