VARG-Tools: Browser-Based Software to Streamline Tephra Correlation and Bayesian Age Modelling

Volcanic ash (tephra) layers are potent tools for correlating sedimentary records and constructing robust Quaternary chronologies. However, rapidly growing glass geochemical datasets present new challenges. Traditional scatter plots show only two elements at a time, forcing analysts to examine many variable pairs to see the whole picture. Many samples may contain poor analyses or mixed glass populations from complex eruptions, reworked sediment, or temporally close events, which require careful filtering before correlation. In addition, building the sophisticated Bayesian age models now standard in the field remains time-intensive, even for experienced researchers. These challenges motivate tools that streamline repetitive tasks while preserving the expert judgment essential for accurate tephra work.

To address these needs, we developed VARG-Tools, an open-source software suite for glass compositional data analysis and the generation of age model (OxCal) code, accessible entirely through a web browser. The software guides researchers through the complete tephrochronological workflow via three interconnected modules:

• The Processing Module prepares glass geochemical data for examination. It handles missing values, applies compositional data transformations, and uses Gaussian Mixture Modelling (GMM) to identify distinct glass populations and automatically flag statistical outliers. It then applies a dimensionality reduction technique called Uniform Manifold Approximation and Projection (UMAP) to project multi-element oxide chemistry into a simplified two-dimensional space where compositionally similar samples plot together. UMAP enables rapid visual assessment of likely volcanic sources or other compositional groupings. Users can also interactively select points directly in plots and assign custom labels, such as population groups or quality flags (e.g., “feldspar-contaminated analysis”).
• The Visualization Module generates publication-quality figures interactively. Features include custom plotting of point data and density fields (with options for filtering, variable selection, and styling), automatic identification of the most discriminating element pairs, and tools for visualizing UMAP-projected values against depth or age. Tie points identified between sites can be exported directly for chronological modelling.
• The Chronology Module automates the generation of Bayesian age-model code for OxCal. From simple spreadsheet inputs, VARG-Tools generates code for depositional models and linked multi-site models that use tephra correlations to calculate shared ages, potentially significantly reducing age-model setup time.

VARG-Tools also introduces the “VARG26 UMAP,” a pre-calculated compositional coordinate system built from tephra across the northern Pacific Ring of Fire (Kamchatka, Alaska, and Japan). Researchers can project their data onto this standardized coordinate system, providing a stable comparative baseline for future studies. Users can also create and save custom reference projections based on their uploaded datasets. Fixed randomization seeds ensure identical results when analyses are repeated, supporting reproducible, FAIR-compliant tephra research. We demonstrate the workflow using peat records from Anchor Point, Alaska, where linking sites through tephra correlations substantially improves chronological precision.