warming up your workspace

The planetary energy budget, or how to compute Earth's temperature with one equation

Here is a claim that sounds absurd until you check it: you can predict a planet's average temperature with about five lines of arithmetic, no supercomputer, no climate model. All you need is how much sunlight reaches it, how shiny it is, and one law that connects temperature to radiation. Do this for Earth and you get an answer that is confidently, precisely wrong, too cold by 33 degrees. That wrongness is not a mistake. It is the greenhouse effect, and this calculation is how you measure it.

The one idea

A planet sitting in space is a checking account for energy. Sunlight comes in; heat radiates out. When the two balance, the temperature holds steady. So to find the temperature, you write down the incoming and outgoing sides and set them equal.

Incoming is easy: the Sun delivers a known flux, the planet reflects some fraction (its albedo) straight back, and it absorbs the rest. Outgoing uses the Stefan-Boltzmann law: any warm object radiates power proportional to its temperature to the fourth power. Set absorbed equal to radiated, solve for temperature, and you are done. The fourth-power law is the whole engine, it means a planet has a strong tendency to settle at one equilibrium temperature, because if it warms even slightly it radiates much harder and cools back down.

Build it

Two subtleties go into the incoming side. Sunlight hits the planet as a disk but the planet radiates from its whole sphere, so the average absorbed flux is the solar constant divided by four. And you multiply by one minus the albedo, the part that is not reflected away.

sigma = 5.670374419e-8      # Stefan-Boltzmann constant, W/m^2/K^4
S = 1361.0                  # solar constant, W/m^2
albedo = 0.30               # Earth reflects about 30 percent of sunlight

absorbed = (S / 4) * (1 - albedo)       # average absorbed flux over the sphere
T_eff = (absorbed / sigma) ** 0.25      # invert F = sigma*T^4 to get temperature
print(f"absorbed: {absorbed:.1f} W/m^2, effective T: {T_eff:.1f} K")

Three details that matter:

  • The divide-by-four is geometry, not fudge. A sphere presents a circular shadow of area pi r squared to the Sun but has a surface of 4 pi r squared to radiate from. The ratio is exactly four. Miss it and your planet runs far too hot.
  • Albedo is a single number standing in for clouds, ice, ocean, and desert all at once. Earth's is about 0.30. Crank it toward 1 (a frozen white planet) and the world absorbs almost nothing and freezes harder, a feedback loop with its own consequences.
  • The fourth root is why climate is stable on the short term. To radiate twice the power a planet only needs to warm by about 19 percent in absolute temperature, so the outgoing side responds steeply to any imbalance and pulls the system back toward equilibrium.

Proof: the answer is 33 degrees too cold, on purpose

absorbed flux: 238.2 W/m^2
effective (no-atmosphere) temperature: 254.6 K  (-18.6 C)
actual surface temperature: 288.0 K  (14.9 C)
greenhouse warming: 33.4 K

The physics says Earth's surface should sit at 255 kelvin, about minus 19 Celsius, a frozen ball. The real average is 288 kelvin, a livable 15 Celsius. The model is not broken; it left something out. It computed the temperature Earth radiates to space, which really is about 255 kelvin, measured from orbit. The 33-degree gap between that and the ground is the atmosphere trapping outgoing heat: the greenhouse effect, isolated as a single number by the simplest possible model. Add greenhouse gases and that gap widens, which is the whole of climate change in one line.

Where this shows up

This zero-dimensional model is the seed of every climate model that exists. The next step gives the planet latitude bands so the equator and poles differ, then an atmosphere with layers, then feedbacks: ice that changes albedo as it melts, water vapor that amplifies warming, clouds that cut both ways. Each layer of realism is bolted onto this same balance, energy in equals energy out. Get this equation and you understand what every one of those models is arguing about.

If you want to build the layers, latitude bands, an ice-albedo feedback that can flip the planet into a snowball, and a real radiative atmosphere, that is what the climate-modeling track on IWTLP constructs step by step.