Coupling CICE/ICEPACK to GSFC GEOS model

[bzhao]

Dear consortium,

We have started the process of incorporating CICE6 in our coupled model framework GEOS. The original approach as described here had to be abandoned because of various issues. We have charted a new approach which is more in line with the coupling strategy adopted by other modeling centers. We'd like to get some ideas as to the following 2 main tasks:

1. Build a driver for GEOS: we need the driver which sort of follows the old ESMF protocol (now depreciated), but some existing ones could be used as templates for implementation. I see there is a github issue related to the driver development and wonder discussions could happen there.

2. Partially related to 1, GEOS surface scheme requires the thermo interface in ICEPACK to be done in a different way. Again, a guithub issue touches upon this issue.

So shall we continue the dialog on github or here in the forum?

Thank you for your input.

Regards,

Bin

 

[eclare@lanl_gov]

Hi Bin, nice to hear from you. Please provide more details of what you need here in the forum, and then we can move the discussion if/when it makes sense.

The driver github issue is more about unifying drivers so there aren't so many to maintain. That might not happen for a while, and in the meantime I suggest that you look through the various ones and see which one most closely matches your needs, then make a copy that can be 'personalized' for your needs. @dbailey might be able to point you to a good starting place re ESMF.

Alternative thermo coupling is a bigger problem to tackle. Do you have a write-up of some sort that describes what you need? Some other groups are doing things differently, and perhaps we can coordinate a larger discussion.

 

[bzhao]

Hi Elizabeth, I went through the existing drivers and the two nuopc ones seem to be closer to ours. I 'll start with cmeps which looks very comprehensive. Of course, David please feel free to suggest any other paths.

As to the thermo coupling, we have the same interface change requirement as in GFDL SIS2 as described here (only item 1). Additionally, the evaporation/sublimation flux needs to be handled a bit differently as well. I'll try to have more details about this later.

Is it a good idea to start the driver development using CICE/dg_driver.rst at main · CICE-Consortium/CICE as guidance?

Thank you for the reply.

Bin

 

[dbailey]

I guess I am confused. Are you talking about a standalone driver? Or just a coupling interface for the NASA sea ice component? If you look in the CICE code you'll see that we have a few calls to icepack routines.

icepack_step_therm1
icepack_step_therm2
icepack_step_radiation

....

One could potentially break these down further and call directly into some of the icepack subroutines. The gist of the github issue is to try to simplify the calling interfaces to icepack for easier coupling to other sea ice components. We certainly have a NUOPC (ESMF+) cap to call CICE in CESM, but not Icepack directly.

 

[bzhao]

David, sorry for not being clear. We need a driver to couple CICE to our AGCM/OGCM components. We have got a wrapper/container for CICE which calls driver routines (init, run and finalize) and this is very similar to what is in CESM.

As to ICEPACK thermo coupling, I am talking about the potential changes to the interface of temperature_changes_salinity routine. I need to revisit the interface again to get me better oriented.

Thank you.

Bin