cosima-recipes Xgcm operations

First pass at a notebook that pulls in all the metrics required to create an xgcm grid object.

It's all pretty rough and ready at the moment. The MOM6 example at the end is especially rough.

closes #164

Jan 24 '23 03:01 edoddridge

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

Jan 24 '23 03:01 review-notebook-app[bot]

I'll have a look!

Jan 24 '23 04:01 navidcy

Just had a quick look at this -- will look carefully soon.

Seems like we can ditch the Relative vorticity example and keep this one? What do you think?

Jan 26 '23 21:01 navidcy

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-01-30T01:10:44Z ----------------------------------------------------------------

Line #2.    rel_vort_region.plot()

Change to

rel_vort_region.plot(vmin=-2e-5, vmax=2e-5, cmap='RdBu_r');

otherwise it's like a blank plot.

julia-neme commented on 2023-09-04T04:39:49Z ----------------------------------------------------------------

I would probably delete this map and go straight to the circumpolar plot

Jan 30 '23 01:01 review-notebook-app[bot]

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-01-30T01:10:45Z ----------------------------------------------------------------

Line #1.    foo = xr.open_dataset('/g/data/ik11/outputs/mom6-panan/panant-v2/output090/19980701.ocean_daily.nc')

let's not foo but rather use the cookbook to load some variables?

Jan 30 '23 01:01 review-notebook-app[bot]

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-01-30T01:10:46Z ----------------------------------------------------------------

Line #3.    ds_panant = ds_panant.rename({'latq':'yq'}).rename({'lath':'yh'}).rename({'lonq':'xq'}).rename({'lonh':'xh'})

why do you rename? is it needed? perhaps we should explain? (would cf-xarray be useful here?)

Jan 30 '23 01:01 review-notebook-app[bot]

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-01-30T01:10:47Z ----------------------------------------------------------------

Line #22.    grid = xgcm.Grid(ds_panant, coords={'X':{'center':'xh', 'left':'xq'},

I feel that the grid construction belongs to its own cell with a bit of a remark on how this is a bit different than before making the connection with MOM5 being B-grid while MOM6 C-grid?

Jan 30 '23 01:01 review-notebook-app[bot]

I had a quick go and put a few comments. It's a bit rough... but I'm keen to work to polish it.

I'm wondering whether we should incorporate pint/cf-xarray practices.
Also wondering whether this should become a tutorial.

Jan 30 '23 01:01 navidcy

Thanks Navid

I had a quick go and put a few comments. It's a bit rough... but I'm keen to work to polish it.

Absolutely. I didn't spend time polishing it, because I wanted to have a conversation about what we wanted the final result to look like before doing that. (And there were other people working on xgcm things who wanted to look at it, so I pushed it before it was ready for merging.)

I'm wondering whether we should incorporate pint/cf-xarray practices.

Oh definitely. Now that I've learnt about cf-xarray, I'm keen to use it.

Also wondering whether this should become a tutorial.

I think there's scope for it to be a tutorial. Performing grid-aware operations is a sufficiently basic task that that most users will need to do it and xgcm elegantly deals with the intricacies. But, it would be good to have a streamlined process for creating the grid objects before promoting it as the go-to method. Is that currently blocked by #212 ?

Jan 30 '23 03:01 edoddridge

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-02-10T10:16:20Z ----------------------------------------------------------------

Line #1.    sim = '0.1RYF'

Would the tutorial here break if we load 0.25-degree output?

julia-neme commented on 2023-09-04T04:29:43Z ----------------------------------------------------------------

Also, why wouldn't this work with IAF? This could be probably generalised to whatever access experiment we want right? An option could be adding a line that defines the experiment, and explaining that it can be changed to whatever experiment is of interest?

Feb 10 '23 10:02 review-notebook-app[bot]

View / edit / reply to this conversation on ReviewNB

navidcy commented on 2023-02-10T10:16:21Z ----------------------------------------------------------------

Line #4.    ocean_grid = xr.open_dataset('/g/data/ik11/outputs/access-om2-01/01deg_jra55v13_ryf9091/output000/ocean/ocean_grid.nc')

can we use cosima cookbook to load this?

I don't like hardcoding the path of the .nc.file? how would the users know the path the file their experiment is?

@Angus-g?

Feb 10 '23 10:02 review-notebook-app[bot]

@edoddridge should we aim to finish this during this hackathon? do you wanna push this or you have something else in mind for Monday's hackathon?

Aug 31 '23 14:08 navidcy

We definitely should. I'll be in transit for much of Monday :( so I'm not sure I should lead it, but am happy to pick it up when I arrive if nobody else has.

Sep 01 '23 01:09 edoddridge

Also, why wouldn't this work with IAF? This could be probably generalised to whatever access experiment we want right?

View entire conversation on ReviewNB

Sep 04 '23 04:09 julia-neme

I would probably delete this map and go straight to the circumpolar plot

View entire conversation on ReviewNB

Sep 04 '23 04:09 julia-neme

View / edit / reply to this conversation on ReviewNB

julia-neme commented on 2023-09-04T04:44:27Z ----------------------------------------------------------------

I would maybe add a description here with what this notebook does, and maybe divide in sections so that it is easier to browse the notebook. For example: a section that loads necessary variables, a section that creates the grid info, and a section with example calculations