CAM support for writing 10m wind component diagnostics

[IngoBethke]

A long standing issue has been that CESM (and by extension NorESM) is missing 10 m U,V diagnostic output. These fields are desired by downscaling activities and many MIPs.

I have prepared an implementation based on NorESM2.0.6 and attached SourceMods to this discussion thread.

The implementation of UAS and VAS is quite complex because U10 is not computed in the atmosphere itself but is computed in the sea ice, land and coupler and then sent to the atmosphere. I am afraid I won't find the time to do more extensive testing and to properly add the implementation to the respective NorESM code repositories (i.e., CIME, CAM, CICE, CLM). Nevertheless, I want to share it with you in case you need to produce UAS and VAS output in some future project.

In order to write 10 m wind components, one has to add "UAS" and "VAS" via fincl in CAM's namelist.

The UAS and VAS patterns look very similar to UBOT and VBOT but have a smaller magnitude (as expected) according to a visual comparison with ncview.
SourceMods_uasvas_NorESM2.0.6.tar.gz

[TomasTorsvik]

Thanks @IngoBethke , this looks to be a very useful addition to the NorESM diagnostic output.

Having the sourcemods available here in the discussions is nice, but it won't be easily visible for model users, so I think we should try to incorporate these code changes in the model components.

[TomasTorsvik]

Hi @IngoBethke ,
I have copied in your SourceMods files to my personal repositories, you can see the commit changes below. If this is correct, I can make pull requests to CAM, CICE, CTSM and CIME. One question: should these changes go to the main/release branch of the respective components, or should it be kept in a separate development branch? Equivalently, should the 10m wind diagnostic be calculated in the default model setup, or should it be an optional output?

CAM : TomasTorsvik/CAM-TTfork@18d76f8
CICE5 : TomasTorsvik/CESM_CICE5-TTfork@18311a2
CTSM : TomasTorsvik/CTSM-TTfork@bcceca5
CIME : TomasTorsvik/cime-TTfork@3939757

[IngoBethke]

Hi @TomasTorsvik

Big thanks for doing that and sorry for late reply.

It is maybe safest to have the 10m wind components as optional diagnostic output. Otherwise, users might run into problems when branching from existing simulations that do not include these diagnostics (but I'm not 100% sure if that can be a problem).

Before committing the changes to the main/release branches we probably should do some more verification.

Most importantly, the addition of the diagnostic code should not cause the model state itself to change, cause any compiler error or model crash. Therefore, it would be good to check bit-reproducibility (with vs without SourceMods) and also to test compiling with DEBUG set to true (to detect if all fields are declared properly and array bounds are not violated).

Also, I have only tested a NorESM compset that uses TT's COARE flux implementation. We therefore should do a short test simulation using the CESM's default flux computation, just to check that the model compiles and runs properly and the resulting UAS, VAS diagnostics have physical values over the ocean.

Concerning verification of the UAS, VAS diagnostics, I think it is sufficient to have a look using ncview and visually compare the fields with UBOT, VBOT.

I would have liked to help with the above but am too busy these days.