Update xclim

[Zeitsperre]

Overview

This PR fixes [issue id]

Changes:

• Added ...

Related Issue / Discussion

Additional Information

Links to other issues or sources.

[tlogan2000]

Moving forward I will need to use xscen for additions to the climatedata.ca analysis page : I have started a branch https://github.com/bird-house/finch/tree/use_xscen. I suggest we merge the two branches togther as they have more or less the same problems right now. Using xscen requirement list becomes simpler as many of the dependencies are bundled together (xclim, clisops, xesmf). However, changes to xclim are currently breaking the creation of wps processes

[aulemahal]

@tlogan2000 Indeed, there was a dict-like access on a Parameter object, which lost that feature in recent xclim. However, there is also a bug in xclim.

The indicator "frost_free_spell_max_length" has the default ">=" for its parameter op. But in its docstring, it lists the possible choices for op as <, <=, le, lt... A faulty copy-paste I presume. I added a workaround that spits a log error.

[Zeitsperre]

@aulemahal There's still time to fix this bug upstream before the patch release. I'll see if we can get everything together for a release later today.

[aulemahal]

Yish. I'll leave you the stats / discharge issues, but cf-xarray's reshape_unique_geometries seems completely broken with the newest xarray...

[tlogan2000]

@Zeitsperre @aulemahal I am starting to implement changes to the spatial averaging section of the ensembles code (using xscen but with this branch as a starting point). Opinions whether I should push these "xscen" changes in this branch directly?

[Zeitsperre]

@tlogan2000 I would make a branch based off of this work. We don't want to mix up the conversion work with the new features if we can. It's bad enough haha