Skip to content

Conversation

@jakobwes
Copy link
Collaborator

@jakobwes jakobwes commented Aug 18, 2025

This accounts for the temporal evolution of trends. It was already implemented for CDFt and QDM and is extended to SDM, ECDFM , QuantileMapping. It is not necessary for LinearScaling nor for ISIMIP.

jakobwes added 9 commits July 2, 2025 13:54
A running window over the future period is useful to smooth out preserved trends when bias adjusting. Functionality for this existed in the QDM and CDFt classes, but not in others. Updates:

- New general class: SeasonalAndFutureRunningWindowDebiaser, inheriting from the SeasonalRunningWindowDebiaser (formerly RunningWindowDebiaser) implementing seasonal running window and one over the future period.
- Changed CDFt, SDM, QDM, ECDFM and QM to inherit from that class in order to add this functionality to the classes. For QDM the implementation is slightly slower than earlier versions as pre-computation was done before the future running window application, but tests showed that this is in the order of ~1-2% and the code is strongly simplified that way.
Use length of 31 years and step length of 9 years as default. Default: off. Updates do docs.
@jakobwes jakobwes changed the title Added a running window over the future for most debasers Added a running window over the future for most debiasers Aug 18, 2025
@jakobwes jakobwes merged commit f230d3e into main Aug 19, 2025
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants