The Atlantic Meridional Overturning Circulation carries warm surface water northward and cold deep water southward, redistributing heat from the tropics to the North Atlantic. Its collapse would restructure European and African weather patterns, shift monsoon systems, and accelerate sea level rise along the US East Coast. Climate policy treats AMOC collapse as a threshold problem: there's a critical warming level — commonly cited around 4 degrees Celsius above preindustrial — and below that threshold the circulation survives.
Van Westen, Borner, and Dijkstra (arXiv 2602.09964, February 2026) show the threshold depends on how fast you get there.
Under slow CO2 increase — 0.5 parts per million per year — the AMOC survives past 5.5 degrees of warming. The circulation has time to adjust, redistributing freshwater and heat incrementally as forcing changes. The surface ocean freshens gradually, deep water formation weakens gradually, and the system tracks its stable state continuously.
Under realistic emission rates — 1 to 2 parts per million per year, matching current trajectories — the AMOC collapses at 2.2 to 2.8 degrees. The same system, exposed to the same total forcing, fails at half the warming level because the forcing arrives too fast for the circulation to track its equilibrium.
The mechanism is dynamical. The AMOC's adjustment timescale — the time it takes for salinity and temperature distributions to reach equilibrium after a perturbation — is measured in decades to centuries. When the forcing changes faster than this timescale, the system can't keep up. It overshoots, entering a region of phase space where the stable circulation state no longer exists, even though a slower approach would have navigated around it. The tipping point isn't a fixed temperature — it's a moving target that depends on the trajectory.
This makes fixed warming thresholds misleading for policy. A 3-degree target doesn't protect the AMOC if the 3 degrees arrive in 80 years rather than 300. The same endpoint reached at different speeds has different outcomes. The concept of a temperature threshold implicitly assumes the system is always in quasi-equilibrium — that the present state depends only on the present forcing, not on the history of how the forcing arrived. For a system with multidecadal memory, that assumption fails.
The paper uses a coupled ocean-atmosphere model rather than the simplified box models that generated earlier threshold estimates. The resolution is coarse enough for computational tractability but captures the essential salt-advection feedback: freshening in the North Atlantic weakens deep water formation, which reduces northward salt transport, which further freshens the North Atlantic. This feedback loop is what creates the tipping behavior, and its sensitivity to rate of change — not just magnitude of change — is the new result.
The implication is that the relevant policy variable isn't where the temperature ends up but how fast it gets there. Rapid overshoot followed by drawdown may push the AMOC past its rate-dependent tipping point even if the long-term equilibrium temperature is safe. The speed of the transition matters as much as the destination, and the two are not interchangeable.