grids in /g/data/ik11 don't have the same format

COSIMA / access-om2

Deprecated ACCESS-OM2 global ocean - sea ice coupled model code and configurations.

21 stars 23 forks source link

grids in /g/data/ik11 don't have the same format #233

Open navidcy opened 3 years ago

navidcy commented 3 years ago

BUT

aekiss commented 3 years ago

I've been able to make do with a workaround for area_t (see loaddata in ice_maps.ipynb here, which uses grid paths defined in exptdata.py here), but it would be better to avoid this kludge and fix the 0.1deg grid files.

navidcy commented 3 years ago

I can do hocus-pocus to go around it but we should fix it. It makes it cumbersome to compare things across the three resolutions.

(Unfortunately I don't have any idea how to produce these files so when I say "we should fix it" in principle I just hope somebody else will do :()

aekiss commented 3 years ago

Agreed, and I also don't know how to do it. I'm guessing @aidanheerdegen does?

FYI I'm abstracting away these cumbersome differences by accessing the data via exptdata.py and loaddata, so I have a uniform interface to all resolutions and cycles

navidcy commented 3 years ago

hm... can you elaborate on this (eg point to a notebook)? Are these methods in cosima-recipes?

aekiss commented 3 years ago

see ice_maps.ipynb in https://github.com/aekiss/ice_analysis/

rmholmes commented 3 years ago

Stop me if I'm wrong, but all of our runs include an ocean_grid.nc file in the output directories. These have all the variables you should need. Why not just use those?

E.g. for the 1/10th-degree: /g/data/ik11/outputs/access-om2-01/01deg_jra55v13_ryf9091/output***/ocean/ocean_grid.nc

navidcy commented 3 years ago

Because those files they have nans for land e.g. in geolon, geolat and this creates trouble in plotting or interpolation.

navidcy commented 3 years ago

see ice_maps.ipynb in https://github.com/aekiss/ice_analysis/

Seems very complicated... :(

navidcy commented 3 years ago

see ice_maps.ipynb in https://github.com/aekiss/ice_analysis/

Seems very complicated... :(

Don't bother though. I managed to do what I needed. Just wanted to point out the asymmetry in the grid files.

aidanheerdegen commented 3 years ago

Seems @AndyHoggANU made them:

$ ls -l
total 514020
-rw-r--r-- 1 amh157 ik11  44099122 Apr 12  2020 Antarctic_slope_contour_1000m.npz
-rw-r----- 1 amh157 ik11 466576179 Mar 23  2020 ocean_grid_01.nc
-rw-r----- 1 amh157 ik11  14521988 Mar 20  2020 ocean_grid_025.nc
-rw-r----- 1 amh157 ik11   1146336 Apr  6  2020 ocean_grid_10.nc

navidcy commented 3 years ago

deep inside we all know @AndyHoggANU's fault..

@AndyHoggANU, can you comment on this asymmetry or recreate the 0.10 deg file to have the same structure as the other ones?

AndyHoggANU commented 3 years ago

I just copied them there as a hack. @russfiedler created the 01deg one:

// global attributes: :history = "Fri Mar 20 16:34:09 2020: ncrename -v x_T,geolon_t -v x_C,geolon_c -v y_T,geolat_t -v y_C,geolat_c /g/data/v14/pas548/cosima/enkf/grid/old_grid_spec.nc /scratch/v45/raf599/old_grid_spec.nc" ;

aidanheerdegen commented 3 years ago

Old school the way to get a grid without missing values was to run the model for a time step unmasked.

The ice grid has all the lat and lon unmasked, so that is an option too.

russfiedler commented 3 years ago

That looks like a file created with a tool that I created to generate info for our assimilation suite which wants info in the form of older style grid_spec.nc. files https://github.com/russfiedler/grid_spec_tools Should work for any mosaic style set up.

navidcy commented 3 years ago

I've been running into more troubles with this asymmetry...

I can fix it if nobody is willing to put their hand up! But I have no idea how? Could you enlighten me? (I don't know Fortran nor I can run Fortran though...)

aekiss commented 1 year ago

we are working on fixing this here: https://github.com/COSIMA/cosima-recipes/issues/212