Ouranosinc / xscen / 25073828750

90%

push

github

Drop nans before interpolating WL period (#722)

<!-- Please ensure the PR fulfills the following requirements! -->
<!-- If this is your first PR, make sure to add your details to the
AUTHORS.rst! -->
### Pull Request Checklist:
- [ ] This PR addresses an already opened issue (for bug fixes /
features)
    - This PR fixes #xyz
- [ ] (If applicable) Documentation has been added / updated (for bug
fixes / features).
- [ ] (If applicable) Tests have been added.
- [x] This PR does not seem to break the templates.
- [x] CHANGELOG.rst has been updated (with summary of main changes).
- [x] Link to issue (:issue:`number`) and pull request (:pull:`number`)
has been added.

### What kind of change does this PR introduce?
It seems that `get_period_from_warming_level` can't handle timeseries
with nans at the beginning, scipy's interpolator returns the first value
when it is nan.

This fixes the issue by dropping any nans in the timeseries, letting the
interpolator work only on "known" data points.

I think we didn't see the issue before because it would only happen when
the beginning of a tas timeseries is missing, which only happens when a
window larger than 50 is requested (as we always have 1850-1900 for all
simulations).

Another way to do this would be `bfill('time')` as NaNs can only be
present at the beginning of the series. Both approach seem equivalent.

### Illustration of issue
```python3
xs.get_period_from_warming_level("CMIP6_NorESM2-MM_ssp126_r1i1p1f1", 0.69, window=55)
# ['1823', '1877']
```

```python3
ds = xr.open_dataset('/home/pbourg/Projets/xscen/src/xscen/data/IPCC_annual_global_tas.nc')

t = ds.tas.where((ds.experiment == 'ssp126') & (ds.source == 'NorESM2-MM'), drop=True).squeeze('simulation')
r = t.sel(time=slice('1850', '1900')).mean('time')

rd = (t - r).rolling(time=55, center=True, min_periods=55).mean().shift(time=1).cumulative('time').max()
```
Where "rd" is the rolling diff, the yearly max of warming level reached
by the model. For ... (continued)

1 of 1 new or added line in 1 file covered. (100.0%)

4389 of 4875 relevant lines covered (90.03%)

10.52 hits per line