Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Science / diagnostics plans for 1/40° config #21

Open
adele-morrison opened this issue Jan 20, 2023 · 31 comments
Open

Science / diagnostics plans for 1/40° config #21

adele-morrison opened this issue Jan 20, 2023 · 31 comments

Comments

@adele-morrison
Copy link
Contributor

What are our science plans for the 1/40° config?

We will have limited run length and storage capability.

We definitely want 1 year after spinup of high temporal frequency data for pretty movies. Daily u, v, T, S, sea ice, what else do we want movies of?

@willaguiar and I have been chatting about looking into how the ASC and cross-slope heat transport changes from 1/10° to 1/20° to 1/40°.

@AndyHoggANU suggests saving transports and heat fluxes on isopycnals.
@aekiss
Copy link

aekiss commented Jan 23, 2023

What's our storage budget for this data? We could start with that and work out what will fit.

Could get away with 2d daily data for movies if we can't store 3d.

@adele-morrison
Copy link
Contributor Author

adele-morrison commented Jan 23, 2023 via email

@aekiss
Copy link

aekiss commented Jan 23, 2023

I meant, how much are we prepared to store within our existing quotas?

Longer duration with fewer cores sounds good - fewer uncertainties around scaling.

@adele-morrison
Copy link
Contributor Author

I'm not sure how much storage we'll have available. Let's clean up ik11 and see how much space there is? We can also keep data temporarily on /scratch/ if we think we can analyse it quickly.

@aekiss
Copy link

aekiss commented Apr 3, 2023

The 1/40th will be under RYF forcing, right?

@adele-morrison
Copy link
Contributor Author

adele-morrison commented Apr 3, 2023 via email

@aekiss
Copy link

aekiss commented Apr 6, 2023

Alex Fraser would like 3hrly sea ice U and V just for the month of July in the final year if possible

@aekiss
Copy link

aekiss commented Apr 12, 2023

Daily mean sea ice concentration and sea ice thickness would be good too. They're 2d fields, so are probably affordable for the whole run.

@schmidt-christina
Copy link
Collaborator

If we want to calculate SWMT, we need monthly mean salt_flux (in addition to wfo and hfds).

@schmidt-christina
Copy link
Collaborator

schmidt-christina commented Apr 18, 2023 •
edited
Loading

Regarding the density binning for DSW export diagnostics.

I plotted the meridional transport in each layer at different latitudes for the last year of the 1/20th simulation. On the right of the plots are the density bins as grey lines. The abyssal transport is within the bins with the highest resolution. But I am not sure about the transport off the ANtarctic shelf because I haven't calculated the transport across the 1000-m isobath yet and I can't really compare the density of the peak SWMT with the these density bins as they are $\rho2$ and I calculated the SWMT in $\rho1$.

image

@adele-morrison
Copy link
Contributor Author

adele-morrison commented Apr 18, 2023 via email

@AndyHoggANU
Copy link
Contributor

Yes, it's trivial to change it. It's controlled by diag_rho2.nc in the input directory, so you could just add more levels in there. I think the code figures out how many diagnostic levels to include based on the dimensions in that file.

I think I originally made that file. Will try to track down the script ...

@willaguiar
Copy link
Collaborator

willaguiar commented Apr 19, 2023 •
edited
Loading

Since we have extended walltime limit on normalsr now, I can already start running the 1/40th ryf. Up until now the diags requested are listed below. Please edit this comment and add any diag you need, or let me know so I can add them.

Update (April 20th,2023): I separated into 2 sections ( Variables for spin up, and for the last years of the ryf experiment)

For Spinup:

Monthly 2D variables in file ocean_month:

  • Surface mass flux: wfo
  • Net surface boundary heat flux: hfds
  • Net salt flux into ocean at surface (restoring + sea-ice): salt_flux
  • MLD: mlotst
  • SSH: zos
  • Depth integrated volume transports: umo_2d, vmo_2d
  • Wind stress: tauuo, tauvo
  • Bottom temperature: tob
  • Bottom salinity: sob

Monthly 3D variables in file ocean_month_z:

  • Temperature: thetao
  • Salinity: so
  • Velocity: uo,vo
  • Grid cell volume: volcello
  • Density: rhopot0, rhopot2
  • Ideal age: agessc

Monthly 2D ice variables in file ice_month:

  • Sea ice concentration, thickness siconc, sithick

Monthly 3D variables on density layers in file ocean_month_rho2:

  • Mass transports: vmo, umo
  • Cell volume of density layers: volcello

Once off grid variables in file ocean_static (save for first month, then switch these diagnostics off):

  • All the usual ones areacello,deptho,Coriolis,geolon,geolat,geolon_c,geolat_c,geolon_u,geolat_u,geolon_v,geolat_v, geolat_v,wet,wet_c,wet_u,wet_v,dxt,dyt,dxCu,dyCu,dxCv,dyCv,areacello_cu,areacello_cv,areacello_bu

