Comparison Report Release

edit from 2022/03/15: re-uploaded pdf with some rewording 7914b95

This release aims to provide a first version of the comparison between icclim(xclim) and Climpact(climdex.pcic).
Attached are a pdf report and its companion zip.
Within zip file, there are the 49 indices included in the comparison. They are computed on annual frequency and drawn using icclim v4, icclim v5 and Climpact results.

Report's TL;DR:

• Most indices' results are similar regardless of the software used.
• Indices relying on "consecutive days" have varying results due to how icclim v4 and xclim handles spells lasting between resampling periods.
• Percentile bootstrap algorithm effects can be observed in WSDI and CSDI because it is missing in Climpact for these two indices.
• The computed percentiles are a bit different between the 3 software, but I haven't found why.
• For some indices, icclim v4 could give strange results. In Python we recommend switching to icclim v5 (or xclim directly).