Presets
Playback
Climate forcing
Run
Camera
Layers
Quality
Model check
The science
What is modelled
A one-dimensional energy-balance climate model (Budyko-Sellers / North 1975): the planet is split into 36 latitude bands, each with one surface temperature. Sunlight warms each band, the planet radiates heat to space, and the atmosphere and oceans carry heat from the warm tropics towards the poles. Wherever a band is colder than -10 C it is covered by ice and snow, which reflects much more sunlight (albedo 0.62 instead of 0.30); the ice edge you see on the globe is computed from the temperatures, not painted by hand.
Equations
- Heat budget per band, with x = sin(latitude):
C dT/dt = Q s S(x) (1 - α(T)) - (A - ΔF + B T) + D d/dx[(1 - x²) dT/dx]. - Q = 340.25 W/m² (a quarter of the solar constant), s = the solar-constant slider, and the sunlight distribution
S(x) = 1 - 0.482 P2(x)withP2(x) = (3x² - 1)/2(more sun at the equator than at the poles). - Outgoing infrared is linearised:
OLR = A + B Twith A = 203.3 W/m² and B = 2.09 W/m²/K (North 1975 values, typical). More CO2 lowers A by the standard logarithmic forcingΔF = 5.35 ln(CO2/280 ppm). - Ice-albedo feedback:
α(T) = αopen + (αice - αopen) (1 - smoothstep(-12, -8 C, T)). Colder surface, more ice, more reflection, even colder: a positive feedback. Untick the layer to freeze the albedo at the open value and compare. - Heat capacity C = 2×108 J/m²/K (about 50 m of ocean), so the climate relaxes in a few years; the integrator is explicit with a 3-day step.
Checks run in the page
- With constant albedo the global mean temperature must equal the closed form
(Q(1 - α) - A)/B(diffusion cannot change the mean). - The equator-to-pole profile must match the analytic Legendre solution
T2 = Q(1 - α) S2 / (B + 6D)of the linear model. - At equilibrium the absorbed sunlight must equal the outgoing infrared (energy budget closure).
Hysteresis
The graph on the right sweeps the solar constant down from a warm planet and up from a frozen one. Because of the feedback the two branches differ: there is a range of sunlight where both a warm and an ice-covered climate are possible, and a frozen planet needs a much brighter sun to thaw than it took to freeze it. The sweep is computed live with the same equations.
Honest simplifications
- The model is annual-mean and has no seasons, clouds-as-a-feedback, water vapour, land-sea contrast or ocean dynamics; heat transport is a simple diffusion. It is a teaching model, not a forecast.
- The numbers A, B, D, the ice threshold and the albedos are typical textbook values, not fitted to today's observations; the resulting ice edge is plausible, not a prediction. A real climate is more sensitive in some ways and less in others.
- The sunlight arrows are illustrative: they show the fraction reflected or absorbed at the latitude they hit, but the globe is lit as at an equinox while the model uses the annual-mean sunlight.
- The Sun is not drawn to scale.
What to try
- "Doubled CO2": watch the ice edge retreat and the albedo fall; then untick the feedback and compare the warming.
- "Ice age (dimmer sun)": the ice edge moves towards the equator.
- "Snowball start" at today's sunlight: does the planet thaw? Then compare with the red and blue branches in the hysteresis graph.