Presets
Playback
Scenario
Climate physics
Carbon cycle and sea level
Time
Layers
Model check
The science
Energy balance by latitude (Budyko-Sellers / North)
The Earth is cut into 36 equal-area latitude bands (equal steps of x = sin(latitude)). Each band obeys
C dT/dt = Q S(x) (1 - α(T)) - (A - F + B T) + D d/dx[(1 - x2) dT/dx] - γ (T - Tref - Td).
Q = S0/4 = 340 W/m2 is the mean sunlight, S(x) = 1 - 0.482 P2(x) its latitude distribution, A + B T the outgoing infrared radiation (a linearised Stefan-Boltzmann law that already contains the water-vapour feedback), D the heat carried poleward by air and ocean, and F the extra forcing. A is calibrated so that 280 ppm gives a 14 C global mean. The page integrates this with an implicit scheme (stable at any step).
Ice-albedo feedback and the snowball
Albedo falls smoothly from αice to 0.30 as a band warms through the freezing line (-2 C): α(T) = αice + (0.30 - αice)/(1 + e-(T-Tice)/w). Less ice absorbs more sunlight, which melts more ice. The bottom-right graph sweeps the sunlight down and then back up: the climate falls into a snowball near 0.9x sunlight but only climbs out again at about 1.0x. That gap is hysteresis, a genuine feature of this model class.
Greenhouse forcing and the ocean
CO2 forcing is F = 5.35 ln(C/280) W/m2 (Myhre et al. 1998). Without ice feedback the equilibrium warming for doubled CO2 is ΔT = 5.35 ln2 / B; the ice-albedo feedback adds to it (see "ECS" in the readouts). The surface layer (mixed-layer depth sets C) leaks heat to a deep-ocean box, Cd dTd/dt = γ (ΔT - Td), which is why warming lags the forcing by decades. Sulphate aerosols are a cooling term that follows fossil burning, so they vanish when emissions stop.
Carbon cycle boxes
Atmosphere, land, surface ocean and deep ocean exchange carbon (GtC; 1 ppm = 2.12 GtC). Air-sea flux k (A - Aeq) with Aeq = A0(1 + R (O - O0)/O0), where the Revelle factor R about 10 says the ocean takes up much less CO2 than its size suggests. Land absorbs in proportion to the excess CO2 (fertilisation) but warming makes soils respire carbon back. Emissions follow measured history to 2020, then your scenario.
Sea level
Two parts: thermal expansion of seawater, about 1.9 mm per W yr/m2 of heat absorbed, and land-ice melt dh/dt = a (ΔT - T0) with a = 3.4 mm/yr/K (semi-empirical form of Rahmstorf 2007, halved here because only part of the ice responds).
Honest simplifications
This is a teaching model, not a forecast. There is no longitude, no clouds or circulation changes, one hemisphere-symmetric ocean, a single response time for ice sheets and a stylised world map (continents are random noise, not geography). Sensitivity here (about 3.9 K per doubling at the default setting) is near the top of the IPCC range 2.5-4. Real projections use 3-D coupled models. The checks below the controls test the code against analytic results and conservation laws; they do not validate the climate itself.