ajout nb variabilite interannelle

[martinleduc]

No description provided.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[tlogan2000]

Line #3.    ds = xr.open_dataset(url, chunks={"time": -1, "lat": 50, "lon": 50})

Missed this before but this time chunking could potentially cause issues if used in more advanced calculations or on longer time series

Suggestion : chunks={"time": (365*4)+1, "lat": 50, "lon": 50}

This has become a pretty standard chunk size on PAVICS datasets (4 years in time and 50, 50 in spatial dims)

Reply via ReviewNB

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

tlogan2000 commented on 2025-01-08T14:58:07Z
----------------------------------------------------------------

Line #3.    ds = xr.open_dataset(url, chunks={"time": -1, "lat": 50, "lon": 50})

Suggest to change chunks to the same as mentioned in comment added to the 'Climate Average' notebook

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

tlogan2000 commented on 2025-01-08T14:58:08Z
----------------------------------------------------------------

Line #2.    std_ann = tas_region.resample(time='YE').mean().std(dim='time',ddof=1)

Use time=YS (year-start) here instead. The time=YE (year end) frequency will potentially cause problems if code is eventually applied to climate simulations with a mix of 360_day and other calendars (time coord anchor point will be 12-30 for 360day and 12-31 for the rest). Combining data in an ensemble will then be problematic as time coord do not align.

In this case nothing changes (and even here the output has no time dim) but this just avoids potential confusion elsewhere if people get used to using 'YS' (or 'MS', 'QS-DEC' etc) as a default.