Impacts of North Atlantic Model Biases on Natural Decadal Climate Variability

Abstract Increasing the horizontal resolution of an ocean model is frequently seen as a way to reduce the model biases in the North Atlantic, but we are often limited by computational resources. Here, a two‐way nested ocean model configuration (VIKING10) that consists of a high‐resolution (1/10°) component and covers the northern North Atlantic, is embedded in a 1/2° ocean grid as part of the global chemistry‐climate model, FOCI (called FOCI‐VIKING10). This configuration yields a significantly improved path of the North Atlantic current (NAC), which here reduces the North Atlantic cold bias by ∼50%. Compared with the coarse‐resolution, non‐eddying model, the improved thermal state of upper ocean layers and surface heat fluxes in a historical simulation based on FOCI‐VIKING10 are beneficial for simulating the subdecadal North Atlantic Oscillation (NAO) variability (i.e., a period of 8 years). A northward drift of the NAO‐forced ocean thermal anomalies as seen in observations and the eddying FOCI‐VIKING10, provide a lagged ocean feedback to the NAO via changes in the net surface heat flux, leading to the NAO periodicity of 8 years. This lagged feedback and the 8 years variability of the NAO cannot be captured by the non‐eddying standard FOCI historical simulation. Furthermore, the argumentative responses of the North Atlantic to the 11‐year solar cycle are re‐examined in this study. The reported solar‐induced NAO‐like responses are confirmed in the 9‐member ensemble mean based on FOCI but with low robustness among individual members. A lagged NAO‐like response is only found in the nested eddying simulation but absent from the non‐eddying reference simulation, suggesting North Atlantic biases importantly limit climate model capability to realistically solar imprints in North Atlantic climate.