Every time a planet is discovered orbiting another star, people ask if it is habitable. The habitable zone, where terrestrial water-based life could exist, is defined as the region of a stellar system where liquid water can persist. The limits of the habitable zone depend on the composition of gases in the planet's atmosphere, surface gravity and the abundance of water. They also depend on the star's power output, the wavelength of light that it emits and the history of the star-planet system. Sun-like stars brighten over time, so if a planet starts out in a snowball state with its surface nearly covered by ice, the ice will eventually melt as the star's power increases — the habitable zone comes out to meet the planet. Indeed, previous climate models have suggested that the icy worlds common in our Solar System and extra-solar systems could enter a habitable stage with sufficient solar radiation. However, writing in Nature Geoscience, Yang et al.1 find that some planets transition from Mars-like snowballs to Venus-like hothouses as their host stars brighten without ever experiencing Earth-like habitable conditions.

The climate of water-rich worlds depends on two positive feedbacks (Fig. 1). First, there is the ice-albedo feedback2,3. Ice has a high albedo: ice on a planet's surface reflects most of the sunlight (or starlight) back to space. This acts to cool the planet, which leads to more ice and makes the planet even colder. Second, there is the water vapour feedback. Water vapour in a planet's atmosphere acts as a greenhouse gas and traps infrared light emitted by the planet. This acts to warm the planet, which evaporates more surface water and makes the planet even warmer. Both ice-albedo and water vapour feedbacks operate on present-day Earth and result in a climate that is sensitive to imposed changes in both solar output and greenhouse gases.

Figure 1: Climate feedbacks in series.
figure 1

The positive ice-albedo and water vapour feedbacks act to amplify changes in a planet's temperature. The ice-albedo feedback can cause a planet to transition to or from an icy snowball state, and the water vapour feedback can lead to a runaway greenhouse state. Yang et al.1 show that some icy bodies abruptly transition between these two states as they warm without passing through a habitable state.

On cold icy planets, the ice-albedo feedback dominates. This was probably the state of early Earth when the young Sun's power output was only 70% of its present value. The most recent snowball episode occurred 600–800 million years ago4. Earth was probably able to escape its snowball state due to volcanic outgassing of greenhouse gases into the atmosphere and increased solar radiation from the Sun5. However, many small icy bodies like Jupiter's moon Europa and Saturn's moon Enceladus lack substantial volcanic outgassing of greenhouse gases. So, there is only sunlight to warm the planet.

Yang et al.1 apply a comprehensive 3D global climate model to the climatic evolution of an icy planet lacking the emission of greenhouse gases other than water vapour. Compared to previous simulations, the 3D model is able to accurately simulate transport of water vapour to the upper atmosphere — a key process in a planet's transition to a Venus-like hothouse state.

In the simulations, the ice-albedo feedback initially keeps the planet cold and retards melting as stellar power increases. Without atmospheric greenhouse gases, the stellar flux at which melting occurs is high — about 10% to 40% more than that received by the Earth's atmosphere. When the ice finally does melt, the resulting drop in albedo makes the planet much warmer very quickly. If the planet is sufficiently warm for the water vapour feedback to kick in, the planet transitions into a moist greenhouse or a runaway greenhouse state6. In a runaway greenhouse the oceans vaporize completely. In a moist greenhouse some oceans remain. But in both cases the atmosphere contains so much water that substantial water vapour reaches the upper atmosphere where it is destroyed by sunlight and lost to space7. The planet never passes through a long-lived period of habitability.

To explain why an abrupt climate transition did not occur on Earth, Yang et al. call on the stabilizing feedback that CO2 has on Earth's long-term climate5. Higher global temperatures increase weathering rates of silicate rocks. Increased weathering delivers more calcium and magnesium to the oceans, which remove dissolved CO2 from the atmosphere and precipitate carbonate rocks. Enhanced removal of atmospheric CO2 reduces its greenhouse effect and results in cooling. Lower global temperatures, by contrast, result in slower silicate weathering and slower drawdown of CO2, allowing atmospheric CO2 outgassed by volcanism to build up and warm the planet. The Earth's active carbonate–silicate cycle acts to stabilize the climate. This negative feedback has long been regarded as the mechanism that nudged Earth out of its most recent snowball state and into the present stable temperate climate between 600–800 million years ago5.

CO2 as a stabilizer of long-term climate seemingly contrasts with CO2 as the instigator of global warming, but these effects operate on different timescales. The natural response of atmospheric CO2 to temperature changes driven by the carbonate–silicate cycle acts over millions of years. By contrast, the CO2 build-up over the past century is forced by human activities. The long-term negative feedback driven by increased weathering rates has not kicked in yet. Other feedbacks involving sudden turnover of the oceans and surges of the ice sheets8 act on intermediate timescales, and how they would operate on a small icy planet as its stellar power increases remains unknown.

Compared with earlier planetary climate models, the Yang et al. model is sophisticated. It is 3D and covers the globe with multiple vertical levels. It calculates the winds and their effects on temperatures, clouds, water vapour, snow and ice. It contains a realistic radiative transfer scheme — the part that calculates the absorption of sunlight and the emission of infrared radiation. What is remarkable is that climate models developed for Earth can now be transferred to other planets with relative ease.

Yang et al.1 show that a small, uninhabitable icy world can bypass a habitable climate and go straight to an equally uninhabitable greenhouse as its host star brightens. Still, our own Solar System hints that the answer may be more complicated. Geological evidence shows that early Earth had both warm episodes and snowball episodes. And early Mars once had liquid water flowing on its surface, even though Mars today is dry and frozen. The warm episodes are not consistent with the steady increase of solar power — an enigma known as the faint young Sun paradox9. The near absence of water on Venus today suggests that a runaway greenhouse caused it to lose a large amount of water7. But whether there was an earlier snowball phase or a habitable phase is unknown. The lesson for the search for habitable worlds beyond our Solar System is that our models, based on Earth experience, still have large uncertainties.