Docs on Sea Ice Diagnostics

[anton-seaice]

This change adds some documentation and notes on the CMIP7 style diagnostics used in ACCESS-ESM1.6

The preview is here:

https://access-esm16-configs--234.org.readthedocs.build/diagnostics/sea_ice/

The page is generated from a jupyter notebook, the process to generate the page is:

• run the notebook,
• export the notebook as html
• rename the notebook to markdown and replace the markdown file added in this PR.

I tried using the mkdocs-jupyter plugin to render the notebook, which rendered fine and was convenient. However it wasn't compatible with the https://github.com/ACCESS-NRI/access-esm1.6-configs/blob/214-main/documentation/hooks/hide_pages_url_segment.py hook to remove pages/ from our urls. The pages always remained only for the rendered notebook.

[github-actions]

PR Preview
🚀 Preview of PR head f7cd2e1 deployed to https://access-nri.github.io/access-esm1.6-configs/pr-preview-234
2025-10-16 14:25 AEDT
Preview generated through the Deploy to GitHub Pages workflow run 18549306598.

[ofa001]

Hi @anton-seaice just spotted some of this documentation which I cant seem to access, put one of the changes we are using a value of and CICE salinity there is this discrepancy with the setting in the ocean code, Martin has just switched it to 0.005 in the latest run to try and resolve some closure issues (didnt change what we wanted).

[anton-seaice]

Ah ok thanks - don't worry too much about this doc yet. I just put it here to make sure the docs website builds ok

[chrisb13]

Given this is now merged, I'd suggest you rebase or resolve (as below)

Thanks for your help on sorting out the structure!

[anton-seaice]

@blimlim - When the um calculates the sea ice surface temperature (fld_s00i508), what does it use for open water ? It looks like maybe -1.8C ?

[anton-seaice]

@ofa001 - if you have time to look though https://access-esm16-configs--234.org.readthedocs.build/diagnostics/sea_ice/ that would be great

@blimlim - this is ready for review! :)

[blimlim]

@blimlim - When the um calculates the sea ice surface temperature (fld_s00i508), what does it use for open water ? It looks like maybe -1.8C ?

I think that's right. It's a bit confusing because it gets set differently in several different places in the UM. From what I can tell, the last time it's changed in a timestep is here, where it gets set to ltfs = access_tfs = -1.8

[ofa001]

I haven't done the python sections I noticed rlus had a question mark next to it. I would check what is in MOPPER but in APP4 it was a group of fields for ESM1.5. I guess all of these are still being saved in ESM1.6, @bimlim will know.
rlus,yes,fld_s02i207 fld_s02i201 fld_s03i332 fld_s02i205,var[0]-var[1]+var[2]-var[3],W m-2,,up,both,atmos,

Its a bit confusing what is included in relation to snow in siflfwbot which is the term furthest from closure. Should it contain melting snow? IIts a source of freshwater. The snow budget seems to be the bit we have the most issues with and it was with CMIP6 CM2 runs we dropped it out of our comparison with the Met office and their runs. I guess its too late now but something to keep in mind for CM3.

[anton-seaice]

I haven't done the python sections I noticed rlus had a question mark next to it. I would check what is in MOPPER but in APP4 it was a group of fields for ESM1.5. I guess all of these are still being saved in ESM1.6, @bimlim will know. rlus,yes,fld_s02i207 fld_s02i201 fld_s03i332 fld_s02i205,var[0]-var[1]+var[2]-var[3],W m-2,,up,both,atmos,

Ah yes - I thought i looked at it wasn't being saved but that doesn't seem very likely. I have fld_s02i207 marked as downwelling, not upwelling ?

Its a bit confusing what is included in relation to snow in siflfwbot which is the term furthest from closure. Should it contain melting snow? Its a source of freshwater. The snow budget seems to be the bit we have the most issues with and it was with CMIP6 CM2 runs we dropped it out of our comparison with the Met office and their runs. I guess its too late now but something to keep in mind for CM3.

I was convinced that what we do is internally consistent, the cice internal field "fresh" is used for "siflfwbot" and for the coupled melt and form water fluxes.

[ofa001]

@anton-seaice
in ice_therm_vertical.F90 there is an accounting in 'fresh' for the change in ice and snow, surely this includes the snow melt in spring.
dhi = hin(ij) - worki(ij)
dhs = hsn(ij) - works(ij) - hsn_new(ij)

     freshn(i,j) = freshn(i,j) + &
                   evapn(i,j) - &
                   (rhoi*dhi + rhos*dhs) / dt

The new snow hsn_new is not included. But during melting there will be unlikely to be snowfall.

