This repository contains a Jupyter notebook that trains a U-Net to reconstruct missing values in RADOLAN precipitation fields.
uv sync --locked --group dev
uv run python -m ipykernel install --user --name hackfill --display-name "Python (hackfill)"Open infill_radolan.ipynb and select the Python (hackfill) kernel. Update
the data, observation and mask paths in the notebook.
The RADOLAN data, model checkpoints, logs, and local data paths are not included.