Author note (2026-11-02). The paper cited here, arXiv:2603.11577 (“Regular and chaotic Welander oscillations in a four-dimensional conceptual model for the Atlantic Meridional Overturning Circulation”), is by John Bailie, Priya Subramanian, Bernd Krauskopf. The name(s) this essay attributes it to (Boot, Baars, Dijkstra) do not appear among the listed authors, so the attribution below is wrong or refers to someone else's earlier work; I have not worked out which. Found by a mechanical check of each essay's named authors against the arXiv author list.
The same perturbation that weakens the Atlantic overturning also makes it unpredictable.
The Atlantic Meridional Overturning Circulation (AMOC) carries warm water north, where it cools, sinks, and returns south. Climate models project it will weaken as warming drives more freshwater into the subpolar North Atlantic. The question is whether the weakening will be gradual or abrupt — and whether we'd see it coming.
Boot, Baars, and Dijkstra (arXiv:2603.11577) study a four-dimensional box model with three compartments: tropical surface, subpolar surface, and deep water. A comprehensive bifurcation analysis reveals something the simpler two-box models miss entirely: up to four coexisting overturning states, with regimes of bistability and tristability. Two distinct oscillatory modes emerge — slow millennial cycles modulated by faster decadal-to-centennial shutdowns of subpolar convection.
As freshwater input increases, three things happen simultaneously: the background overturning weakens, the convective shutdown episodes become more frequent, and — past a threshold — the dynamics become chaotic. The oscillations lose their regularity. Convection switches on and off irregularly, driven not by external forcing but by the system's own nonlinear dynamics.
This means freshwater forcing doesn't just push the AMOC toward a weaker state. It pushes it toward a state where predictions about when convection will shut down become fundamentally harder. The same perturbation that reduces the mean also destroys the periodicity. You get less circulation and less ability to forecast its fluctuations — the signal weakens and the noise increases at the same time.
Gradual forcing producing chaotic response. The linear intuition — that a slow change in input produces a slow change in output — breaks exactly where it matters most.
Boot, Baars, and Dijkstra, "Regular and chaotic Welander oscillations in a four-dimensional conceptual model for the AMOC," arXiv:2603.11577 (2026).