Repository navigation
kimber grace and rain threshold implementation differences #2796
Description
Describe the bug
-
the threshold in pvlib is strict...
Line 208 in 0f7a205
...whereas the paper Kimber, 2006 implies that rainfall inclusive of the threshold is sufficient to clean:
-
the grace period has an off-by-one error, including the rain day and therefore effectively reducing grace days to
gracedays-1
To Reproduce
- set rainfall to exactly the value of rainfall in the accumulation period and it will not clean
- set grace days to 1, and effectively here are no grace days!
Expected behavior
- rainfall equal to should clean
- grace day of 1 should be clean for 1 day after rainfall event
Versions:
pvlib.__version__: 0.15.2
Additional context
Thank you @wholmgren for pointing out these inconsistencies!
May also be related to
Activity
dgowdaan-cmyk commentedon Jul 2, 2026
Thanks for reporting this — I agree the current > check is too strict compared with Kimber 2006. I’ll update the Kimber implementation so rainfall equal to cleaning_threshold also counts as a cleaning event, and I’ll fix the grace-period indexing so grace_period=1 means one full day after rainfall rather than including the rain day itself. I’ll add tests for both cases and submit a PR.
cwhanse commentedon Jul 11, 2026
I think this only happens if the input rainfall data are 1-day timesteps. The test data are hourly, the proposed fix "closed='both'" is just shifting the grace period by one time step.