Added cubed_sphere target grid test to utility/combine_topo
#861
Conversation
…ine_topo, nearing complete workflow for pull request
…nd cubed_sphere remapping cases, and successfully ran for 0.1 deg and ne30 target grids
There was a problem hiding this comment.
@bmooremaley, this is excellent work! I'm very happy with the way you have reorganized things and the clean-up you've done along the way.
I have some suggested changes that we can discuss today.
| # Build combined filename | ||
| combined_filename = '_'.join([ | ||
| antarctic_filename.strip('.nc'), global_filename.strip('.nc'), | ||
| self.resolution_name, datetime.now().strftime('%Y%m%d.nc'), | ||
| ]) |
There was a problem hiding this comment.
This is a nice improvement compared to the config option we had before.
| masks = { | ||
| 'ice_mask': 0., | ||
| 'grounded_mask': 0., | ||
| 'ocean_mask': ocean_mask, | ||
| } | ||
| for field in masks: | ||
| combined[field] = bedmachine[field].where( | ||
| bedmachine[field].notnull(), masks[field], | ||
| ) |
There was a problem hiding this comment.
I think this is missing some logic for putting in fill values for ice_draft and thickness that was in the original code.
| # Remove PETxxx.RegridWeightGen.Log files | ||
| for f in glob('*.RegridWeightGen.Log'): | ||
| os.remove(f) |
There was a problem hiding this comment.
I like this. The log files will be there if the code crashes but get removed if it ran successfully. That should help with debugging.
| gebco.bathymetry.where(gebco.bathymetry < 0, 0.) - \ | ||
| bedmachine.bathymetry | ||
| diff.to_netcdf('gebco_minus_bedmachine.nc') | ||
| logger.info(' Done.') |
There was a problem hiding this comment.
We're missing a newline here at the end of the file.
|
|
||
| # the number of tiles in lat and lon for GEBCO remapping | ||
| lat_tiles = 3 | ||
| lon_tiles = 6 |
There was a problem hiding this comment.
Same here, we're missing a newline here at the end of the file.
| grounded_mask = np.logical_or(np.logical_or(mask == 1, mask == 2), | ||
| mask == 4).astype(float) | ||
| mask == 4).astype(float) |
There was a problem hiding this comment.
I think the correct indentation according to the pep8 style guide is as it was before, since logical_or() is a method.
| '--file', f'{self.resolution_name}.g', | ||
| ] | ||
| logger.info(f' Running: {" ".join(args)}') | ||
| subprocess.run(args, check=True) |
There was a problem hiding this comment.
In compass, there's a strong preference to call mpas_tools.logging.check_call() and pass self.logger. If you do this, the output from these calls goes to the log file, not to the "terminal" or the log for the job script. We try to keep that output really minimal so it's obvious if the task succeeded and how long it took. The same goes for all the subprocess.run() calls.
However, I know you were seeing hanging so we will have to see if that recurs with calls to check_call().
|
|
||
| # Generate weights file | ||
| args = [ | ||
| 'srun', '-n', f'{self.ntasks}', |
There was a problem hiding this comment.
We want to let compass take care of this, rather than hard-coding srun and its flags. The preferred way to do this it to call compass.parallal.run_command() something like the way it's done for the model:
Line 60 in 2e06f2e
|
Presumably, you'll need to do a commit to make the linter happy. The easiest way to do that is probably to add an extra newline somewhere (maybe at the end of the file) and then fix whatever it complains about. Make sure you do that with the conda environment loaded so pre-commit can happen. |
The
ocean/utility/combine_topotest has been split into two separate tests: one for alat_lontarget grid and one for acubed_spheretarget grid. Thecubed_spheretest allows the migration tombtempestfor remapping bathymetry to MPAS grids by removing the pole singularities. The workflow for the existinglat_lontarget grid has also been updated to use remapping tools more consistently between the GEBCO and BedMachineAntarctica data sets, and between thelat_lonandcubed_spheretarget mesh workflows. Specifically, the GEBCO dataset is separated into tiles prior to remapping,ESMF_RegridWeightGenis used for building weights files, andncremapis used for the final remapping step.Checklist
Testingin this PR) any testing that was used to verify the changes