Flat-earth weather, modeled honestly
Plane Old
Weather
Take the flat Earth seriously as physics — a disc, an ice wall, a dome, a small close sun — and build the most honest weather model those rules allow. Then watch, precisely, where it breaks.
The Earth is not flat. This is a physics thought experiment exploring what weather would look like if it were — and why that model fails.
Chapter 02 — The premise
The rules of a flat world
Take the claim literally and it becomes a set of physical assumptions. A disc about forty thousand kilometers across, with the North Pole at its center and the continents fanning outward. A wall of ice — the thing maps call Antarctica — ringing the edge. A solid firmament dome overhead. And a small, close Sun, only a few thousand kilometers up, circling above the plane.
We take those rules seriously, because a model you can test teaches far more than one you simply mock. The point is not that any of this is true — the Earth is an oblate spheroid — but that honest physics, applied to a flat world, breaks in specific, measurable places. The rest of this is finding them.
Chapter 03 — The projection
A real map, read too literally
The flat map is not invented. It is the azimuthal-equidistant projection, centered on the North Pole — the same one drawn on the United Nations emblem. It places the pole at the center and keeps distances from that center true, so the lines of latitude become perfect concentric circles and the South Pole smears out into the entire outer rim.
It is a genuinely useful way to flatten a globe, and near the center it is nearly perfect. The distortion only grows with distance, stretching the southern hemisphere wider and wider. The flat-earth move is to mistake the map for the territory — to treat a drawing of a sphere as the shape of the world. So we will re-map real latitude and longitude data onto the disc, and keep score of what that costs.
Chapter 04 — Thermodynamics
A spotlight, not a season
On a sphere, sunlight arrives in nearly parallel rays. Latitude sets the angle they strike the ground, so the tropics bake and the poles freeze in steady, predictable bands. The gradient barely moves; it is the backbone of the whole climate.
A small, nearby Sun cannot do that. It behaves like a spotlight — a bright bull’s-eye of heat directly beneath it that fades steeply with distance, the inverse-square law sharpened by the low angle of the rays, sweeping a full circle every day and spiraling inward and outward to make the seasons. Our model heats the disc exactly this way and balances it with a uniform cooling sink so the run reaches equilibrium. What comes out is a moving hot ring that looks nothing like Earth’s fixed latitude gradient — and it is the first place the physics begins to strain.
Chapter 05 — Dynamics
Storms that cannot spin
A region of low pressure pulls air inward from every direction. On a rotating sphere the Coriolis force bends that inflow sideways — counter-clockwise in the northern hemisphere, clockwise in the southern — and the storm winds up into a spiral. Every hurricane, every winter low, turns for this one reason.
A flat, still disc has no such force. With the Coriolis parameter set to zero, air simply slides straight down the pressure gradient and piles into the center without ever turning. You can try to rescue rotation by spinning the entire disc — an f-plane, with one constant rotation everywhere — but then every storm spins at the very same rate, a cyclone at the rim turning exactly as fast as one at the pole. Earth shows a clean climb of rotation with latitude instead. This is the single cleanest sign that something is wrong.
Chapter 06 — Boundary conditions
The wall at the edge of the world
A disc has to end somewhere. We close it with the ice wall and enforce no-penetration: wind can slide along the rim but never cross it — a free-slip boundary applied with a ghost-cell mirror. A sphere has no such edge; weather systems circle it endlessly and come home.
Against a wall, they cannot. Waves run out to the rim, reflect, and the reflections interfere into standing waves — basin resonances pinned in place by the walls, the way a closed pond sloshes at its own fixed set of frequencies. A boundaryless sphere has no edge to reflect from; its waves simply travel on around the planet. Mass driven outward heaps up against the ice into a permanent ring of high pressure, a rim anticyclone that no one has ever observed, and the sharp corner at the rim churns up turbulence with no real-world counterpart. None of this is a numerical mistake. It is simply what a bounded world is forced to do.
Chapter 07 — The model
Explore it yourself
Everything so far is computed and waiting. The interactive viewer — temperature, pressure, wind and rain across all six scenarios, scrubbed through time and laid beside an idealized real Earth — arrives next.
Chapter 08 — Falsification
Where the disc meets the data
Set the disc beside the sphere and the disagreements stop being rhetorical. They become numbers. Take cyclone rotation: the spherical Earth predicts a rate that climbs smoothly with latitude and vanishes at the equator, exactly as observed. The flat disc predicts a flat line — the same spin everywhere, or none at all. The two curves cannot both be right, and only one matches the sky.
Add the rim’s phantom anticyclone, the basin resonances of a walled world that a boundaryless sphere never produces, and antipodal journeys that the flat map stretches to impossible lengths, and the model fails not with a shrug but with a tally. That is the real lesson hiding inside the joke: the spherical model is not merely one option among many. It is the thing that makes the measurements come out right.
Chapter 09 — Methodology & source
How this was built
The weather on these pages is a single-layer shallow-water model, the same lineage as the first numerical forecasts of the 1950s, solved on a Cartesian grid masked to a disc. It conserves mass to within rounding error, respects the CFL stability limit, and runs entirely offline — the site only ever serves the pre-computed result, never a live calculation. The real-Earth fields it is set against are, for now, an idealized zonal climatology — a stand-in for full ERA5 reanalysis, which is the next dataset to fold in.
None of it is hidden. The solver and this website and the full-resolution model output are public to read, fork, run locally, and download — source-available rather than open source, so a public redeployment needs written permission (see the license). And so there is no mistaking the exercise for a claim: the Earth is not flat. This is a physics thought experiment exploring what weather would look like if it were — and why that model fails. The Earth is an oblate spheroid; the worth of a flat model lies entirely in the precision with which it breaks.