The tropical Pacific (TP) shows strong variability of sea surface temperature (SST) associated with El

Sea surface temperature (SST) variability in the North Pacific (NP) has broad impacts on climate, ecology, and fishing, but its decadal prediction remains challenging. Some studies have highlighted the importance of tropical Pacific forcing on NP SST variability, while others suggested a non-negligible role of North Atlantic forcing. The contribution of each region to NP SST decadal prediction needs to be carefully evaluated. Here, we apply linear inverse models (LIMs) to assess decadal prediction skill of observed NP SST (based on predictions averaged over 2–5 and 6–9 years in advance). By using model experiments that restore to observed SST variability, we quantify tropical Pacific and North Atlantic forcing, and find that the latter exhibits stronger contribution than the former to NP SST decadal prediction. Specifically, by removing the North Atlantic forcing (tropical Pacific forcing), the area ratio of the NP region with significant prediction skill reduces from 49% to 29% (43%) for years 2–5 and from 48% to 26% (44%) for years 6–9. Further analyses reveal the possible reason: NP SST decadal prediction is strongly modulated by two leading non-orthogonal patterns of the LIMs with long decay time (although higher-order patterns also play a role for years 2–5); the North Atlantic forcing largely and robustly contributes to the long decay time, while the tropical Pacific forcing plays a secondary role. Our study calls for more efforts on improving the simulation of the Atlantic trans-basin effect for better predicting NP climate.