Abstract
In this paper we introduce GeoChemNet as a novel tool for visualizing and analyzing high-dimensional geochemical datasets. Unlike traditional approaches such as principal component analysis and pairplots, GeoChemNet uses a non-linear methodology that does not require predefined number of elements, allowing it to reveal highly multidimensional anomalies that are not elucidated using other methods. We demonstrate its advantages using an artificial dataset and compare its performance to conventional techniques. We then apply the tool to two case studies from the same region. The first illustrates how GeoChemNet can work synergistically with clustering algorithms to uncover structure within lithological units. The second demonstrates its utility as a standalone tool for reducing the search space in mineral exploration of sediment-hosted copper deposits. Ultimately, GeoChemNet's interactive, geospatial visualizations enhance communication between geoscientists and data scientists and uncover geological insights from geochemical datasets.
| Original language | English (US) |
|---|---|
| Article number | 106712 |
| Journal | Applied Geochemistry |
| Volume | 199 |
| DOIs | |
| State | Published - Feb 2026 |
ASJC Scopus subject areas
- Environmental Chemistry
- Pollution
- Geochemistry and Petrology
Keywords
- Anomaly detection
- Data science
- Data visualization
- Geochemical data analysis
- Mineral exploration
- Soil geochemistry
Fingerprint
Dive into the research topics of 'GeoChemNet: An interactive tool for visualizing and interpreting outliers in geochemical data using networks'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver