eCite Digital Repository

Unsupervised clustering of LA-ICP-MS raster map data for geological interpretation: A case study using epidote from the Yerington district, Nevada

Citation

Ahmed, AD and Hood, SB and Cooke, DR and Belousov, I, Unsupervised clustering of LA-ICP-MS raster map data for geological interpretation: A case study using epidote from the Yerington district, Nevada, Applied Computing and Geosciences, 8 Article 100036. ISSN 2590-1974 (2020) [Refereed Article]


Preview
PDF (Published version)
8Mb
  

Copyright Statement

ę 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/).

DOI: doi:10.1016/j.acags.2020.100036

Abstract

Raster element concentration maps created using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) can be used to interpret microscale compositional and textural domains within mineral grains. Raster maps are typically evaluated element by element; however, application of statistical techniques (such as cluster analysis) can enhance the generation of geochemical domains to support interpretation of growth zones, core-rim relationships, sector zones, and compositional-textural associations. Clustered LA-ICP-MS map data can be assessed within individual samples and between multiple samples, and can extend insight from the microscopic scale to the regional scale to better understand geological paragenesis of an area.

Our workflow (1) applies a centred log transformation to selected elements in a raster map dataset; (2) uses principal component analysis (PCA) applied to the multi-sample, mono-mineralic dataset to group similar elements in epidote based on geochemical character; (3) applies unsupervised clustering to separate different types and generations of epidote in chemical feature space; (4) presents clustered LA-ICP-MS raster map results for interpretation of inter- and intra-mineral chemical zones; and (5) plots results spatially, across a regional map area, to investigate geological paragenesis.

The workflow is illustrated using samples of epidote from the Yerington porphyry-skarn Cu (MoľAu) district. In the case study area, six clusters are defined by unique mineral compositions: (1) low U; (2) elevated Pb, Mn and low Fe and Sr; (3) elevated Ce, U and low Mn and Pb; (4) elevated U, Ce; low Mn, Pb; (5) elevated Sr, Fe and low Mn, Pb; and (6) elevated Mn, Sr, and Fe and low Ce and U. The regional distribution of these groups is presented as indicating proximity to the porphyry environment (lower concentrations of Ce, U, As and Sb and higher concentrations of Mn, Sr and Fe) versus retrograde skarn (elevated Ce, U, As and Sb).

Item Details

Item Type:Refereed Article
Keywords:geochemistry, PCA, unsupervised learning, LA-ICP-MS, k-means clustering, epidote, Ann Mason, Nevada
Research Division:Earth Sciences
Research Group:Geochemistry
Research Field:Exploration geochemistry
Objective Division:Mineral Resources (Excl. Energy Resources)
Objective Group:Mineral exploration
Objective Field:Copper ore exploration
UTAS Author:Ahmed, AD (Ms Ayesha Ahmed)
UTAS Author:Hood, SB (Mr Shawn Hood)
UTAS Author:Cooke, DR (Professor David Cooke)
UTAS Author:Belousov, I (Dr Ivan Belousov)
ID Code:141185
Year Published:2020
Funding Support:Australian Research Council (IH130200004)
Deposited By:CODES ARC
Deposited On:2020-09-30
Last Modified:2020-10-07
Downloads:5 View Download Statistics

Repository Staff Only: item control page