Enable UFS-GOCART integration

[rmontuoro]

This PR introduces changes required to integrate the NASA GOCART model into the UFS (#663) along with a few bug fixes.

Added features:

• Exported 10-meter wind components
• Exported instantaneous 3D non-convective liquid and ice precipitation fluxes from GFDL cloud microphysics

Fixes:

• Properly handle error flag in PBL generic scheme when aerosol diffusion is enabled. This bug prevented aerosols from being diffused within the PBL
• Initialize work arrays in SAMF deep and shallow convection schemes
• Initialize wet deposition work array and ensure only proper elements of cloud base mass flux array are used in SAMF aerosol deep and shallow convective transport schemes
• Export correct values for the upward sensible heat flux over the ocean when cplflx is .true. and coupling with aerosol component is enabled.

Regression test log files for the following platforms available at https://github.com/rmontuoro/ufs-weather-model/tree/feature/gocart:

• hera.intel
• orion.intel
• wcoss_dell_p3

[climbfuji]

Looks good, just one tiny change to revert.

[climbfuji]

Looks good, will approve when the regression testing is done. Thanks!

[grantfirl]

Just a question: I understand that it is better for memory access to have the nested loop in k,i order, but does inserting the IF statement within the inner loop change the computational performance in practice such that the loop order should be reversed? I.e. Does the benefit of ordering the loop k,i overcome the expense of running IF i*k times? (Also repeated below)

[rmontuoro]

The cost (CPU cycles) associated with memory access is usually much larger than the cost of branching inside loops due to optimized branch prediction available in modern processors.

[grantfirl]

Thanks for the explanation!

[grantfirl]

Looks good. One comment about inserting an i-dependent IF statement within a nested k,i loop.

[climbfuji]

@SMoorthi-emc please take a look at the RAS changes: 7f6e45e - thanks!

[SMoorthi-emc]

Dom, I assume "itc" is the starting address and "ntc" is the ending address of chemistry variables, to be subjected to scavenging. So, itc is >2 and ntc could be <=ntr and the dimension of "fscav) would be (ntc-itc+1). If so, don't you think it should be "fscav_(itc:ntc) = fscav"? For example, say there is only "one" chemistry variable. Then itc=ntc right? Moorthi

…

On Wed, Jun 30, 2021 at 2:00 PM Dom Heinzeller ***@***.***> wrote: @SMoorthi-emc <https://github.com/SMoorthi-emc> please take a look at the RAS changes: 7f6e45e <7f6e45e> - thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#664 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALLVRYW5BKNSMKNUMEY3TM3TVNLVLANCNFSM45MSYEOA> .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718 e-mail: ***@***.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[rmontuoro]

I agree with @SMoorthi-emc - this should be

My previous email included a typo. Apologies

[climbfuji]

Dom, I assume "itc" is the starting address and "ntc" is the ending address of chemistry variables, to be subjected to scavenging. So, itc is >2 and ntc could be <=ntr and the dimension of "fscav) would be (ntc-itc+1). If so, don't you think it should be "fscav_(itc:ntc) = fscav"? For example, say there is only "one" chemistry variable. Then itc=ntc right? Moorthi
…

Of course, Moorthi. Thanks for catching it, I updated the code.