Arctic Warming: Model Insights and Challenges

The Arctic has heated up much faster in recent decades than any other region in the world. Analysis of the Arctic energy budget provides important clues about the processes behind this pronounced warming trend.

Figure 1: Map showing the Arctic and the study area bounded by the red lines in the main Arctic gateways.

Climate models are expected to reliably predict and quantify future changes in the Arctic climate system. When applied to the recent past, however, significant biases can often be found, which likely affect projections into the future. In a recent study we aim to determine which climate models can best represent the complex processes of the Arctic energy budget. We analyze the simulated key energy budget variables and their connections to understand typical model biases.

Climate models are sophisticated computer simulations that replicate Earth’s complex climate system to forecast future climatic conditions, evaluate the potential impacts of human activities on the climate and assess areas in time and space where no direct observations are available. Climate models are essential tools for scientists to project and comprehend the dynamics of Earth’s climate system and changes therein, to inform policymakers and the public. The complex interactions between atmosphere, ocean and sea ice pose a significant challenge to Arctic climate simulations and introduce large uncertainties and biases. This raises the need for a thorough evaluation of historical climate model simulations against observations to detect potential shortcomings and improve our confidence in future projections of Arctic change.

We focus here on the 6th generation of models participating in the Coupled Model Intercomparison Project (CMIP6), a project that has been initiated by the World Climate Research Program (WCRP), fostering collaboration and ensuring the robustness of climate projections. To date, these models provide the most comprehensive and accurate description of the present and future climate and its subsystems on the global scale.

The vertically integrated energy budget comprises vertical energy fluxes at the surface (F_S) and the top of the atmosphere (Rad_TOA), lateral transports of heat through the atmosphere (F_A), ocean (F_O) and sea ice (F_I) and changes in the energy storage in ocean (OHCT), sea ice (MET) and atmosphere (AET) as shown in Figure 2. The Arctic climate system is generally characterized by a net energy loss to space throughout most of the year, whereby sustained poleward heat transports by atmosphere and ocean balance this radiative imbalance. As the fluxes strongly depend on insolation, they feature pronounced seasonalities.

Due to various feedback mechanisms the effects of global warming are especially pronounced in the Arctic, where the surface is warming at a rate about 2-4 times as fast as globally (termed Arctic Amplification). Robust observational estimates of the Arctic energy budget (such as provided by Mayer et al. 2019) are essential to understand this pronounced warming trend and predict possible future changes.

To deliver reliable projections of the future change and improve our understanding of the underlying processes, climate models need to realistically capture the intricate interactions within the Arctic energy budget.

Figure 2: Schematic of the Arctic energy budget, adapted from Mayer et al. (2019). The size of the arrows is scaled by their relative magnitude.

Results of our new study revealed that while the models perform reasonably well in some aspects there are also some systematic errors in certain major energy budget components. All CMIP6 models consistently underestimate the upward energy fluxes at the Arctic surface (Fs), as can be seen from Fig. 3. Differences are particularly large in the Nordic Seas where the CMIP6 multi-model mean shows a 30% lower net upward flux compared to observational estimates (REF). Deviations in Fs over the central Arctic are small, but substantial discrepancies are observed near the sea ice edge between Greenland and Svalbard, the Barents Sea, and the Norwegian Sea. The positioning of the sea ice edge varies among CMIP6 models, whereby they generally tend to underestimate the area of open water resulting in smaller net outgoing energy fluxes. Apart from sea ice, the sea surface temperature also has major effects on the vertical energy fluxes.

Figure 3: Surface energy fluxes from observations (REF, left) and their difference to the CMIP6 multi-model mean (right). Note that fluxes are counted positive if directed from the atmosphere into the ocean.

The loss of energy to space over the Arctic is balanced by northward heat transports in atmosphere and ocean. The CMIP6 models show a large range of simulated oceanic heat transports (Figure 4, y-axis) and tend to underestimate the net inflow of energy into the Arctic. Those biases are mainly driven by biases in the inflow of Atlantic waters via the Norwegian Coastal Current and lower temperatures in the Barents Sea Opening. Heat transports and biases therein have some major effects on the state of and changes within the Arctic system. We see clear connections between heat transports and sea ice extent (Figure 4, left) as well as ocean temperatures and changes thereof (Figure 4, right). Any biases subsequently also propagate into the atmosphere through vertical surface fluxes. The impact of heat transports on other components of the Arctic system highlights the importance of detecting the source of any possible biases therein. The climate models deviating most strongly from the observed states and transports should likely be omitted from climate projection ensembles as used by the Intergovernmental Panel on Climate Change (IPCC), since they tend to inflate the projection uncertainty estimates.

Figure 4: Connection between oceanic heat transports (OHT) and sea ice (left) as well as the warming rate of the ocean (right). Colorful symbols show different CMIP6 models.

In the challenge of a changing climate the accurate representation of the complex processes in the Arctic in climate models is vital for reliable future projections of Arctic change. Yet the climate models exhibit systematic errors in some of the major energy budget components. They are likely to affect also climate sensitivity (amount of warming in association with the increase of atmospheric CO2) and therefore we propose to include oceanic transports as quality metrics in addition to basic state quantities. Such process-based metrics could better constrain future model projections and therefore reduce the spread of future projections of Arctic change.