The atmosphere of Venus

Venus's air is about 96.5 percent carbon dioxide and 3.5 percent nitrogen by volume near the ground, according to NASA's Venus fact sheet. It presses on the surface at about 92 bar and holds the ground at an average of 737 K (464 °C), with almost no difference between day and night. The sulfuric acid clouds lie roughly between 47 and 70 km, and below their base the acid evaporates, so no liquid droplets survive in the hot lower air. Little sunlight gets to the bottom. Summing up Akatsuki in January 2026, JAXA's scientists give the budget as 76 percent of incoming sunlight reflected, 21 percent absorbed in the clouds and 3 percent reaching the surface, against 31, 20 and 49 percent on Earth. NPO Lavochkin reports that illumination measured by the Venera 9 and 10 landers in 1975 showed 5 to 10 percent of the Sun's energy reaching the surface as light scattered by the clouds. Those were readings at two landing sites, while the ISAS figure is a whole-planet budget, so the two measure different things.

People inferred the atmosphere long before anything flew there, and who did it first is still argued. On 6 June 1761 (26 May in the calendar Russia then used) Mikhail Lomonosov watched Venus cross the Sun from his house in St Petersburg. He saw the Sun's edge lose its sharpness as Venus arrived and a bright bulge swell from the solar limb as it left, and his report, printed by the Academy of Sciences that July, concludes that Venus 'is surrounded by a significant air atmosphere' like Earth's, the bulge being sunlight refracted in that air. The commentary in the 1955 Soviet edition of his works defends his priority and treats the French observer Chappe d'Auteroche's account of a ring of atmosphere as unoriginal. In 2012 Jay Pasachoff and William Sheehan concluded that Lomonosov was describing the 'black drop', an optical effect unrelated to any atmosphere, and that observers in 1761 and 1769, Chappe d'Auteroche among them, may have seen the real arc. The same year Alexandre Koukarine, Vladimir Shiltsev and colleagues watched the 2012 transit through antique refractors and concluded that Lomonosov's telescope was fully adequate to see the arc. The transits of Venus page follows the observing campaigns.

Telescopes also worked out what the clouds are made of before any probe touched them. In June 1927 Frank Ross photographed Venus at Mount Wilson and found markings in ultraviolet light that red and infrared exposures do not show; he read them as disturbances in the atmosphere. Bernard Lyot had already measured by eye at Meudon, between 1922 and 1924, how Venus polarises sunlight at different phases, and found the result puzzling and unlike the Moon, Mars or Mercury. In May 1974 James Hansen and J. W. Hovenier showed that the curves require an optically thick cloud of micron-sized liquid droplets with a refractive index of 1.43 to 1.45, consistent with sulfuric acid of about 75 percent strength. With infrared spectra, that made a strong case for sulfuric acid, which the Pioneer Venus probes later confirmed in place. A century after Lyot, Jeremy Bailey and colleagues repeated the measurement (accepted by Icarus in 2026), found the particle sizes in good agreement with earlier determinations, and saw the polar regions behave differently in ultraviolet, most simply explained by a lower cloud top there.

Probes began sampling in 1967. Venera 4's descent capsule, NPO Lavochkin writes, made the first direct measurements of the temperature, density, pressure and chemical composition of Venus's atmosphere on 18 October 1967, and its gas analysers found mostly carbon dioxide, about 90 percent, with very little oxygen or water vapour. Nephelometers on Venera 9 and 10 and then Venera 11 and 12 placed the lower boundary of the clouds at about 48 to 49 km; the 2026 haze study discussed below uses about 47 km.

The clouds' four-day circuit was spotted from the ground by an amateur. In August 1957 Charles Boyer, President of the Court of Appeal in Brazzaville, began photographing Venus through a violet filter with his own 256 mm telescope. Over 68 days he noticed the markings repeating every four days, and Henri Camichel found the same period in plates from the 60 cm reflector at Pic du Midi. Their joint paper of 1961 was cautious: 'similar appearances return approximately every four days', a period that 'could be that of the rotation of Venus, though it cannot be affirmed'. By 1969 Boyer, writing with Pierre Guérin, titled his paper a study of the four day retrograde rotation of the outer cloud layer. The ground itself turns backwards once every 243 days, so the cloud tops outrun it hugely: about 60 times the planet's spin in JAXA's words, more than 50 times in IKI's. ISAS gives the ground's equatorial speed as 1.6 m/s and the wind near 70 km as 100 m/s, a ratio of about 62.

