Update conditions for Southern Ocean ice and river runoff removal

[cbegeman]

This PR modifies the conditions for river and ice runoff removal in E3SM cases. This changes the behavior of E3SM cases where data icebergs (DIB) or ice-shelf melting (ISMF) were active and thus doesn't not affect water cycle simulations.

River runoff

• Previous behavior: never removed
• New behavior: removed when atmosphere is not active and ice shelf melt fluxes are on

Ice runoff

• Previous behavior: removed when DIB is on and atmosphere is not active
• New behavior: removed when DIB is on

[BFB] when DIB and DISMF are inactive (all WC configurations)
[non-BFB] when DIB or DISMF are active

[xylar]

Great, thanks @cbegeman! I'll run a few quick tests with various combinations of G and B cases and with or without DIB and DISMF

[cbegeman]

Testing

I have run a comparison between this branch and master with the e3sm_cryo_developers suite on chrys with intel, impi.

The result was NLFAIL ${TEST_NAME} (but otherwise OK) RUN for all tests. In each case, the master namelist included

config_remove_ais_ice_runoff = .false.
config_remove_ais_river_runoff = .false.

whereas for this branch those options are

config_remove_ais_ice_runoff = .true.
config_remove_ais_river_runoff = .true.

[cbegeman]

@xylar If you could run a few tests other than the ones in the cryo dev suite or tests with a different machine, compiler combo than chrys, intel, impi that would be great!

[xylar]

Okay, I'll do that.

I don't think I would have expected the cryo developer tests to be BFB. Was that a surprise to you, too?

[cbegeman]

I don't think I would have expected the cryo developer tests to be BFB. Was that a surprise to you, too?

Yes, it was. There are both CORE-IAF and JRA-forced runs in the suite and both should have river runoff and JRA should have ice runoff. Do you think these tests could just be too short?

[xylar]

Do you think these tests could just be too short?

I wouldn't have thought so. I think they run for ~5 days, which should be long enough for removed runoff to affect ocean T and S. But maybe not fields the coupler sees? If not, that would be disconcerting!

This may point to another situation (G cases) where the fact that we don't look at MPAS-Ocean or -Seaice history files is a disaster waiting to happen.

@jonbob, do you have thoughts on this?

[jonbob]

We usually see impacts in the cpl fields, but not always. But I would expect those tests to be non-BFB. Have we run longer comparisons? I don't see any referenced in this PR

[xylar]

No, we haven't. We presumably need to run a longer Icos G case with master and with this to show that we're not seeing unexpected differences. That would be certain to be non-BFB.

[xylar]

I can get that going.

[jonbob]

I think it would be necessary to document the impact of this PR

[xylar]

@jonbob, would a 10-year G-case be sufficient? I would think so.

[jonbob]

@xylar -- I agree, a 10-year G-case comparison would be perfect

[cbegeman]

@xylar Thanks for doing that test!