For the last few years:

  • Daily temperature: thetao
  • Daily salinity: so
  • Daily u,v: uo,vo only in the upper 1500m
  • Monthly T advection zonal, meridional T_adx,T_ady
  • Daily T advection zonal, meridional, vertically integrated T_adx_2d,T_ady_2d
  • Monthly and Daily surface mass flux wfo
  • Daily mass transports in density space vmo, umo
  • Monthly vertically integrated mass transports in density space umo_2d,vmo_2d
  • Monthly and Daily net surface boundary heat fluxhfds
  • Monthly and Daily Net salt flux into ocean at surface (restoring + sea-ice) salt_flux
  • Daily mean SIC,SIT siconc,sithick faster than daily for movies
  • Monthly surface forcing components (precip,lrunoff,frunoff)
  • Total diapycnal diffusivity Kd_interface (Monthly?Daily?)
  • Monthly ice transports ix_trans,iy_trans
  • 3-hourly Sea ice U,V for the last July siu,siv
  • near-surface vorticity for movies (say, at ~30m to avoid Ekman effects), faster than daily (e.g. 3-hourly)
  • mixed layer depth faster than daily for movies mlotst
  • daily SSH zos
  • daily SSH snapshots zos (Only for the 1/20th)
  • Daily wind stress tauuo,tauvo

PS: for the last year we might need 3-hourly T,S,U,V (or even bigger frequency)
PS: for the Monthly surface forcing components, check if we have runoff values. if so, keep saving it, if not remove from outputs.

@AndyHoggANU
Copy link
Contributor

So, it seems to me that we might not need all of these daily variables during spin up, just the last few years? Also, if we want to use this for animations, daily won't be enough - we'll probably need 3-hourly or something insane ...

@willaguiar
Copy link
Collaborator

For the SWMT, is the depth summed 2d-salt budget enough, or de we want to have the individual components too? @schmidt-christina

@adele-morrison
Copy link
Contributor Author

2d salt fluxes are fine, the variable is salt_flux (not osalttend_2d). We'll need this for the whole simulation for SWMT.

@adele-morrison
Copy link
Contributor Author

Let's also save monthly zonal and meridional wind stress, and mass transports on density levels (vmo, umo). Depth integrated volume transports might be useful for quickly computing Drake Passage transport etc (umo_2d, vmo_2d). Saving density will also save us a lot of computation time (both rho0 and rho2 - haven't checked the exact diagnostic names sorry). Ideal age is also handy. And we'll need volcello for volume weighting when we do volume integrals.

@adele-morrison
Copy link
Contributor Author

And sea level, and mixed layer depths.

@aekiss
Copy link

aekiss commented Apr 19, 2023

I've added near-surface vorticity and MLD to the comment above. Vorticity and sea ice thickness and concentration should be faster than daily for movies. Daily is a bit jerky/blurry. See https://cosima.org.au/index.php/gallery/

@adele-morrison
Copy link
Contributor Author

adele-morrison commented Apr 20, 2023 •
edited
Loading

We need to set
USE_IDEAL_AGE_TRACER = True in MOM_input to save the age diagnostic. @schmidt-christina would also be worth switching this on in your 1/20th run now.

@schmidt-christina
Copy link
Collaborator

I don't see this option in MOM_input. Which line would that be?

@willaguiar
Copy link
Collaborator

I don't see this option in MOM_input. Which line would that be?

It is not in there, I think you have to add it.

@AndyHoggANU
Copy link
Contributor

AndyHoggANU commented Apr 20, 2023 •
edited
Loading

One of the most useful things in MOM6 is the MOM_parameter_doc.all file, which is written every time you run the model. It lists all the possible parameter choices for your configuration. If you look in a recent run, you should see a line like:

USE_IDEAL_AGE_TRACER = False    !   [Boolean] default = False
                                ! If true, use the ideal_age_example tracer package.

@schmidt-christina
Copy link
Collaborator

I think we should increase the density resolution at least below 1037, maybe even below 1036.5. Here is the yearly mean 1995 transport across the 1000-m isobath in panan-01-zstar-v13. We resolve the DSW export manly with only 4 density bins.

@AndyHoggANU Did you find your script?

image

And here is a zoom into the layer of DSW export:

image

@AndyHoggANU
Copy link
Contributor

I think it is this one:
https://github.com/COSIMA/mom6-diagnostics/blob/main/PanAntarctic/inputs/RFNC1.ipynb
My suggestion is that you (a) download, but copy to a more sensible place; (b) check that it produces a file which matches the diag_rho2.nc that we have and (c) add some documentation and explanation.
Note that it is based on my interpretation of the RFNC1 function which is included in MOM6 -- and then the grid is embellished in the dense regions. Should be easy to modify for what you want, but let me know if anything doesn't make sense!

