Did Venus have oceans?

Venus, in the words of a 2024 Nature paper, is extremely dry. All the water vapour in its atmosphere, condensed and spread evenly over the planet, would make a layer about 3 centimetres deep; M. S. Chaffin and colleagues at the University of Colorado set that against the roughly 3 kilometre layer an Earth-like start would have supplied. NASA's planetary fact sheet lists 20 parts per million of water near the surface, in an atmosphere of 96.5 per cent carbon dioxide pressing at 92 bars on ground averaging 464 °C. Its hydrogen isotopes say it once held far more. What remains open is the form the water took: whether it ever fell as rain and gathered into seas, or stayed aloft as steam until sunlight split it and the hydrogen leaked away to space.

Soviet engineers were allowing for seas in 1965. When Venera 3 left for Venus on 16 November that year, its descent craft had been tested for buoyancy in case the planet had seas and oceans, as K. Lantratov recorded in Novosti Kosmonavtiki in 1996. On 18 October 1967 the gas analysers of Venera 4 sampled the air on the way down and, in NPO Lavochkin's summary, found about 90 per cent carbon dioxide with 'a quite insignificant content of oxygen and water vapour'. Ingersoll's 1969 paper was titled 'The Runaway Greenhouse: A History of Water on Venus', and in 1970 Rasool and de Bergh argued in Nature that Earth would be as hostile as Venus if it were only 6 to 10 million miles closer to the Sun. Older imaginings of swamps and seas are on the imagined Venus page.

Warm a wet planet and its air fills with water vapour, itself a greenhouse gas. Past a certain point, Colin Goldblatt and colleagues explained in 2013, only the upper atmosphere can radiate heat to space, which caps what the planet can shed at about 282 watts per square metre; absorb more sunlight than that and it heats uncontrollably until the entire ocean has evaporated. Venus receives 1.91 times the sunlight Earth does. Once steam reaches high altitude, ultraviolet light breaks the molecules apart and the hydrogen escapes, as NASA summarised the standard account in 2016. Without liquid water, carbon dioxide stays in the atmosphere. Earth keeps much of its carbon in crustal carbonates; Venus holds a comparable inventory almost entirely as gas, and the processes that built Earth's store (biology, liquid water and plate tectonics) are all absent there today, Tereza Constantinou, Oliver Shorttle and Harrison Nicholls noted in 2026.

Evidence that Venus lost water came down with the Pioneer Venus Large Probe on 9 December 1978. Its neutral mass spectrometer made the key measurement while its inlets were coated with sulfuric acid from the clouds, and in May 1982 Donahue, Hoffman, Hodges and Watson published the result in Science as 'Venus Was Wet': a deuterium to hydrogen ratio of (1.6 ± 0.2) × 10⁻², a hundredfold enrichment over Earth. Deuterium is hydrogen with an added neutron. Ordinary hydrogen is lighter and escapes more readily, so water that has been losing hydrogen for a long time grows richer in the heavy form. The team concluded that at least 0.3 per cent of a terrestrial ocean had been outgassed on Venus, a floor the paper calls 'consistent with a much greater production'.

Later measurements agree that the water is strongly enriched but not on how strongly, and each samples a different height. In 1991 de Bergh and colleagues used night-side spectra taken from Earth, which sound 32 to 42 kilometres above the surface, and found 120 ± 40 times the terrestrial ratio. Venus Express measured water and its deuterated form, HDO, 70 to 95 km up between April 2006 and August 2007; Fedorova and colleagues reported 240 ± 25 times terrestrial over the northern polar region, about 1.5 times the 157 ± 30 they quote for the lower atmosphere, and suggested that HDO is split less readily, or deuterium escapes more slowly, at those altitudes. For the bulk atmosphere Chaffin's 2024 paper uses about 120 times Earth, and Constantinou's 2024 paper 150 ± 30. The same spacecraft found hydrogen and oxygen ions leaving in a ratio of 1.9 to 1, which with the estimated escape of neutral atoms is close to the two-to-one of water (Barabash and colleagues, Nature, 2007).

What the ratio cannot do is say how much water there was, or whether it was ever liquid. Everything above Donahue's 0.3 per cent floor depends on escape rates that are still being revised. In 2024 Chaffin's group showed that a process missing from the standard account, the recombination of HCO⁺ ions with electrons, nearly doubles the present rate of hydrogen escape. That doubles the water that volcanoes or infalling impactors must supply to hold the atmosphere steady, and would let a late ocean, if one existed, dry out faster.

