Convert the Surface Water Mass Transformation example to be model-agnostic #402
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
We definitely don't want our examples to be dependent on data that lives on user's personal home or g/data directories |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:42Z Line #1. def save_SWMT(expt, session, start_time, end_time, outpath,model='mom6', lat_north = -59, n = None): space after commas |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:42Z Line #215. ds[SST.cf['latitude'].name].attrs= {'standard_name':'latitude'}
space after colons |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:43Z Line #228. resolution = str(0.01/(pme_net.cf['longitude'].size/3600))[2:] what is this here doing? reads strange to me willaguiar commented on 2024-07-07T06:37:56Z This is an attempt to automatically get the resolution of the model, assuming a global square-cell grid. This is necessary so we can identify the correct name of the isobath file for panan (contains 3 resolutions) . So this lines calculates there resolution and convert it to string to use it on... contour_dir = "/home/142/cs6673/work/mom6_comparison/Antarctic_slope_contours/Antarctic_slope_contour_1000m_MOM6_" + resolution + "deg.nc" Happy to improve if you have a suggestion adele-morrison commented on 2024-07-07T06:49:58Z Maybe a dictionary with experiment names mapped to resolution is best? This wouldn't work for a regional model right? willaguiar commented on 2024-07-07T07:01:10Z True, it wouldn't if the regional model is not circumpolar. How would we map the dict to the experiment name? something like...
if expt='panant-01-zstar-ACCESSyr2': resolution='01' ? Or maybe only extracting the 01 string from the name for experiments starting with 'panant'?
schmidt-christina commented on 2024-07-08T00:54:08Z I extract the resolution from the experiment name, but that assumes we keep the naming convention:
willaguiar commented on 2024-08-15T05:01:48Z I haven't changed that one |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:44Z what are all these willaguiar commented on 2024-07-07T06:41:09Z This is a print that only appears when I run ... cc.querying.getvar ...through VScode ssh system. It doesn't appear when I run it on ARE as it is usually done. Not sure why tho ( but other then the 0.3.0 print itself, no other difference). The new edits/submission will be run on ARE so this will not appear |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:44Z Line #1. def shelf_mask_isobath(var,model_dict): space after comma |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:22:45Z space after commas |
There was a problem hiding this comment.
Great job! Thanks @willaguiar!
Made a few minor comments.
My major comment is that you claim that it works with MOM6 output now. Let's make an example calculation and a figure using MOM6 output though to make sure! Unless it was there and I missed it?
|
@willaguiar could you rename the PR to something more descriptive? E.g., "Convert the ..... example to be model agnostic"? Don't use SWMT... Personally I'm not 100% sure I know what that is? Is it another acronym similar to CDW, AABW, NADW, etc? |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:29:28Z
|
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:29:29Z Line #19. import cf_xarray as cfxr this is under "plotting"! let's remove the |
|
View / edit / reply to this conversation on ReviewNB navidcy commented on 2024-07-07T04:29:30Z Rephrase to "We suggest..." Always use plural because we want the users to feel that the community is guiding them, not a particular individual. |
|
Seems that the PR closes #341, right @willaguiar? |
|
Also the phrase |
willaguiar
changed the title
willaguiar
changed the title
|
This is an attempt to automatically get the resolution of the model, assuming a global square-cell grid. This is necessary so we can identify the correct name of the isobath file for panan (contains 3 resolutions) . So this lines calculates there resolution and convert it to string to use it on... contour_dir = "/home/142/cs6673/work/mom6_comparison/Antarctic_slope_contours/Antarctic_slope_contour_1000m_MOM6_" + resolution + "deg.nc" Happy to improve if you have a suggestion View entire conversation on ReviewNB |
|
This is a print that only appears when I run ... cc.querying.getvar ...through VScode ssh system. It doesn't appear when I run it on ARE as it is usually done. Not sure why tho ( but other then the 0.3.0 print itself, no other difference). The new edits/submission will be run on ARE so this will not appear View entire conversation on ReviewNB |
|
Maybe a dictionary with experiment names mapped to resolution is best? This wouldn't work for a regional model right? View entire conversation on ReviewNB |
|
True, it wouldn't if the regional model is not circumpolar. How would we map the dict to the experiment name? something like...
if expt='panant-01-zstar-ACCESSyr2': resolution='01' ? Or maybe only extracting the 01 string from the name for experiments starting with 'panant'?
