We use the last 10 years of nest-domain data from the VIKING20CJRAshort data pulished on WDCC: 10.26050/WDCC/VIKING20XJRAshort
In scripts/calculate_VIKING20XJRAshort_w.sh, scripts/compress_VIKING20XJRAshort_w.sh, and scripts/download_VIKING20XJRAshort.sh, there's a sequence of batch scripts for downloading the UVT fields and the mesh files, for calculating the vertical velocities, and for compressing the resulting grid_W files.
The W calculation is done with CDFTOOLS in a container built according to https://github.com/willirath/cdftools_container.
There's a notebook notebooks/test_viking_data.ipynb run with notebooks/run_test_viking_data.sh (results in notebooks/test_viking_data.exe.ipynb) which uses all the downloaded data to make sure there are no corrupted files. (Jblob tends to stall sometimes and is not really good at letting the user know what went wrong...)
In notebooks/test_W.ipynb, there's a notebook extracting a time series from all the W files.
Kernel installation is handled in scripts/setup_parcels_env.sh. It essentially follows https://docs.dkrz.de/doc/software&services/jupyterhub/kernels.html#use-your-own-kernel but uses mamba for the env resolution (because it's faster), and creates a Juptyer kernel spec which uses conda run to make sure the kernel is within a properly activated conda env, because we want to use the gcc installed in the env etc.
In notebooks/test_parcels.ipynb, there's a simple Brownian motion test run in the newly installed env.