Climate models from NASA's Goddard Institute for Space Studies carry the case for oceans. In August 2016 Michael Way and colleagues took the topography mapped by Magellan, filled the lowlands with a shallow ocean, gave the planet an Earth-like atmosphere and its present slow rotation, and dimmed the Sun by up to 30 per cent to match its youth. The slow spin, Anthony Del Genio explained, exposes the day side for almost two months at a time and raises a thick umbrella of rain cloud. With rotation slower than about 16 Earth days, the Geophysical Research Letters paper found moderate temperatures despite 46 to 70 per cent more sunlight than Earth receives. Its two public summaries differ: NASA's release said the ocean could have lasted 'for up to 2 billion years of its early history'; the abstract says that at its current rotation Venus 'could have remained habitable until at least 0.715 Gya', 715 million years ago. One is a duration, the other a latest date for one case. In 2020 Way and Del Genio speculated that vast lava outpourings and the global resurfacing hundreds of millions of years ago ended the clement period. Whether anything lived there is on the life on Venus page.

Arguments against oceans reach further back, to the molten aftermath of formation. In 2013 Keiko Hamano, Yutaka Abe and Hidenori Genda of the University of Tokyo argued in Nature that rocky planets split into two types as their magma oceans freeze. A type I planet, beyond a critical distance from its star, solidifies within several million years and keeps most of its water, which forms the earliest oceans. A type II planet, inside that distance, can stay molten for as long as 100 million years while escaping hydrogen dries it out. Earth is type I. Venus formed close to the boundary, so the authors could not assign it with certainty, but its dry surface and mantle are what type II predicts.

Martin Turbet and colleagues at the University of Geneva asked whether steam from that molten phase could ever have rained out. Their 2021 Nature paper ran three-dimensional climate models of the young Earth and Venus with the water still airborne, and found that water clouds form preferentially on the night side, where they have a strong net warming effect. That warming blocked condensation even at sunlight down to 0.95 times what Earth receives today, and the paper concludes that water never condensed on Venus and oceans never formed. The same models say Earth kept its water because the young Sun was faint, which Emeline Bolmont called 'an unhoped-for opportunity' for a very hot young Earth. David Ehrenreich added a caveat: 'we will not be able to answer this question definitively on our computers'.

Way's and Turbet's models also ask different questions. Way's begin with an ocean already on the surface and ask whether it could stay stable. Turbet's begin with a hot planet whose water is all steam and ask whether it could condense at all. If Turbet's night-side clouds are right, Way's starting point was never reached; if they warm less than modelled, an ocean could have formed and then lasted as Way describes. Both depend on how clouds behaved on the young planet, which no instrument can observe now.

Chemistry offers a second line of attack, and it too points both ways. Water that is split and loses its hydrogen leaves oxygen behind, and Venus's air holds very little. In 2021 Krissansen-Totton, Fortney and Nimmo modelled interior, atmosphere and climate together from the magma ocean onward and found that both a habitable past and a never-wet past reproduce the atmosphere's composition, the inferred heat flow and the measured argon-40 and helium-4, even allowing a global ocean about 100 metres deep as late as a billion years ago. Warren and Kite, following the oxygen in PNAS in 2023, reproduced the modern atmosphere in fewer than 0.4 per cent of their runs, a narrow range favouring habitable eras that ended before 3 billion years ago. In December 2024 Constantinou, Shorttle and Paul Rimmer of the University of Cambridge worked backwards from today's air: the water, carbon dioxide and carbonyl sulfide that chemistry destroys must be restored by volcanism, and on their accounting those volcanic gases carry at most 6 per cent water. A dry interior, they argued in Nature Astronomy, means Venus 'has never been liquid-water habitable'. The argument depends on today's atmosphere being held in balance by volcanic resupply.

Work published in 2026 keeps both histories alive. Haskelle White-Gianella and Joshua Krissansen-Totton reported in the Planetary Science Journal that a rocky planet needs at least 20 to 50 per cent of Earth's ocean mass to keep its carbon cycle balanced over 4.5 billion years, and that limited surface water could have tipped Venus from temperate to uninhabitable. Constantinou, Shorttle and Nicholls found in Monthly Notices of the Royal Astronomical Society that a habitable Venus could have stored about 20 bar of carbon dioxide in carbonate rock and released it over tens of millions of years once the runaway began, but that magma ocean outgassing and some volcanic histories can build the same atmosphere with no temperate phase, so it 'does not uniquely diagnose a temperate past'. The CHILI intercomparison (a June 2026 preprint benchmarking interior and atmosphere codes, with Hamano among its authors) found that Venus's magma ocean can last 50 million years in some scenarios, with substantial disagreement between codes.

Rocks may hold a clue, although global resurfacing, as Turbet's paper notes, has obscured most of the planet's history. Tesserae, deformed highland terrains, are the exception: NASA calls one of them, Alpha Regio, possibly among the oldest surfaces on Venus. In 2008 Hashimoto and colleagues read night-side infrared emission recorded by Galileo during its Venus flyby and found that most highlands seem to emit less efficiently than most lowlands. One interpretation, they wrote, is felsic rock, the family that includes granite, and since large bodies of granitic magma need water to form, such terrain would imply that Venus may have had an ocean and a way to recycle water into the mantle. They noted the uncertainty from poorly known atmospheric parameters. Germany's DLR, whose Venus Emissivity Mapper is the infrared instrument planned for VERITAS, goes further on its project page: Venus Express data suggest the great highland plateaus are comparable to continents, so water may have been present when they formed, and Venus 'could hide more water in its interior than Earth'. That reading sits directly against the Cambridge dry-interior result, and DLR says VERITAS will test it, although NASA's fiscal 2027 request leaves it out; its team now aims to launch in November 2032.

