use built in clip

[raybellwaves]

Closes errors seen in #308

The latest xarray did some refactoring on the built in clip method. It broke something in xskillscore but it helped identify areas of improvement

we were calling np.clip(xarray object, min, max). There is the built in clip method which is preferred:

• https://xarray.pydata.org/en/v0.18.0/generated/xarray.DataArray.clip.html
• https://xarray.pydata.org/en/v0.18.0/generated/xarray.Dataset.clip.html

We were calling np.abs(xarray object). We can use the python abs method to do the same:

• https://stackoverflow.com/a/46981308/6046019

For what it's worth here's how the clip refactor showed up in xskillscore.

Latest:

conda remove --name test_env --all --y
conda create -n test_env python=3.9 --y
conda activate test_env
mamba install -c conda-forge xarray netCDF4 pooch --y
python

import xarray as xr
import numpy as np

ds = xr.tutorial.open_dataset("air_temperature")
np.clip(ds, 0, 1) # -> TypeError: clip() got an unexpected keyword argument 'out'

exit()
conda deactivate

xarray v0.17.0:

conda remove --name test_env --all --y
conda create -n test_env python=3.9 --y
conda activate test_env
mamba install -c conda-forge xarray==0.17.0 netCDF4 pooch --y
python

import xarray as xr
import numpy as np

ds = xr.tutorial.open_dataset("air_temperature")
np.clip(ds, 0, 1) # Works

[codecov]

Codecov Report

Merging #309 (54ff8f2) into main (e2e3e7f) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main     #309   +/-   ##
=======================================
  Coverage   94.50%   94.50%           
=======================================
  Files          23       23           
  Lines        2566     2566           
=======================================
  Hits         2425     2425           
  Misses        141      141           

Impacted Files	Coverage Δ
xskillscore/core/probabilistic.py	90.94% <100.00%> (ø)
xskillscore/tests/test_probabilistic.py	96.64% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 67b47b9...54ff8f2. Read the comment docs.

[aaronspring]

isnt np faster?

[raybellwaves]

I see no difference in speed

[max-sixty]

It calls __abs__ (unlike using the builtin sum, which would be much slower)

[raybellwaves]

any idea why this got closed @aaronspring ?

[aaronspring]

any idea why this got closed @aaronspring ?

Ups. Sorry. No.

[aaronspring]

I used fixed and then mentioned this PR in my climpred PR. Didn’t know this works across repos

[aaronspring]

I don’t see how I could reopen…

[raybellwaves]

No worries. Seems crazy that GitHub allows that