The AMOC's tipping point behaves like a speed limit, not a temperature line

Researchers at Utrecht University's Institute for Marine and Atmospheric Research report in Nature Climate Change that the Atlantic Meridional Overturning Circulation collapsed at around +2C of warming when atmospheric CO2 rose quickly in their model, but stayed intact past +5C when it rose slowly ^[1][6][7]. That inverts the usual framing: the number to plan against is not a temperature level but a rate of change, and the previously assumed collapse point of about +4C turns out to be an artefact of how fast you get there ^[4][5].

The experiment is simple in design. The group ran one climate model twice with gradually increasing CO2, changing only the speed: 0.5 ppm per year in one run, 2.5 ppm per year in the other, the faster figure comparable to today's rate ^[6]. That is a fivefold difference in pace ^[14]. Under the slow ramp the circulation remained stable well beyond +4C and did not collapse even at +5C; under the fast ramp it went at roughly +2C ^[7][8]. Co-author Reyk Börner says the deliberately slow scenario was chosen to isolate the effect of the warming rate alone, independent of how warm it eventually gets ^[9].

The mechanism the authors offer is adjustment time in the deep ocean. Under slow warming the whole water column, surface to depth, has time to reorganise; under faster warming, in co-author Henk Dijkstra's phrasing, the ocean simply cannot keep up ^[10]. Lead author Rene van Westen puts the conclusion plainly: there is not necessarily a fixed temperature beyond which the AMOC inevitably collapses, and stability depends on how fast the climate is changing ^[5]. The critical warming rate, according to the researchers, sits around 0.3C per decade, a pace the world is already approaching ^[11].

This is the same group's fourth pass at the problem, and the new paper is partly a reconciliation of its own earlier numbers ^[12]. A 2024 study showed meltwater in the North Atlantic destabilises the circulation and identified a critical meltwater threshold, but one high enough to be unrealistic on its own; it did not account for warming or its pace ^[15]. A later study put the tipping point at around 2060 under both intermediate- and high-emissions scenarios, at roughly 2.5C of warming ^[16]. The new work argues that the spread in published temperature thresholds across models and scenarios is explained by rate: the meltwater threshold is real, a universal temperature threshold is not ^[17].

Read it as model output, not observation. It is one model run twice, and the reported result is a rate dependence rather than a calibrated date ^[6]. The operational consequence is still awkward for anyone whose climate risk register is indexed to a peak-warming figure. A 2C-compatible pathway reached quickly is not, on this evidence, safer for the AMOC than a 5C pathway reached slowly ^[7][8], and overshoot-then-remove strategies inherit that problem directly.

What to watch: whether other modelling groups reproduce the rate dependence in different models, whether the observed decadal warming rate crosses the 0.3C figure the Utrecht team flags ^[11], and whether the group's own 2060 estimate is restated once rate is treated as the controlling variable ^[16][17].