Impacts of North Atlantic Model Biases on Natural Decadal Climate Variability.

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. Plain Language Summary: A long‐standing cold bias in the North Atlantic in climate models could be reduced by increasing the horizontal resolution, but we are often limited by computational resources. Here we embedded a nest with 1/10° resolution in the ocean in the North Atlantic in a global chemistry‐climate model (called FOCI‐VIKING10). It can largely reduce the North Atlantic cold bias (roughly 50%) and correct the path of the North Atlantic current (NAC). North Atlantic Oscillation (NAO) subdecadal variability (a period of 8 years) can be simulated by FOCI‐VIKING10 when the representations of the NAO‐forced anomalies and the ocean feedback are improved by alleviating the biases. The reported NAO‐like responses to the 11‐year solar cycle are confirmed in the 9‐member ensemble mean with FOCI (non‐eddying). Although the solar signals are also found in a single member with FOCI‐VIKING10, we cannot rule out the aliasing of the internal variability in this single short member. For detecting a weak or varying signal of external forcings like the 11‐year solar cycle in this study, a large ensemble with a "coarse" resolution model is favorable over a single realization with a "presumably" better model. Key Points: North Atlantic biases are alleviated by an eddying nested ocean configuration embedded in a global climate model, FOCI‐VIKING10It is indicated that reduction of the North Atlantic biases could improve the representation of NAO sub‐decadal (8 years) variabilityFor detecting weak external imprints with limited computational resources, an ensemble with a coarse‐resolution model is favorable [ABSTRACT FROM AUTHOR]