GPlatesReconstructionModel

Prototyping for GPlates plate reconstructions as a python class

• Usage examples are available in the test_notebooks directory

Installation

Install directly from GitHub with pip:

pip install git+https://github.com/siwill22/GPlatesReconstructionModel

This pulls in every core dependency automatically, including pygplates and PlateTectonicTools, both of which are ordinary PyPI packages.

To also install the optional dependencies for plotting, geophysics helpers, and additional spatial algorithms, install the all extra:

pip install "gprm[all] @ git+https://github.com/siwill22/GPlatesReconstructionModel"

A plain install needs no system packages — every core dependency has wheels on PyPI for Linux, macOS and Windows. That means the data-preparation and analysis path (fetching reconstruction models, assigning plate IDs, reconstructing data, plate snapshots, and all proximity calculations) runs on a headless machine or in CI with nothing but pip install.

pygmt is deliberately not a core dependency, because it is a wrapper around the GMT command-line library and does not bundle GMT itself — that is the one thing pip cannot install for you. It is only needed for plotting and for a few grid-sampling helpers, all of which import it at the point of use and tell you what to install if it is missing.

Python Dependencies

Required (installed automatically, all pip-installable):

• numpy, scipy, pandas
• geopandas, shapely, rasterio
• matplotlib, xarray, cartopy
• pygplates
• PlateTectonicTools (https://github.com/EarthByte/PlateTectonicTools) — this itself imports cartopy unconditionally, which is why cartopy is required rather than optional
• pooch, requests, tqdm, pyyaml

Optional extras:

• pip install "gprm[viz]" — pygmt (plotting, grid sampling). pygmt also needs GMT >=6 installed separately: conda install -c conda-forge gmt, Homebrew's gmt, or your Linux package manager.
• pip install "gprm[geophysics]" — pyshtools, litho1pt0, pmagpy
• pip install "gprm[spatial]" — astropy-healpix, scikit-image, scikit-learn (if astropy-healpix is not installed, some functions fall back on precomputed point distributions in the Data folder)
• pip install "gprm[all]" — everything above

Where downloaded data goes

Datasets are cached with pooch, in a platform-dependent location. Ask for it rather than writing it out by hand:

from gprm.datasets import cache_path
cache_path()                                    # the cache root
cache_path('TorsvikCocks2017', 'CEED6_LAND.gpml')  # a file inside a fetched bundle

Note on stripy: gprm no longer uses stripy directly, but geophysics still pulls it in transitively via litho1pt0. stripy has no prebuilt wheel for Apple Silicon macOS on any Python version, and none at all for Python 3.13+. On those platforms pip will try to build it from source, which needs a Fortran compiler (e.g. brew install gcc on macOS, or conda install -c conda-forge stripy as an alternative to pip for just that package). This means pip install "gprm[geophysics]" and "gprm[all]" can fail on those platforms even though the bare install and [spatial]/[viz] do not.