CMIP6/OMIP simulations of the Arctic Ocean and the impact of resolutions

Underlying the polar climate system are a number of closely coupled processes that are interconnected through complex feedbacks on a range of temporal and spatial scales. Observations are limited in these inaccessible and remote areas, and understanding of these processes often relies on regional and global climate modelling. However, large uncertainties remain due to unresolved key processes in both the regional and global contexts.

In this presentation, we first show that large model spread and biases exist in simulating the Arctic Ocean hydrography from the latest CMIP6/OMIP experiments. Our results indicate that there are almost no improvements compared with the previous CORE-II experiments (with similar OMIP-like protocol) which were thoroughly assessed by the ocean modelling community. The model spread and biases are especially conspicuous in the simulation of subsurface halocline and Atlantic Water, the latter often being too warm, too thick, and too deep for many models. The models largely agree on the interannual/decadal variabilities of key metrics, such as volume/heat/salt transport across main Arctic gateways, as dictated by the common atmospheric forcing reanalysis.

We then examine a hierarchy of global models with horizontal resolutions of the ocean on the order of 1-deg, 0.25-deg, and 0.1-deg. For the 0.1-deg resolution, we take advantage of a recent unprecedented ensemble of high-resolution CESM simulations, as well as NorESM simulations of similar ocean resolution but of shorter integration. High(er) resolutions show signs of improvements and advantages in simulating the Arctic Ocean, but certain biases remain, which will be discussed together with the challenges of high-resolution simulations in the region.