“…Therefore, focusing on the same projected target (September SIE) as Knutti et al (2017), we considered both SIE and SIC to estimate the weights for each model member. Although SAT and SST are closely related to sea-ice variability (Alekseev et al, 2021;Screen & Deser, 2019;Senftleben et al, 2020;Wang et al, 2021), the two variables are not used in the weighting scheme because of their intense co-variation and their close relation to SIC, which may lead to overfitting in the weighting. Given that the dynamics are also crucial to sea-ice variability (Lukovich et al, 2021;Park et al, 2018;Ricker et al, 2021), we considered an additional climate variable, the summer (June-July-August, JJA) mean SLP, which can drive the variability of September Arctic sea ice (Ding et al, 2017(Ding et al, , 2019.…”