Sea level rise, a major consequence of anthropogenic warming, poses significant threats to coastal communities and ecosystems globally. The Intergovernmental Panel on Climate Change (IPCC) projects substantial global mean sea level rise by 2100, with varying regional impacts. Observations reveal non-uniform spatial patterns, with some regions experiencing significantly faster sea level rise than the global average. The Subpolar North Atlantic (SPNA) is one such region, projected to experience accelerated sea level rise linked to a slowing Atlantic Meridional Overturning Circulation (AMOC). The AMOC slowdown reduces northward heat and salinity transport, contributing to SPNA sea level rise. While considerable uncertainty remains in the exact spatial and temporal patterns, climate models generally agree on AMOC weakening under global warming, consistent with the accelerated sea level rise observed in the SPNA. Given the significant populations and economic activity concentrated along the northeastern coast of North America, understanding sea level changes in the SPNA is crucial. The Paris Agreement goals and concerns about their achievability have spurred research into the reversibility of climate change through CO2 removal. While global mean temperature and precipitation show significant reversibility, subsystems like the AMOC exhibit nonlinear responses and hysteresis. Previous studies have explored global mean sea level reversibility; however, a regional approach, focusing on areas like the SPNA, is necessary due to the localized nature of sea level rise hazards. This study investigates the spatio-temporal evolution of sea level, particularly in the SPNA, under an idealized CO2 ramp-up and -down scenario to determine the potential for regional sea level recovery under climate mitigation.

Extensive research documents sea level rise as a consequence of global warming, primarily attributed to thermal expansion and melting ice. The IPCC's AR6 report provides projections of global mean sea level rise under different emission scenarios. However, observations reveal significant regional variability in sea level change. Studies show that certain regions, including the SPNA, experience sea level rise exceeding the global mean. This accelerated rise in the SPNA is largely associated with the AMOC slowdown, which reduces northward heat and salinity transport. Climate models generally agree on the weakening of the AMOC under enhanced greenhouse gas emissions, further supporting the observed SPNA sea level rise. Past studies have focused on the global mean perspective of sea level reversibility, but the localized nature of sea level hazards necessitates a regional focus, particularly for vulnerable coastal areas like those bordering the SPNA. Prior research highlights the hysteresis of the AMOC in response to CO2 changes, with the potential for overshoot after CO2 reduction. However, the specific regional sea level response in the SPNA to CO2 removal remains relatively understudied, prompting the need for this investigation.

This study employs the Community Earth System Model, Version 1.2 (CESM1.2), a coupled climate model incorporating atmospheric, land, ice, and ocean components. The model has a horizontal resolution of ~1°x1° for atmospheric and land components and a higher vertical resolution for ice and ocean components. Two types of idealized simulations were conducted: a 900-year present-day control experiment with constant atmospheric CO2 concentration (367 ppm) and a 500-year CO2 ramp-up and ramp-down experiment. The CO2 ramp-up experiment increased CO2 by 1% annually for 140 years to quadruple the initial concentration (1468 ppm), followed by a symmetric decrease to the original level over another 140 years, with an additional 220 years at the original CO2 level. To ensure robustness, 28 ensemble simulations were run with varying initial conditions. Sea level was calculated and decomposed into steric (density changes) and mass redistribution contributions. Steric sea level was further divided into thermosteric (temperature changes) and halosteric (salinity changes) components. Statistical significance was assessed using a bootstrap analysis, considering values significant at the 95% confidence level. Additionally, a two-fold CO2 ramp-up and -down scenario and CMIP6 model data based on the CDRMIP protocol were also utilized to verify the robustness of findings. The sea surface height (SSH) tendency equation under hydrostatic balance was used to decompose the sea level changes into density and mass contributions. A Boussinesq approximation was used in the model, requiring a correction for the global steric effect. Thermosteric and halosteric contributions were approximated using established equations.

The study reveals significant sea level fluctuations in the SPNA in response to CO2 forcing. During the CO2 ramp-up period, SPNA sea level rose approximately 1.5 times faster than the global mean, reaching up to 0.9m increase. Conversely, during the ramp-down period, SPNA sea level declined at a rate about 4.5 times faster than the global mean, falling below the global mean around 73 years after its peak. These rapid fluctuations are strongly associated with the AMOC response to CO2 forcing. The AMOC showed clear hysteresis, with a slower weakening during CO2 increase but a faster recovery and even overshoot during CO2 reduction. The timing of AMOC overshoot coincided with the cessation of SPNA sea level decline. The spatial patterns of SPNA sea level rise and fall were skewed to the northwest, closely linked to changes in deep convection and subpolar gyre circulation. Decomposition of sea level changes revealed significant contributions from steric changes (density variations), particularly the halosteric component (salinity changes). The halosteric component dominated steric sea level changes, especially the rapid decline during the recovery period. Inter-ensemble analysis revealed a significant negative relationship between AMOC strength and SPNA sea level, consistent across both the rising and recovery periods. The analysis using a more moderate two-fold CO2 change scenario produced consistent results, indicating that rapid SPNA sea level changes are potential even under low-end emissions. CMIP6 model simulations, based on the CDRMIP protocol and initialized from a pre-industrial baseline, further validated the findings, showing that these results are not uniquely dependent on initializing from the present-day state or using a single model.

The findings demonstrate the substantial sensitivity of the SPNA sea level to CO2 forcing, highlighting the region's vulnerability to rapid sea level fluctuations. The rapid sea level decline during the CO2 ramp-down period is attributed to both reduced global ocean thermal expansion and changes in meridional heat and salt redistribution driven by the AMOC hysteresis. The dominant process is the restoration of the northward transport of heat and salinity resulting from the AMOC recovery and overshoot. The halosteric component plays a crucial role in the observed sea level changes, particularly the rapid decline. These results underscore the importance of considering ocean circulation dynamics, in addition to thermal expansion, when projecting sea level changes in the SPNA. The study's results are consistent with and extend previous research by providing a more detailed spatial and temporal breakdown of sea level change mechanisms. The spatially skewed patterns are particularly relevant for assessing risks to densely populated coastlines. While this study used an idealized experiment, it provides valuable insights into potential impacts of CO2 reduction. The fast sea level recovery in the SPNA, even under a challenging CO2 reduction scenario, underlines the importance of continued and enhanced global CO2 mitigation efforts. The findings highlight the disproportionate impact on the SPNA and its coastal communities.

This study demonstrates the potential for rapid recovery of North Atlantic sea level in response to atmospheric CO2 removal. The Subpolar North Atlantic shows a strong sensitivity to CO2 changes, with sea level rise and fall rates significantly exceeding the global mean. This rapid reversibility is mainly attributed to the AMOC's response and the dominance of halosteric effects. The findings underscore the northeastern coast of North America's vulnerability to intense sea level fluctuations and the crucial role of CO2 reduction efforts. Future research should incorporate interactive land-based ice processes, particularly the Greenland Ice Sheet, to improve the accuracy of sea level projections under climate mitigation scenarios.

This study utilizes an idealized CO2 ramp-up and -down scenario, which may not perfectly represent real-world CO2 reduction pathways. The model does not include interactive land-based ice processes, particularly the Greenland Ice Sheet, which could influence SPNA sea level through mass input and AMOC dynamics. The initialization from a present-day state, rather than a pre-industrial baseline, may influence the simulated climate system responses. However, these limitations are addressed by utilizing both alternative CO2 scenarios and CMIP6 models initialized from pre-industrial states, which strengthens the robustness of the findings.