An observational answer, for low latitudes, came six decades later. As ISAS's 2026 retrospective notes, friction with the surface should keep the air turning roughly with the ground, and the slow overturning that carries heat from the equator toward the poles also carries angular momentum away from the equator. A team led by Takeshi Horinouchi of Hokkaido University tracked clouds in Akatsuki's ultraviolet images, added temperatures from the LIR camera, and reported in Science on 24 April 2020 that the fastest flow, at the cloud tops near the equator, is sustained by thermal tides: waves raised as sunlight heats the clouds by day and they cool by night. The tides carry angular momentum toward low latitudes, while other waves and turbulence, long suspected of doing that job, turned out to work weakly against it there. JAXA's release added that observation has limits and uncertainties remain, though the team considered the basic picture secure. The retrospective ties the tides to the 21 percent of sunlight absorbed in the clouds, calls them the main mechanism at least at low latitudes, and leaves the role of other waves at middle and high latitudes unfinished.

While that question was open, Venus Express showed the winds changing. Igor Khatuntsev and colleagues at the Space Research Institute (IKI) in Moscow tracked cloud features in Venus Monitoring Camera images of the southern hemisphere near 70 km from 2006 to 2012, deriving more than 45,000 wind vectors by hand and, in IKI's 2014 account, more than 390,000 automatically (ESA's 2013 release said more than 350,000). IKI puts the mean low-latitude wind at 85 m/s at the start of the mission and 110 m/s by mid 2012. ESA announced the result on 18 June 2013 alongside a parallel study led by Toru Kouyama in Japan, and quoted Khatuntsev: 'Such a large variation has never before been observed on Venus, and we do not yet understand why this occurred.' Both teams also saw faster swings, with low-latitude clouds lapping the planet in 3.9 days on one orbit and 5.3 days on another.

Why the winds change has two competing answers. Yeon Joo Lee and colleagues found in 2019 that the planet's brightness at 365 nm, close to the absorber's peak, changed by a factor of two between 2006 and 2017, enough to alter solar heating at low latitudes by 25 to 40 percent, and their circulation model suggested this could explain the wind changes. Khatuntsev's team, adding Akatsuki's ultraviolet images to cover 2006 to 2021, reported in 2022 a cycle of 12.5 ± 0.5 years in the cloud-top winds, correlated with the ultraviolet albedo and close to the roughly 11 year solar cycle. The Akatsuki team reads its own long series differently. Its 2026 retrospective reports a brown-noise spectrum with no dominant frequency, which suggests the variation may come from the atmosphere's own self-excited behaviour, and calls the link between wind speed and albedo interesting but premature. A cycle paced from outside and fluctuation generated inside cannot both be the whole story; the question is open.

The markings Ross photographed are still unexplained. Something near the cloud tops absorbs between at least 280 and 500 nm, and it plays a critical part in how much sunlight the clouds take in. Akatsuki's UVI camera watched it through a 365 nm filter, and JAXA's image gallery still called it an unidentified substance in September 2026. Clancy Zhijian Jiang and colleagues showed in Science Advances in January 2024 that ferric iron reacts with sulfuric acid to form the minerals rhomboclase and acid ferric sulfate, good candidates for the 200 to 300 nm and 300 to 500 nm features. Jan Spacek and colleagues argued in a 2025 preprint, revised in August 2026, that absorption peaking this strongly near 375 nm needs a very efficient absorber such as conjugated organic molecules, and noted that the planned Rocket Lab mission is meant to test the organic idea. In August 2026 Rakesh Mogul and colleagues posted a preprint, still under review, proposing mixtures of polycyclic aromatic hydrocarbons and iron compounds, the altered remains of cometary dust, which they say match spectra from the cloud tops to below the clouds. None is confirmed.

Dust from space turns up in a second puzzle lower down. Below the cloud base near 47 km the sulfuric acid evaporates, yet Venera, Pioneer Venus and Vega probes all found a persistent layer of solid submicrometre particles there, the lower haze, whose origin Tohoku University calls a fifty-year-old problem. On 13 April 2026 Hiroki Karyu, then a graduate student at Tohoku, and an international team of colleagues published a model in Nature Astronomy. In it, meteor smoke enters at 100 km as grains about a nanometre in radius, is swept up by acid droplets and carried down, and is left behind when the droplets evaporate at the cloud base, where the grains stick together into haze. Influxes of 3 to 300 tonnes a day reproduce the particle sizes the probes measured; Earth receives an estimated 5 to 60 tonnes. The grains also seed the main cloud deck, and their iron is a plausible source of the ultraviolet absorber. Tohoku's Japanese release presents this as the world's first demonstration that the haze is cosmic dust; the paper's wording is narrower, that the dust is sufficient to sustain the haze, a model result still to be tested by measurement. The authors suggest EnVision could check the iron link by following where the ultraviolet absorber goes.

David Allen and John Crawford, writing from the Anglo-Australian Observatory, reported in Nature in January 1984 that the dark side of Venus is anomalously bright in parts of the 1.5 to 2.5 µm band and shows structured cloud patterns whose rotation period, they wrote, is longer than that of any other clouds. The glow is heat from the hot air near the surface, and the middle and lower clouds show up as silhouettes against it. Galileo's Near Infrared Mapping Spectrometer mapped these lower clouds through the night side windows as it approached Venus on 10 February 1990. Akatsuki's IR2 camera imaged the night side in 2016 and, ISAS says, found an equatorial jet in the middle and lower clouds in data from July and August, before the electronics shared by IR1 and IR2 malfunctioned in early December 2016 and left both cameras dormant. ISAS adds that the night side images allow the emissivity of the ground to be estimated, a thread the mapping Venus page takes up.

