MLPP Features

Define, track, share, and discover features for ML-based postprocessing of weather forecasts.

⚠️ The code in this repository is currently work-in-progress and not recommended for production use. ⚠️

Installation

Clone the mlpp-features repository

git clone [email protected]:MeteoSwiss/mlpp-features.git
cd mlpp-features

Install the project via

pip install -e .

Getting started

The repo is organised mainly in three parts.
Features are defined in nwp.py, obs.py, terrain.py and time.py depending on their nature.
Some features are not natively produced by NWP models but derived from other variables. Relevant function are defined in the calc module. Geographical calculations are handled in the geo_calculations.py module.
Finally, accessors.py and selectors.py are used to define utility functions.