Export evaporative flux from ATM

[DeniseWorthen]

Description

Adds the evaporative flux to the available export fields from the ATM.

Also adds a new procedure within the block_data_copy interface to prevent scaling of export values where the field contains a special value.

Issue(s) addressed

• fixes Add Export of evaporative flux for use by ocean #671
• provides an incremental fix to resolution of CMEPS Issue CMEPS is forced to use custom merges because sign of field imported from ATM is wrong or unit conversion is required CMEPS#93

Testing

Testing results will be posted in the associated UFS PR ufs-community/ufs-weather-model#1812

In brief, coupled tests using an active ATM and MOM6 will change baselines.

Dependencies

If testing this branch requires non-default branches in other repositories, list them.
Those branches should have matching names (ideally)

• depends on use evap from atm CMEPS#96

Do PRs in upstream repositories need to be merged first?
If so add the "waiting for other repos" label and list the upstream PRs

• associated with Send Evap rate from ATM to OCN ufs-community/ufs-weather-model#1812

[junwang-noaa]

Just to confirm the scaled fields are only used for coupling with ocn, not by aerosol models. Otherwise the code looks good to me.

[DeniseWorthen]

@junwang-noaa The scaling is only applied to the "mean" fields, which I don't believe are used anywhere else (for coupling itself). The full RTs have passed for all tests with only coupled tests changing. I will double check the first fields passed out of the ATM when aerosols are included.

[DeniseWorthen]

Fixing the scaling of the special value (huge) by rtime does change the fields coming out of the ATM at the first coupling timestep (dt=720s) when aerosols are included, but only at those points which contain the special value. They are otherwise identical. The four fields affected are the mean taux,y, sensible and latent. These fields have different maximum values (1.384612495609288E+34 vs 9.969209968386869E+36).

The sfcf000.tileX.nc files are identical.

[junwang-noaa]

Do you still have those files that I can take a look? Thanks

[DeniseWorthen]

Yes, the develop output is in /scratch1/NCEPDEV/stmp2/Denise.Worthen/FV3_RT/rt_9995/s2swa.dev. This is a sandbox of the /scratch1/NCEPDEV/stmp2/Denise.Worthen/FV3_RT/rt_9995/cpld_control_p8_intel with only the executable changed.

[junwang-noaa]

I am curious, does the coupled test without aerosol have the same changes in the mean taux,y, sensible and latent?

[DeniseWorthen]

I've done another test using the current develop branch, but pointing to my FV3 branch. In this case, CMEPS is still using the laten_heat exported by the ATM so no answer changes are expected, but the exported fields will contain the impact of the scaling change in atmos_model. In my sandbox, s2swa.dev and s2swa.scale.dev have idential sfc and restart files.

The mediator files (history and restart) show the only difference is the max value---the actual values are the same, they differ only in the areas which have a value huge. In one case it is the actual value huge, in the other it is huge*rtime.

[junwang-noaa]

Thanks for the testing and confirming the change is only on the max values.