Lightning has been claimed and doubted since the 1970s. The Groza instrument on Venera 11 and 12, which descended in December 1978, recorded bursts that NPO Lavochkin summarises as electrical discharges in the cloud layer, about as energetic as Earth's lightning but far more frequent, up to 30 a second; L. V. Ksanfomaliti reported them in Nature in 1980 as a discovery of frequent lightning discharges. Lavochkin's Venera 9 and 10 page goes further, stating that optical measurements by the 1975 orbiters showed Venusian lightning 25 times more energetic than Earth's, a claim later optical searches have not borne out. Venus Express's magnetometer found circularly polarised bursts near 100 Hz, which C. T. Russell's team reported in 2007 as whistler waves with the properties expected from lightning in the clouds. Against that, Cassini listened at 0.125 to 16 MHz during two close flybys and heard nothing, though the same receiver easily picked up Earth's lightning, and its team concluded in 2001 that Venus lightning is 'either extremely rare, or very different from terrestrial lightning'. In 2023 Parker Solar Probe data showed night side whistlers travelling toward the planet alongside Langmuir waves, a source that rules out lightning for those waves and suggests whistler-based flash rates may be overestimated. A 2025 chemistry study pulled the other way: the nitric oxide estimated for the lower atmosphere needs lightning that releases at least three times the power Earth's does. These results cannot all be right.

Akatsuki carried a dedicated instrument, the Lightning and Airglow Camera (LAC), built to record a flash's light curve at high time resolution. From October 2016 to August 2017 it saw nothing, although earlier flash-rate estimates gave it a 35 percent chance of catching at least one, and ISAS says that null result led to a reassessment of how often Venus lightning happens. On 1 March 2020 LAC recorded a single flash. Yukihiro Takahashi of Hokkaido University presented it at the American Geophysical Union meeting that December, and National Geographic's Japanese edition reported that his team thought it was either powerful lightning with about ten times the energy of Earth's or a large meteor exploding in the atmosphere. The team's April 2021 preprint describes a transient ten times brighter than a typical terrestrial flash and lasting a few hundred milliseconds, against about a millisecond for ordinary lightning, and judges a bolide possible but improbable given how often meteoroids should arrive. Crossref listed no journal version of that preprint as of 26 September 2026. LAC kept watching and caught nothing more, and the January 2026 retrospective says the upper limit on the flash rate is being tightened.

Even the second ingredient is uncertain. The standard 3.5 percent nitrogen rests on probe measurements below about 45 km in 1978, with a recommended uncertainty of 1.8 percentage points either way; NPO Lavochkin's summary of Venera 11 and 12 says 'about 5%', and the Pioneer Venus gas chromatograph measured 4.60 ± 0.14 percent at 55 km. When MESSENGER passed over the night side on 5 June 2007, its neutron spectrometer gave 5.0 ± 0.4 percent between 60 and 100 km, and Patrick Peplowski and colleagues concluded in Nature Astronomy in 2020 that the upper and lower atmosphere are chemically distinct. Near the ground, the Vega 2 lander's 1985 temperature profile, which Sebastien Lebonnois and Gerald Schubert call the only reliable one for the deep atmosphere so far, looks unstable below 7 km. In 2017 they proposed that nitrogen thins to near zero at the surface, separated from carbon dioxide by density in a layer so hot and dense that it behaves as a supercritical fluid. Daniel Cordier and colleagues (2019) found diffusion far too slow and phase separation unlikely, and suggested diffuse volcanic degassing instead. In November 2025 Rakesh Mogul, Sanjay Limaye and Michael Way published in Geophysical Research Letters a reanalysis of the Pioneer Venus Large Probe's 1978 mass spectra: 2.49 ± 0.10 percent nitrogen, uniformly mixed from 0.2 to 15 km, rising about twofold to 5.21 ± 0.18 percent at 51 to 59 km. It disagrees both with the 3.5 percent standard and with a near-zero surface.

What is settled is the bulk: carbon dioxide air at crushing pressure, sulfuric acid cloud, and heat that barely changes from day to night. Most of what moves or reacts inside it is still argued: why the winds speed up and slow down, what absorbs the ultraviolet, where the haze comes from, whether lightning flashes, and how the gases are layered in the bottom 15 km. The threads connect: meteor dust may feed both the haze and the absorber, and the absorber sets heating that may pace the winds. The same clouds carry the arguments on life on Venus, and the greenhouse that keeps the ground so hot has its own page, oceans and greenhouse.