Senisitivity test on mixed layer depth params

[PaulSpence]

Test impact on mixed layer params on sea ice area and thickness

[adfraser]

Hey @dpath2o I couldn't find the URL to the animations you showed - only the still pngs. Could you please let me know the address to the html? Ta.

[adfraser]

Example target plot

[dpath2o]

Hi @adfraser, sorry for the delay, I did not see your comment until now (I think I just noticed @PaulSpence and not yours on Monday). Anyhow, here are the two webpages: http://131.217.175.138/~ec2-user/thickness_comparisons.html and http://131.217.175.138/~ec2-user/sea_ice_concentration_comparisons.html

[dpath2o]

Northern Hemisphere

Southern Hemisphere

[willrhobbs]

Can't see the yellow/green lines - are the JRA55-forced runs all doing the same thing?

[dpath2o]

Can't see the yellow/green lines - are the JRA55-forced runs all doing the same thing?

Pretty much. Here's the raw data.

jra55_hmix_50m_SIA: [16.169842 17.613031 17.79402 16.981176 14.144441 10.604894 7.387684 4.618927 4.10125 6.1371818 9.217053 11.6080065 14.05937 15.808064 16.471622 15.939794 13.304051 10.303132 7.595977 4.973775 4.447061 6.252906 9.094266 11.138029 14.095788 15.690297 16.347557 15.709509 13.708787 10.631416 7.44328 4.530213 3.6007574 5.112146 8.538609 11.66096 ]

jra55_modBath_SIA: [16.159918 17.6038 17.79216 16.97375 14.124769 10.668326 7.494419 4.6552296 4.136887 6.1212444 9.285069 11.612766 14.055577 15.804845 16.467264 15.925379 13.28124 10.3442 7.7095227 5.1292515 4.4967957 6.315997 9.136775 11.153026 14.090532 15.685882 16.346376 15.718026 13.715186 10.688131 7.5429125 4.517687 3.6109009 5.0724144 8.597887 11.676709 ]

jra55_hmix_20m_SIA: [16.169842 17.613031 17.79402 16.981176 14.144441 10.604894 7.387684 4.618927 4.10125 6.1371818 9.217053 11.6080065 14.05937 15.808064 16.471622 15.939794 13.304051 10.303132 7.595977 4.973775 4.447061 6.252906 9.094266 11.138029 14.095788 15.690297 16.347557 15.709509 13.708787 10.631416 7.44328 4.530213 3.6007574 5.112146 8.538609 11.66096 ]

era5_hmix_20m_SIA: [16.987568 18.74642 18.829979 17.527113 14.198496 10.659855 6.9031715 2.815488 3.0819383 6.753491 10.621066 14.031336 17.327517 18.832298 18.822023 17.504723 14.0758705 10.4474945 6.5724754 1.7991941 2.2478998 6.5822377 10.599411 14.022533 17.32285 18.828592 18.818476 17.50401 14.079066 10.4379 6.4703836 1.563451 2.02929 6.5389614 10.588608 14.02235 ]

era5_hmix_60m_SIA: [16.000612 17.403828 17.653439 16.959938 14.252499 10.943455 7.441215 3.5212777 2.7410052 5.8780885 9.041332 11.102043 13.36763 14.848048 15.762027 15.665377 13.538446 10.534092 7.3095527 2.919826 2.094179 5.727508 8.973732 11.015937 13.302339 14.79111 15.678868 15.612678 13.5089 10.517571 7.256289 2.7554076 1.9124215 5.6698694 8.939565 10.99868 ]

AOM2_SIA: [13.239415 13.971027 14.224967 13.553355 11.464819 8.991395 6.4456778 4.591633 4.6193833 6.5100837 9.586332 11.678825 12.994289 13.848988 13.974069 13.269288 11.099564 8.666927 6.413695 4.745217 5.00214 6.726763 9.24843 11.41044 13.270933 13.863171 13.864149 13.171139 11.504571 9.041229 6.4094744 4.4723315 3.8170483 5.536771 9.236349 11.666017 ]

