A Python library to write individual tasks in Numerical Weather Prediction pipelines.
Experiments in Numerical Weather Prediction (NWP) and related fields consist in a series of computational tasks that can depend on each other's output data. Each task is typically made of three successive steps:
- Fetch required input data.
- Execute a program.
- Make the program's output data available to subsequent tasks in the pipeline.
Tasks have historically been written in some variant of the UNIX shell, which was convenient to interact with the file system, manage environment variables and execute programs. As NWP pipelines and tasks grow more and more complex, however, there is a need for a language providing more abstraction and code reuse mechanisms.
On top of the popular Python language, vortex provides abstractions that encapsulate running -- potentially distributed -- programs as well as fetching and storing the data they consume and generate.
The documentation is available at vortex-nwp.readthedocs.io.
Vortex can be installed using pip like most Python packages:
pip install vortex-nwpSee CONTRIBUTING.md.