renku datasets and initial notebooks

• git lfs track nc and zarr

• add renku datasets notebook

• add renku datasets:

  • hindcast-like-observations_2000-2019_biweekly_deterministic
  • forecast-like-observations_2020_biweekly_deterministic
  • hindcast-like-observations_2000-2019_biweekly_tercile-edges
  • hindcast-like-observations_2000-2019_biweekly_terciled
  • forecast-like-observations_2020_biweekly_terciled
  • ecmwf_forecast-input_2020_biweekly_deterministic
  • ecmwf_hindcast-input_2000-2019_biweekly_deterministic

• add verification notebook

• add ML notebook based on weatherbench

• link submission file to website

• add reproducibility section below in training

• how to deal with this dry mask? provide as renku dataset? now implicitly masked in categorized observations obs_p

• benchmark working on renku: wait for climetlab release: week 5-6, tp and 2000-2019 dont wait

• dont commit nc and zarr, only with git lfs

• dont commit figures

• create nc for files <100MB, zarr if >100MB?

• justify if training takes more than a week

• show RPS for all years

• test & dry run

• create one renku dataset with all files: s2s-ai-challenge

• add mean bias reduction dataset