Smoke/Dust updates for RRFS code freeze

[haiqinli]

1. Include convective transport and wet removal of smoke/dust in the GF convection.
2. A new dry deposition option for smoke/dust.
3. Update the diurnal cycle of biomass burning emissions.
4. Clean the smoke code.

[lisa-bengtsson]

Hi Haiqin, should these updates be in C3 as well?

[haiqinli]

Hi Haiqin, should these updates be in C3 as well?

Hi Lisa, yes, we will include these updates in C3 when we have more time to test it.

[dustinswales]

Some changes are needed wrt the precision, but otherwise looks fine to me.

[dustinswales]

Since this touches GPU functionality in the GF scheme, I think it's a good idea to have one of the GPU developers review the proposed changes (@timsliwinski-noaa).

[danielabdi-noaa]

@dustinswales I have made a few obvious suggestions to make it GPU complaint, but this is an extensive set of changes that really need to be tested to figure out all the GPU related issues. New smoke related modules may need to be ported to the GPU as well. Adding OpenACC pragmas, even incorrect ones, will not break the CPU code so it is safe in that regard.

[grantfirl]

@haiqinli When is the RRFS code freeze and when does this need to be merged? I'm concerned that there are a lot of requested changes and testing the GPU fixes isn't even accomplished through the normal RT process.

[dustinswales]

@danielabdi-noaa Thanks for your review.
@grantfirl As you mentioned, the GPU concerns are a whole other can of worms, which we are still figuring out how to test.
From my understanding, the GF scheme's GPU functionality is currently not working and Daniel is working on that.
Should we accept these recommendation from Daniel in this PR? Or include them in a follow up PR with other tested (in the SCM) changes? (And maybe(?) we can harden up the scripts to easily run the SCM GPU tests on Hera?)

[mkavulich]

A few suggested changes, mostly minor

[haiqinli]

@haiqinli When is the RRFS code freeze and when does this need to be merged? I'm concerned that there are a lot of requested changes and testing the GPU fixes isn't even accomplished through the normal RT process.

@grantfirl In the RRFS-SD PR, which uses the vegfrac from RUC LSM. However, the conus13km_debug_intel case crashed when using vegfrac from RUC LSM. We just figured it out that the vegfrac is not available in the regression test run case of conus13km_debug_intel/INPUT/oro_data.nc. I will redo the regression test and mark the reason of crashed cases like conus13km_debug_intel in the PR. It is expected to be ready tomorrow morning when the regression test is done. Thanks.

[jkbk2004]

@haiqinli can you resolve the suggestions to get final approvals on this pr?

[haiqinli]

@haiqinli can you resolve the suggestions to get final approvals on this pr?

Sure, I am working on it, and should be done soon. Thanks.

[grantfirl]

@haiqinli As long as you push the change related to the intent of nchem, that is good enough. I'm sorry to review so late. I'll approve and resolve the comments so that we can move this along for your deadline. Thanks for getting the GPU changes in, although we (CCPP code managers) have no way to test this right now.

Comments have been addressed.

[haiqinli]

@grantfirl Thank you very much for your help. The PR has been updated to include nchem intent declaration.

[jkbk2004]

All tests are done and confirmed at ufs-community/ufs-weather-model#2024. @grantfirl @Qingfu-Liu @dustinswales Can you merge in this pr?