Nothing that has flown can settle the question. NASA's DAVINCI probe is designed to try: its descent sphere should deliver, in the words of the 2022 mission paper, 'definitive chemical and isotopic composition' of the atmosphere above Alpha Regio, including key stable isotopes and oxygen fugacity, plus near-infrared images of a tessera from below the clouds. NASA says the aim is to help determine whether Venus was once wet and habitable. The 2022 paper planned a 2029 launch and a NASA abstract of May 2026 gave late 2031; in October 2026 Ars Technica reported a new target a year earlier, late December 2030, and probe release on 3 October 2032. The fiscal 2027 request does not mention DAVINCI either, but Congress funded it for fiscal 2026, and on 24 September 2026 NASA's DAVINCI blog reported a probe engineering model surviving Venus-like heat and the mission working toward approval for final design and fabrication. ESA's EnVision is to carry a comparable near-infrared mapper, VenSpec-M, designed to map rock types from orbit. None of the three has launched.

If Venus once had seas, could people give them back? In a 1961 review for Science, before the planet's extreme surface conditions were known, Carl Sagan suggested seeding the clouds with photosynthetic algae to break carbon dioxide down into oxygen and organic material. Alex Howe of NASA Goddard, reviewing the proposals in 2022, explains why that fails: Venus is too dry and its atmosphere too thick for chemistry alone, and splitting all the carbon dioxide would leave tens of bars of oxygen. Adelman proposed in 1982 that asteroid impacts strip the atmosphere, which would need impactors heavier than the roughly 5 × 10²⁰ kg of air itself. Birch's 1991 plan, which he argued could take as little as 200 years, spends most of that time behind a sunshade while the atmosphere cools, condenses and freezes, and brings in an ocean by breaking up Saturn's moon Enceladus. Bullock and Grinspoon, in 1996 work not aimed at terraforming, noted that calcium minerals on the surface could lock carbon dioxide away, over a natural timescale of 10 to 100 million years.

Numbers explain why every proposal remains a paper study. Howe's inventory puts about 1,027 tonnes of carbon dioxide above each square metre of the surface, against 8.6 kilograms of water. His own scheme leaves the surface alone: a shell of interlocking tiles at about 50 km, built largely from carbon extracted from the air, floated on the atmosphere's nitrogen and watered from Mars. He estimates about 200 years and calls that an extreme best case, since the materials exist in laboratories and nowhere on an industrial scale. Geoffrey Landis judged in 2011 that settling the cloud layer, where temperature and pressure resemble Earth's, is much more feasible; those plans are on the crewed Venus page.

Beyond the Solar System, Venus's water matters for a third reason. Transit surveys such as Kepler measure a planet's size, and at 95 per cent of Earth's radius Venus could not be told from Earth on size alone, as Stephen Kane and colleagues pointed out in 2014. They defined a Venus Zone around other stars, between the runaway greenhouse limit and about 25 times the sunlight Earth receives, where a planet is likely either to be in a runaway greenhouse or to be losing its atmosphere, and identified 43 potential Venus analogues among Kepler's planet candidates. The sunward edge of the habitable zone depends on the model: in 2013 Kopparapu and colleagues put the Solar System's water-loss limit at 0.99 astronomical units and an optimistic 'recent Venus' limit at 0.75 AU, an empirical edge resting on the inference that Venus has had no surface liquid water for at least the last billion years. Venus orbits at 0.723 AU, inside both.

Webb has begun to test the idea on real planets. TRAPPIST-1 c is 'basically a Venus twin', Laura Kreidberg of the Max Planck Institute for Astronomy said in NASA's June 2023 release: about the size of Venus, and receiving a similar amount of radiation from its small red star. The telescope measured a day side of 380 ± 31 K, and Sebastian Zieba's team found that this disfavours a thick carbon dioxide atmosphere, disfavours a Venus-like atmosphere with sulfuric acid clouds at 2.6 sigma, and points to a planet that formed with less than about 9.5 Earth oceans of water. In 2025 Daisy Turner's team weighed what their paper calls 'the exo-Venus Gliese 12 b': about Earth's size and mass, around a quiet M dwarf.

Because it can be visited, Venus is the reference for all of these worlds. A team led by Kane wrote in August 2026 that Venus provides the only accessible analogue for rocky exoplanets with thick carbon dioxide atmospheres, and calculated that DAVINCI's atmospheric profile would narrow the spread in a modelled Venus-like spectrum by factors of about 4 to 5 in transmission and 5 to 15 in thermal emission. A second 2026 paper from the group expects ESA's PLATO telescope to find about 170 to 280 terrestrial planets in Venus Zones during a four-year mission, 40 to 80 of them close to Earth's size.