Sea ice-air interactions amplify multidecadal variability in the North Atlantic and Arctic region

The Arctic has experienced rapid warming since the early 20th century, superimposed on larger multidecadal variations in surface air temperature (Tas). These multidecadal variations cannot be solely explained by increasing greenhouse gases (GHGs). Model simulations suggest that internal variability plays a significant role in generating these low-frequency variations in Arctic Tas. Multidecadal variations in poleward energy transport, associated with the Atlantic Multidecadal Oscillation (AMO) and Atlantic Meridional Overturning Circulation (AMOC), are identified as major contributors to Arctic Tas multidecadal variations. It has been hypothesized that increased oceanic heat transport from the North Atlantic to the Barents-Kara Seas (BKS) and other Arctic regions reduces sea-ice cover (SIC), allowing the Arctic Ocean to absorb more energy in summer and release more heat to the air in winter, resulting in warmer winter Tas. Conversely, Arctic sea-ice loss has been shown to weaken the AMOC in coupled model simulations. Sea ice in the subpolar Atlantic and Nordic Seas may also influence Atlantic multidecadal variability (AMV). This study addresses the critical question of whether multidecadal variations in AMV (or AMO) and AMOC, a major source of multidecadal climate variability, and the oceanic heat transport itself, are influenced by sea-ice changes. The study also investigates whether poleward energy transport alone can cause large multidecadal variations in Arctic Tas without the involvement of sea ice. Answering these questions is crucial for understanding the formation mechanisms of the AMOC and AMV, and for predicting future climate, given the projected significant sea-ice loss over the BKS and other Arctic regions in coming decades and centuries.

Previous research has highlighted the significant role of internal climate variability in driving multidecadal fluctuations in Arctic temperature. Studies have pointed to the influence of the Atlantic Multidecadal Oscillation (AMO) and Atlantic Meridional Overturning Circulation (AMOC) on Arctic temperature variability, suggesting a link between oceanic heat transport and Arctic warming. Some research suggests that above-normal oceanic heat transport to the Arctic reduces sea ice, leading to a feedback loop of increased energy absorption and heat release, impacting surface air temperatures. Conversely, the role of Arctic sea ice loss in weakening the AMOC has also been explored, with studies suggesting a connection between sea ice in the subpolar Atlantic and Atlantic multidecadal variability. However, a crucial gap remains in understanding the potential influence of sea ice changes on the AMV, AMOC, and the oceanic heat transport itself. The extent to which poleward energy transport directly causes large multidecadal variations in Arctic temperature independent of sea ice is also unclear.

The study analyzes observations and model simulations to investigate the role of sea ice-air interactions in multidecadal climate variability. Monthly mean surface air temperature (Tas), sea surface temperature (SST), and sea ice data from 1950 to 2020 were obtained from the ERA5 reanalysis. Data from twelve CMIP5/CMIP6 models were also used, including historical and RCP8.5 simulations extended to the 23rd century. The Community Earth System Model version 1.2.1 (CESM1) was used for further analysis, with two types of simulations: one with fully coupled dynamic sea ice (CTL), and another with fixed sea ice concentration in flux calculations (CTL_FixedIce). To isolate internally generated decadal-multidecadal variability, the externally forced signal was removed from all datasets using a method based on global mean surface temperature. A 10-90 year Lanczos band-pass filter was applied to isolate decadal-multidecadal variations. The standard deviation of the detrended and filtered anomalies was calculated to quantify variability. The AMV index was defined using SST anomalies in the North Atlantic, and the AMOC index using meridional stream function anomalies. F-tests were used to assess the significance of changes in standard deviations, and Student's t-tests for composite differences. A resampling method was used to test the significance of correlations between autocorrelated time series.

Observations reveal significant multidecadal variations in sea ice cover (SIC) and Tas along Arctic sea-ice margins, which are anti-correlated and related to Atlantic multidecadal variability (AMV). The AMV-associated SST anomalies are much smaller than the Tas anomalies in the Arctic, suggesting that direct heating by AMV is insufficient to explain the large Arctic Tas anomalies. CESM1 simulations reproduce observed multidecadal variations in SIC, Tas, and AMV. However, when sea ice-air two-way interactions are removed in the CTL_FixedIce simulations, multidecadal variations in Arctic Tas weaken substantially, especially over the GNS and BKS. Multidecadal SST variations also reduce moderately over the northern North Atlantic and Nordic Seas. The recent decadal warming trend over the BKS is significantly reduced when sea ice-air interactions are removed. The study reveals a positive feedback loop where multidecadal sea-ice decline allows the warm Arctic Ocean to release more heat, further warming the air and melting sea ice. This feedback is weakened or reversed when fixed SIC is used in flux calculations. Multidecadal variations in winter SST from the northern North Atlantic to the BKS weaken significantly when Arctic sea ice is fixed. The CESM1 CTL run simulates large multidecadal AMOC variations, which weaken when sea ice-air interactions are cut off. The AMOC index is correlated with SIC and latent heat flux (LHF) over the Labrador Sea on multidecadal timescales, with SIC and LHF leading the AMOC index by 4-5 years. Multidecadal SIC decrease increases LHF, surface salinity, upper-ocean density, and NADW formation, leading to a stronger AMOC. When SIC is fixed, the SIC-associated multidecadal LHF variability decreases, weakening SST variations and AMOC variability. Similar results are observed in other climate models, indicating that SIC variations amplify multidecadal variability in AMOC and NASST. When subpolar sea ice melts, the lack of sea ice-air interactions weakens multidecadal variability in surface fluxes and upper-ocean conditions, leading to weaker and shallower AMOC variations.

The findings highlight the crucial role of sea ice-air two-way interactions in amplifying multidecadal variability in the North Atlantic and Arctic. While poleward heat transport from the North Atlantic contributes to Arctic warming, the local sea ice-air interactions are primarily responsible for the large Tas and SIC multidecadal variations. Without these interactions, not only do Tas variations disappear, but multidecadal variations in SIC, SSTs, AMV, and AMOC weaken considerably. The reduced AMOC multidecadal variability under weakened sea ice-air interactions is consistent with projected future weakening. Winter sea ice-air interactions are a major mechanism for generating or amplifying multidecadal variability in AMOC and AMV, complementing previous notions that the AMOC’s multidecadal variability is generated by stochastic atmospheric variability. Models need to accurately simulate sea ice cover and its interactions to reliably simulate AMOC responses to future GHG increases. Biases in model-simulated AMOC strength could be linked to biases in model-simulated sea-ice extent. The northward shift of sea ice margins under warming affects AMOC and AMV, but further research is needed to quantify this impact. The reduced variability of the AMOC and AMV in an equilibrium warmer climate when subpolar North Atlantic sea ice disappears remains to be investigated.

This study demonstrates that sea ice-air interactions play a crucial role in amplifying multidecadal climate variability in the North Atlantic and Arctic regions. The findings underscore the importance of accurately representing these interactions in climate models to improve predictions of future climate change, particularly concerning the AMOC and its variability. Future research should focus on investigating the long-term impacts of diminishing sea ice-air interactions on climate variability and exploring the recovery of AMOC variability in equilibrium warmer climates.

While the study uses multiple datasets and models, uncertainties remain in the representation of sea ice processes in climate models. The fixed sea ice concentration approach, while useful for isolating the effects of sea ice-air interactions, is an idealized representation. Future work could explore more sophisticated methods to examine the role of sea ice-air interactions in a fully coupled climate system. Also, although the study considers other factors that may also affect the results, the exact influence of each of the identified processes requires further quantitative investigation.