Animations with panan configs

[AndyHoggANU]

ACCESS-NRI is looking to have some spinup projects to develop capacity in producing animations.

Our suggestion here is that we could attempt to re-make the first half of the Circulation in the Southern Ocean movie using output from higher resolution panan simulations -- ultimately the 1/40° case. This work will be done by Owen Kaluza (@OKaluza) who has recently joined the ACCESS-NRI team.

I'm proposing this issue to kick off the discussion. The process I would be thinking of would be:

• Get used to the data format using existing snapshots or time averages.
• Learn how to plot density interface on top of bathymetry
• @willaguiar save some data showing speed on an isopycnal surface to produce still images, and plot that
• Start to save high-frequency output for a year
• Produce a full animation.

Suggestions on the viability of this, or where to start, are very welcome.

[willaguiar]

Very interested in seeing how these movies would look like in the 1/40th. We would use the branched last year simulation of panan0025-ryf with high frequency output speed (u,v) for these movies, right?

[AndyHoggANU]

Actually, I think you might actually be able to write speed itself on the isopcynal. But we can start by playing around with just showing isopycnal surfaces in 3D?

[StephenGriffies]

In addition to the speed on density surfaces, another cool animation would be to take a single snapshot of the zonal velocity and animate movement through longitude around Antarctica. That is, to show meridional-depth sections of u(y,z) with x-animated. Or, one might be able to do a 3d perspective with different shading to depict the current in all of its jets and filaments.

The structure of the ACC is is fascinating, with huge jet-like currents extending from the top to the bottom. I can imagine that seeing it in high detail would produce "wow" responses, at least from oceanographers who study the current.

Indeed, there is interesting science related to the ACC jets, which can be both eastward and westward, along with the Antarctic Slope Current. I am genuinely curious to see what these jets look like in this model.

[willaguiar]

Here is some plots with possible density layers we could output in high frequency for the ACCESS-NRI animations and/or video. @AndyHoggANU

---

Ross Sea:
From top to bottom: top: Monthly transects of density along -190 West ($\sigma_2=37.14\ Kg\ m^{-3}$ , gold line), bottom: Bottom Depth of the $\sigma_2=37.14\ Kg\ m^{-3}$ surface, in June 2000 of panan0025, superimposed over bathymetry surface (Shades of black/grey). I could only find an isopycnal that connects the shelf to deeper layers in June ($\sigma_2=37.14\ Kg\ m^{-3}$ , gold line), because if we go much lighter than that, the isopycnal just don't go deep enough offshore. This is probably an issue of the monthly mean, as I suspected we would be able to see denser isopycnal along the slope in outputs with higher frequency.

---

Weddell Sea:

Plots are the same as the ones done to the Ross Sea. Here a deeper isopycnal seem to have no trouble representing the overflow along the slope( $\sigma_2=37.17\ Kg\ m^{-3}$).


Based on that, I think we could try to output maybe ~10 layers between 37.14 and 37.175 $Kg\ m^{-3}$. What do you all think? Let me know if you think we could explore additional plots/locations

[AndyHoggANU]

Just to recap on today's discussion:

• We think that 37.18 is probably the right isopycnal (based on looking at a super-slow ncview version of daily $\sigma_2$ volumes).
• Just in case, we will also save 37.16 and 37.2 as well.
• Plan is to save layer thickness and (if possible) speed on the isosurface at 3-hourly temporal resolution.
• Also, will save speed at ~30m depth, SSH, SST(?) and SIC.
• Is that all??
• Initially, we'll just try a 5-day segment.

@OKaluza - Once this done this will give you something to start playing with.

[OKaluza]

Thanks @AndyHoggANU for starting this conversation and thanks for the contributions everyone. I'm planning on looking at this in depth first thing next year after finishing off my current vis project, in the meantime I haven't got anything to add but will have plenty of questions then, looking forward to getting some data to start working on.

[willaguiar]

Update....
I have ran the model, with 3 hourly output for about a month now, and was able to output 10 days at a time, with all variables above. It would take about a week to run a full model year. I am stopping the run now to save some resources, but as the week goes by if we have too many MSU's left on oz91 or x77 before the end of the quarter, I will continue the run.

A few additional info in case you want to plot something while I am running...

1- There was no output for speed itself in the model, so I saved U and V in the $\sigma_2$ levels we mentioned above, and in the following z levels 25.6, 29.4, 33.6 m.

files should be in /scratch/oz91/wf4500/panan_Hifreq_formovies/wf4500/mom6/archive/mom6-panan/output{108:110}. Later outputs (>110) will have the tiles collated as I run, so I recommend to only look at these rn, while I continue the run.

or, few free to use the database below (I will update with the collated files as I run)

database:/home/156/wf4500/databases/access/panan_HF.db , experiment name mom6-panan

[AndyHoggANU]

Nice - thanks @willaguiar.
One wrinkle that has me worried. We go up to 1037.2 as our lowest layer, but we don't know how thick the layer below that is. Likewise for the upper layer. So, I'm worried that we won't have the diagnostics saved to re-reate the total height of the interface (relative to either the surface or the bottom).
What do you think?

[willaguiar]

Make sense. I looked up for other diagnostics like mean depth of isopycnal but couldn't find any for MOM6. I guess the alternative is to save all layer thicknesses till the maximum $\sigma_2$ (21 levels), or save daily $\sigma_2(z)$ for conversion (Which would likely use much more memory)?

Ill run a test to check if we can save 21 $\sigma_2$ levels on the frequency we need without hitting the memory ceiling

[AndyHoggANU]

I think if you just save a very deep isopycnal (like 1038) then you will get all the depths below the deepest interface -- you can then add this to topo to get total height.

[willaguiar]

update: I added a deeper 1038 layer as suggested, and it seems to work ( shows a thickness of the layer between 1037.25 until 1038, in all model domain).

[willaguiar]

Update @OKaluza

I just finished running ~4 months of the 3 hourly data. The data should be in /scratch/oz91/wf4500/panan_Hifreq_formovies/wf4500/mom6/archive/mom6-panan/ , or if you prefer to access through a Cosima database, you can use /home/156/wf4500/databases/access/panan0025_HF.db, experiment mom6-panan.

Let me know if you have any trouble accessing the files