Xarray Extension for working with spherical harmonic data

This extension adds functionality to work with Spherical Harmonics to Xarray.

Features and functionality

• Gravity functionals: (convert from Stokes coefficients to various gravity functionals, such as equivalent water heights, geoid changes, etc.)
• Filter (e.g. Gaussian or anisotropic filter applied in the spectral domain)
• The use of Xarray-like operations allow for applying functionality to multi-dimensional datasets
• A spectral sea level equation solver

Getting started

The tutorials in the documentation provide Jupyter Notebooks with examples of how to make use of the module. The notebooks can also be found on the github repository.

The functionality of shxarray becomes available when importing the module together with Xarray:

import shxarray
import xarray as xr

after which the shxarray accessor becomes available for use, e.g.:

nmax=20
nmin=2
dazeros=xr.DataArray.sh.ones(nmax=nmax,nmin=nmin)

Installation

You can install this package from PyPi using:

pip install shxarray

Backends

Shxarray comes with a default shlib backend written in C++ and Cython. In addition, a very fast 'shtns' backend can be used when SHTns is installed. The backends can be specified in enabled routines as the options: engine='shlib' or engine='shtns'.

Development Installation

If you want to help in the development of this package, it's best to clone the repository to allow for modifications and pull requests. The extension makes use of Cython generated code to speed up spherical harmonic synthesis and analysis.

1. Create your own virtual environment with venv or Anaconda (Optional but recommended, when a user installation is desired)
2. Clone this repository git clone https://github.com/ITC-Water-Resources/shxarray.git
3. Change to the repository directory cd shaxarray
4. Set the environment variable export USE_CYTHON=1 (Optional and only in the case Cython code is being developed or needs to be regenerated)
5. Install using pip pip install . or use pip install -e . for an editable install

Cython build tip on an editable install

From the repository root directory, regenerating the shared library running

python ./setup.py build_ext

will be much faster than using

pip install -e .

This will build the shared library in for example ./build/lib.linux-x86_64-cpython-3xx/shxarray/shlib.cpython-3xx-x86_64-linux-gnu.so. To make sure changes are picked up in your editable install you should create a symbolic link in the Python part of the library e.g. :

cd src/shxarray/
ln -sf ../../build/lib.linux-x86_64-cpython-311/shxarray/shlib.cpython-311-x86_64-linux-gnu.so

Numpy version issues

The provided c++ files are cythonized against numpy > 2. When building against older numpy versions (<2), the cpp files are re-cythonized upon install, this requires a working cython installation.

Contributing

This repository is under development and contributions and feedback is welcome.

Contributors

• Main developer: Roelof Rietbroek (r.rietbroek@utwente.nl)
• Kiana Karimi