APWP Strat Models

Site-based apparent polar wander path (APWP) workflows for the Keweenawan track, combining paleomagnetic virtual geomagnetic pole (VGP) compilation, stratigraphic height constraints, Bayesian age-depth modeling, and uncertainty propagation through bootstrap APWP construction.

This repository brings together:

• site- and pole-level paleomagnetic compilations
• stratigraphic height calculations for key successions
• Bayesian age-model notebooks built on Chron.jl
• Python utilities for APWP calculation, visualization, and uncertainty propagation

Overview

The core goal of this project is to build APWPs that are informed not only by paleomagnetic directions and pole statistics, but also by stratigraphic ordering and age-model uncertainty. In practice, the repository supports a workflow that:

1. compiles VGP- and site-level paleomagnetic data for the Keweenawan track
2. assigns or reconstructs stratigraphic heights for samples and formations where possible
3. generates section-specific Bayesian age-depth models in Julia using Chron.jl
4. propagates age and directional uncertainty through bootstrap or running-mean APWP calculations
5. exports tabular APWP products and figures for interpretation and paleogeographic analysis

Repository Layout

Path	Purpose
code/	Main notebooks for compilation, APWP construction, and paleogeographic analysis
code/age_models/	Julia-based age-model notebooks for individual stratigraphic successions
code/strat_height/	Notebooks and helper code for stratigraphic height calculations
code/vgptools/	Reusable Python utilities for APWP statistics, compilation, and visualization
data/pmag_compiled/	Compiled paleomagnetic datasets and study-level source folders
data/age_models_output/	Posterior age-model outputs exported from Chron.jl workflows
data/models_shapes/	Rotation files and paleogeographic shape files used for reconstructions
code/apwp_output/	Exported APWP tables
figure/	Generated figures
environment_APWP_age_model.yml	Recommended conda environment specification
environment_APWP_age_model_full.yml	More tightly pinned environment for reproducibility

Notebook Guide

The notebooks are organized roughly in workflow order.

Notebook	Role
code/00_MagIC_files.ipynb	Prepare or inspect MagIC-format inputs
code/01_VGP_compilation.ipynb	Compile Keweenawan-track VGP and site-level paleomagnetic data
code/02_APWP_age_model.ipynb	Construct APWPs while propagating age-model uncertainty
code/03_Keweenawan_VGP_paleogeography.ipynb	Paleogeographic interpretation and visualization of the resulting track

Age-model notebooks in code/age_models/ are organized by stratigraphic succession, including Mamainse Point, Michipicoten, NSVG, Oronto Group, Osler Volcanics, and PLV.

Data and Methods Notes

Several workflows in this repository extend earlier Keweenawan-track compilation efforts by incorporating site-based information and stratigraphic context. The VGP compilation notebooks explicitly track sample or site heights where available so that posterior age-depth models can be sampled during APWP bootstrap resampling.

The Python APWP utilities in code/vgptools/ calculate, among other quantities:

• Fisher mean poles and associated statistics
• moving-window or weighted APWP estimates with a triangular kernel
• apparent polar wander rates between successive mean poles
• shape descriptors for pole distributions
• pseudo-VGP or directional resampling products used in uncertainty propagation

The Julia helpers in code/age_models/MCR_age_models.jl wrap Chron.jl workflows for building stratigraphic age models, exporting posterior age distributions, and plotting age-depth and accumulation-rate results.

Environment Setup

This repository includes two conda environment files:

• environment_APWP_age_model.yml: recommended environment for most users
• environment_APWP_age_model_full.yml: more tightly pinned environment for higher reproducibility on similar platforms

The recommended environment currently specifies Python 3.14 and includes the main scientific stack plus pmagpy==4.3.11.

Create the Python environment

conda env create -f environment_APWP_age_model.yml
conda activate APWP_age_model

Create from the fully pinned spec

conda env create -f environment_APWP_age_model_full.yml
conda activate APWP_age_model

Update an existing environment

conda env update -n APWP_age_model -f environment_APWP_age_model.yml --prune

Launch notebooks

jupyter lab

After launching Jupyter, open notebooks in code/ and select the APWP_age_model kernel.

Julia Age-Model Environment

The age-model construction notebooks in code/age_models/ use Julia rather than Python. The Julia project files live in code/Project.toml and code/Manifest.toml and include Chron, SubsidenceChron, CSV, DataFrames, and related dependencies.

If you want to run the Julia notebooks locally, instantiate the Julia environment from the code/ directory:

julia --project=code -e 'using Pkg; Pkg.instantiate()'

The checked-in manifest was generated with Julia 1.11.6.

You will also need a Julia notebook kernel available in your Jupyter environment if you want to execute the notebooks interactively.

Typical Workflow

For a fresh reproduction of the analysis, the practical order is:

1. create the Python environment and verify the Jupyter kernel
2. compile or inspect the paleomagnetic data in code/01_VGP_compilation.ipynb
3. build or review section-specific age models in code/age_models/
4. confirm posterior age tables in data/age_models_output/
5. run the code/02_APWP_age_model notebook to generate APWPs
6. inspect exported path tables in code/apwp_output/ and figures in figure/
7. use code/03_Keweenawan_VGP_paleogeography.ipynb for paleogeographic visualization

Outputs

The repository already includes representative derived products, including:

• posterior age tables in data/age_models_output/
• compiled paleomagnetic datasets in data/pmag_compiled/
• APWP export tables in code/apwp_output/

One example output table currently present is code/apwp_output/Keweenawan_track_VGP_APWP.csv.

Reproducibility Notes

• Python and Julia workflows are intentionally separated; both environments are needed for full end-to-end reproduction.
• The repository contains both code and intermediate data products, so some notebooks can be explored without rerunning every upstream step.
• Relative paths in the notebooks assume they are run from within the checked-out repository structure.

Authors

• Yiming Zhang
• Nicholas Swanson-Hysell
• Leandro Gallo
• Mathew Domeier
• Facundo Sapienza
• Diego Alberto Osorio Afanador