2024-08-29 Community meeting

Move to Slack

• Everyone present in the meeting on slack
    - If you cannot install slack, then you can use the web version
    - The meetings will still be on teams

Comparing weather sources
- Alliander: Found that Meteomatics(paid source) provides better forecasts than KNMI. DWD ICON provides long term forecasts of very similar quality as KNMI short term
- Jan Maarten: Uses Local KNMI equivalent (SMHI, 10 day ahead), netgroup (company that makes forecast) and NET
- Maxime: Found ICMWF to be good, a global weather forecast model with lower resolution compared to the non-global ones. Harome did not provide good forecasts compared to other models. The global weather model works for every country, so can be used by everyone! It is a paid model. Will switch from meteo-france to ICMWF
- RTE-Int -> uses openmeteo, no comparison done to other sources

Customer forecasts are bad because curtailment:
- Idea JM: use pvlib/physical model with park specs. Already being explored by Alliander
- Idea Maxime: restore based on customer provided missed generation.
  - Data comes in too late for production process, but could be used to evaluation of the past
- other idea by maxime: combine multiple signals to... [this part was not included in the minutes]

Maxime: are we going to include Dagster in openstef?
- Alliander currently has a recorded demo we did on dagster
- Alliander probably could include dagster implementation in openstef
- Larger question: do we want openstef to include infrastructure?
  - JM: could be definitely useful as a separated infrastructure example repo, good to check with Jonas