What if Earth were covered in ice?

Snow and ice are bright and white. Sunshine bounces off them.

What if ice covered most of our planet? Would Earth get warmer or colder?

On a sunny day, fresh snow can be almost too bright to look at. That brightness is sunlight bouncing straight off it. Scientists measure how much a surface reflects with a number called albedo, from 0 (reflects nothing) to 1 (reflects everything).

NASA says Earth reflects about 29 percent of the sunlight that reaches it. So what would happen if ice spread over most of the planet and made it far more reflective? In this lab you can turn up Earth's brightness and watch its temperature.

Albedo is the fraction of incoming solar energy a surface reflects. NSIDC gives about 0.06 for open ocean, about 0.5 to 0.7 for bare sea ice, and as much as 0.9 for thick sea ice covered with snow. Averaged over the whole planet, NASA puts the reflected share at about 29 percent, mostly from clouds, particles and bright surfaces such as snow and sea ice.

This lab asks how sensitive a planet's temperature is to that one number. If Earth's albedo doubled to 0.6, roughly what an ice-dominated surface might give, where would its energy balance settle?

Most of Earth is dark blue ocean and green or brown land. They soak up lots of sunshine.

Only the poles have big white ice. So Earth stays nice and warm.

Most of Earth's surface is darker than ice. NSIDC says the open ocean reflects only about 6 percent of the sunlight that hits it and soaks up the rest. Clouds, deserts and the icy poles are the bright parts.

Add it all up and Earth reflects about 29 percent of incoming sunlight; our model rounds this to 30 percent. In the model that leaves about 238 watts of sunshine for every square meter of the planet, averaged over day, night and the whole globe. With today's greenhouse gases, that gives an average of about 15 °C.

NASA's energy budget gives about 340 W/m² of sunlight at the top of the atmosphere, averaged over the globe, with about 240 W/m² absorbed. The model uses S = 1,361 W/m² and α = 0.30: S(1 − α)/4 = 238.2 W/m² absorbed and 102.1 W/m² reflected.

Balancing absorption against εσT⁴ with the model's teaching emissivity of 0.612 gives 287.8 K, or 14.7 °C. The emissivity is chosen to match Earth's observed average; it is not calculated from gas amounts.

In our pretend lab, Olly makes Earth white with ice.

Now lots of sunshine bounces away. Earth gets colder and colder. It freezes!

Choose 'mostly ice' to raise Earth's reflectiveness from 0.30 to 0.60. Now Earth bounces back twice as much sunlight, and only about 136 watts per square meter are left to warm it.

The model planet cools until it balances at about −23 °C, far below freezing. Watch the graph: the oceans hold heat, so the cooling takes years. About 5 years in it is near −13 °C, and after about 10 years it is about −20 °C.

With α = 0.60 the absorbed flux drops to 136.1 W/m² and the balance temperature to [136.1/(0.612σ)]^¼ = 250.3 K, −22.9 °C: 37.6 K below the baseline. Starting from 288 K with a 70 m ocean mixed layer, the run reaches −13.2 °C at 5 years, −20.1 °C at 10 and −22.7 °C at 20, with a response time near 4.2 years.

The greenhouse effect is still there: the warming readout shows the model's gas blanket adds about 29 °C, yet the planet stays frozen.

Dark things soak up sunshine and get warm. White things bounce it away and stay cool.

An icy Earth bounces most sunshine away. So it stays cold.

A planet warms up until the heat it sends out matches the sunlight it soaks up. If ice bounces away more sunlight, there is less to soak up, so the planet settles at a colder temperature.

In real life the story doesn't stop there. NSIDC explains that ice and darker ocean work in a loop: when ice melts, dark water soaks up more sunlight, which warms it and melts more ice. NASA scientists describe the same loop running the other way: when ice spreads, more sunlight bounces away and it gets colder still. Scientists call this a feedback. Our model doesn't do this by itself, so you play the feedback: raise the reflectiveness and see.

At balance S(1 − α)/4 = εσT⁴, so T ∝ (1 − α)^¼. Going from α = 0.30 to 0.60 multiplies T by (0.40/0.70)^¼ ≈ 0.869, from 287.8 K to 250.3 K.

The ice-albedo feedback amplifies such changes: NSIDC describes melting sea ice exposing dark ocean, which absorbs more energy and causes more melting, a positive feedback loop. The model holds α fixed, so the feedback appears only when you move the slider yourself; it also has one temperature, no seasons or latitudes, and only a 70 m ocean layer.

Scientists think Earth may have been very icy a very, very long time ago.

They study old rocks to learn about it. They are still finding out!

Scientists have found signs that, around 700 million years ago, Earth may have come close to freezing over. They found deposits left by glaciers in rocks that formed near the equator. This idea is called Snowball Earth, and scientists are still testing it.

NASA scientists explain that the ice's brightness could have helped the freeze run away, because spreading ice reflected more sunlight. Some computer models suggest the oceans might never have frozen completely. Scientists think the ice finally melted after millions of years, once carbon dioxide from volcanoes built up enough warming.

The Snowball Earth hypothesis rests mainly on glacial deposits at low latitudes. NASA GISS describes at least two Cryogenian glaciations, roughly 710 and 640 million years ago, each lasting about 10 million years, and notes it is still uncertain whether ice covered the whole planet; its models kept half the ocean ice-free even with a global mean 12 °C below freezing.

Penn State's Earth 104 course adds that a snowball planet's high albedo would require a few million years of volcanic carbon dioxide before melting, and that the science is evolving rapidly. Our model's −22.9 °C is a teaching result, not a reconstruction.