Flexible and extensible bulk flux formulae

[sanAkel]

Goal:

Seek inputs from @DeniseWorthen, @NickSzapiro-NOAA and other contributors to this repository.

Topic:

• Aerobulk provides a clean interface to many bulk-flux formulae.
• Adding hooks to it will be a good benefit to tuning models.
• Also aids community collaboration, via addition of COARE since its developers have also expressed willingness to @sanAkel on pooling together.

What has been done so far?

raphaeldussin hooked it (= aerobulk) to ocean-only MOM6 - which is a very good starting point for an ocean-only configuration.

Specific questions:

• (Please add/modify as needed).
• What would have to be done to provide necessary connections between Aerobulk and CMEPS - ideally with a markdown list?
• So we can scope out the work for developing the code and testing it. @sanAkel would like to first use it in RTOFS, i.e., UFS DATM.
• Will this effort aid us with the exchange grid testing (@uturuncoglu ) ?
• How will it relate to MPAS coupling issues (@DusanJovic-NOAA ) ?

Related issue(s):

• refactor ufs/ flux_atmocn_mod.F90 #160

cc: @JacobCarley-NOAA, @yangfanglin

[yangfanglin]

@sanAkel Santha, Can you provide a bit more background information about AeroBulk? From what I can tell, it is a Fortran90 library developed around 2015–2017 to provide a unified implementation of bulk aerodynamic algorithms (e.g., COARE, ECMWF, NCAR) for computing turbulent air–sea fluxes. I only see two contributors so far. Who are the current maintainers/owners, and is the code still actively maintained? Also, what is the intended integration path for us: copying selected code/subroutines into our system, or treating it as an external dependency (e.g., submodule/library)? We should be cautious before introducing additional external dependencies, especially for something that may have limited maintenance or community support.

[JacobCarley-NOAA]

I concur with @yangfanglin . What is missing that this library is providing? In Issue #160 it seems it's proposed that this should be put in MOM6 and not CMEPS?

[sanAkel]

Thanks for your comments, please see ⬇️ for point-by-point replies.

@yangfanglin

Can you provide a bit more background information about AeroBulk? From what I can tell, it is a Fortran90 library developed around 2015–2017 to provide a unified implementation of bulk aerodynamic algorithms (e.g., COARE, ECMWF, NCAR) for computing turbulent air–sea fluxes. I only see two contributors so far. Who are the current maintainers/owners, and is the code still actively maintained?

• Please see Brodeau et al., 2017 for full details.
• True that it has 2 contributors and I do not know @brodeau's plans for future development/active maintenance, though he seems to updating it.

Also, what is the intended integration path for us: copying selected code/subroutines into our system, or treating it as an external dependency (e.g., submodule/library)?

Add it as a submodule to MOM6 - which has support from MOM6 developers (at GFDL: @marshallward and @Hallberg-NOAA AFAIK).

We should be cautious before introducing additional external dependencies, especially for something that may have limited maintenance or community support.

👍 💯 agreed!

Having (options for) different bulk fluxes will be beneficial, but if there is a different alternative, we can consider it. 🙏

---

@JacobCarley-NOAA

What is missing that this library is providing?

At present, we have access to a single bulk flux algorithm,, with a hook to aerobulk there will be more choices.

In Issue #160 it seems it's proposed that this should be put in MOM6 and not CMEPS?

Yes, that was the plan @raphaeldussin and myself had. Once MOM6 includes it as a submodule, CMEPS may (or may not) need any changes- which is why I thought its good to start coordination now.

[sanAkel]

MOM6 sub-module addition https://github.com/raphaeldussin/MOM6/tree/05349834beb004bc29a192b000a418d027a4be29/pkg

[NickSzapiro-NOAA]

I think it's a good idea to have modular transfer coefficients and enable more near-surface physics/coupling/tuning/sensitivities...

It's nice for these bulk flux options to come in well supported via MOM6. I'm not sure if an application without MOM6 would want them

Happy to help set up example regression tests, if it helps