Sea level rise (SLR) is a major consequence of climate change, causing coastal flooding and threatening lives and infrastructure. Coastal sea level fluctuates across various timescales, with low-frequency changes forming a background state that modulates the impact of high-frequency events like storms and tides. The superposition of short-term SLR with decadal and longer-term changes can lead to extreme sea level events. Both external radiative forcing and internal variability influence SLR. The U.S. Southeast Coast (USSEC) has experienced rapid SLR acceleration since 2010, with a linear trend of ~10.8 mm/year (2010-2022), significantly higher than the 1920-2009 trend (~2.6 mm/year). This contrasts with a weak deceleration of SLR north of Cape Hatteras. Several hypotheses have been proposed to explain this acceleration, including changes in the Gulf Stream (GS) system and the AMOC, warming of the Florida Current/GS waters, large-scale heat convergence in the North Atlantic subtropical gyre, the combined effects of external forcing and wind-driven Rossby waves, and atmospheric condition changes via the inverse barometer effect. The mechanisms driving this acceleration remain debated. The accelerated SLR along the densely populated USSEC elevates storm surge and exacerbates coastal flooding, highlighting the need to understand and predict this phenomenon to mitigate and adapt to its impacts. This study combines observations and model simulations to explore the mechanisms and multiyear predictability of sea level along the USSEC.

Previous research has identified the USSEC as a hotspot for rapid SLR in the North Atlantic. Studies have suggested various explanations, including changes in the Gulf Stream and AMOC strength and position, warming of the Florida Current/Gulf Stream waters, large-scale heat convergence in the North Atlantic subtropical gyre related to the North Atlantic Oscillation (NAO), and combined effects of external forcing and wind-driven Rossby waves. The role of atmospheric conditions via the inverse barometer effect has also been considered. However, a consensus on the primary mechanisms responsible for the observed acceleration remains elusive.

This study utilizes a combination of observational data and model simulations to investigate the mechanisms and predictability of sea level rise along the USSEC. Observational data include monthly gridded sea surface height (SSH) data from the Copernicus Marine and Environment Monitoring Service (CMEMS) starting in 1993 and monthly sea level observations from tide gauge (TG) stations processed by the Permanent Service for Mean Sea Level (PSMSL) dating back to the 20th century. The inverse barometer effect is corrected in TG observations using the 55-year Japanese Reanalysis (JRA-55) SLP dataset. The primary model employed is the Geophysical Fluid Dynamics Laboratory's Seamless system for Prediction and Earth system Research (SPEAR), specifically the low-ocean resolution version (SPEAR_LO). This fully coupled model includes components for the ocean, ice, atmosphere, and land. A SPEAR reanalysis is developed by restoring atmospheric temperature and winds towards JRA-55 and SSTs towards ERSSTv5. The retrospective decadal prediction system based on SPEAR_LO is initialized from the SPEAR reanalysis, producing hindcasts with 20 ensemble members initialized on January 1st of each year from 1961 to 2022. The hindcasts are integrated for 10 years with realistic time-evolving radiative forcings. Lead-time-dependent climatology is subtracted to remove model drift. Empirical Orthogonal Function (EOF) analysis is used to identify dominant modes of sea level variability. Linear detrending is applied to satellite and TG data before EOF analysis to account for global warming trends. Steric and thermosteric/halosteric sea level components are calculated to understand the contributions of temperature and salinity changes. The average predictability time (APT) method is used to identify the most predictable components of sea level in the SPEAR decadal prediction system, maximizing predictability rather than variance. Monte Carlo methods are used to test the statistical significance of correlations and APT.

