CMIP7 Diagnostics from CICE5

[anton-seaice]

I ported the CICE6 diagnostics (into CICE5), and add the new names for CMIP7 diagnostics in ACCESS-NRI/cice5#35

There should be example output in

/g/data/tm70/as2285/experiments/esm1.6-cmip7_vars

The goal here is to provide all core and high 2d diagnostics as native output from CICE5. Native output is of the form where the diagnostic and variable are correct to the CMIP7 data request.

Some notes:

• sisnconc - snow area fraction. This is not prognostic in cice/only the ocean grid with the UM coupling, but is listed as essential on the data request. So I've added it so its recalculated in the same way as the UM calculates is, but we could also not provide it. I am note sure the aice weighting is correct yet.
• sitempsnic - Snow-ice interface temperature. This is not calculated in our CICE5, but I can add following the CICE6 method Edit: nevermind, its not requested
• The example output has aicenmin = puny = 1e-11
• Variable names should be correct, but other metadata needs postprocessing (e.g. Cell Methods, global attributes etc). I assume we do these in Mopper @rbeucher
• The 1d diagnostics (siareaacrossline, simassacrossline, siarean, siareas, sisnmassn, sisnmasss, sivoln, sivols, siextentn, siextents) will need post processing - I assume we do these in Mopper @rbeucher

@ofa001 @blimlim - If you have time to have a look at the example output that would be great
@rbeucher - If you have time to look and confirm the metadata can be added and 1d diags can be calculated that would be great. (I think the post processing calculations might need to use the new variable names).
@aidanheerdegen - I think all the questions about compression, number of files etc from the previous lot of example output should be resolved.

Please share errors found below - and any validation done too :)

[anton-seaice]

A few notes to follow up:

• should tarea be renamed to areacello?
• We haven't been saving the ice_age tracer, so ice can't be older than 1843. Maybe we should reset to 0 at the start of the official PI control ??
• sitemptop isn't available (its calculated in the UM), but is a core variable requested on the ocean grid Its probably too late to fix this?

[aidanheerdegen]

@aidanheerdegen - I think all the questions about compression, number of files etc from the previous lot of example output should be resolved.

I can't recall the context for this, so am at a bit of a loss to make any meaningful comment.

[anton-seaice]

@aidanheerdegen - I think all the questions about compression, number of files etc from the previous lot of example output should be resolved.

I can't recall the context for this, so am at a bit of a loss to make any meaningful comment.

ACCESS-NRI/cice5#58

[ofa001]

@anton-seaice I suggested that we save sitemptop in the UM as it matches the UM fluxes we are calculating there we would don't capture it easily during the zero sea layer sea code.

[ofa001]

Yes @anton-seaice saving ice age tracer from the start of the pre-industrial control that we are going to submit makes sense, running it though the long spin up doesn't make sense, it could build up a lot in places where the ice is "stuck".

[aidanheerdegen]

ACCESS-NRI/cice5#58

Oh yes, thanks for the reminder. Assuming the file sizes are similar to what the nco concatenation produces: yes, all good.

[har917]

@anton-seaice @ofa001 Following up on the discussion at the ESM1.6 standup today - fld_s03i298 maps onto UM variable ei (in diagnostics_bl) - this is the land/sea weighted average of ei_land and ei_sice (in sf_impl)

ei_sice is a copy of fqw_ice - the flux of water to/from ice over ice-parts of the grid cell - as calculated by UM/MOSES. It can be either sign (sublimation >0 or frosting < 0).

Similarly ei_land is the flux of water to/from land (mainly to/from snow but can also be frost onto non-snow surfaces) - passed by CABLE into the UM/MOSES variable. It can also be either sign.

I would expect that you can get quite a bit of sublimation and frosting occurring when 'warm' and moist ocean air travels over cold, dry land in the coastal regions - so this is likely why you see high values there.

Note if you are interested in the sea-ice component of sublimation/frosting then this comes out in fld_s03i353 (may need a STASH-C change) - note this is the average over the sea-ice proportion of a grid cell not the grid cell average.

[ofa001]

Thanks @har917 yes it was the sublimation field fld_s03i298 in the UM that we saw the link to the CICE field, glad that even in the the JULES sf_impl routines sublimation/evap could have both signs and the output was the combination of the ice and land outputs. I also plotted the latent heat field from the Um and again a sign change is present close to the coast but its less clear that it can be the combination of land and ice/ocean effects.

The plots are in the middle of this discussion which Chris didnt link to this mornings minutes it too many other things in it.

ACCESS-NRI/cice5#35 (comment)

[anton-seaice]

Thanks @har917 - i will make sure to note this in the docs about the sea ice diagnostics