View entire conversation on ReviewNB |
|
I extract the resolution from the experiment name, but that assumes we keep the naming convention:
View entire conversation on ReviewNB |
|
View / edit / reply to this conversation on ReviewNB schmidt-christina commented on 2024-07-08T07:59:23Z Line #132. pressure = xr.DataArray(p_from_z(depth_tile, lat_t), coords = [yt_ocean, xt_ocean], dims = [SST.cf['latitude'].name, SST.cf['longitude'].name], name = 'pressure', attrs = {'units':'dbar'})
Here, and whenever you use the gsw functions: There is no need to explicitly convert it into xr.DataArray anymore as the result is already a DataArray.
pressure = p_from_z(depth_tile, lat_t)
pressure.attrs = {'units': 'dbar'}
|
|
Some accidental newlines (e.g. th e water mass transformation in each model simulation) View entire conversation on ReviewNB |
navidcy
added
🛸 updating
First time looking at it now! And indeed found it useful ( especially the histogram). I classify it as VERY helpful ( made me realize I needed to first updated my fork before creating the pull request). Still not sure it went the right way tho |
|
where are we with this? (I admit I didn't see closely what changes were suggested.... just saw there were a few review comments and I'm just wondering about the status of the PR.) |
I think for the last review, most suggestions were met, except the splitting of the SWMT function into heat and salt components, which I think might make the code prolix (happy to try if there is a wide disagreement) So I'm not sure if something else is required to be done.... @schmidt-christina @julia-neme ? |
|
Look, regarding the splitting if it's something you wanna embark do it. But if you half-do it then it might not be useful. An other option is to merge the PR and open an issue documenting a potential improvement for future (e.g. splitting the SWMT function into heat and salt components etc). |
|
View / edit / reply to this conversation on ReviewNB julia-neme commented on 2024-09-11T01:27:27Z This is not running for me, computing failing due to the following line in T = cc.querying.getvar(expt, model_vars[model]['temperature'], session, frequency='1 monthly',chunks=model_vars[model]['chunks']) I think using start_time and end_time solves this, but maybe there is a reason behind opening all times within the experiment?
|
|
Sorry for another request of changes Wilton! I'm happy to add the start and end times, but I'm not sure whether all dates where needed! |
There was a problem hiding this comment.
@julia-neme I tried to run the notebook from top to bottom and got an error. I pushed the notebook with the error so you can have a look.
Can you ensure that before you push you restart the kernel and run all cells sequentially?
|
Right, fixing those paths now so it worked for me. You might have to correct to point to your own directory. |
|
oh we can't merge something that has your local directories :) |
|
Question: I see this line: shelf_mask = xr.where(shelf_mask == 0, 1, 0)#+1; shelf_mask = shelf_mask.where(shelf_mask == 1, 0)what's this commented code? I'll delete it... I consider commented code like putting something in the recycle bin. If it's useful let's add an explanation about it? |
Not useful! I forgot to delete. Sorry! |
|
@julia-neme you need to approve because at some point you requested changes so merging is blocked ;) |
Closes #277
Closes #341
This is an attempt to make the surface water mass transformation partially model agnostic. I used cf-xarray to deal with operations along dimensions. To deal with the different diagnostics required by each model I added an embedded dictionary to the save_SWMT function. The function also spits the specific dictionary for that model so it can be used later for shelf masking.
Shelf masking requires the input isobath file. Currently we have these isobath files only for OM2 and Panans. Panan isobath is on @schmidt-christina disk. Perhaps we can move them to ik11 to avoid permission issues?
For mom5 I also changed the diag to use pme_net instead of pme_river, according to issue #341