Turning off liquid runoff for when ice shelves are included and spreading the iceberg flux

[claireyung]

For the access-rom3 panantarctic with ice shelves, we will need to turn off liquid runoff, which includes the ice shelf basal melt flux. We still want to keep the iceberg calving term, ideally spread a bit at the coast so it's not dumping a huge amount of freshwater in front of the cavity. Just wondering
a. what are the steps involved in turning off the JRA liquid runoff/basal melting flux with nuopc? (I think it is reasonable to neglect river terms from South America/NZ/Tassie that are in the domain, but presumably in the future, the Antarctic basal melt will also need to be separated from real river runoff)
b. what are the steps involved in iceberg flux spreading? I heard @kieranricardo has been doing this in CM3?

[dougiesquire]

a. what are the steps involved in turning off the JRA liquid runoff/basal melting flux with nuopc?

I haven't tested, but I think this should be as simple as removing the liquid runoff section from drof.streams.xml

[chrisb13]

Vaguely related 1 and 2.

On b, om3 iceberg-y options have been discussed but not yet implemented 1 and 2.

[claireyung]

Awesome, thanks @dougiesquire, so just deleting these lines: https://github.com/claireyung/access-om3-configs/blob/dev-MC_4km_jra_ryf%2Bregionalpanan/drof.streams.xml#L8-L27

I will have a go...

[kieranricardo]

@claireyung not sure whether the CM3 iceberg flux spreading will be useful, but we also have a river runoff spreading scheme that might be more applicable.

The CM3 "iceberg" flux scheme calculates the amount of snow accumulated over the Antarctic and Greenland ice sheets (I think this could be the total SH and NH iceberg flux in OM3), and then spreads this over a given spatial distribution which is input as a netcdf file. It's a kinda hacky way of maintaining SSH in CM3, otherwise the snow over the icesheets would never make it to the ocean, accumulating forever!

We also do a conservative exponential smoothing of the river runoff that might be more useful. This script calculates regridding weights for spreading the runoff on the atmosphere grid (maps atm grid to atm grid). It's based on this CESM runoff spreading scheme, which basically calculates the regridding weights we have, and then composes this with a conservative regridding to the ocean grid. You might be able to use this directly, or use include this extra step in the CM3 script.

Hope this helps!

[claireyung]

Hi @anton-seaice , I was just wondering if theoretically it would work to spread the iceberg flux by replacing the JRA forcing file with an equivalent file from a different source. E.g. this Merino et al 2016 product has an iceberg flux climatology from a model with icebergs. If we remapped this Merino file conservatively onto the JRA grid to make an equivalent netdcf file as /g/data/vk83/configurations/inputs/JRA-55/RYF/v1-6/data/RYF.licalvf.1990_1991.nc (and maybe did some scaling to get the same total flux as JRA) would it get remapped correctly with the current rof mesh mask?

This is a plot of the Merino product vs JRA with the same latitude extent, unfortunately the Merino product doesn't go south enough so I don't think we can use it. But maybe there are others.

[anton-seaice]

It shouldn't be too hard to make a mesh file and streams file for the Merino product, that would be easier long term than regridding to the JRA grid.

Does Merino include both basal melt and iceberg melt ? (or are we ok without basal melt) e.g. for a panantarctic, is it ok to only use the Merino product for all the runoff ?

unfortunately the Merino product doesn't go south enough so I don't think we can use it. But maybe there are others.

This might not matter for the iceberg fluxes ? Would be more challenging to fix for basal melt.

For the global om3 configs, the question gets more complicated, because we probably want to spread the JRA volume of runoff using the pattern in the merino climatology and would end up needing to mix and match a bit

[claireyung]

I think the Merino product is just iceberg flux, so if you don't have ice shelf cavities you'll still need the JRA liquid runoff from Antarctica for the basal melt component, which would be just added to the first cell next to Antarctica.

Agree the iceberg flux not going to the calving front is probably not a huge deal (looks like it was just the southern end of their model domain in their paper), but it would have been a bit cleaner if it did!

[anton-seaice]

We can specify meshfiles for each variable independently, so this may actually be possible without code changes !

[anton-seaice]

Is Merino flux in ik11 somewhere ?

[adele-morrison]

Oh, now I'm remembering we did a dodgy version of this iceberg spreading for Pedro's project Claire! We took the Merino fluxes and shifted anything over land to the ocean and also spread the fluxes further south in the Weddell/Ross Seas:
https://github.com/pedrocol/basal_mom5-collaborative-project/blob/main/notebooks/ICB/iceberg_flux_adjustments.ipynb

I have the original Merino files here: /g/data/g40/akm157/data/Nacho_iceberg_fluxes/
Iceberg_Interannual_Merino.nc was the one we used before.
The modified one (as per the script above) is:
Merino_adjusted_fluxes.nc

But also I think I agree with you Claire that most of the flux is still over the shelf, so perhaps it won't make too much difference?