The study finds that the accelerated SLR along the USSEC after 2010 is a result of the combined effects of long-term global warming, buoyancy-driven AMOC variations, and wind-driven circulation adjustments linked to the NAO. A prominent North Atlantic sea level tripole pattern, characterized by out-of-phase sea level variability north and south of Cape Hatteras, is identified in both observations and model simulations. This tripole pattern is linked to AMOC variations, showing a strong upward trend from 2010 to 2022, coinciding with a transition phase in the AMOC and positive NAO-like atmospheric circulation. The SPEAR reanalysis and control simulation both reproduce the sea level tripole, although the reanalysis shows stronger variability. The tripole is not merely a reflection of Ekman pumping, but also represents horizontal/meridional circulation adjustments. Analysis suggests that the buoyancy-driven AMOC and NAO-like wind-driven circulation adjustments both contribute to the SLR acceleration, with the latter driving high sea levels across the subtropical region. EOF analysis of the SPEAR control simulation separates low-frequency variability (AMOC fingerprint) and high-frequency variability (NAO-like wind response). The accelerated SLR in the SPEAR reanalysis is mainly caused by meridional heat transport convergence associated with the circulation adjustments. This convergence is a key driver for the SLR acceleration, as shown by the strong correlation between monthly time derivatives of sea level and meridional heat transport convergence. The buoyancy-driven AMOC plays a crucial role in this heat transport convergence. When the AMOC transitions from a negative to a positive phase, southward-propagating AMOC anomalies lead to a northward shift in the Gulf Stream path and heat convergence near the Gulf Stream, elevating sea levels. Initialized decadal hindcasts successfully capture the rising sea level trend along the USSEC from 2010-2022, showing a prediction skill of up to 5 years for AMOC-induced SLR and 2 years for wind-driven variations. Analysis extending back into the 20th century using SPEAR reanalysis reveals that both AMOC and wind-driven circulation play roles in shaping sea level variability, with multidecadal AMOC fluctuations and decadal-scale variability in the NA tripole mode influencing sea level. The prediction skill reduces to three years for the longer 1958-2022 period, and to two years when considering high-frequency sea level variability (<15 years). The two-year skill is likely linked to the propagation of Rossby waves associated with the NAO-driven tripole mode. Forecasts indicate a likely slowdown of the rapid SLR acceleration along the USSEC in the next five years.

This study demonstrates that the rapid acceleration of SLR along the USSEC after 2010 is not solely attributable to external forcing, but also reflects a significant contribution from internal climate variability related to AMOC and NAO fluctuations. The combined effects of long-term global warming, multidecadal AMOC variations, and wind-driven responses to the NAO provide a comprehensive explanation for the observed changes. The findings highlight the importance of considering both external and internal factors when assessing coastal flood risk. The ability to predict SLR on multiyear timescales is a significant advancement, offering valuable information for coastal management and adaptation strategies. The skill in predicting AMOC-induced SLR up to 5 years in advance and wind-driven changes up to 2 years in advance provides a basis for improved forecasting of coastal sea levels and associated hazards. While the study focuses on identifying potential drivers of SLR and predictability, the precise separation of forced and internal variability remains challenging. Future research should strive to improve the model's fidelity in simulating the interactions between atmospheric circulation, ocean circulation, and sea level. The incorporation of additional factors such as tide and land ice components, and non-climate factors, like vertical land movements into future analyses is also warranted.

This study reveals the complex interplay between long-term global warming, buoyancy-driven AMOC variations, and wind-driven circulation linked to the NAO in driving the rapid acceleration of SLR along the USSEC. The demonstrated multiyear predictability of SLR, particularly the 5-year skill for AMOC-induced changes and 2-year skill for wind-driven changes, presents a significant opportunity for improved coastal flood risk assessment and adaptation strategies. Future work should focus on refining models to better separate forced and internal variability and include additional processes to enhance forecasting accuracy.

The study acknowledges several limitations. The SPEAR model used has low ocean resolution and lacks tide and land ice components, and does not account for non-climate factors such as vertical land movement. These limitations may affect the interpretation and generalizability of the results. The relationships proposed require validation from other climate models and should be revisited as more advanced models and observations become available. The precise separation of forced and internal variability remains challenging, and the study does not aim to fully resolve this issue.