Define, track, share, and discover features for ML-based postprocessing of weather forecasts.
Clone the mlpp-features
repository
git clone git@github.com:MeteoSwiss/mlpp-features.git
cd mlpp-features
Install the project via
pip install -e .
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.