[H1] Create CM2 atmosphere field files to force ACCESS-OM2-01

[willaguiar]

Read H1 for more information

[willaguiar]

What atmospheric fields do we need to run OM2?

[wghuneke]

Here a link to a forum entry that discusses the forcing fields: https://forum.access-hive.org.au/t/what-are-the-inputs-access-om-needs-to-run/458

[adele-morrison]

We need 3 hourly temporal frequency.

Should we think a bit about which year we want? Ideally a year where the winds look similar to the RYF9091 in JRA55? How reproducible is CM2 though? If we look at the winds for existing simulations and choose a year, will the rerun have the same winds? I guess it's also a cheap model, we could run a bunch of years with high frequency output and choose afterwards.

Maybe as a start we could plot the winds compared to JRA55 RYF9091 and see how much variability there is in the strength / structure of the winds in CM2 over different years?

[rmholmes]

@wghuneke I'm not sure our ERA5 attempt can help you directly here. All the details are at https://github.com/COSIMA/access-om2/issues/242. Maybe the best approach is to give it a go and see what happens? If you run into problems we can try to help (and maybe it'll be something we've run into before)?

Also note that this kind of approach has been flagged for speeding up paleo-climate spinup times in coupled models (see https://forum.access-hive.org.au/t/decoupling-the-access-esm-for-long-spin-up/446). That community might be interested in this as well.

With regards to @adele157's comment above - maybe you'd want to cycle over a 10-year block or similar to avoid being subject to a specific year? Also, are you planning to run a spin-up with the CM2 forcing? I imagine the simulation could be quite out of equilibrium with that forcing (although maybe you don't want a spin-up so the simulation doesn't drift too far)?

[rmholmes]

Also, did you consider just smoothing the JRA-55 fields as an attempt to remove the katabatic's?

[wghuneke]

Thanks Ryan.

• We were planning to run for 10 years, only to see if the DSW formation changes drastically or not. So no proper spin-up.
• Amending the JRA55 fields is an option we talked about, depending how tricky it is to get the CM2 forcing working, we might end up doing that instead.
• Given that we'll need 3-hourly data to force OM2, I think 1 yr is more achievable. Picking a year with stronger winds might increase the chances of DSW production, happy to hear opinions from others. A plot of 10yr mean and std winds in CM2-025 is here.

[aekiss]

Other approaches you could consider:

• reduce the JRA55-do wind strength in DSW formation regions https://github.com/COSIMA/access-om2/wiki/Tutorials#Scaling-the-forcing-fields
• tweak cice parameters to see if that will alter whether polynyas form, e.g. turning angle (cosw and sinw), ice-ocean drag (dragio), ice-ocean heat transfer (chio), etc.

Were you planning on using just the CM2 winds, or all the forcing fields?

[PaulSpence]

I have a feeling that modifying the JRA winds will be easier than adapting the CM2 or ERA5 winds ... but I guess it depends on what CM2 outputs we can get. It would be a nice functionality to be able to apply CM2 winds to ACCESS-om2.

[wghuneke]

Hi all, I prepared perturbed wind forcing files. The details and some figures are in this script, in a nutshell:

• monthly climatology of 100 years of CM2-025 wind fields
• 30 d running mean of the monthly bias (monthly bias = 3hr JRA winds - monthly CM2 winds)
• apply the CM2 running mean bias to the winds south of the easterly/westerly dividing line (following the approach from the easterly experiment), keep JRA winds elsewhere

If everyone is happy with this approach of perturbing the wind field, we can start running a test case...

[adele-morrison]

What do you think about applying the wind bias everywhere globally? That’s probably what we’d do if we apply the actual CM2 winds or run a coupled model right? The change in westerlies probably would have minimal effect on the DSW production on these timescales, but you never know, it could impact sea ice advection or CDW upwelling and have an impact.

On Tue, Apr 18, 2023 at 2:13 PM, Wilma Huneke @.***> wrote:

Hi all, I prepared perturbed wind forcing files. The details and some figures are in this script https://github.com/willaguiar/DSW-collaborative-project/blob/main/JupyterCodes/CM2_bias_perturbation_setup.ipynb, in a nutshell:

• monthly climatology of 100 years of CM2-025 wind fields
• 30 d running mean of the monthly bias (monthly bias = 3hr JRA winds
• monthly CM2 winds)
• apply the CM2 running mean bias to the winds south of the easterly/westerly dividing line (following the approach from the easterly experiment https://github.com/adele157/easterlies-collaborative-project/blob/master/notebooks/forcing_perturbation/JRA_wind_perturbation_setup.ipynb), keep JRA winds elsewhere

If everyone is happy with this approach of perturbing the wind field, we can start running a test case...

— Reply to this email directly, view it on GitHub https://github.com/willaguiar/DSW-collaborative-project/issues/8#issuecomment-1512408338, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACA44U7SS74OF5JOSAAKNO3XBYIFHANCNFSM6AAAAAAVOAL7MA . You are receiving this because you were mentioned.Message ID: @.***>

[wghuneke]

To me, changing the winds everywhere seems to be a bigger step. If we don't get DSW formation, is it because the model aims for a different equilibrium state or because of the "local" winds? If we don't get DSW formation in the SH winds case, then there wouldn't be any value in running the global case. If there is still DSW formation in the SH wind case, then it would be interesting in running the global case.

[willaguiar]

If we apply the bias only south of the westerlies/easterlies line, then wouldn't we likely find results somewhat similar to what was found in the easterlies project? If so a global wind bias would make more sense to separate the results. We could run the model with a global wind bias and see if the changes in DSW are consistent with what would be predicted based on the results of the easterlies project. If so, then we have already a simple explanation. If not, then we can analyze what happens with the westerlies. What do you all think?

Also, it would make the run useful in terms of diagnose biases other than DSW formation in CM2 created by the winds.

[PaulSpence]

I think I support @wghuneke arguments, if the goal is to identify if CM2 winds at the margin are why they don't get DSW in CM2.6.

[dkhutch]

I kind of like the idea of doing a global perturbation, since that gets closer to what a CM2 run would do. However, since Wilma has nicely laid out a method and a wind bias field for restricting below the easterly line, it might make sense to do both.

[adele-morrison]

Maybe it’s more of a decision of whether we have enough hours to do both then. If we only have enough hours to do one wind perturbation, which one would we do?

On Wed, Apr 19, 2023 at 9:59 AM, dkhutch @.***> wrote:

I kind of like the idea of doing a global perturbation, since that gets closer to what a CM2 run would do. However, since Wilma has nicely laid out a method and a wind bias field for restricting below the easterly line, it might make sense to do both.

— Reply to this email directly, view it on GitHub https://github.com/willaguiar/DSW-collaborative-project/issues/8#issuecomment-1513925037, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACA44UYUUW33JZTSQTNXTZLXB4TFLANCNFSM6AAAAAAVOAL7MA . You are receiving this because you were mentioned.Message ID: @.***>

[willaguiar]

Update April 28th: We have by now decided to run the global wind perturbation experiment. @wghuneke has sent me the atmospheric u,v forcing for the experiment, and I put 01deg_jra55v13_ryf9091_CM2atm to run already ( completing two months now)