Accept float in days_over_precip_thresh

[huard]

Pull Request Checklist:

• This PR addresses an already opened issue (for bug fixes / features)
• Tests for the changes have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)
• HISTORY.rst has been updated

After merging to master and before closing the branch:

• bumpversion (minor / major / patch) has been called on master branch
• Tags have been pushed

• What kind of change does this PR introduce?

• Add support for simple annual percentiles.

• Does this PR introduce a breaking change?
  No

• Other information:
  TODO

• Review code and naming.
• Apply same logic to other relevant functions

This PR fixes #79

[coveralls]

Pull Request Test Coverage Report for Build 1094

• 11 of 11 (100.0%) changed or added relevant lines in 2 files are covered.
• 19 unchanged lines in 3 files lost coverage.
• Overall coverage increased (+0.5%) to 92.989%

---

Files with Coverage Reduction	New Missed Lines	%
xclim/indices/_multivariate.py	1	99.35%
xclim/indices/_simple.py	2	97.65%
xclim/utils.py	16	95.03%

Totals
Change from base Build 1041:	0.5%
Covered Lines:	1313
Relevant Lines:	1412

---

💛 - Coveralls

[huard]

Is the name resample_doy sufficiently informative ? resample_from_doy ?

[huard]

@Zeitsperre Are the azure failures expected ? Is this something we really want to invest in ?

[Zeitsperre]

@huard Yeah, I'm thinking it might be worthwhile to move from Azure as well. The goal was simply to perform the same sort of platform checks as macOS. I learned a few weeks ago that Travis CI recently added a Windows image. I'd much rather use their infra.

I'll open an issue for replacing Azure with Travis CI for Windows builds.

[huard]

@tlogan2000 Could you take a look ?

[Zeitsperre]

@huard The Azure builds are no longer valid. There was a misconfiguration in their .yaml and I didn't see the value in investing time into making them all work. I feel like we can wait for Travis CI to create some Windows images.

For this PR, I would simply ignore those checks. Everything is passing on Travis.

[tlogan2000]

@tlogan2000 Could you take a look ?

I will take a look at this soon... sorry it slipped through a crack I think

[huard]

@tlogan2000 Removed support for str arguments, thanks for catching this.

Found a small bug in xarray while writing new tests. A PR is submitted pydata/xarray#3305

[tlogan2000]

@huard with the roll back of float/string input what new features does this PR still include? Is it the doy_resample()?

[huard]

Yes, that's mostly it. I'd call it a clean-up of how doy percentiles are handled.

Temperature indicators require a doy percentile, while precip percentiles also accept a single "annual" value.

[tlogan2000]

Would an example of using an annual percentile DataArray (i.e. not using percentile_doy()) be useful
Currently the help shows :

p75 = percentile_doy(historical_pr, per=0.75)
r75p = days_over_precip_thresh(pr, p75)

but users could also send in a 2d array of the annual percentile (NB* not sure of my syntax here)

p75ann = historical_pr.quantile(dim-'time',q=0.75)

[tlogan2000]

Yes, that's mostly it. I'd call it a clean-up of how doy percentiles are handled.

Temperature indicators require a doy percentile, while precip percentiles also accept a single "annual" value.

Sounds good

[huard]

Made change in docstring. Won't really work as-is however until xarray 0.13 because keep_attrs does not yet work for DataArray.quantiles.

[Zeitsperre]

@huard if you want to update the history and bumpversion patch master, I can pull the new master and send it to PyPI/conda-forge.

[huard]

@Zeitsperre
So bumpversion does not actually change the tag anymore ? I had to commit manually the change and call git tag myself... weird.

[Zeitsperre]

The tagging is re-enabled on another PR since I realize it doesn't cause commits to fail (only commit = true seems to do that). I'll inform people when that is in master.

I had it turned off since it was conflicting with pre-commit. It turns out that the bumpversion python library has a few bugs with automatically adding lines (causing code formatting to fail). This might be fixed in a fork that Birdhouse uses called bump2version but I need to look into it.