NSIDC_SIA: [12.81416879 13.45801106 13.7703871 13.1898324 11.86504902 9.7634933 7.36227007 5.34838167 4.98299816 6.78892786 9.62827392 11.51623841 12.61072031 13.41062957 13.50867523 12.91886672 11.38678042 9.61136111 7.28058951 5.51856707 5.21524998 6.96041697 9.09790098 11.16137082 12.86977374 13.66821851 13.60061348 12.8602832 11.75132021 9.59952621 6.58430977 4.292362 3.75032229 5.42861842 9.10845131 11.22695767]

[adfraser]

So you fixed the mixed layer bug for ERA5 but it's still present in your JRA55 runs?

Separate question - looks like with ERA5, a ~100 m MLD might be required to knock the modeled area down to observed. How does 100 m compare to reality during winter? Too deep yeah?

Another separate one: Why do you think ERA5 grows more ice than JRA?

NB of course as Siobhan mentioned ACCESS-OM2 uses JRA55-do but I guess you used JRA55 non-do.

[dpath2o]

So the JRA55-do runs that are represented above, and the only ones I have done back in Jul/Aug, are all with the same hmix, hmix = 50m . This will be reflected in the JRA55-do labels in subsequent figures and mislabelled in the above figure.

As an aside, at the time of those runs (Jul/Aug JRA55-do runs) hmix was not as much of a consideration for my work as other aspects of standalone CICE6 testing were. Now that there are three ERA5 runs with hmix set at three different depths over the same duration set against the JRA55-do runs (with a fixed hmix of 50m) we can see the importance of hmix parameter on the standalone system. Yes, we knew this, and, yes, Noah and I had discussed this back in late Aug with Noah going on to show this in his work with CICE6 and waves in the MIZ in Sep. So this is building upon that and showing the implied rationale for https://tc.copernicus.org/articles/8/761/2014/ (Petty et. al, 2014). I believe it underlines/shows the importance of having a fully coupled ocean in my pursuit of modelling Antarctic fast ice. I think the Petty approach is interesting if MOM6-CICE6 (NUOPC) appears to be a 2025/26 timeframe, but I think this question is outside the scope of this thread (maybe!).

The other, potentially interesting question, is, why is ERA5 growing significantly more sea ice at hmix=20m? Firstly, I would like to understand if ERA5 is really growing more sea ice than JRA55-do? (As the three JRA55-do runs shown above are all set to hmix=50m and do not differ that much from ERA5 run with hmix=60m. To look at this I am running an equivalent JRA55-do runs. All other inputs at to the model configuration and setup will be equivalent (with the only exception being the thermodynamic timestep being 1800s for JRA55do and 600s for ERA5, the reason for this is due to the higher temporal resolution of ERA5).

[Siobhan]

Hello. You have mentioned the wrong person.

On Thu, 30 Nov 2023, 22:45 Alex Fraser, @.***> wrote:

So you fixed the mixed layer bug for ERA5 but it's still present in your JRA55 runs?

Separate question - looks like with ERA5, a ~100 m MLD might be required to knock the modeled area down to observed. How does 100 m compare to reality during winter? Too deep yeah?

Another separate one: Why do you think ERA5 grows more ice than JRA?

NB of course as @Siobhan https://github.com/Siobhan mentioned ACCESS-OM2 uses JRA55-do but I guess you used JRA55 non-do.

— Reply to this email directly, view it on GitHub https://github.com/dpath2o/AFIM/issues/3#issuecomment-1834678154, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAESJ4TPDHUFJQGHPCJHOR3YHED77AVCNFSM6AAAAAA5YHRJLKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZUGY3TQMJVGQ . You are receiving this because you were mentioned.Message ID: @.***>

[dpath2o]

Northern Hemisphere

Southern Hemisphere

[adfraser]

Interesting @dpath2o. 60 and 100 align for ERA5 and JRA55 but 20 is completely different.

