Integration tests for current code

[douglowe]

Run some integration tests on the current code, with the latest wrf-python libraries, to make sure that we are still compatible.

[douglowe]

@fherreazcue to advise on how to do this.

[fherreazcue]

I think these things would need to be tested:

• WRF Diagnostics (diagnostic task)
• Video generation
• CSV generation (csv task)
• Technically, also the mp4diff, mp4stitch, and wrfcompare tasks, but I don't think these were used in the end, and I don't think they are going to be used... so I will ignore them...
• RIP

Diagnostics

There is a file that I used to test all the diagnostics were being generated correctly in the tests directory. It can be called from the repo's base directory as:

python ./tests/single_image_all_diags_test.py

It expects wrf data in tests/wrfdata/control, so I have for example tests/wrfdata/control/wrfout_d01_2015-07-06_00:00:00, which has a single hour measurement in it, so it produces a single image.

The script currently generates a single combined pdf (tests/results/test_single_image_all_diags....pdf), which is easier to flick through as a human, but if tests were to be automated, it might be better to skip the merging...

Video and CSV generation

if diagnostics are tested, a single mp4 successful video should suffice, as this simply stitches the diagnostic outputs.
This would be best with a data file with at least two time samples, but should be ok to test with the same data as above.

The simplest thing for this and csv generation would be to use the test.py script.
In that script, the "all_args" array is a list of arguments you want to be passed to csf.py.
So for example, replace the contents of all_args in the test.py file with:

f"--task=csv --var=AirTemp2m,DewpointTemp2m,CIN,CAPE    --dir_path={wrfdata}/control/  --domain=full  --outdir={results}/ --lat=51.38  --lon=-2.36 --file_tag=_Bath",
f"--task=diagnostic   --var=DewpointTemp2m        --dir_path={wrfdata}/control/         --outdir={results}/",

and run the script with:

python ./tests/test.py

RIP

I think this is not affected by the libraries you updated @pricet1 , but I am not sure what you updated, so...

Some commands are documented inside the test.py script, which can be used for testing RIP... but for these you do definitely need more than one data point, and I currently have no data of that size, so unfortunately could not check they work... they seem ok though...

[pricet1]

Thanks for these pointers @fherreazcue.

I think this is not affected by the libraries you updated @pricet1 , but I am not sure what you updated, so...

I updated the wrf-py-env dependencies in environment.yml by removing all the version pins, i.e. changing it to:

dependencies:
  - netCDF4
  - matplotlib
  - cmasher
  - cartopy
  - wrf-python
  - imageio
  - imageio-ffmpeg
  - metpy

So it has installed current versions of these, which means it can also install wrf-python for arm64. I have not yet re-pinned these with more current versions, though of course that could be easily done for reproducibility.

[pricet1]

I have created an updated Conda env with python 3.14 (with dependency versions also unpinned in environment.yml so they will be at their current versions), and made some small related updates to the code. This is on branch 56-integration-tests-for-current-code with commit messages which are hopefully sufficiently descriptive.

As an aside, if the objective is ultimately to add automated/semi-automated tests, it would of course be advisable to generate test output for them with the old python version and (pinned) dependencies - as that output should form the basis for comparison.

Coming back to the updated code though, for the tests suggested by @fherreazcue, the current code mostly seems to run with updated versions though there seem to be some issues which would need investigation.

1. tests/single_image_all_diags_test.py

This runs for some variables but for others it is giving these sorts of errors (I haven't yet had a chance to look at these in more detail):

<snip>

Started at: 2026-06-22 14:45:20.191520

python /Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/CSF/csf.py --task=diagnostic --var=cmr --dir_path=/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/tests/wrfdata/control/ --save_pdf_frames=1 --outdir=/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/tests/results/tsiad_2026-06-22_14-40-02/
Traceback (most recent call last):
  File "/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/CSF/csf.py", line 251, in <module>
    if "SkewT" in wvar.outfile and (args.lat is not None or args.lon is not None):
                  ^^^^^^^^^^^^
AttributeError: module 'cmasher' has no attribute 'outfile'

Finished after: 0:00:02.551555

Started at: 2026-06-22 14:45:22.743135

python /Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/CSF/csf.py --task=diagnostic --var=pyplot --dir_path=/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/tests/wrfdata/control/ --save_pdf_frames=1 --outdir=/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/tests/results/tsiad_2026-06-22_14-40-02/
Traceback (most recent call last):
  File "/Users/user/workspace/dev/wrfworkshop/spanishplumeanalysis/CSF/csf.py", line 251, in <module>
    if "SkewT" in wvar.outfile and (args.lat is not None or args.lon is not None):
                  ^^^^^^^^^^^^
AttributeError: module 'matplotlib.pyplot' has no attribute 'outfile'

Finished after: 0:00:02.367665

<snip>

2. tests/test.py

This runs but it's not clear if the results for the .csv are correct as CIN and CAPE are both NaNs:

$ cat tests/results/CSV_Data_51.38,-2.36_Bath.csv
# lat:51.38  lon:-2.36
Timestamp,AirTemp2m,DewpointTemp2m,CIN,CAPE
2026-05-21T00:00:00,283.98737,9.927796106171229,nan,nan

Note that this was run on the example data from the SharePoint share, wrfout_d02_2026-05-21_00; I have not tried running it on any of the other data yet.

[fherreazcue]

@pricet1 could you please share the dependency versions that compiled for you?
If I let it completely unpinned it fails on my side (get_cmap was moved in more recent matplotlib to live within pyplot, rather than cm), and since I do not have a mac, I do not know which versions would work...
I am testing with the "newest" versions, but I had to update the import of get_cmap in the codebase... since you didn't mention having to do that, I am not sure we're using the same versions...

[fherreazcue]

Just as a reference, I've now tested this with pinned dependencies:

• netCDF4=1.7.4
• matplotlib=3.11.0
• cmasher=1.9.2
• cartopy=0.25.0
• wrf-python=1.4.2
• imageio=2.37.0
• imageio-ffmpeg=0.6.0
• metpy=1.7.1
• setuptools=82.0.1

With the above, oututs look correct, and I get no errors... It looks like that would be supported in mac, but let me know if you have very different versions...

[pricet1]

@fherreazcue - querying my version numbers gives the following:

netcdf4                    1.7.4
matplotlib                 3.11.0
cmasher                    1.9.2
cartopy                    0.25.0
wrf-python                 1.4.2
imageio                    2.37.0
imageio-ffmpeg             0.6.0
metpy                      1.7.1
setuptools                 82.0.1

In other words, looks to me like an exact match with your version numbers.

This is indeed running on my Mac (with the issues that I mentioned previously, but this is of course in the original SpanishPlumeAnalysis repo and not your updated repo).