Impact of ocean resolution on the North Atlantic bias in the FOCI-OpenIFS climate model

Abstract Understanding and addressing biases in climate models is essential for reliable climate projections. This study evaluates the mean state of the North Atlantic in three configurations of the FOCI-OpenIFS coupled climate model, varying in horizontal and vertical ocean resolution. Using 20-year hindcast simulations and observational datasets, we assess the impact of resolution on common biases and their potential drivers. Results show that increasing horizontal resolution in the ocean from $$\sim 50~\textrm{km}$$ to $$\sim 8~\textrm{km}$$ improves the pathways of the Gulf Stream extension and North Atlantic Current which reduces persistent surface biases in temperature and salinity. The high-resolution configuration also produces an AMOC in line with observational estimates. However, the model simulates an unrealistically deep mixed layer in the Labrador Sea. Conversely, increasing only the vertical resolution from 46 to 75 levels leads to an unrealistically shallow mixed layer in the Labrador Sea, weaker dense overflows in the Nordic Seas and a poorly simulated AMOC. In fact, the simulation with low horizontal and high vertical resolution in the ocean performs the worst across most analyzed metrics.