Dynamical downscaling simulations and future projections of the Arctic Ocean based on FESOM and FIO-ESM.

The Arctic is one of Earth’s regions most susceptible to climate change. Climate models show that in a warming climate, the Arctic Ocean warms much faster than the global ocean mean, mainly due to the rapid warming of the Atlantic layer, which is called 'Arctic Ocean Amplification.' However, climate models still encounter challenges with large biases and considerable inter-model spread in the Arctic Ocean. For example, the Atlantic layer in the Arctic Ocean, simulated by the climate models, is too thick and too deep. This leads to the warming trend, and inter-annual variability of the simulated Atlantic Water that are too small compared to the observations. Here, we present Arctic Ocean dynamical downscaling simulations and projections based on a high-resolution ice-ocean coupled model, FESOM, and a climate model, FIO-ESM. The historical results demonstrate that the root mean square errors of temperature and salinity in the downscaling simulations are much smaller than those from CMIP6 climate models. The common biases, such as the overly deep and thick Atlantic layer in climate models, are significantly reduced by dynamical downscaling. Dynamical downscaling projections show that the Arctic Ocean may warm faster than the projections made by CMIP6 fully-coupled climate models.