And ERA5 always makes more ice than JRA55-do.

[dpath2o]

Considering the 20m MLD is akin to 'constraining' the thermodynamics of icepack, and also remembering that I am using the default CICE6 thermodynamics "mushy" solver not the more reported (in the literature) Bitz and Lipscomb model, I am thinking that it would be interesting to re-run the ERA5-20m-MLD run and change the thermodynamics solver to the Bitz and Lipscomb solver. However, that pursuit is getting tangential to the main thrust of this effort, which is to establish that I have a reasonable sea ice model. This one-dimensional analysis does not fully answer that but it does give me a certainty that ERA5 and JRA55do forcings are providing results that are explainable without the presence of a fully coupled ocean -- specifically that the heat fluxes into the mixed-layer are not accounted for in this setup. We now can clearly see short-comings of this setup without a coupled ocean, but I think we can also now 'see' how to move forward constraining/accounting for this and focusing on understanding the importance of the atmosphere on Antarctic fast ice.

So moving forward from this, and keeping in-line with my first paper's pursuit that we've previously discussed in September, I think: 1.) modify CICE6-standalone with ERA5 forcing , with advice/help from JF, to allow for a 'tune-able' grounded icebergs scheme (AFIM) 2.) consider/discuss an additional modification to CICE6 that allows for 'table' MLD input; what I envision is simple table that sets the MLD (or "hmix" as CICE calls it) per month per hemisphere. Possibly to take this one step further would be to create a static MLD NetCDF file that contains MLD on the same grid as the CICE6-grid and contains values for MLD for each month of the year. A module/sub-routine reads this in and applies this to the relevant CICE and ICEPACK sub-routines. What do others think?

[adfraser]

Hi Dan: For 1) I agree to move forward with ERA5 for paper 1 - there are things that we need to answer which JRA55 can't provide. I'm not sure if we need the JF-idea of sophisticated grounded iceberg stress for paper 1 - what do you think? It will certainly be needed for when the model is coupled to the ocean. But since the main advantage of JF-spec icebergs is more realistic ocean flow, I feel like we don't really need it for paper 1. Interested in your thoughts. For 2) I think you've shown that MLD is an effective way of tuning the model. Are you thinking to proceed with 100 m MLD with ERA5? Or go even deeper than 100 m? Is that crazy? If I understand your proposal above, you're thinking of a monthly updating "mask" of MLD (spatially varying, and monthly varying), is that right? We could use ACCESS-OM2 - then we will probably approach OM2's seasonal cycle, right? Do you have any sense on how much work it would be to implement this? Thanks!

[PaulSpence]

Both 1 and 2 as suggested are more technical model dev work. I suggest we evaluate this base case and understand it a bit more, especially in our region of interest for paper 1. Using the om2 mld as a 3d(xyt) input is an interesting idea.

[dpath2o]

Thanks Alex.

• with regard to (1), a sophisticated grounded iceberg scheme, it will no doubt be time-consuming. The current four (4) ocean parameters that my CICE is restoring to every 10 days are: SST, SSS, U & V. So I'm also in line with your thinking that advection of surface conditions is imparting a weak impact to the dynamics of any introduced ground icebergs. I think a first approach to this will be to more carefully edit the bathymetry file for the entirety of the Antarctic coast, and perform the same runs.
• with regard to (2), yes, 100m MLD with ERA5 as a baseline. Going deeper as a static setting seems a bit extreme. And yes, that is what I'm thinking -- a monthly latitude-based MLD "mask". Not ideal, but without going with a mixed-layer model (i.e. Petty et. al) it is likely the poorest, yet, least-time-consuming system-physical-realistic approach that can be implemented.

[adfraser]

Totally in agreement with 1st point :) Let's discuss pt 2 today.