PPE results - sea ice area

[adagj]

This figure shows the SH sea ice area averaged over the last 10 years of each model simulation (note that this is not the exact same time period for all runs since some have been extended).
All PPE members exhibit a similarly extensive positive bias and a skewed seasonal cycle.
The ongoing Beta06 experiment appears to be among the most biased. It may be worth examining members with relatively lower sea ice area, such as 03, 65, 69, and 64.

And this figure shows the NH sea ice area averaged over the same period and for the same members (and even with the same colors :)

We can see that the ongoing #277 (beta06) has very extensive sea ice, also in the Arctic. It is actually worse than the semi-reduced PPE ensemble with 21 members

To be considered for #280

@JensBDebernard @matsbn

[adagj]

This might be relevant. They used to have a similar problem in CESM3dev NCAR/cesm_dev#12
The question is how they fixed it, because in the ongoing simulations the SH and NH sea ice areas are OK

[adagj]

I asked :) and this is a summary of the main factors that warmed the Southern Ocean and reduced sea ice extent in the current development version of CESM3. The main effect was increased horizontal mixing in the ocean to enhance heat transport, and the second was the introduction of a floe size distribution scheme in CICE, which enhances lateral melting of sea ice.

What they did:

• Increased Gent–McWilliams (GM) coefficients in MOM6 to enhance horizontal mixing and import more heat from lower latitudes.
• Activated the floe size distribution (tr_fsd) in CICE with two-way coupling to WW3 which cause wave–ice interactions to break up floes and enhance lateral melting (we can't do this since we are not running with WW3) , but we will test with increased floe diameter

[maritsandstad]

Is it easy (and possibly useful), to look at this and the Arctic cycle that you posted here #280 (comment) @adagj maybe even with the same per member colour scheme?

[kjetilaas]

Thanks for this, @adagj! Do you know when the obs period is? And roughly how much sea ice has reduced since then?

[adagj]

Antarctica obs period: 1993 - 2023
source = "GLORYS2V4; C-GLORS; ORAS5"
references = "http://marine.copernicus.eu/"
file on NIRD: /projects/NS9560K/diagnostics/ppe_obs/ANTARCTIC_OMI_SI_extent.nc

Arctic obs period: 1978 - Nov. 2025
source: NSIDC
reference: NSIDC website: http://nsidc.org/data/g02135"
file on NIRD: /projects/NS9560K/diagnostics/ppe_obs/n_icearea.mon.nc

[adagj]

So Arctic is probably fine, but Antarctic is not

[oyvindseland]

Several ensemble member has frozen over Barents sea and sea-ice visible from NorthCape so there are problems in the North Atlantic even so if it is not already decided I would consider using the cice settings (and ocean?) from the warmest member.

[kjetilaas]

About the bias in the latest b06 run (#277), I think this might be a spinup issue. For most parameters, this simulation and the reference simulation in the PPE are very similar:

The biggest changes are

• temperatures, which is probably because n1850.ne16pg3_tn14.noresm3_0_beta06.Best_c8_4p7.2025-12-01 #277 started from a simulation that had drifted downwards
• vegetation C and related variables (fire C loss and GPP, and possibly LH, Qrunoff, soilwater and dust). This is probably just because it has been running for a longer time, and therefore has more developed vegetation.

The question is if sea ice is also a spinup thing, connected to the colder temperatures. I would suspect that it is. @adagj

[adagj]

I also made some maps. I only included the 5 extended members, the Baseline, the ongoing beta06 and obs. The maps are showing the 15% and the 1% contourline, and the NH and SH for March and September

[adagj]

@JensBDebernard You asked about sea ice area and global mean temperature - here is an attempt to collect the dependency in 4 figures

[adagj]

And here are the PPE regression over the ocean / sea ice parameters which were tested and SH sea ice area in September.

EGC and EGIDFQ control the strength of eddy-driven transport and mixing in the BLOM. EGC sets the overall diffusivity (eddy-induced transport (GM) and neutral mixing), while EGIDFQ controls the balance between the two. It seems like EGIDFQ has a strong effect on the SH sea ice extent

[adagj]

It appears that all the PPEV# runs you conducted, @kjetilaas , exhibit a sequential increase in Southern Hemisphere sea-ice extent. We should take this into account when tuning the next runs
@matsbn