Table 2. Implications of models’ evaluation.

Subset of models	Averaged skill score	Unweighted occurrence	Weighted occurrence	Occurrence (S>0.8)	Occurrence (S>0.9)
Non-abrupt	0.53±0.36	77.5% (31/40)	70.1%	66.7% (12/18)	54.5% (6/11)
SPG convection collapse	0.90±0.08	17.5% (7/40)	26.6%	33.3% (6/18)	45.5% (5/11)
AMOC disruption	0.39±0.30	5.0% (2/40)	3.3%	0% (0/18)	0% (0/11)

Summary of the averaged skill scores featured by the different sub-ensembles. The probability of occurrence of the event associated with each subset of models is shown, respectively, if no weighting criterion is applied to the models, if a weighting criterion is applied to the models, if only models satisfying the condition S>0.8 are considered, if only models satisfying the condition S>0.9 are considered.