Climate Tipping Points: Bifurcation, Hysteresis and Cascading Risk

When a system stops responding smoothly

Most of the climate system behaves the way intuition expects: push it a little, it responds a little, and if you undo the push it goes back to where it started. A tipping point is where that stops being true. Past a critical threshold, a small additional nudge triggers a large, self-reinforcing change that continues even if you remove the original push - and in the worst cases the system will not return to its old state even if you later reverse the forcing that caused it. That asymmetry between the path forward and the path back is called hysteresis, and it is the mathematical signature every real tipping element shares.

The mathematics: bifurcation and bistability

A simple way to model a tipping element is a system with two stable equilibria separated by an unstable one - a double-well potential. As a control parameter (global mean temperature, say) shifts, the shape of that potential changes: the well the system currently sits in gets shallower, and at a saddle-node bifurcation it disappears entirely, forcing the system to fall - often abruptly - into the remaining well.

dx/dt = -dV/dx,  V(x) = x⁴/4 - x²/2 - μx     (canonical bistable model)
μ < μ_c   → two stable wells + one unstable barrier between them
μ = μ_c   → saddle-node bifurcation: one well vanishes
μ > μ_c   → system falls into the remaining well (the "tip")
reversing μ back below μ_c does NOT undo the tip if the old well is gone → hysteresis

Reversing the forcing after the tip does not automatically restore the original state, because by the time you push back down through mu_c, the system may already be sitting in a basin that has its own, different escape threshold - the classic hallmark of hysteresis seen in magnets, ecosystems and, the evidence suggests, in parts of Earth's climate.

Arctic sea ice: the ice-albedo feedback

Ice reflects roughly 50-70% of incoming sunlight while open ocean absorbs most of it, so any ice loss exposes darker water that absorbs more heat, melts more ice, and exposes more dark water - the ice-albedo feedback. Whether Arctic summer sea ice has a genuine bifurcation-style tipping point or instead a smoother, more reversible response is still actively debated in the literature; several studies find it can decline and recover within a few years once forcing is removed, making it more reversible than earlier headlines suggested, though the feedback loop itself is real and well measured.

The Amazon and the AMOC

The Amazon rainforest partly makes its own rain: roughly half its precipitation is recycled moisture transpired by the trees themselves and carried onward by wind. Large-scale deforestation and warming both shorten the wet season, and past a threshold the forest can no longer sustain enough recycled rainfall to remain rainforest, degrading toward a drier savanna-like state that, once established, does not simply reforest when rainfall recovers, because the trees that sustained the old moisture cycle are gone.

The Atlantic Meridional Overturning Circulation (AMOC) is a conveyor belt of ocean water: warm, salty surface water flows north, cools, becomes dense enough to sink in the Norwegian and Labrador Seas, and returns south at depth. Melting Greenland ice and increased rainfall add fresh water, which is less dense and resists sinking - weakening the overturning, which then imports less salt northward, which weakens sinking further: another self-reinforcing loop. Paleoclimate records show the AMOC has collapsed and restarted before; recent studies flag rising, though still uncertain, risk of a 21st-century weakening or collapse under continued warming, with major consequences for European winters and tropical rainfall belts.

Why cascades are the real worry

No tipping element sits in isolation. AMOC weakening would cool the North Atlantic and shift tropical rain belts south, which could stress the Amazon further; Arctic warming amplifies permafrost thaw, which releases more greenhouse gas. Researchers describe this risk as a cascade - one tipping event making its neighbours more likely to tip too - which is exactly why current assessments treat these thresholds not as independent switches but as a coupled network where the total risk can exceed the sum of the parts.

Frequently asked questions

Try it live

Everything above runs in your browser — open Climate Tipping Points and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.