adding lai_dtlimit to namelist

[wwieder]

Description of changes

Allow dtlimit to be set for the LAI streams

Specific notes

Contributors other than yourself, if any:

CTSM Issues Fixed (include github issue #):

Are answers expected to change (and if so in what way)? Only potentially if used.

Any User Interface Changes (namelist or namelist defaults changes)?
New lai_dtlimit namelist item.

Testing performed, if any:

[ekluzek]

Hey @wwieder nice work here!

There's one critical change that has to happen and that's to add the shr_mpi_bcast call for lai_dtlimit. And this was missing for lai_mapalgo so you should also add it for it as well.

The other things I have are suggestions. One is suggestion about more text for the description in the namelist_definition file. Another is to add lai_dtlimit and lai_mapalgo to the log output. The others are linked together and it's for setting the default value in the FORTRAN code rather than the namelist.

The reason I suggest setting the default in the FORTRAN code rather than in the namelist, iis because this is likely something that will be rarely changed by users. So it's nice to hide it. If you add the default in the namelist defaults file, it will ALWAYS show up in the lnd_in file. This is useful to do for things that you want to remind users that they are available -- or for items that are often changed for different setups.

[ekluzek]

Above this there needs to be a shr_mpi_bcast call for lai_dtlimit as well (and it looks like this is missing for lai_mapalgo). I also suggest adding to the logging write statements as well (and same for lai_mapalgo).

[wwieder]

done

[wwieder]

@ekluzek I think I've done this correctly, notably the critical part to "add shr_mpi_bcast call for lai_dtlimit.", but can you please confirm?

If this look OK, I think we can merge this PR to @olyson's PR and maybe @slevis-lmwg work to bring this to b4b-dev? Having more functional usermods would be helpful for @TeaganKing to make progress on the larger PLUMBER project.

[ekluzek]

Just one change here, lai_dtlimit is a real rather than character, so remove the trim() function call around it.

[ekluzek]

Everything else looks great. So I'll mark as good to go...

[wwieder]

Thanks for your suggestions, @ekluzek. I'll get to those tomorrow.

Here I'm posting changes to usermods that are a bit cleaner and seem to work for spinup of SP and BGC cases. Transient cases also run as expected, but fail with this error associated with cycling back to the start of the datm file

forrtl: severe (61): format/variable-type mismatch, unit -131, file /glade/u/home/wwieder/scratch/PLUMBER2_tests/ZM-Mon_hist_BGC_test1/run/atm.log.4116681.desched1.240414-083718

[olyson]

I was looking over the differences between your BGC/SP hist cases and my own. One thing I see is that you are running NO_LEAP, while I ran GREGORIAN. The datm forcing and lai streams both have data for 2008 leap day. I'm not sure if that has to do with the errors you are seeing though

[olyson]

Another thing to try is to set taxmode to 'extend'. My case seems to come of the box with that setting. The forcing at the end of the run is unique (the last two timesteps don't use the same forcing), but maybe it needs to read the the last time sample in the file one more time even though it doesn't use the data.
I was able to rerun my case using release-clm5.0.37, which has been ported to derecho. It runs fine with taxmode set to extend, but fails if I set it to 'cycle'.
On the other hand, the taxmode for the lai streams is 'cycle' and I haven't changed it. But I do have dtlimit for the lai streams set very high (1.e30; following some other example I had). So you could try changing that also.

[wwieder]

ah, @olyson this is brilliant! I've add this to the PR so that transient runs (using a HIST compset) are set with taxmode=extend and dtlimit=1e30 (required for some reason), but spinup (using an I2000 compset are cycle and 30, respectively).

Now (I think) we can run both spinup and historical cases using usermods that are customized for each site!

One additional notes here, I had to set lai_dtlimit=1e30 for the transient SP cases too.

I'm going to do a few more tests, but I think this is looking good!

[wwieder]

@slevis-lmwg I realize that merging this part of the PR onto the CTSM b4b-dev or master will require additional testing, as aspects of the code have changed with addition of dtlimit to lai_streams

[slevis-lmwg]

@ekluzek @wwieder
I see at the top of the page that merging this PR will merge to olyson:PLUMBERcsv

Is this the intent, or does this PR need rebasing to b4b-dev?