Have CISM abort with an error when NOEVOLVE is chosen in the compset?

[ekluzek]

In CTSM we ran into this problem when trying to run with CISM2%NOEVOLVE.

https://github.com/ESCOMP/CTSM/issues/2561

Could this be made into something that is disallowed with an error? I suspect it wouldn't be hard to do. And certainly should be as NOEVOLVE is currently being removed and moved to DGLC in CDEPS. So NOEVOLVE should be marked as deprecated and something that shouldn't be used.

Does that sound like something that the CISM team could do?

[billsacks]

Thanks for opening this @ekluzek .

There are some subtleties here: NOEVOLVE may still be needed in cases where there are multiple ice sheets (e.g., Greenland and Antarctica in the most typical case), one or more of which are evolving and one or more of which are non-evolving.

That said, @ekluzek 's suggestion makes a lot of sense as long as it's implemented via a check that at least one ice sheet is evolving. I haven't checked the buildnml logic carefully, but I think this would fit in pretty easily to the existing logic. Basically, the logic that checks for the overall CISM_EVOLVE / GLC_EVOLVE (I forget which it's called now) can / should be changed to confirm that this overall variable should end up True.

[mvertens]

@billsacks - DGLC is written to handle multiple ice sheets - so do not thing we need to continue to have a CISM%NOEVOLVE mode.

[billsacks]

The potential issue is that there may be situations where, for example, we want an evolving Greenland but a non-evolving Antarctica. CISM can handle that situation currently; I'm not sure how we can do it without having CISM NOEVOLVE for Antarctica. (Note that, in a situation like that, the overall GLC_EVOLVE is currently set to True, so from the overall system perspective it looks like an evolving CISM.)

[mvertens]

@billsacks - thanks for the clarification on the use case. That makes total sense.