On the realism of Arctic Ocean transports in CMIP6

This contribution evaluates key components of the Arctic energy budget as represented by the Coupled Model Intercomparison Project Phase 6 (CMIP6) against reanalyses and observations.

The Arctic regions are characterized by a net energy loss to space, which is balanced by northward heat transports in atmosphere and ocean. Mean and variability in the oceanic northward heat transports have major impacts on the state and change of the Arctic Ocean and sea ice. Therefore, an accurate representation of oceanic transports in climate models is a key feature to realistically simulate the Arctic climate. However, the nature of curvilinear ocean model grids and the variety of different grid types used in the CMIP ensemble, make the calculation of oceanic transports on their native grids difficult and time consuming. We developed new tools that enable the precise calculation of volume, heat, salinity and ice transports through any desired oceanic sections or straits for a large number of CMIP6 models as well as ocean reanalyses. Our tools operate on native grids and hence avoid biases that often arise from interpolation to regular grids. Those tools will be made available as open-source Python package enabling easy and effortless calculations of oceanic transports.

In the work presented here, we use the newly developed tools to compare oceanic heat transports (OHT) through the main Arctic gateways from CMIP6 models and reanalyses to those gained from observations and analyze them concerning their annual means, seasonal cycles and trends. We find strong connections between the Arctic’s mean state and lateral OHT, with variations in OHT having major effects on the sea ice cover and ocean warming rate.

Results help us to understand typical model biases. For instance, many models feature systematic biases in oceanic transports in the Arctic main gateways, e.g., some models feature to high sea ice extents due to the underestimation of heat transports entering the Arctic through the Barents Sea Opening. Using those results it is possible to generate physically based metrics to detect outliers from the model ensemble, which may be useful in reducing the spread of future projections of Arctic change.