@schmidt-christina
Copy link
Collaborator

I have increased the resolution below 1036.5. We have now have 100 rho layers, instead of 80 for the diagnostics. When it's running well, I will point you to the file @willaguiar, so you can use it for th1/40th run.

image

@adele-morrison
Copy link
Contributor Author

adele-morrison commented Apr 21, 2023 via email

@schmidt-christina
Copy link
Collaborator

Here is the new file for the rho diagnostics with 100 levels: /g/data/ik11/inputs/mom6/panan/005deg/20230424/diag_rho2.nc

@schmidt-christina
Copy link
Collaborator

And the script to generate the file is this /g/data/ik11/inputs/mom6/panan/005deg/scripts/create_diag_rho2.ipynb

@willaguiar
Copy link
Collaborator

willaguiar commented Jul 19, 2023 •
edited
Loading

Since we have extended walltime limit on normalsr now, I can already start running the 1/40th ryf. Up until now the diags requested are listed below. Please edit this comment and add any diag you need, or let me know so I can add them.

Update (April 20th,2023): I separated into 2 sections ( Variables for spin up, and for the last years of the ryf experiment)

For Spinup:

Monthly 2D variables in file ocean_month:

  • Surface mass flux: wfo
  • Net surface boundary heat flux: hfds
  • Net salt flux into ocean at surface (restoring + sea-ice): salt_flux
  • MLD: mlotst
  • SSH: zos
  • Depth integrated volume transports: umo_2d, vmo_2d
  • Wind stress: tauuo, tauvo
  • Bottom temperature: tob
  • Bottom salinity: sob

Monthly 3D variables in file ocean_month_z:

  • Temperature: thetao
  • Salinity: so
  • Velocity: uo,vo
  • Grid cell volume: volcello
  • Density: rhopot0, rhopot2
  • Ideal age: agessc

Monthly 2D ice variables in file ice_month:

  • Sea ice concentration, thickness siconc, sithick

Monthly 3D variables on density layers in file ocean_month_rho2:

  • Mass transports: vmo, umo
  • Cell volume of density layers: volcello

Once off grid variables in file ocean_static (save for first month, then switch these diagnostics off):

  • All the usual ones areacello,deptho,Coriolis,geolon,geolat,geolon_c,geolat_c,geolon_u,geolat_u,geolon_v,geolat_v, geolat_v,wet,wet_c,wet_u,wet_v,dxt,dyt,dxCu,dyCu,dxCv,dyCv,areacello_cu,areacello_cv,areacello_bu

For the last few years:

  • Daily temperature: thetao
  • Daily salinity: so
  • Daily u,v: uo,vo only in the upper 1500m
  • Monthly T advection zonal, meridional T_adx,T_ady
  • Daily T advection zonal, meridional, vertically integrated T_adx_2d,T_ady_2d
  • Monthly and Daily surface mass flux wfo
  • Daily mass transports in density space vmo, umo
  • Monthly vertically integrated mass transports in density space umo_2d,vmo_2d
  • Monthly and Daily net surface boundary heat fluxhfds
  • Monthly and Daily Net salt flux into ocean at surface (restoring + sea-ice) salt_flux
  • Daily mean SIC,SIT siconc,sithick faster than daily for movies
  • Monthly surface forcing components (precip,lrunoff,frunoff)
  • Total diapycnal diffusivity Kd_interface (Monthly?Daily?)
  • Monthly ice transports
  • 3-hourly Sea ice U,V for the last July siu,siv
  • near-surface vorticity for movies (say, at ~30m to avoid Ekman effects), faster than daily (e.g. 3-hourly)
  • mixed layer depth faster than daily for movies mlotst
  • daily SSH zos
  • daily SSH snapshots zos (Only for the 1/20th)
  • Daily wind stress tauuo,tauvo

PS: for the last year we might need 3-hourly T,S,U,V (or even bigger frequency) PS: for the Monthly surface forcing components, check if we have runoff values. if so, keep saving it, if not remove from outputs.

Update.... The year 10 of panan0025 is just about to start (next run). Just for costing reasons, let me know in care you need me to save any additional diagnostic that is not already listed here.

@AndyHoggANU
Copy link
Contributor

OK, so I would like to propose that we do a year's worth of simulation with some super-high frequency (3-hourly) output for movies. The main movie I'm thinking of is a simple re-make of the first half of our existing Southern Ocean movie. For this, we need speed on two isopycnal surfaces - carefully chosen to connect the abyssal region with the Weddell and Ross/Adelie Shelves. We would also want 30m speed, SST and siconc for the same year.

Note that, because we would be using a different rho2 axis for this, it would be incompatible with the main run, I think we would be doing a 1-year branch just for the movie files? So it could be done at any time...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@adele-morrison @AndyHoggANU @aekiss @schmidt-christina @willaguiar