This is the classic Stommel two-box model (Stommel 1961; simplified form after Cessi 1994) of the thermohaline "conveyor belt". A polar box and an equatorial box exchange heat and salt with a flow driven by their density difference:
dx/dt = (η1 − x(1+|x−y|)) / ε (pole↔equator temperature contrast x)
dy/dt = η2 − y(η3+|x−y|) (pole↔equator salinity contrast y)
q = x − y (density-driven overturning flow)
Here x is the temperature difference and y the salinity difference between the boxes; both are restored toward forcing values η1 (thermal) and η2 (freshwater) but eroded by the flow itself. ε ≪ 1 (fixed at 0.1) makes x relax in years while y — controlled by slow freshwater/evaporation input — drifts over centuries, exactly like the real ocean's heat vs. salt memory.
- q > 0 ("ON") — temperature dominates density: cold, salty water sinks at the pole, driving a strong overturning cell (today's real AMOC, ~15–20 Sv).
- q < 0 ("OFF / reversed") — enough fresh water dilutes the polar box that salinity dominates instead, sinking stops or reverses, and heat transport to high latitudes collapses.
- For a range of η2 the ON and OFF states are both stable at the same forcing (bistability) — a big enough freshwater pulse (the button) can flip the system permanently, even if you never change η2 again. This hysteresis is the toy-model analogue of real concerns that Greenland/Arctic meltwater could trigger an irreversible AMOC weakening.
The 3D loop shows water parcels advected along the conveyor at a speed and direction set by q; colour shifts from warm (surface) to cold (deep) as they travel, and the loop reverses direction when the circulation flips to the OFF state.