Feedback Processes causing an AMOC Collapse in the Community Earth System Model

TL;DR

This paper investigates the mechanisms behind AMOC collapse in the CESM under quasi-equilibrium freshwater forcing, focusing on the role of the salt-advection feedback. By reconstructing the AMOC from the meridional density contrast via thermal wind balance, it shows that AMOC stability is primarily governed by the Atlantic freshwater budget, with $F_{\text{ovS}}$ serving as a key stability indicator. The dominant destabilising mechanism is the salt-advection feedback, whose strength grows as the velocity-weighted salinity contrast $\Delta_v S$ becomes more negative, while gyre circulations and sea-ice/ocean–atmosphere fluxes modulate this path to tipping. The findings imply that many modern climate models may underestimate tipping risk due to positive $F_{\text{ovS}}$ biases, underscoring the need to reassess stability indicators in CMIP7.

Abstract

The Atlantic Meridional Overturning Circulation (AMOC) is recognized as a tipping element within the global climate system. Central to its tipping behavior is the salt-advection feedback mechanism, which has been extensively studied in box models and models of intermediate complexity. However, in contemporary, highly complex climate models, the importance and functioning of this feedback mechanism is less clear due to the intricate interplay of numerous ocean-atmosphere-sea ice feedbacks. In this study, we conduct a detailed mechanistic analysis of an AMOC collapse under quasi-equilibrium forcing conditions using the Community Earth System Model (CESM). By reconstructing the AMOC strength from the meridional density contrast across the Atlantic Ocean, we demonstrate that AMOC stability can be related to the Atlantic freshwater budget, revealing several important feedbacks. The dominant contribution is the destabilising salt-advection feedback, which is quantified through a negative sign of the overturning freshwater transport at 34$^{\circ}$S, indicated by $F_{\mathrm{ovS}}$. Other feedbacks are related to changes in North Atlantic sea-ice melt (destabilising), ocean-atmosphere freshwater fluxes (destabilising) and gyre transports (stabilising). Our study clarifies the role of $F_{\mathrm{ovS}}$ as an indicator of the background state stability of the AMOC. As many modern climate models have a positive $F_{\mathrm{ovS}}$ bias this implies that their AMOC is too stable which leads to an underestimation of the risk of an AMOC collapse under climate change.

Figures (14)

• Figure 1: (a): Time-mean and zonally-averaged potential density of the Atlantic Ocean (full zonal extent) and Southern Ocean (zonal extent between 53$^\circ$W -- 20$^\circ$E), for the first 100 model years. (b): Time-mean meridional potential density with depth ($\Delta_y \rho$ thick black) and vertical average ($\langle\Delta_y \rho\rangle$ dashed red).
• Figure 2: (a): Time-mean depth dependence of the interior overturning streamfunction (black curve) and TWB reconstruction (blue curve) over the first 100 model years. (b): Interior AMOC strength at 1,000 m under varying hosing strengths $F_H$, as simulated by the CESM (thick black curve) and the TWB reconstruction (thick blue curve).
• Figure 3: Decomposition of change in $\hat{\Psi}_{\mathrm{int}}$ (compared to the initial state) driven by changes in vertically-averaged density contrast (thick green line) and stratification (thick red line). Inset shows time series of $\langle\Delta_y\rho\rangle$ (thick green line).
• Figure 4: (a): Decomposition of AMOC perturbations by vertically-averaged potential density changes into thermal (orange) and haline (magenta) by the North (dash-dot) and South Atlantic (dotted). (b & c): The North Atlantic freshwater and heat balances (see equation (\ref{['Balance']})).
• Figure 5: (a): Interior AMOC streamfunction (thick black) and $\psi$ (equation (\ref{['MStrF']})) evaluated at 34$^\circ$S and 1,000 depth (thin black). (b): Sketch illustrating the rationale behind equation (\ref{['Physical_Rel']}). Arrows indicate the volume transport of the overturning circulation, with the abbreviation below each arrow corresponding to the transported Atlantic water mass. (c): Relation between $\Delta_v S$ and $\langle S_s-S_n\rangle$ anomaly. The slope $\mu$ is calculated from a linear regression. Each dot represents a 10-year average. (d): AMOC anomaly resulting from change in $\Delta_v S$ (thick cyan) and $\langle\Delta_y S\rangle$ (thick magenta). Inset shows that time series of $\Delta_v S$ and $\langle S_s-S_n\rangle$.