A low resolution version of NorESM2.5

[adagj]

Please list what features are wanted for a low resolution version of NorESM2.5 .
Please share good arguments for having a low resolution versions and good arguments for only maintaining one version.
Please be pragmatic, but keep a friendly tone :)

[rosiealice]

(Good?) arguments for low -res are a lot to do with allowing parameter exploration (which is important), investigation of alternative scenarios, longer scenarios (incl paleo) and initial condition ensembles.

Arguments for only-one-version: Two versions are harder, some people are skeptical about low-res models, particularly wrt extreme events.

Do both need to be ready on the same timescale? Seems likely that the low res version would not be the CMIP7 version...

[MichaelSchulzMETNO]

From the discussion in Bergen:

• many required a lower resolution version of NorESM2.5
• Arguments for having a lower resolution mentioned were quicker science and development testing, paleo runs, possible bottlenecks at sigma2 if a higher resolved version would be very costly, potential to downscale lower resolution run to higher resolution with ML, less storage required
• Problems identified: Much more work to set up and tune a second version, likely no NCAR support for a 2 degree CAM and CTSM version,
• Wall clock time gain should be estimated (eg 1 degree CAM&CTSM+ 1/4 BLOM versus 2 degree CAM&CTSM+ 1/2 BLOM)
• Unclear additional work amount needed for forcings & setup for low resolution version