The uncertainty of the ensemble mean is generally similar when computed with an historical ensemble and the corresponding piControl experiment.
This remains true for other epoch lengths: 30, 45, 60, 75-year epochs
and other variables:
Net Surface Heat flux: 30, 45, 60, 75, 90-year epochs
NHF regressed on SST: 30, 45, 60, 75, 90-year epochs
PR: 30, 45, 60, 75, 90-year epochs
SST regressed on SSH: 30, 45, 60, 75, 90-year epochs
SST: 30, 45, 60, 75, 90-year epochs
Taux: 30, 45, 60, 75, 90-year epochs
Taux regressed on SST: 30, 45, 60, 75, 90-year epochs
Tauy: 30, 45, 60, 75, 90-year epochs
project: CMIP6
experiments: piControl, historical
epoch length: 90-year
regions: Niño4 (5N-5S, 160E-150W), Niño3.4 (5N-5S, 120-170W), Niño3 (5N-5S, 90-150W)
variable: Sea Surface Height (seasonal cycle removed for variance and skewness)
statistics: mean, variance, skewness
Computed using equation (9) from Planton et al. (preprint)