The Planet Builder — Climate Evolution Across Geological Time
Set axial tilt, orbital distance, atmospheric CO2, water fraction. Run 100 million years. Watch ice caps swing with Milankovitch cycles, biomes shift, climate find new equilibria. A real energy-balance model.
What this is
Five sliders define a planet: axial tilt, orbital distance, CO2, water fraction, rotation rate. Press a preset for Earth, Mars, Venus, Snowball, or Hothouse. The page runs a 36-band energy-balance climate model with ice-albedo feedback, finds the equilibrium, then time-steps the planet through one orbital year showing seasonal heat transport. Procedural Perlin continents, sea level shifts with the water slider, ice caps grow and shrink in real time. "Run 100 Myr" applies a ±2° Milankovitch tilt oscillation on a 41,000-year cycle and you watch ice ages come and go on the bottom graph. Click anywhere on the globe for that location's annual temperature curve.
Why this is mind-blowing
Climate science is hard to teach because the timescales are wrong for human attention. This compresses 100 million years into one minute and lets you turn the knobs. The same model that produces a stable Earth produces a snowball planet if you nudge the parameters — the same bistability climate scientists worry about for real.
Build a single-file planet simulator. Inputs: axial tilt, orbital
distance (AU), atmospheric CO2 (ppm), water fraction, rotation rate.
The page runs a simplified energy-balance climate model with latitudinal
heat transport, finds the temperature equilibrium, then evolves the
planet over 100 million simulated years showing seasonal cycles,
Milankovitch wobbles, ice-age oscillations. Render the planet as a 3D
globe with biomes drawn from local temperature and precipitation. Click
anywhere to see that location's climate timeline.Paste this into Claude, Cursor, or Copilot. Change one thing that matters to you.
What I learned shipping it
- A 1D Budyko/Sellers energy-balance model with 36 latitude bands is enough to reproduce snowball-Earth bistability and Milankovitch cycles. The ice-albedo feedback does most of the dramatic work.
- Greenhouse forcing scales logarithmically with CO2 concentration, not linearly. That single equation explains why the climate response to doubling CO2 is the relevant metric in IPCC reports.
- Click-anywhere interrogation (raycast lat/lon → local climate timeline) turns a globe from a static visualization into a tool. The whole simulation becomes addressable.