[anton-seaice]

Yes - that is correct. Fresh includes snow melt (and presumably melt pond drainage) but not rain (unless it goes into a melt pond)

[ofa001]

There is this line in ice_step_mod,F90 that includes frain.

do i = 1, nx_block
fresh (i,j,iblk) = fresh(i,j,iblk) &
+ frain(i,j,iblk)*aice(i,j,iblk)

Still not clear why the budget doesn't close nor why it doesn't close whilst advecting terms are considered.

One other thought I had yesterday, when you did put out a sisnconc in those test diagnostics what formula did you use wasit the snow patch one from CICE or the one from the UM radiation code, the values looked reasonable that was the point I was planning to make on how the data looked. Not sure how the UM formulae would be but I guess it would be a modification of 100% cover.

[blimlim]

I've tried taking another look at the budgets, as it would be nice if we could mention likely reasons that they aren't closing. I've probably misunderstood a few things but have a couple of candidates:

Water and thermodynamics

For the comparison of siflfwbot in the plot below, I think there might be a few things going on:

• Our gridbox-mean+intensive processing of the siflfwbot mean. This means when we multiply by the cell area, we don't recover the total kg/s of water transferred, and it might not be possible to get it back.
• I'm pretty unsure, but can't figure out if this change to the snow mass gets accounted for in the melt terms? https://github.com/ACCESS-NRI/cice5/blob/5828e0cd3785ba38d0a51bf0ad92d449024fa95d/source/ice_therm_vertical.F90#L1921-L1928
• There's a zap small areas call in the dynamics, could this add a tiny bit to fresh?
• The coupling can lead to sublimation or condensation over ice free cells, this also gets added to fresh but wouldn't affect the thermodynamics diagnostics https://github.com/ACCESS-NRI/cice5/blob/5828e0cd3785ba38d0a51bf0ad92d449024fa95d/drivers/access/ice_coupling.F90#L48-L53
• As mentioned in the PR, the snow melt from the ridging gets added to fresh but I don't think it goes into melts? https://github.com/ACCESS-NRI/cice5/blob/5828e0cd3785ba38d0a51bf0ad92d449024fa95d/source/ice_mechred.F90#L687-L697

wfimelt and wfiform

I think the above reasons could also contribute to the melt/formation plots not matching exactly:

The melt and form fields from MOM should be made up of the values of fresh at the timesteps where it is positive/negative respectively: https://github.com/ACCESS-NRI/cice5/blob/5828e0cd3785ba38d0a51bf0ad92d449024fa95d/drivers/access/cpl_forcing_handler.F90#L1135-L1138

In a single timestep could it also be possible (maybe unlikely?) to have bottom formation and top melt at the same time? If this happened wfiform would also not line up perfectly with the formation terms.

[blimlim]

Salt:

For the differences in the figure:

It might be due to the mom mean and cice mean using data that's offset by one timestep, because of the coupling. Saving instantaneous values from the two models, they match really well if you add an offset in the comparison:

Snow loss in dynamics:

Do you think the non-conservation of snow during the dynamics could be due to the melt in the ridging step?

[blimlim]

Thanks @anton-seaice, I think this is looking good. I've just added a few more questions/comments

[ofa001]

THanks @bimlim your additional points on the potential conversation errors in "fresh" and melts are useful.

[blimlim]

My replies ended up in a review accidentally

[anton-seaice]

I think I addressed the comments @blimlim ?

[anton-seaice]

• Our gridbox-mean+intensive processing of the siflfwbot mean. This means when we multiply by the cell area, we don't recover the total kg/s of water transferred, and it might not be possible to get it back.

So we will change this - see #88

Shall we wait until that issue is resolved, or finish this PR and then update siflfwbot subsequently

[blimlim]

Thanks @anton-seaice, happy to finish this one off first and add the siflfwbot changes later. I've just added a few very small comments, and also added some notes on the UM's sea ice surface temperature

[blimlim]

Thanks @anton, I think this is looking good – do you need to re-export the markdown file?

[anton-seaice]

Should be all done @blimlim

[blimlim]

Thanks @anton-seaice, all looks good to me!

[anton]

Thanks @anton, I think this is looking good – do you need to re-export the markdown file?

@blimlim you mentioned my username?

[blimlim]

Apologies @anton! Pinged the wrong user!